YOLOv5 - Anwendungsbeispiel
Contents
YOLOv5 - Anwendungsbeispiel#
!git clone https://github.com/ultralytics/yolov5.git
Cloning into 'yolov5'...
remote: Enumerating objects: 14566, done.
remote: Counting objects: 100% (136/136), done.
remote: Compressing objects: 100% (65/65), done.
remote: Total 14566 (delta 84), reused 116 (delta 71), pack-reused 14430
Receiving objects: 100% (14566/14566), 13.50 MiB | 8.74 MiB/s, done.
Resolving deltas: 100% (10061/10061), done.
from google.colab import drive
drive.mount('/content/drive')
Mounted at /content/drive
%cd yolov5
%pip install -qr requirements.txt
/content/yolov5
|████████████████████████████████| 1.6 MB 33.9 MB/s
?25h
import torch
from IPython.display import Image
import matplotlib.pyplot as plt
import shutil
from tensorflow.keras.preprocessing.image import load_img
for i in [("/content/drive/MyDrive/Exkurs_B2Spirit/stealth_bomber.yaml","/content/yolov5/data/stealth_bomber.yaml"),("/content/drive/MyDrive/Exkurs_B2Spirit/hyp.stealth_bomber.yaml","/content/yolov5/data/hyps/hyp.stealth_bomber.yaml")]:
src_file = i[0]
dst_file = i[1]
shutil.copy(src_file, dst_file)
Vorbemerkung#
Bilder von intakten Militärgerät sind i.d.R. schwer zu beschaffen, weshalb die Datensätze für professionelle Use-Cases oft mit Bildern von 3D-Modellen angereichert werden. Aktuell wird zum Beispiel daran gearbeitet mit Object Detection verschiedene Kampfpanzer-Modelle zu unterscheiden und nicht nur die Typen, wie beispielsweise Schützen- von Kampfpanzern. [Doll and Uysal, 2022]
Der in der Introduction angekündigte Zufallsfund von Tarnkappenbombern des Typs B2 Spirit, welche auf einer US Airbase in Hawaii stationiert sind, wird verwenden um diese Situtation nachzubilden und anhand dieses simplifizierte Umfangs die Arbeitsweise mit Yolov5 und die Metriken zu erklären. Das Vorgehen und die Erkenntnisse aus dem einfachen Modell werden im Anschluss auf den eigentlichen UseCase, die Erkennungen von Flugobjekten auf Militärebasen, angewendet.
Datenbeschaffung#
Aufnahme von Trainingsbildern#
Die vorliegenden Screenshots und wenige andere Internetfunde sind nicht ausreichend, um das Erkennen des Tarnkappenbombers zu trainieren. Das oben erwähnte Modellieren von 3D-Modellen war zeitlich nicht umsetzbar. Stattdessen werden Bilder von einem 3D-gedruckten Modell verwendet. Die Internetfunde werden als Anwendugnsdatein genutzt.
Die Bilder von dem 3D-gedruckten Modell weisen folgende Merkmale auf:
Grau lackiertes Modell ähnlich zum Orginal
Aufnahme in notwendiger Vogel-/Satellitenperspektive
Unterschiedliche Hintergründe (hell und dunkel)
Hinzufügen von lackierten Dreiecken als Hintergrundbilder für eine bessere Abstraktion
Beliebige Anzahl an Hintergrundbildern ohne Objekt
Aufnahmen mit unterschiedlicher Beleuchtung (Innenraum)
Aufnahmen zu unterschiedlicher Tageszeit (Außenbereich)
Nutzung von Schatten als Wolkenersatz
Fig. 3 Eigene Darstellung: Modell B2#
Datenerzeugung#
Es gibt verschiedene Methoden zur Datenanreicherung. Diese sind teilweise in YOLOv5 integriert und werden als Hyperparameter für das Training mitgegeben. Die dafür notwendige Hyperparameter-YAML-Datei wurde wie folgt angepasst.
Image Rotation um +/- 90 Grad
Skalierung um +/- 0.5
Wahrscheinlichkeit für Spiegelung Oben/Unten: 0.5
Wahrscheinlichkeit für Spiegelung Links/Rechts: 0.5
Wahrscheinlichkeit für Erzeugen eines Mosaics: 1.0
Wahrscheinlichkeit für Erzeugen eines MixUp: 0.5
Folgende Abbildung von ultralytics erläutert die Funktionsweise der Mosaik-Erzeugung. Aus einem Bild werden mehrere Bilder erzeugt und das Mosaic für das Trainig verwendet.
Fig. 4 Image Augmentation in YOLOv5 [Jocher, 2021]#
Der MixUp beschreibt die Überlagerung von mehreren Bildern. Dies wird im folgenden Bildausschnitt aus dem selbst erzeugten Trainingsdatensatz verdeutlicht. Es handelt sich um eine Datei, welche von YOLOv5 beim Trainingsstart erzeugt und abgelegt wird. Die beschriebene Überlagerung ist im mittleren Bildabschnitt besonders gut dargestellt. In der linken oberen Ecke wird ein Flugzeugbild aus dem DOTA-Datensatz abgebildet.
Fig. 5 Eigene Darstellung: MixUp aus Trainingsdurchlauf#
Training#
Modell 1#
Ein Basismodell wird trainiert in dem die train.py-Datei mit verschiedenen Paramtern im Terminal (siehe auch Erläuterung zum Working Directory) aufgerufen werden.
Zur Bestimmung eines Basis-Modells müssen folgende 5 Parameter verändert:
data
weights
img
epoch
batch
Mit dem Parameter –data wird die .yaml-Datei angegeben, welche für das Training herangezogen wird. In dieser Datei sind die Pfade zu den Trainings- und Validierungsdaten angegeben, wie auch die zu prognostizierenden Klassenbezeichnungen und ihre Nummer.
#TODO Richtiges YAML eintragen
import re
start = 0
with open("C:\dev\yolov5\data\stealth_bomber.yaml") as f:
for lines in f:
if (re.match("^path", lines)):
start = 1
if (start == 1):
print(lines)
Durch --weights können dem Modell vorgelernte Gewichte mitgeben werden. Die verschiedene Modelle werden in der YOLOv5 Erläuterung vorgestellt. Für Modell 1 wird das S-Modell gewählt. Dieses wird über den weights-Parameter als .pt-Datei aufgerufen. Mit dem Parameter --img wird die Größe der Input-Bilder in Pixel vorgegeben und die Bilder entsprechend skaliert. Im Fall vom Modell 1 sind es 640 Pixel. Die Anzahl an Epochen bestimmt wie oft das neuronale Netz mit den gesamten Trainingsdaten trainiert wird. Der Wert wird über den Parameter --epochs definiert. Mit --batch wird die Anzahl der gemeinsam gelernten Bilder definiert. Dies wird durch die Rechenleistung der Hardware limitiert. Durch den Parameter batch -1 wird die für die Hardware optimale Batchsize automatisch ermittelt. Gemäß des nachfolgenden Trainingsoutputs wurde dies auf 157 Bilder festgelegt. Dies ist insbesondere interessant, da nur 84 Trainingsbildern zur Verfügung gestellt wurden.Die Ursache für die hohe Bildanzahl könnte die oben erläuterte Datenanreicherung sein.
Zur Vereinfachung der Arbeit werden mit project und name die Results unter eigenen Dateipfaden abgespeichert. Mit dem Parameter cache werden die Daten in den Arbeitsspeicher geladen.
Über die Basisparamter hinaus werden für das Modell 1 noch die Parameter freeze und hyp verwendet. Mit dem Parameter freeze 10 wird der Backbone des S-Modells (siehe YOLOv5 Erläuterung) eingefroren und nur die Gewichte im Head trainiert. Dies reduziert die Laufzeit und bietet sich wegen der mangelnden Datenlage an. Die im YOlOV5 Erläuterung angesprochene Hyperparameter-YAML-Datei verwendet standardmäßig folgende wesentlichen Hyperparameter (keine vollständige Liste):
LR0 = 0.01
LRF = 0.01
Weight Decay = 0.0005
IoU Threshold = 0.2 (geändert auf 0.5, für bessere Confidence bei wenig Daten)
!python train.py --project /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports --name No2_S_Train --freeze 10 --hyp data/hyps/hyp.stealth_bomber.yaml --weights yolov5s.pt --epochs 500 --batch -1 --img 640 --data stealth_bomber.yaml --cache
train: weights=yolov5s.pt, cfg=, data=stealth_bomber.yaml, hyp=data/hyps/hyp.stealth_bomber.yaml, epochs=500, batch_size=-1, imgsz=640, rect=False, resume=False, nosave=False, noval=False, noautoanchor=False, noplots=False, evolve=None, bucket=, cache=ram, image_weights=False, device=, multi_scale=False, single_cls=False, optimizer=SGD, sync_bn=False, workers=8, project=/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports, name=No2_S_Train, exist_ok=False, quad=False, cos_lr=False, label_smoothing=0.0, patience=100, freeze=[10], save_period=-1, seed=0, local_rank=-1, entity=None, upload_dataset=False, bbox_interval=-1, artifact_alias=latest
github: up to date with https://github.com/ultralytics/yolov5 ✅
YOLOv5 🚀 v6.2-216-g6e544d5 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
hyperparameters: lr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.1, box=0.05, cls=0.5, cls_pw=1.0, obj=1.0, obj_pw=1.0, iou_t=0.5, anchor_t=4.0, fl_gamma=0.0, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=90.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.5, fliplr=0.5, mosaic=1.0, mixup=0.5, copy_paste=0.0
ClearML: run 'pip install clearml' to automatically track, visualize and remotely train YOLOv5 🚀 in ClearML
Comet: run 'pip install comet_ml' to automatically track and visualize YOLOv5 🚀 runs in Comet
TensorBoard: Start with 'tensorboard --logdir /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports', view at http://localhost:6006/
Overriding model.yaml nc=80 with nc=1
from n params module arguments
0 -1 1 3520 models.common.Conv [3, 32, 6, 2, 2]
1 -1 1 18560 models.common.Conv [32, 64, 3, 2]
2 -1 1 18816 models.common.C3 [64, 64, 1]
3 -1 1 73984 models.common.Conv [64, 128, 3, 2]
4 -1 2 115712 models.common.C3 [128, 128, 2]
5 -1 1 295424 models.common.Conv [128, 256, 3, 2]
6 -1 3 625152 models.common.C3 [256, 256, 3]
7 -1 1 1180672 models.common.Conv [256, 512, 3, 2]
8 -1 1 1182720 models.common.C3 [512, 512, 1]
9 -1 1 656896 models.common.SPPF [512, 512, 5]
10 -1 1 131584 models.common.Conv [512, 256, 1, 1]
11 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
12 [-1, 6] 1 0 models.common.Concat [1]
13 -1 1 361984 models.common.C3 [512, 256, 1, False]
14 -1 1 33024 models.common.Conv [256, 128, 1, 1]
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 4] 1 0 models.common.Concat [1]
17 -1 1 90880 models.common.C3 [256, 128, 1, False]
18 -1 1 147712 models.common.Conv [128, 128, 3, 2]
19 [-1, 14] 1 0 models.common.Concat [1]
20 -1 1 296448 models.common.C3 [256, 256, 1, False]
21 -1 1 590336 models.common.Conv [256, 256, 3, 2]
22 [-1, 10] 1 0 models.common.Concat [1]
23 -1 1 1182720 models.common.C3 [512, 512, 1, False]
24 [17, 20, 23] 1 16182 models.yolo.Detect [1, [[10, 13, 16, 30, 33, 23], [30, 61, 62, 45, 59, 119], [116, 90, 156, 198, 373, 326]], [128, 256, 512]]
Model summary: 214 layers, 7022326 parameters, 7022326 gradients, 15.9 GFLOPs
Transferred 343/349 items from yolov5s.pt
AMP: checks passed ✅
freezing model.0.conv.weight
freezing model.0.bn.weight
freezing model.0.bn.bias
freezing model.1.conv.weight
freezing model.1.bn.weight
freezing model.1.bn.bias
freezing model.2.cv1.conv.weight
freezing model.2.cv1.bn.weight
freezing model.2.cv1.bn.bias
freezing model.2.cv2.conv.weight
freezing model.2.cv2.bn.weight
freezing model.2.cv2.bn.bias
freezing model.2.cv3.conv.weight
freezing model.2.cv3.bn.weight
freezing model.2.cv3.bn.bias
freezing model.2.m.0.cv1.conv.weight
freezing model.2.m.0.cv1.bn.weight
freezing model.2.m.0.cv1.bn.bias
freezing model.2.m.0.cv2.conv.weight
freezing model.2.m.0.cv2.bn.weight
freezing model.2.m.0.cv2.bn.bias
freezing model.3.conv.weight
freezing model.3.bn.weight
freezing model.3.bn.bias
freezing model.4.cv1.conv.weight
freezing model.4.cv1.bn.weight
freezing model.4.cv1.bn.bias
freezing model.4.cv2.conv.weight
freezing model.4.cv2.bn.weight
freezing model.4.cv2.bn.bias
freezing model.4.cv3.conv.weight
freezing model.4.cv3.bn.weight
freezing model.4.cv3.bn.bias
freezing model.4.m.0.cv1.conv.weight
freezing model.4.m.0.cv1.bn.weight
freezing model.4.m.0.cv1.bn.bias
freezing model.4.m.0.cv2.conv.weight
freezing model.4.m.0.cv2.bn.weight
freezing model.4.m.0.cv2.bn.bias
freezing model.4.m.1.cv1.conv.weight
freezing model.4.m.1.cv1.bn.weight
freezing model.4.m.1.cv1.bn.bias
freezing model.4.m.1.cv2.conv.weight
freezing model.4.m.1.cv2.bn.weight
freezing model.4.m.1.cv2.bn.bias
freezing model.5.conv.weight
freezing model.5.bn.weight
freezing model.5.bn.bias
freezing model.6.cv1.conv.weight
freezing model.6.cv1.bn.weight
freezing model.6.cv1.bn.bias
freezing model.6.cv2.conv.weight
freezing model.6.cv2.bn.weight
freezing model.6.cv2.bn.bias
freezing model.6.cv3.conv.weight
freezing model.6.cv3.bn.weight
freezing model.6.cv3.bn.bias
freezing model.6.m.0.cv1.conv.weight
freezing model.6.m.0.cv1.bn.weight
freezing model.6.m.0.cv1.bn.bias
freezing model.6.m.0.cv2.conv.weight
freezing model.6.m.0.cv2.bn.weight
freezing model.6.m.0.cv2.bn.bias
freezing model.6.m.1.cv1.conv.weight
freezing model.6.m.1.cv1.bn.weight
freezing model.6.m.1.cv1.bn.bias
freezing model.6.m.1.cv2.conv.weight
freezing model.6.m.1.cv2.bn.weight
freezing model.6.m.1.cv2.bn.bias
freezing model.6.m.2.cv1.conv.weight
freezing model.6.m.2.cv1.bn.weight
freezing model.6.m.2.cv1.bn.bias
freezing model.6.m.2.cv2.conv.weight
freezing model.6.m.2.cv2.bn.weight
freezing model.6.m.2.cv2.bn.bias
freezing model.7.conv.weight
freezing model.7.bn.weight
freezing model.7.bn.bias
freezing model.8.cv1.conv.weight
freezing model.8.cv1.bn.weight
freezing model.8.cv1.bn.bias
freezing model.8.cv2.conv.weight
freezing model.8.cv2.bn.weight
freezing model.8.cv2.bn.bias
freezing model.8.cv3.conv.weight
freezing model.8.cv3.bn.weight
freezing model.8.cv3.bn.bias
freezing model.8.m.0.cv1.conv.weight
freezing model.8.m.0.cv1.bn.weight
freezing model.8.m.0.cv1.bn.bias
freezing model.8.m.0.cv2.conv.weight
freezing model.8.m.0.cv2.bn.weight
freezing model.8.m.0.cv2.bn.bias
freezing model.9.cv1.conv.weight
freezing model.9.cv1.bn.weight
freezing model.9.cv1.bn.bias
freezing model.9.cv2.conv.weight
freezing model.9.cv2.bn.weight
freezing model.9.cv2.bn.bias
AutoBatch: Computing optimal batch size for --imgsz 640
AutoBatch: CUDA:0 (Tesla T4) 14.76G total, 0.10G reserved, 0.05G allocated, 14.61G free
Params GFLOPs GPU_mem (GB) forward (ms) backward (ms) input output
7022326 15.95 0.149 49.98 8.96 (1, 3, 640, 640) list
7022326 31.89 0.201 20.11 8.16 (2, 3, 640, 640) list
7022326 63.78 0.312 19.75 7.966 (4, 3, 640, 640) list
7022326 127.6 0.659 29.46 10.91 (8, 3, 640, 640) list
7022326 255.1 1.239 49.69 18.64 (16, 3, 640, 640) list
AutoBatch: Using batch-size 157 for CUDA:0 11.83G/14.76G (80%) ✅
optimizer: SGD(lr=0.01) with parameter groups 57 weight(decay=0.0), 60 weight(decay=0.0012265625), 60 bias
albumentations: Blur(p=0.01, blur_limit=(3, 7)), MedianBlur(p=0.01, blur_limit=(3, 7)), ToGray(p=0.01), CLAHE(p=0.01, clip_limit=(1, 4.0), tile_grid_size=(8, 8))
train: Scanning '/content/drive/MyDrive/Exkurs_B2Spirit/labels/train.cache' images and labels... 51 found, 33 missing, 0 empty, 0 corrupt: 100% 84/84 [00:00<?, ?it/s]
train: Caching images (0.1GB ram): 100% 84/84 [00:17<00:00, 4.90it/s]
val: Scanning '/content/drive/MyDrive/Exkurs_B2Spirit/labels/validation.cache' images and labels... 6 found, 4 missing, 0 empty, 0 corrupt: 100% 10/10 [00:00<?, ?it/s]
val: Caching images (0.0GB ram): 100% 10/10 [00:05<00:00, 1.94it/s]
AutoAnchor: 4.96 anchors/target, 1.000 Best Possible Recall (BPR). Current anchors are a good fit to dataset ✅
Plotting labels to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/labels.jpg...
Image sizes 640 train, 640 val
Using 2 dataloader workers
Logging results to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train
Starting training for 500 epochs...
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
0/499 5.1G 0.1224 0.03036 0 125 640: 100% 1/1 [00:03<00:00, 3.88s/it]
/usr/local/lib/python3.7/dist-packages/torch/optim/lr_scheduler.py:136: UserWarning: Detected call of `lr_scheduler.step()` before `optimizer.step()`. In PyTorch 1.1.0 and later, you should call them in the opposite order: `optimizer.step()` before `lr_scheduler.step()`. Failure to do this will result in PyTorch skipping the first value of the learning rate schedule. See more details at https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate
"https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate", UserWarning)
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.78it/s]
all 10 6 0.000667 0.333 0.000858 8.58e-05
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
1/499 5.14G 0.1203 0.03151 0 144 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:01<00:00, 1.31s/it]
all 10 6 0.000667 0.333 0.000971 0.000118
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
2/499 5.96G 0.1204 0.03125 0 153 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:01<00:00, 1.02s/it]
all 10 6 0.000667 0.333 0.000981 0.000119
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
3/499 5.96G 0.1185 0.02987 0 150 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.40it/s]
all 10 6 0.000667 0.333 0.000993 0.000148
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
4/499 5.96G 0.118 0.02961 0 156 640: 100% 1/1 [00:00<00:00, 1.62it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.56it/s]
all 10 6 0.000667 0.333 0.000987 0.000121
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
5/499 5.96G 0.1142 0.02777 0 139 640: 100% 1/1 [00:00<00:00, 1.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.41it/s]
all 10 6 0.000667 0.333 0.00103 0.000125
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
6/499 5.96G 0.1135 0.02676 0 139 640: 100% 1/1 [00:00<00:00, 1.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.59it/s]
all 10 6 0.000667 0.333 0.00107 0.000129
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
7/499 5.96G 0.1067 0.02648 0 137 640: 100% 1/1 [00:00<00:00, 1.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 1.44it/s]
all 10 6 0.000667 0.333 0.00102 0.000149
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
8/499 5.96G 0.1054 0.02595 0 149 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.73it/s]
all 10 6 0.000667 0.333 0.00115 0.000136
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
9/499 5.96G 0.1005 0.02582 0 149 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.10it/s]
all 10 6 0.001 0.5 0.00207 0.000286
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
10/499 5.96G 0.0981 0.02648 0 169 640: 100% 1/1 [00:00<00:00, 1.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.69it/s]
all 10 6 0.000667 0.333 0.0024 0.000424
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
11/499 5.96G 0.09736 0.02494 0 155 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.10it/s]
all 10 6 0.001 0.5 0.00467 0.000746
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
12/499 5.96G 0.09276 0.02452 0 153 640: 100% 1/1 [00:00<00:00, 2.03it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.77it/s]
all 10 6 0.001 0.5 0.00299 0.000895
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
13/499 5.96G 0.09385 0.02483 0 164 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.56it/s]
all 10 6 0.001 0.5 0.00365 0.00104
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
14/499 5.96G 0.08736 0.0243 0 149 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.39it/s]
all 10 6 0.001 0.5 0.00453 0.00122
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
15/499 5.96G 0.08656 0.02303 0 139 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.52it/s]
all 10 6 0.001 0.5 0.00304 0.001
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
16/499 5.96G 0.0847 0.02405 0 152 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.75it/s]
all 10 6 0.00133 0.667 0.00605 0.00162
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
17/499 5.96G 0.08357 0.02342 0 149 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.61it/s]
all 10 6 0.00133 0.667 0.00611 0.00124
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
18/499 5.96G 0.0826 0.02287 0 150 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.57it/s]
all 10 6 0.001 0.5 0.00322 0.000963
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
19/499 5.96G 0.08491 0.02278 0 149 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.55it/s]
all 10 6 0.00133 0.667 0.00685 0.00198
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
20/499 5.96G 0.07793 0.02521 0 160 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.54it/s]
all 10 6 0.00167 0.833 0.00783 0.00153
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
21/499 5.96G 0.0781 0.02509 0 154 640: 100% 1/1 [00:00<00:00, 1.35it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.33it/s]
all 10 6 0.00167 0.833 0.00703 0.00133
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
22/499 5.96G 0.07704 0.02378 0 148 640: 100% 1/1 [00:00<00:00, 1.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.45it/s]
all 10 6 0.00167 0.833 0.00644 0.00155
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
23/499 5.96G 0.07363 0.02026 0 121 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.74it/s]
all 10 6 0.00133 0.667 0.0158 0.002
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
24/499 5.96G 0.071 0.02408 0 147 640: 100% 1/1 [00:00<00:00, 1.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.05it/s]
all 10 6 0.002 1 0.0198 0.0078
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
25/499 5.96G 0.07 0.02532 0 156 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.20it/s]
all 10 6 0.002 1 0.0223 0.00677
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
26/499 5.96G 0.07031 0.02243 0 134 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.46it/s]
all 10 6 0.00575 0.833 0.0149 0.00584
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
27/499 5.96G 0.07034 0.02449 0 156 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.82it/s]
all 10 6 0.00167 0.833 0.00807 0.00294
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
28/499 5.96G 0.06846 0.02437 0 151 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.23it/s]
all 10 6 0.00978 0.333 0.0246 0.00591
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
29/499 5.96G 0.06829 0.02355 0 151 640: 100% 1/1 [00:00<00:00, 1.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.57it/s]
all 10 6 0.00651 0.333 0.0206 0.00264
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
30/499 5.96G 0.06732 0.02154 0 129 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.61it/s]
all 10 6 0.013 0.333 0.028 0.00619
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
31/499 5.96G 0.06656 0.0241 0 147 640: 100% 1/1 [00:00<00:00, 1.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.19it/s]
all 10 6 0.002 1 0.00905 0.00196
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
32/499 5.96G 0.0657 0.02473 0 153 640: 100% 1/1 [00:00<00:00, 1.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 2.89it/s]
all 10 6 0.00689 0.167 0.0115 0.00397
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
33/499 5.96G 0.06711 0.0245 0 153 640: 100% 1/1 [00:00<00:00, 1.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.11it/s]
all 10 6 0.002 1 0.0136 0.00336
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
34/499 5.96G 0.06663 0.02761 0 175 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.67it/s]
all 10 6 0.0179 0.5 0.0218 0.00497
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
35/499 5.96G 0.06821 0.02065 0 134 640: 100% 1/1 [00:00<00:00, 1.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.23it/s]
all 10 6 0.0114 0.333 0.0282 0.00554
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
36/499 5.96G 0.06083 0.02291 0 140 640: 100% 1/1 [00:00<00:00, 1.99it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.59it/s]
all 10 6 0.0214 0.333 0.0165 0.00483
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
37/499 5.96G 0.06378 0.02607 0 162 640: 100% 1/1 [00:00<00:00, 1.51it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.98it/s]
all 10 6 0.0145 0.333 0.0222 0.00606
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
38/499 5.96G 0.06067 0.02375 0 152 640: 100% 1/1 [00:00<00:00, 1.03it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.69it/s]
all 10 6 0.0212 0.667 0.0289 0.00602
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
39/499 5.96G 0.05959 0.02384 0 150 640: 100% 1/1 [00:00<00:00, 1.22it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.81it/s]
all 10 6 0.0248 0.833 0.0299 0.00601
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
40/499 5.96G 0.06211 0.02478 0 158 640: 100% 1/1 [00:00<00:00, 1.35it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.98it/s]
all 10 6 0.0174 0.333 0.0197 0.00662
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
41/499 5.96G 0.06154 0.02085 0 130 640: 100% 1/1 [00:00<00:00, 1.25it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.20it/s]
all 10 6 0.0231 0.5 0.0357 0.00898
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
42/499 5.96G 0.06258 0.02429 0 155 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.54it/s]
all 10 6 0.0187 0.333 0.0211 0.00628
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
43/499 5.96G 0.06071 0.02486 0 164 640: 100% 1/1 [00:00<00:00, 1.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.53it/s]
all 10 6 0.016 0.5 0.0236 0.00592
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
44/499 5.96G 0.05969 0.02315 0 154 640: 100% 1/1 [00:00<00:00, 1.37it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.64it/s]
all 10 6 0.0172 0.333 0.0174 0.00589
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
45/499 5.96G 0.06168 0.02558 0 177 640: 100% 1/1 [00:01<00:00, 1.44s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.26it/s]
all 10 6 0.0206 0.5 0.0258 0.0079
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
46/499 5.96G 0.05922 0.02248 0 148 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.73it/s]
all 10 6 0.0127 0.167 0.0134 0.00479
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
47/499 5.96G 0.06214 0.02261 0 151 640: 100% 1/1 [00:00<00:00, 1.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.24it/s]
all 10 6 0.0189 0.176 0.0245 0.00903
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
48/499 5.96G 0.06146 0.02368 0 163 640: 100% 1/1 [00:00<00:00, 1.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.98it/s]
all 10 6 0.0151 0.167 0.0161 0.00495
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
49/499 5.96G 0.05796 0.0211 0 140 640: 100% 1/1 [00:00<00:00, 1.07it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.65it/s]
all 10 6 0.0418 0.5 0.0331 0.0127
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
50/499 5.96G 0.05973 0.0218 0 148 640: 100% 1/1 [00:00<00:00, 1.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.50it/s]
all 10 6 0.0282 0.167 0.0227 0.00634
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
51/499 5.96G 0.05838 0.02405 0 159 640: 100% 1/1 [00:00<00:00, 1.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.36it/s]
all 10 6 0.0366 0.5 0.0421 0.0152
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
52/499 5.96G 0.05796 0.02065 0 144 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.21it/s]
all 10 6 0.0276 0.167 0.0216 0.00622
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
53/499 5.96G 0.05882 0.01987 0 140 640: 100% 1/1 [00:00<00:00, 1.20it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.46it/s]
all 10 6 0.0631 0.5 0.0612 0.02
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
54/499 5.96G 0.05793 0.01895 0 127 640: 100% 1/1 [00:00<00:00, 1.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.35it/s]
all 10 6 0.0759 0.333 0.0662 0.0132
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
55/499 5.96G 0.05711 0.02143 0 148 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.34it/s]
all 10 6 0.0409 0.333 0.0576 0.0183
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
56/499 5.96G 0.0613 0.02081 0 147 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.97it/s]
all 10 6 0.0457 0.5 0.0428 0.0147
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
57/499 5.96G 0.05359 0.02156 0 140 640: 100% 1/1 [00:00<00:00, 2.01it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.37it/s]
all 10 6 0.0622 0.667 0.0605 0.0245
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
58/499 5.96G 0.05914 0.02205 0 150 640: 100% 1/1 [00:00<00:00, 1.61it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.64it/s]
all 10 6 0.0746 0.167 0.0507 0.0116
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
59/499 5.96G 0.05499 0.02071 0 139 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.42it/s]
all 10 6 0.0851 0.167 0.101 0.0451
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
60/499 5.96G 0.05683 0.02407 0 161 640: 100% 1/1 [00:00<00:00, 1.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.25it/s]
all 10 6 0.0833 0.167 0.0589 0.0153
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
61/499 5.96G 0.05875 0.01961 0 142 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.96it/s]
all 10 6 0.115 0.333 0.118 0.0464
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
62/499 5.96G 0.05698 0.0227 0 159 640: 100% 1/1 [00:00<00:00, 1.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.61it/s]
all 10 6 0.0933 0.167 0.0684 0.0214
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
63/499 5.96G 0.057 0.02292 0 162 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.25it/s]
all 10 6 0.146 0.167 0.0891 0.0336
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
64/499 5.96G 0.05821 0.02073 0 149 640: 100% 1/1 [00:00<00:00, 2.00it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.83it/s]
all 10 6 0.0938 0.167 0.0611 0.0281
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
65/499 5.96G 0.05417 0.01731 0 122 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.10it/s]
all 10 6 0.119 0.333 0.089 0.0302
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
66/499 5.96G 0.05906 0.02072 0 142 640: 100% 1/1 [00:00<00:00, 1.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.89it/s]
all 10 6 0.371 0.333 0.184 0.0741
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
67/499 5.96G 0.05515 0.0198 0 138 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.11it/s]
all 10 6 0.203 0.333 0.13 0.0464
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
68/499 5.96G 0.05823 0.02089 0 154 640: 100% 1/1 [00:00<00:00, 1.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.62it/s]
all 10 6 0.178 0.167 0.0907 0.023
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
69/499 5.96G 0.05687 0.02053 0 142 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.69it/s]
all 10 6 0.237 0.333 0.167 0.0711
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
70/499 5.96G 0.05393 0.02156 0 151 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.16it/s]
all 10 6 0.236 0.167 0.22 0.0471
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
71/499 5.96G 0.05878 0.01851 0 130 640: 100% 1/1 [00:00<00:00, 1.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.82it/s]
all 10 6 0.572 0.333 0.441 0.231
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
72/499 5.96G 0.0526 0.02177 0 157 640: 100% 1/1 [00:00<00:00, 1.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.52it/s]
all 10 6 0.293 0.333 0.281 0.0361
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
73/499 5.96G 0.05713 0.0223 0 166 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.49it/s]
all 10 6 0.73 0.333 0.466 0.236
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
74/499 5.96G 0.05365 0.02033 0 148 640: 100% 1/1 [00:00<00:00, 1.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.89it/s]
all 10 6 0.498 0.333 0.312 0.0905
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
75/499 5.96G 0.05534 0.02208 0 164 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.23it/s]
all 10 6 0.837 0.5 0.537 0.293
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
76/499 5.96G 0.05884 0.0192 0 147 640: 100% 1/1 [00:00<00:00, 1.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.80it/s]
all 10 6 0.561 0.333 0.311 0.0409
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
77/499 5.96G 0.05254 0.01826 0 138 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.89it/s]
all 10 6 0.926 0.5 0.54 0.242
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
78/499 5.96G 0.05776 0.01809 0 137 640: 100% 1/1 [00:00<00:00, 1.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.05it/s]
all 10 6 0.562 0.333 0.342 0.0685
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
79/499 5.96G 0.05177 0.01985 0 147 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.79it/s]
all 10 6 0.95 0.5 0.553 0.176
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
80/499 5.96G 0.05432 0.01967 0 152 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.99it/s]
all 10 6 0.524 0.333 0.346 0.178
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
81/499 5.96G 0.05492 0.01934 0 152 640: 100% 1/1 [00:00<00:00, 1.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.18it/s]
all 10 6 0.514 0.5 0.526 0.139
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
82/499 5.96G 0.05586 0.01857 0 138 640: 100% 1/1 [00:00<00:00, 1.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.42it/s]
all 10 6 0.5 0.333 0.322 0.0785
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
83/499 5.96G 0.05101 0.01675 0 127 640: 100% 1/1 [00:00<00:00, 1.29it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.11it/s]
all 10 6 0.519 0.5 0.562 0.193
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
84/499 5.96G 0.05091 0.01832 0 145 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.85it/s]
all 10 6 0.264 0.5 0.205 0.0288
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
85/499 5.96G 0.05531 0.02011 0 151 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.88it/s]
all 10 6 0.468 0.833 0.565 0.212
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
86/499 5.96G 0.05275 0.02258 0 171 640: 100% 1/1 [00:00<00:00, 1.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.28it/s]
all 10 6 0.285 0.5 0.301 0.0855
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
87/499 5.96G 0.05355 0.01911 0 147 640: 100% 1/1 [00:00<00:00, 1.54it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.90it/s]
all 10 6 0.46 0.5 0.56 0.238
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
88/499 5.96G 0.05322 0.01857 0 141 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.52it/s]
all 10 6 0.631 0.5 0.463 0.0748
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
89/499 5.96G 0.05176 0.0176 0 134 640: 100% 1/1 [00:00<00:00, 1.10it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.23it/s]
all 10 6 0.695 0.5 0.618 0.21
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
90/499 5.96G 0.05047 0.01539 0 125 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.31it/s]
all 10 6 0.561 0.333 0.318 0.0877
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
91/499 5.96G 0.05116 0.02071 0 164 640: 100% 1/1 [00:00<00:00, 1.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.57it/s]
all 10 6 0.936 0.5 0.733 0.189
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
92/499 5.96G 0.05503 0.02089 0 167 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.24it/s]
all 10 6 0.377 0.5 0.436 0.192
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
93/499 5.96G 0.05316 0.0193 0 149 640: 100% 1/1 [00:01<00:00, 1.02s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.76it/s]
all 10 6 0.563 0.647 0.626 0.17
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
94/499 5.96G 0.05154 0.01811 0 155 640: 100% 1/1 [00:00<00:00, 1.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.88it/s]
all 10 6 0.418 0.5 0.41 0.178
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
95/499 5.96G 0.0513 0.01589 0 134 640: 100% 1/1 [00:00<00:00, 1.12it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.93it/s]
all 10 6 0.416 0.83 0.392 0.119
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
96/499 5.96G 0.05307 0.01652 0 132 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.10it/s]
all 10 6 0.67 0.5 0.671 0.18
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
97/499 5.96G 0.05227 0.01729 0 140 640: 100% 1/1 [00:00<00:00, 1.23it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.45it/s]
all 10 6 0.634 0.584 0.672 0.218
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
98/499 5.96G 0.04806 0.01651 0 133 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.70it/s]
all 10 6 0.658 0.644 0.737 0.375
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
99/499 5.96G 0.05135 0.01875 0 153 640: 100% 1/1 [00:00<00:00, 1.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.12it/s]
all 10 6 0.634 0.667 0.599 0.22
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
100/499 5.96G 0.05542 0.01831 0 160 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.78it/s]
all 10 6 0.912 0.5 0.724 0.33
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
101/499 5.96G 0.05323 0.01938 0 159 640: 100% 1/1 [00:00<00:00, 1.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.97it/s]
all 10 6 0.616 0.333 0.457 0.123
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
102/499 5.96G 0.05015 0.01673 0 150 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.88it/s]
all 10 6 0.54 0.98 0.774 0.408
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
103/499 5.96G 0.0514 0.01706 0 146 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.35it/s]
all 10 6 0.74 0.333 0.485 0.0861
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
104/499 5.96G 0.05024 0.01871 0 154 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.19it/s]
all 10 6 0.941 0.5 0.726 0.356
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
105/499 5.96G 0.04805 0.01894 0 156 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.27it/s]
all 10 6 0.59 0.333 0.426 0.0942
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
106/499 5.96G 0.04916 0.02101 0 177 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.66it/s]
all 10 6 0.734 0.667 0.785 0.37
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
107/499 5.96G 0.05332 0.01899 0 153 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.73it/s]
all 10 6 0.139 0.541 0.115 0.0256
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
108/499 5.96G 0.0491 0.01678 0 144 640: 100% 1/1 [00:00<00:00, 1.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.21it/s]
all 10 6 0.884 0.5 0.687 0.361
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
109/499 5.96G 0.0495 0.01772 0 149 640: 100% 1/1 [00:01<00:00, 1.23s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.17it/s]
all 10 6 0.675 0.333 0.504 0.0642
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
110/499 5.96G 0.04892 0.01593 0 139 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.12it/s]
all 10 6 0.88 0.5 0.723 0.377
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
111/499 5.96G 0.05047 0.0183 0 158 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.00it/s]
all 10 6 0.582 0.333 0.444 0.15
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
112/499 5.96G 0.05063 0.01701 0 152 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.18it/s]
all 10 6 0.374 0.833 0.657 0.367
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
113/499 5.96G 0.05119 0.01624 0 151 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.38it/s]
all 10 6 0.399 0.667 0.531 0.132
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
114/499 5.96G 0.05094 0.01677 0 141 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.32it/s]
all 10 6 0.425 0.863 0.767 0.317
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
115/499 5.96G 0.04647 0.01629 0 141 640: 100% 1/1 [00:00<00:00, 1.09it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.57it/s]
all 10 6 0.175 0.333 0.113 0.0264
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
116/499 5.96G 0.04705 0.01542 0 145 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.28it/s]
all 10 6 0.597 0.987 0.771 0.317
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
117/499 5.96G 0.04824 0.01643 0 143 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.37it/s]
all 10 6 0.542 0.833 0.651 0.122
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
118/499 5.96G 0.04466 0.01641 0 140 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.71it/s]
all 10 6 0.529 1 0.69 0.385
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
119/499 5.96G 0.04748 0.01616 0 146 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.48it/s]
all 10 6 0.618 0.833 0.741 0.224
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
120/499 5.96G 0.04851 0.01761 0 164 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.70it/s]
all 10 6 0.689 0.833 0.685 0.388
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
121/499 5.96G 0.04742 0.01493 0 137 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.21it/s]
all 10 6 0.447 0.333 0.342 0.152
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
122/499 5.96G 0.04796 0.01652 0 149 640: 100% 1/1 [00:00<00:00, 1.54it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.95it/s]
all 10 6 0.46 0.5 0.518 0.284
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
123/499 5.96G 0.04752 0.01683 0 169 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.13it/s]
all 10 6 0.533 0.833 0.674 0.217
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
124/499 5.96G 0.04694 0.01487 0 140 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.02it/s]
all 10 6 0.348 0.833 0.561 0.288
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
125/499 5.96G 0.04499 0.01518 0 142 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.23it/s]
all 10 6 0.404 0.5 0.39 0.15
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
126/499 5.96G 0.0488 0.01742 0 158 640: 100% 1/1 [00:00<00:00, 1.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.61it/s]
all 10 6 0.582 1 0.663 0.286
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
127/499 5.96G 0.04873 0.01827 0 179 640: 100% 1/1 [00:00<00:00, 1.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.03it/s]
all 10 6 0.488 0.333 0.394 0.149
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
128/499 5.96G 0.04575 0.01485 0 139 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.99it/s]
all 10 6 0.634 1 0.805 0.365
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
129/499 5.96G 0.04173 0.01511 0 144 640: 100% 1/1 [00:00<00:00, 1.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.51it/s]
all 10 6 0.623 0.833 0.72 0.229
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
130/499 5.96G 0.04507 0.01409 0 126 640: 100% 1/1 [00:00<00:00, 1.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.47it/s]
all 10 6 0.621 0.667 0.662 0.149
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
131/499 5.96G 0.04857 0.01371 0 133 640: 100% 1/1 [00:01<00:00, 1.25s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.75it/s]
all 10 6 0.807 1 0.995 0.3
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
132/499 5.96G 0.0477 0.01472 0 137 640: 100% 1/1 [00:00<00:00, 1.45it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.93it/s]
all 10 6 0.396 0.656 0.416 0.0844
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
133/499 5.96G 0.04627 0.01729 0 149 640: 100% 1/1 [00:01<00:00, 1.22s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.20it/s]
all 10 6 0.715 0.833 0.885 0.236
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
134/499 5.96G 0.0461 0.01628 0 147 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.93it/s]
all 10 6 0.32 0.333 0.369 0.109
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
135/499 5.96G 0.04246 0.01451 0 143 640: 100% 1/1 [00:00<00:00, 1.60it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.37it/s]
all 10 6 0.864 1 0.995 0.408
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
136/499 5.96G 0.0462 0.01702 0 158 640: 100% 1/1 [00:00<00:00, 1.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.97it/s]
all 10 6 0.603 0.833 0.793 0.181
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
137/499 5.96G 0.04312 0.01748 0 165 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.05it/s]
all 10 6 0.748 0.991 0.83 0.315
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
138/499 5.96G 0.04745 0.01737 0 163 640: 100% 1/1 [00:00<00:00, 1.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.74it/s]
all 10 6 0.687 0.833 0.799 0.207
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
139/499 5.96G 0.04337 0.01587 0 147 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.91it/s]
all 10 6 0.74 1 0.995 0.313
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
140/499 5.96G 0.04616 0.01819 0 166 640: 100% 1/1 [00:00<00:00, 1.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.83it/s]
all 10 6 0.611 0.833 0.687 0.175
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
141/499 5.96G 0.04419 0.01443 0 134 640: 100% 1/1 [00:00<00:00, 1.33it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.27it/s]
all 10 6 0.467 0.5 0.346 0.163
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
142/499 5.96G 0.04241 0.01461 0 142 640: 100% 1/1 [00:00<00:00, 1.44it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.52it/s]
all 10 6 0.646 1 0.828 0.217
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
143/499 5.96G 0.04596 0.01473 0 140 640: 100% 1/1 [00:01<00:00, 1.01s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.31it/s]
all 10 6 0.65 1 0.718 0.188
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
144/499 5.96G 0.04157 0.01505 0 152 640: 100% 1/1 [00:00<00:00, 1.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.26it/s]
all 10 6 0.537 0.667 0.49 0.127
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
145/499 5.96G 0.0411 0.01625 0 164 640: 100% 1/1 [00:00<00:00, 1.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.53it/s]
all 10 6 0.563 0.833 0.614 0.197
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
146/499 5.96G 0.04842 0.01556 0 142 640: 100% 1/1 [00:00<00:00, 1.12it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.82it/s]
all 10 6 0.528 0.667 0.544 0.157
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
147/499 5.96G 0.04247 0.01685 0 164 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.44it/s]
all 10 6 0.799 0.833 0.809 0.22
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
148/499 5.96G 0.0423 0.01541 0 143 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.26it/s]
all 10 6 0.632 0.667 0.64 0.126
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
149/499 5.96G 0.0446 0.01777 0 173 640: 100% 1/1 [00:00<00:00, 1.16it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.91it/s]
all 10 6 0.966 1 0.995 0.375
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
150/499 5.96G 0.04172 0.0162 0 166 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.06it/s]
all 10 6 0.672 0.833 0.672 0.142
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
151/499 5.96G 0.04571 0.015 0 142 640: 100% 1/1 [00:00<00:00, 1.33it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.83it/s]
all 10 6 0.867 1 0.995 0.336
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
152/499 5.96G 0.04247 0.01488 0 149 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.21it/s]
all 10 6 0.732 0.833 0.747 0.234
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
153/499 5.96G 0.04524 0.0168 0 153 640: 100% 1/1 [00:00<00:00, 1.21it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.81it/s]
all 10 6 0.849 1 0.995 0.305
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
154/499 5.96G 0.04171 0.01391 0 133 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.55it/s]
all 10 6 0.679 0.833 0.728 0.241
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
155/499 5.96G 0.04381 0.01572 0 151 640: 100% 1/1 [00:00<00:00, 1.00it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.73it/s]
all 10 6 0.694 0.833 0.73 0.24
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
156/499 5.96G 0.04377 0.01523 0 146 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.44it/s]
all 10 6 0.967 1 0.995 0.409
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
157/499 5.96G 0.04368 0.01504 0 144 640: 100% 1/1 [00:00<00:00, 1.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.88it/s]
all 10 6 0.563 0.667 0.572 0.221
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
158/499 5.96G 0.04294 0.01479 0 146 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.26it/s]
all 10 6 0.976 1 0.995 0.384
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
159/499 5.96G 0.04372 0.01582 0 159 640: 100% 1/1 [00:00<00:00, 1.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.38it/s]
all 10 6 0.562 0.667 0.599 0.222
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
160/499 5.96G 0.04274 0.01453 0 140 640: 100% 1/1 [00:00<00:00, 1.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.01it/s]
all 10 6 0.979 1 0.995 0.368
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
161/499 5.96G 0.04359 0.01769 0 159 640: 100% 1/1 [00:00<00:00, 1.37it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.38it/s]
all 10 6 0.657 0.667 0.686 0.264
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
162/499 5.96G 0.04308 0.01499 0 153 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.39it/s]
all 10 6 0.835 1 0.948 0.436
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
163/499 5.96G 0.0416 0.01406 0 133 640: 100% 1/1 [00:00<00:00, 1.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.42it/s]
all 10 6 0.667 0.833 0.687 0.31
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
164/499 5.96G 0.04276 0.01474 0 137 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.71it/s]
all 10 6 0.824 1 0.972 0.422
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
165/499 5.96G 0.04173 0.01477 0 142 640: 100% 1/1 [00:00<00:00, 1.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.21it/s]
all 10 6 0.615 0.833 0.624 0.203
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
166/499 5.96G 0.04682 0.01397 0 127 640: 100% 1/1 [00:00<00:00, 1.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.37it/s]
all 10 6 0.908 1 0.995 0.48
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
167/499 5.96G 0.04456 0.01497 0 156 640: 100% 1/1 [00:00<00:00, 1.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.98it/s]
all 10 6 0.713 0.833 0.725 0.338
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
168/499 5.96G 0.04252 0.01714 0 171 640: 100% 1/1 [00:00<00:00, 1.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.64it/s]
all 10 6 0.818 0.833 0.752 0.285
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
169/499 5.96G 0.04229 0.01634 0 161 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.70it/s]
all 10 6 0.983 1 0.995 0.386
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
170/499 5.96G 0.04505 0.01308 0 135 640: 100% 1/1 [00:00<00:00, 1.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.31it/s]
all 10 6 0.816 0.833 0.784 0.312
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
171/499 5.96G 0.04004 0.01455 0 147 640: 100% 1/1 [00:00<00:00, 1.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.36it/s]
all 10 6 0.982 1 0.995 0.4
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
172/499 5.96G 0.0435 0.01565 0 163 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.24it/s]
all 10 6 0.811 0.833 0.753 0.324
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
173/499 5.96G 0.04086 0.01442 0 156 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.43it/s]
all 10 6 0.812 0.833 0.802 0.322
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
174/499 5.96G 0.04179 0.01441 0 139 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.11it/s]
all 10 6 0.84 1 0.995 0.426
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
175/499 5.96G 0.03695 0.0118 0 124 640: 100% 1/1 [00:00<00:00, 1.18it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.83it/s]
all 10 6 0.698 0.833 0.782 0.398
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
176/499 5.96G 0.04172 0.01795 0 164 640: 100% 1/1 [00:00<00:00, 1.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.76it/s]
all 10 6 0.848 1 0.995 0.489
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
177/499 5.96G 0.04104 0.01468 0 136 640: 100% 1/1 [00:00<00:00, 1.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.55it/s]
all 10 6 0.759 0.833 0.815 0.407
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
178/499 5.96G 0.04071 0.01622 0 143 640: 100% 1/1 [00:00<00:00, 1.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.50it/s]
all 10 6 0.871 1 0.995 0.594
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
179/499 5.96G 0.04221 0.01389 0 133 640: 100% 1/1 [00:00<00:00, 1.99it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.44it/s]
all 10 6 0.77 0.833 0.826 0.349
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
180/499 5.96G 0.04156 0.01541 0 159 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.15it/s]
all 10 6 0.972 1 0.995 0.559
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
181/499 5.96G 0.04121 0.01531 0 156 640: 100% 1/1 [00:00<00:00, 1.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.48it/s]
all 10 6 0.973 1 0.995 0.432
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
182/499 5.96G 0.04372 0.01466 0 150 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.83it/s]
all 10 6 0.969 1 0.995 0.557
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
183/499 5.96G 0.04328 0.01569 0 161 640: 100% 1/1 [00:01<00:00, 1.16s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.55it/s]
all 10 6 0.806 0.833 0.801 0.365
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
184/499 5.96G 0.04281 0.01469 0 152 640: 100% 1/1 [00:00<00:00, 1.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.03it/s]
all 10 6 0.971 1 0.995 0.5
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
185/499 5.96G 0.04155 0.01374 0 136 640: 100% 1/1 [00:01<00:00, 1.46s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.50it/s]
all 10 6 0.71 0.82 0.742 0.249
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
186/499 5.96G 0.04339 0.01505 0 147 640: 100% 1/1 [00:00<00:00, 1.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.24it/s]
all 10 6 0.975 1 0.995 0.36
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
187/499 5.96G 0.0408 0.01386 0 134 640: 100% 1/1 [00:00<00:00, 1.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.58it/s]
all 10 6 0.978 1 0.995 0.267
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
188/499 5.96G 0.03852 0.01307 0 136 640: 100% 1/1 [00:00<00:00, 1.29it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.13it/s]
all 10 6 0.977 1 0.995 0.309
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
189/499 5.96G 0.03636 0.01223 0 126 640: 100% 1/1 [00:00<00:00, 1.29it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.75it/s]
all 10 6 0.978 1 0.995 0.317
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
190/499 5.96G 0.03957 0.01414 0 139 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.04it/s]
all 10 6 0.813 0.833 0.766 0.286
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
191/499 5.96G 0.04007 0.01385 0 147 640: 100% 1/1 [00:00<00:00, 1.05it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.88it/s]
all 10 6 0.973 1 0.995 0.504
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
192/499 5.96G 0.04224 0.01448 0 145 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.05it/s]
all 10 6 0.812 0.833 0.768 0.36
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
193/499 5.96G 0.04005 0.01558 0 162 640: 100% 1/1 [00:01<00:00, 1.18s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.67it/s]
all 10 6 0.97 1 0.995 0.492
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
194/499 5.96G 0.04059 0.0154 0 151 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.41it/s]
all 10 6 0.809 0.833 0.79 0.36
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
195/499 5.96G 0.03908 0.0153 0 163 640: 100% 1/1 [00:00<00:00, 1.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.86it/s]
all 10 6 0.97 1 0.995 0.44
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
196/499 5.96G 0.04029 0.01545 0 161 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.40it/s]
all 10 6 0.812 0.833 0.793 0.394
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
197/499 5.96G 0.03979 0.01432 0 152 640: 100% 1/1 [00:00<00:00, 1.24it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.74it/s]
all 10 6 0.812 0.833 0.773 0.384
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
198/499 5.96G 0.03961 0.01604 0 159 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.65it/s]
all 10 6 0.809 0.833 0.797 0.411
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
199/499 5.96G 0.03917 0.01421 0 154 640: 100% 1/1 [00:00<00:00, 1.26it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.92it/s]
all 10 6 0.808 0.833 0.757 0.406
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
200/499 5.96G 0.04214 0.01617 0 156 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.17it/s]
all 10 6 0.97 1 0.995 0.562
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
201/499 5.96G 0.04015 0.01519 0 154 640: 100% 1/1 [00:00<00:00, 1.51it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.39it/s]
all 10 6 0.808 0.833 0.79 0.366
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
202/499 5.96G 0.04017 0.01491 0 146 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.72it/s]
all 10 6 0.971 1 0.995 0.531
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
203/499 5.96G 0.04013 0.01215 0 126 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.67it/s]
all 10 6 0.815 0.833 0.79 0.346
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
204/499 5.96G 0.041 0.01369 0 131 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.30it/s]
all 10 6 0.816 0.833 0.789 0.375
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
205/499 5.96G 0.0403 0.01594 0 154 640: 100% 1/1 [00:00<00:00, 1.08it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.75it/s]
all 10 6 0.816 0.833 0.754 0.281
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
206/499 5.96G 0.0404 0.01444 0 148 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.60it/s]
all 10 6 0.981 1 0.995 0.468
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
207/499 5.96G 0.04191 0.01535 0 150 640: 100% 1/1 [00:00<00:00, 1.07it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.57it/s]
all 10 6 0.82 0.833 0.754 0.35
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
208/499 5.96G 0.03811 0.01341 0 143 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.27it/s]
all 10 6 0.977 1 0.995 0.506
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
209/499 5.96G 0.03872 0.01435 0 152 640: 100% 1/1 [00:00<00:00, 1.53it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.05it/s]
all 10 6 0.978 1 0.995 0.36
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
210/499 5.96G 0.04299 0.01532 0 151 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.68it/s]
all 10 6 0.711 0.823 0.657 0.277
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
211/499 5.96G 0.03848 0.01454 0 152 640: 100% 1/1 [00:01<00:00, 1.16s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.41it/s]
all 10 6 0.653 0.667 0.68 0.291
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
212/499 5.96G 0.04203 0.01326 0 142 640: 100% 1/1 [00:00<00:00, 1.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.71it/s]
all 10 6 0.818 0.833 0.927 0.356
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
213/499 5.96G 0.04158 0.01678 0 170 640: 100% 1/1 [00:00<00:00, 1.72it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.84it/s]
all 10 6 0.7 0.5 0.623 0.294
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
214/499 5.96G 0.04345 0.01532 0 156 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.37it/s]
all 10 6 0.819 0.833 0.867 0.357
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
215/499 5.96G 0.04179 0.01577 0 163 640: 100% 1/1 [00:00<00:00, 1.16it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.17it/s]
all 10 6 0.68 0.5 0.518 0.291
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
216/499 5.96G 0.0393 0.01309 0 138 640: 100% 1/1 [00:00<00:00, 1.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.33it/s]
all 10 6 0.819 0.833 0.863 0.346
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
217/499 5.96G 0.04448 0.01544 0 155 640: 100% 1/1 [00:00<00:00, 1.10it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.37it/s]
all 10 6 0.654 0.667 0.577 0.269
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
218/499 5.96G 0.04092 0.01533 0 159 640: 100% 1/1 [00:00<00:00, 1.96it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.31it/s]
all 10 6 0.987 1 0.995 0.352
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
219/499 5.96G 0.04166 0.01598 0 164 640: 100% 1/1 [00:01<00:00, 1.26s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.98it/s]
all 10 6 0.653 0.667 0.579 0.282
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
220/499 5.96G 0.03784 0.01547 0 156 640: 100% 1/1 [00:00<00:00, 1.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.95it/s]
all 10 6 0.653 0.667 0.629 0.325
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
221/499 5.96G 0.03674 0.01483 0 161 640: 100% 1/1 [00:01<00:00, 1.10s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.96it/s]
all 10 6 0.653 0.667 0.603 0.341
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
222/499 5.96G 0.03895 0.01465 0 142 640: 100% 1/1 [00:00<00:00, 1.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.74it/s]
all 10 6 0.817 0.833 0.913 0.409
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
223/499 5.96G 0.03802 0.01485 0 161 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.18it/s]
all 10 6 0.507 0.5 0.589 0.342
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
224/499 5.96G 0.04004 0.01586 0 149 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.68it/s]
all 10 6 0.723 1 0.927 0.435
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
225/499 5.96G 0.03911 0.01812 0 181 640: 100% 1/1 [00:01<00:00, 1.01s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.26it/s]
all 10 6 0.708 0.811 0.733 0.39
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
226/499 5.96G 0.04154 0.01756 0 181 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.28it/s]
all 10 6 0.855 0.984 0.948 0.517
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
227/499 5.96G 0.03608 0.0124 0 132 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.02it/s]
all 10 6 0.712 0.825 0.675 0.343
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
228/499 5.96G 0.04096 0.01672 0 163 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.99it/s]
all 10 6 0.986 1 0.995 0.482
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
229/499 5.96G 0.03877 0.01251 0 135 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.41it/s]
all 10 6 0.623 0.828 0.693 0.353
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
230/499 5.96G 0.03803 0.01474 0 159 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.03it/s]
all 10 6 0.985 1 0.995 0.418
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
231/499 5.96G 0.04001 0.01445 0 150 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.17it/s]
all 10 6 0.82 0.833 0.786 0.396
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
232/499 5.96G 0.03838 0.01455 0 142 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.27it/s]
all 10 6 0.985 1 0.995 0.432
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
233/499 5.96G 0.03955 0.01464 0 152 640: 100% 1/1 [00:00<00:00, 1.31it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.69it/s]
all 10 6 0.656 0.667 0.669 0.335
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
234/499 5.96G 0.03962 0.01547 0 151 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.69it/s]
all 10 6 0.82 0.833 0.871 0.357
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
235/499 5.96G 0.0371 0.01442 0 156 640: 100% 1/1 [00:00<00:00, 1.17it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.83it/s]
all 10 6 0.655 0.667 0.64 0.313
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
236/499 5.96G 0.0409 0.01305 0 140 640: 100% 1/1 [00:00<00:00, 1.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.76it/s]
all 10 6 0.82 0.833 0.894 0.365
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
237/499 5.96G 0.04056 0.01618 0 159 640: 100% 1/1 [00:00<00:00, 1.44it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.33it/s]
all 10 6 0.655 0.667 0.68 0.334
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
238/499 5.96G 0.03883 0.01381 0 157 640: 100% 1/1 [00:00<00:00, 1.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.82it/s]
all 10 6 0.709 0.816 0.707 0.393
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
239/499 5.96G 0.03566 0.01392 0 153 640: 100% 1/1 [00:01<00:00, 1.45s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.40it/s]
all 10 6 0.657 0.667 0.698 0.357
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
240/499 5.96G 0.03671 0.01401 0 151 640: 100% 1/1 [00:00<00:00, 1.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.16it/s]
all 10 6 0.978 1 0.995 0.445
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
241/499 5.96G 0.03753 0.01421 0 150 640: 100% 1/1 [00:01<00:00, 1.39s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.52it/s]
all 10 6 0.82 0.833 0.763 0.364
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
242/499 5.96G 0.03717 0.01265 0 132 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.82it/s]
all 10 6 0.656 0.667 0.592 0.378
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
243/499 5.96G 0.03937 0.01426 0 154 640: 100% 1/1 [00:00<00:00, 1.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.85it/s]
all 10 6 0.712 0.827 0.694 0.352
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
244/499 5.96G 0.03652 0.01409 0 163 640: 100% 1/1 [00:00<00:00, 1.96it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.66it/s]
all 10 6 0.713 0.828 0.661 0.386
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
245/499 5.96G 0.03637 0.01227 0 136 640: 100% 1/1 [00:00<00:00, 1.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.00it/s]
all 10 6 0.713 0.829 0.663 0.358
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
246/499 5.96G 0.03532 0.01413 0 158 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.06it/s]
all 10 6 0.713 0.828 0.694 0.373
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
247/499 5.96G 0.03723 0.01388 0 155 640: 100% 1/1 [00:00<00:00, 1.09it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.88it/s]
all 10 6 0.855 0.988 0.948 0.408
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
248/499 5.96G 0.03527 0.01395 0 154 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.01it/s]
all 10 6 0.701 0.5 0.52 0.276
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
249/499 5.96G 0.0386 0.01421 0 150 640: 100% 1/1 [00:00<00:00, 1.25it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.73it/s]
all 10 6 0.956 0.667 0.867 0.346
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
250/499 5.96G 0.03664 0.01546 0 163 640: 100% 1/1 [00:00<00:00, 1.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.80it/s]
all 10 6 0.71 0.5 0.657 0.317
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
251/499 5.96G 0.03615 0.01492 0 154 640: 100% 1/1 [00:00<00:00, 1.26it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.82it/s]
all 10 6 0.964 0.667 0.782 0.337
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
252/499 5.96G 0.03964 0.01492 0 160 640: 100% 1/1 [00:00<00:00, 1.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.94it/s]
all 10 6 0.708 0.5 0.598 0.315
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
253/499 5.96G 0.03725 0.01388 0 147 640: 100% 1/1 [00:01<00:00, 1.14s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.44it/s]
all 10 6 0.65 0.667 0.682 0.34
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
254/499 5.96G 0.03931 0.01614 0 165 640: 100% 1/1 [00:00<00:00, 1.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.46it/s]
all 10 6 0.729 1 0.927 0.371
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
255/499 5.96G 0.03713 0.01436 0 153 640: 100% 1/1 [00:00<00:00, 1.08it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.03it/s]
all 10 6 0.5 0.832 0.688 0.336
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
256/499 5.96G 0.04028 0.01386 0 148 640: 100% 1/1 [00:00<00:00, 1.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.28it/s]
all 10 6 0.821 0.833 0.913 0.362
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
257/499 5.96G 0.03744 0.01248 0 129 640: 100% 1/1 [00:00<00:00, 1.96it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.34it/s]
all 10 6 0.595 0.833 0.731 0.329
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
258/499 5.96G 0.04026 0.01415 0 134 640: 100% 1/1 [00:00<00:00, 1.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.24it/s]
all 10 6 0.823 0.833 0.903 0.342
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
259/499 5.96G 0.03626 0.01515 0 162 640: 100% 1/1 [00:00<00:00, 1.18it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.81it/s]
all 10 6 0.823 0.833 0.903 0.381
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
260/499 5.96G 0.03627 0.01332 0 152 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.48it/s]
all 10 6 0.657 0.667 0.713 0.323
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
261/499 5.96G 0.03606 0.01456 0 161 640: 100% 1/1 [00:00<00:00, 1.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.41it/s]
all 10 6 0.855 0.985 0.948 0.339
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
262/499 5.96G 0.0371 0.0139 0 139 640: 100% 1/1 [00:00<00:00, 1.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.88it/s]
all 10 6 0.657 0.667 0.731 0.296
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
263/499 5.96G 0.03871 0.01245 0 132 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.15it/s]
all 10 6 0.821 0.833 0.887 0.388
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
264/499 5.96G 0.03613 0.01407 0 151 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.54it/s]
all 10 6 0.656 0.667 0.686 0.288
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
265/499 5.96G 0.04097 0.013 0 133 640: 100% 1/1 [00:00<00:00, 1.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.32it/s]
all 10 6 0.815 0.833 0.837 0.346
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
266/499 5.96G 0.03729 0.01539 0 161 640: 100% 1/1 [00:00<00:00, 1.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.46it/s]
all 10 6 0.818 0.833 0.887 0.351
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
267/499 5.96G 0.03698 0.01412 0 155 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.42it/s]
all 10 6 0.814 0.833 0.832 0.339
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
268/499 5.96G 0.03544 0.01351 0 152 640: 100% 1/1 [00:00<00:00, 1.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.66it/s]
all 10 6 0.706 0.5 0.551 0.279
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
269/499 5.96G 0.03899 0.01293 0 141 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.71it/s]
all 10 6 0.951 0.667 0.798 0.376
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
270/499 5.96G 0.03729 0.01601 0 164 640: 100% 1/1 [00:00<00:00, 1.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.89it/s]
all 10 6 0.952 0.667 0.81 0.329
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
271/499 5.96G 0.03464 0.01593 0 170 640: 100% 1/1 [00:00<00:00, 1.19it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.79it/s]
all 10 6 0.954 0.667 0.867 0.324
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
272/499 5.96G 0.03712 0.01251 0 134 640: 100% 1/1 [00:00<00:00, 1.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.08it/s]
all 10 6 0.95 0.667 0.773 0.364
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
273/499 5.96G 0.03769 0.013 0 145 640: 100% 1/1 [00:00<00:00, 1.20it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.57it/s]
all 10 6 0.705 0.5 0.595 0.307
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
274/499 5.96G 0.03555 0.0147 0 153 640: 100% 1/1 [00:00<00:00, 1.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.70it/s]
all 10 6 0.855 0.987 0.948 0.397
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
275/499 5.96G 0.03654 0.01286 0 138 640: 100% 1/1 [00:00<00:00, 1.37it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.66it/s]
all 10 6 0.656 0.667 0.591 0.289
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
276/499 5.96G 0.03912 0.01418 0 141 640: 100% 1/1 [00:00<00:00, 1.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.45it/s]
all 10 6 0.818 0.833 0.913 0.363
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
277/499 5.96G 0.03512 0.0128 0 147 640: 100% 1/1 [00:01<00:00, 1.24s/it]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.24it/s]
all 10 6 0.448 0.677 0.643 0.282
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
278/499 5.96G 0.03658 0.01402 0 150 640: 100% 1/1 [00:00<00:00, 1.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.22it/s]
all 10 6 0.967 0.667 0.863 0.377
Stopping training early as no improvement observed in last 100 epochs. Best results observed at epoch 178, best model saved as best.pt.
To update EarlyStopping(patience=100) pass a new patience value, i.e. `python train.py --patience 300` or use `--patience 0` to disable EarlyStopping.
279 epochs completed in 0.138 hours.
Optimizer stripped from /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/last.pt, 14.5MB
Optimizer stripped from /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt, 14.5MB
Validating /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt...
Fusing layers...
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.80it/s]
all 10 6 0.871 1 0.995 0.594
Results saved to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train
Das Training wurde nach 279 Epochen abgebrochen, weil seit Epoche 178 keine Verbesserungen gemessen wurden.
Modell 2#
Im Modell 2 werden die besten Gewichte des zuvor trainierten Modell 1 weiter trainiert. Dafür wird der weights-Parameter mit der gespeicherten best.pt-Datei aufgerufen.
Das Training erfolgt mit einer anderen, von YOLO mitgelieferten, Hyperparameter-YAML-Datei, welche geringere Learning Rates und Weight-Decays verwendet. Dadurch kann das zuvor erreichte Minimum der Loss Function weiter reduziert werden.
TODO Loss Funtion
LR0 = 0.01 –> 0.00334
LRF = 0.01 –> 0.15135
Weight Decay = 0.005 –> 0.00025
Die weiter oben erläuterten Hyperparameter wie etwa Mosaic oder Mixup werden auf dem Standardwert dieser YAML-Datei belassen.
Des weiteren wird der oben erläuterte Backbone nicht eingefroren, also das gesamte Netz trainiert. Das Training erfolgt über 300 Epochen. Das ist ein Standardwert, weshalb der Parameter nicht übergeben werden muss. Gleiches gilt für die Batchsize 16.
!python train.py --hyp '/content/yolov5/data/hyps/hyp.VOC.yaml' --project /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports --name No2_S_FineTuning --weights '/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt' --img 640 --data stealth_bomber.yaml --cache
train: weights=/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt, cfg=, data=stealth_bomber.yaml, hyp=/content/yolov5/data/hyps/hyp.VOC.yaml, epochs=300, batch_size=16, imgsz=640, rect=False, resume=False, nosave=False, noval=False, noautoanchor=False, noplots=False, evolve=None, bucket=, cache=ram, image_weights=False, device=, multi_scale=False, single_cls=False, optimizer=SGD, sync_bn=False, workers=8, project=/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports, name=No2_S_FineTuning, exist_ok=False, quad=False, cos_lr=False, label_smoothing=0.0, patience=100, freeze=[0], save_period=-1, seed=0, local_rank=-1, entity=None, upload_dataset=False, bbox_interval=-1, artifact_alias=latest
github: up to date with https://github.com/ultralytics/yolov5 ✅
YOLOv5 🚀 v6.2-216-g6e544d5 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
hyperparameters: lr0=0.00334, lrf=0.15135, momentum=0.74832, weight_decay=0.00025, warmup_epochs=3.3835, warmup_momentum=0.59462, warmup_bias_lr=0.18657, box=0.02, cls=0.21638, cls_pw=0.5, obj=0.51728, obj_pw=0.67198, iou_t=0.2, anchor_t=3.3744, fl_gamma=0.0, hsv_h=0.01041, hsv_s=0.54703, hsv_v=0.27739, degrees=0.0, translate=0.04591, scale=0.75544, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, mosaic=0.85834, mixup=0.04266, copy_paste=0.0, anchors=3.412
ClearML: run 'pip install clearml' to automatically track, visualize and remotely train YOLOv5 🚀 in ClearML
Comet: run 'pip install comet_ml' to automatically track and visualize YOLOv5 🚀 runs in Comet
TensorBoard: Start with 'tensorboard --logdir /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports', view at http://localhost:6006/
Overriding model.yaml anchors with anchors=3.412
from n params module arguments
0 -1 1 3520 models.common.Conv [3, 32, 6, 2, 2]
1 -1 1 18560 models.common.Conv [32, 64, 3, 2]
2 -1 1 18816 models.common.C3 [64, 64, 1]
3 -1 1 73984 models.common.Conv [64, 128, 3, 2]
4 -1 2 115712 models.common.C3 [128, 128, 2]
5 -1 1 295424 models.common.Conv [128, 256, 3, 2]
6 -1 3 625152 models.common.C3 [256, 256, 3]
7 -1 1 1180672 models.common.Conv [256, 512, 3, 2]
8 -1 1 1182720 models.common.C3 [512, 512, 1]
9 -1 1 656896 models.common.SPPF [512, 512, 5]
10 -1 1 131584 models.common.Conv [512, 256, 1, 1]
11 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
12 [-1, 6] 1 0 models.common.Concat [1]
13 -1 1 361984 models.common.C3 [512, 256, 1, False]
14 -1 1 33024 models.common.Conv [256, 128, 1, 1]
15 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
16 [-1, 4] 1 0 models.common.Concat [1]
17 -1 1 90880 models.common.C3 [256, 128, 1, False]
18 -1 1 147712 models.common.Conv [128, 128, 3, 2]
19 [-1, 14] 1 0 models.common.Concat [1]
20 -1 1 296448 models.common.C3 [256, 256, 1, False]
21 -1 1 590336 models.common.Conv [256, 256, 3, 2]
22 [-1, 10] 1 0 models.common.Concat [1]
23 -1 1 1182720 models.common.C3 [512, 512, 1, False]
24 [17, 20, 23] 1 16182 models.yolo.Detect [1, [[0, 1, 2, 3, 4, 5], [0, 1, 2, 3, 4, 5], [0, 1, 2, 3, 4, 5]], [128, 256, 512]]
Model summary: 214 layers, 7022326 parameters, 7022326 gradients, 15.9 GFLOPs
Transferred 348/349 items from /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt
AMP: checks passed ✅
optimizer: SGD(lr=0.00334) with parameter groups 57 weight(decay=0.0), 60 weight(decay=0.00025), 60 bias
albumentations: Blur(p=0.01, blur_limit=(3, 7)), MedianBlur(p=0.01, blur_limit=(3, 7)), ToGray(p=0.01), CLAHE(p=0.01, clip_limit=(1, 4.0), tile_grid_size=(8, 8))
train: Scanning '/content/drive/MyDrive/Exkurs_B2Spirit/labels/train.cache' images and labels... 51 found, 33 missing, 0 empty, 0 corrupt: 100% 84/84 [00:00<?, ?it/s]
train: Caching images (0.1GB ram): 100% 84/84 [00:17<00:00, 4.93it/s]
val: Scanning '/content/drive/MyDrive/Exkurs_B2Spirit/labels/validation.cache' images and labels... 6 found, 4 missing, 0 empty, 0 corrupt: 100% 10/10 [00:00<?, ?it/s]
val: Caching images (0.0GB ram): 100% 10/10 [00:02<00:00, 3.85it/s]
AutoAnchor: 0.00 anchors/target, 0.000 Best Possible Recall (BPR). Anchors are a poor fit to dataset ⚠️, attempting to improve...
AutoAnchor: Running kmeans for 9 anchors on 51 points...
AutoAnchor: Evolving anchors with Genetic Algorithm: fitness = 0.9168: 100% 1000/1000 [00:00<00:00, 2918.05it/s]
AutoAnchor: thr=0.30: 1.0000 best possible recall, 7.57 anchors past thr
AutoAnchor: n=9, img_size=640, metric_all=0.515/0.917-mean/best, past_thr=0.565-mean: 39,99, 106,41, 48,123, 146,57, 78,108, 99,96, 63,153, 196,81, 149,187
AutoAnchor: Done ✅ (optional: update model *.yaml to use these anchors in the future)
Plotting labels to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning/labels.jpg...
Image sizes 640 train, 640 val
Using 2 dataloader workers
Logging results to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning
Starting training for 300 epochs...
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
0/299 3.73G 0.04313 0.004035 0 7 640: 100% 6/6 [00:04<00:00, 1.44it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.40it/s]
all 10 6 0.125 0.667 0.123 0.0194
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
1/299 3.73G 0.04054 0.003972 0 2 640: 100% 6/6 [00:01<00:00, 4.39it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.99it/s]
all 10 6 0.257 1 0.293 0.057
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
2/299 3.73G 0.04019 0.003727 0 9 640: 100% 6/6 [00:01<00:00, 4.53it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.47it/s]
all 10 6 0.291 1 0.314 0.109
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
3/299 3.73G 0.03826 0.003633 0 3 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.50it/s]
all 10 6 0.332 1 0.351 0.157
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
4/299 3.73G 0.03735 0.00307 0 5 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.99it/s]
all 10 6 0.332 1 0.351 0.187
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
5/299 3.73G 0.0373 0.003117 0 5 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.21it/s]
all 10 6 0.376 1 0.412 0.221
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
6/299 3.73G 0.03491 0.00279 0 3 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.57it/s]
all 10 6 0.407 1 0.426 0.249
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
7/299 3.73G 0.03589 0.002915 0 6 640: 100% 6/6 [00:01<00:00, 4.61it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.83it/s]
all 10 6 0.413 1 0.426 0.237
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
8/299 3.73G 0.03441 0.003378 0 6 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.83it/s]
all 10 6 0.422 1 0.497 0.288
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
9/299 3.73G 0.03451 0.002889 0 6 640: 100% 6/6 [00:01<00:00, 4.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 12.58it/s]
all 10 6 0.457 1 0.592 0.352
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
10/299 3.73G 0.03375 0.003068 0 6 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.14it/s]
all 10 6 0.553 1 0.775 0.33
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
11/299 3.73G 0.03173 0.002996 0 6 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.79it/s]
all 10 6 0.646 1 0.83 0.362
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
12/299 3.73G 0.0325 0.002951 0 8 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.04it/s]
all 10 6 0.663 1 0.83 0.406
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
13/299 3.73G 0.03174 0.003024 0 8 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.02it/s]
all 10 6 0.69 1 0.898 0.509
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
14/299 3.73G 0.03157 0.003346 0 5 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.26it/s]
all 10 6 0.692 1 0.948 0.561
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
15/299 3.73G 0.03113 0.002628 0 5 640: 100% 6/6 [00:01<00:00, 4.72it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.12it/s]
all 10 6 0.743 1 0.972 0.63
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
16/299 3.73G 0.03073 0.00313 0 2 640: 100% 6/6 [00:01<00:00, 5.10it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.86it/s]
all 10 6 0.826 1 0.972 0.589
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
17/299 3.73G 0.03012 0.002876 0 4 640: 100% 6/6 [00:01<00:00, 4.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.92it/s]
all 10 6 0.852 1 0.972 0.552
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
18/299 3.73G 0.02907 0.003045 0 2 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.32it/s]
all 10 6 0.85 0.951 0.972 0.586
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
19/299 3.73G 0.03006 0.003174 0 7 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.53it/s]
all 10 6 0.847 0.833 0.972 0.586
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
20/299 3.73G 0.02924 0.002866 0 7 640: 100% 6/6 [00:01<00:00, 5.29it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.02it/s]
all 10 6 0.889 1 0.995 0.637
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
21/299 3.73G 0.02812 0.002617 0 6 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.46it/s]
all 10 6 0.923 1 0.995 0.618
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
22/299 3.73G 0.02781 0.002529 0 8 640: 100% 6/6 [00:01<00:00, 4.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.29it/s]
all 10 6 0.943 1 0.995 0.618
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
23/299 3.73G 0.02832 0.002671 0 2 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.98it/s]
all 10 6 0.943 1 0.995 0.646
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
24/299 3.73G 0.02587 0.002438 0 2 640: 100% 6/6 [00:01<00:00, 5.14it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.11it/s]
all 10 6 0.973 1 0.995 0.646
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
25/299 3.73G 0.02834 0.002739 0 6 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.34it/s]
all 10 6 1 0.998 0.995 0.659
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
26/299 3.73G 0.02864 0.003051 0 6 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.27it/s]
all 10 6 1 0.986 0.995 0.69
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
27/299 3.73G 0.02724 0.003274 0 8 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.49it/s]
all 10 6 0.997 1 0.995 0.704
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
28/299 3.73G 0.02608 0.00267 0 3 640: 100% 6/6 [00:01<00:00, 4.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.35it/s]
all 10 6 1 0.979 0.995 0.684
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
29/299 3.73G 0.02723 0.002555 0 1 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.22it/s]
all 10 6 1 0.998 0.995 0.65
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
30/299 3.73G 0.02611 0.00281 0 1 640: 100% 6/6 [00:01<00:00, 4.99it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.58it/s]
all 10 6 1 0.983 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
31/299 3.73G 0.02559 0.002677 0 2 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.78it/s]
all 10 6 1 0.99 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
32/299 3.73G 0.02557 0.003224 0 8 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.10it/s]
all 10 6 0.83 1 0.995 0.685
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
33/299 3.73G 0.02475 0.00335 0 5 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.08it/s]
all 10 6 0.81 1 0.972 0.667
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
34/299 3.73G 0.02421 0.002885 0 6 640: 100% 6/6 [00:01<00:00, 4.11it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.27it/s]
all 10 6 0.794 1 0.972 0.659
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
35/299 3.73G 0.02399 0.002539 0 2 640: 100% 6/6 [00:01<00:00, 4.37it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.30it/s]
all 10 6 0.966 1 0.995 0.676
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
36/299 3.73G 0.02363 0.004063 0 14 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.57it/s]
all 10 6 0.959 1 0.995 0.7
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
37/299 3.73G 0.02488 0.003282 0 4 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.20it/s]
all 10 6 0.977 1 0.995 0.722
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
38/299 3.73G 0.0239 0.003327 0 6 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.41it/s]
all 10 6 0.983 1 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
39/299 3.73G 0.0225 0.002756 0 5 640: 100% 6/6 [00:01<00:00, 4.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.22it/s]
all 10 6 0.983 1 0.995 0.694
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
40/299 3.73G 0.02509 0.004132 0 9 640: 100% 6/6 [00:01<00:00, 5.12it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.74it/s]
all 10 6 0.983 1 0.995 0.652
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
41/299 3.73G 0.02119 0.002973 0 3 640: 100% 6/6 [00:01<00:00, 4.62it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.39it/s]
all 10 6 0.983 1 0.995 0.643
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
42/299 3.73G 0.02089 0.003062 0 4 640: 100% 6/6 [00:01<00:00, 4.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.97it/s]
all 10 6 0.983 1 0.995 0.649
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
43/299 3.73G 0.02181 0.00363 0 8 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.29it/s]
all 10 6 0.982 1 0.995 0.774
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
44/299 3.73G 0.02085 0.003177 0 4 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.69it/s]
all 10 6 0.981 1 0.995 0.762
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
45/299 3.73G 0.02103 0.003513 0 1 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.93it/s]
all 10 6 0.98 1 0.995 0.746
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
46/299 3.73G 0.02116 0.003736 0 7 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.43it/s]
all 10 6 0.985 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
47/299 3.73G 0.02045 0.004017 0 5 640: 100% 6/6 [00:01<00:00, 4.20it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.02it/s]
all 10 6 0.986 1 0.995 0.71
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
48/299 3.73G 0.01891 0.003514 0 5 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.02it/s]
all 10 6 0.985 1 0.995 0.73
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
49/299 3.73G 0.01997 0.003587 0 5 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.96it/s]
all 10 6 0.985 1 0.995 0.72
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
50/299 3.73G 0.01899 0.003655 0 5 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.86it/s]
all 10 6 0.985 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
51/299 3.73G 0.02048 0.003166 0 7 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.75it/s]
all 10 6 0.987 1 0.995 0.713
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
52/299 3.73G 0.01761 0.002738 0 3 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.96it/s]
all 10 6 0.986 1 0.995 0.713
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
53/299 3.73G 0.0189 0.003273 0 4 640: 100% 6/6 [00:01<00:00, 4.62it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.27it/s]
all 10 6 0.986 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
54/299 3.73G 0.02092 0.002856 0 1 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.49it/s]
all 10 6 0.986 1 0.995 0.732
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
55/299 3.73G 0.01788 0.00326 0 3 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.73it/s]
all 10 6 0.987 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
56/299 3.73G 0.01863 0.004516 0 6 640: 100% 6/6 [00:01<00:00, 4.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.17it/s]
all 10 6 0.988 1 0.995 0.732
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
57/299 3.73G 0.01985 0.003123 0 8 640: 100% 6/6 [00:01<00:00, 4.72it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.11it/s]
all 10 6 0.988 1 0.995 0.706
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
58/299 3.73G 0.0181 0.003809 0 1 640: 100% 6/6 [00:01<00:00, 5.03it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.29it/s]
all 10 6 0.987 1 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
59/299 3.73G 0.01864 0.004115 0 5 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.10it/s]
all 10 6 0.988 1 0.995 0.725
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
60/299 3.73G 0.01835 0.003837 0 6 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.24it/s]
all 10 6 0.987 1 0.995 0.733
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
61/299 3.73G 0.01729 0.00337 0 6 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.09it/s]
all 10 6 0.987 1 0.995 0.702
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
62/299 3.73G 0.018 0.003638 0 2 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.39it/s]
all 10 6 0.987 1 0.995 0.765
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
63/299 3.73G 0.01586 0.004115 0 4 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.33it/s]
all 10 6 0.986 1 0.995 0.78
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
64/299 3.73G 0.01621 0.003493 0 3 640: 100% 6/6 [00:01<00:00, 5.01it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.77it/s]
all 10 6 0.986 1 0.995 0.731
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
65/299 3.73G 0.01511 0.003639 0 3 640: 100% 6/6 [00:01<00:00, 4.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.42it/s]
all 10 6 0.986 1 0.995 0.78
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
66/299 3.73G 0.01447 0.003539 0 1 640: 100% 6/6 [00:01<00:00, 4.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.83it/s]
all 10 6 0.986 1 0.995 0.748
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
67/299 3.73G 0.01582 0.003645 0 11 640: 100% 6/6 [00:01<00:00, 3.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.81it/s]
all 10 6 0.987 1 0.995 0.712
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
68/299 3.73G 0.01589 0.003516 0 3 640: 100% 6/6 [00:01<00:00, 3.04it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.54it/s]
all 10 6 0.989 1 0.995 0.736
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
69/299 3.73G 0.01458 0.003605 0 4 640: 100% 6/6 [00:01<00:00, 4.29it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.52it/s]
all 10 6 0.989 1 0.995 0.717
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
70/299 3.73G 0.01604 0.003895 0 4 640: 100% 6/6 [00:01<00:00, 5.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.66it/s]
all 10 6 0.988 1 0.995 0.747
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
71/299 3.73G 0.01363 0.004405 0 9 640: 100% 6/6 [00:01<00:00, 4.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.38it/s]
all 10 6 0.989 1 0.995 0.716
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
72/299 3.73G 0.01448 0.004061 0 4 640: 100% 6/6 [00:01<00:00, 5.11it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.38it/s]
all 10 6 0.99 1 0.995 0.713
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
73/299 3.73G 0.01395 0.004422 0 4 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.63it/s]
all 10 6 0.989 1 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
74/299 3.73G 0.01262 0.003992 0 6 640: 100% 6/6 [00:01<00:00, 4.99it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.89it/s]
all 10 6 0.989 1 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
75/299 3.73G 0.0145 0.003629 0 2 640: 100% 6/6 [00:01<00:00, 4.72it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.60it/s]
all 10 6 0.989 1 0.995 0.762
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
76/299 3.73G 0.01329 0.003916 0 5 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.91it/s]
all 10 6 0.989 1 0.995 0.744
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
77/299 3.73G 0.01259 0.003496 0 3 640: 100% 6/6 [00:01<00:00, 4.49it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 12.97it/s]
all 10 6 0.989 1 0.995 0.714
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
78/299 3.73G 0.01368 0.004431 0 5 640: 100% 6/6 [00:01<00:00, 5.06it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.09it/s]
all 10 6 0.988 1 0.995 0.726
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
79/299 3.73G 0.01365 0.005059 0 7 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.74it/s]
all 10 6 0.989 1 0.995 0.728
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
80/299 3.73G 0.01185 0.003531 0 3 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.03it/s]
all 10 6 0.989 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
81/299 3.73G 0.01315 0.00363 0 4 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.71it/s]
all 10 6 0.99 1 0.995 0.695
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
82/299 3.73G 0.01271 0.003266 0 4 640: 100% 6/6 [00:01<00:00, 5.08it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.31it/s]
all 10 6 0.99 1 0.995 0.723
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
83/299 3.73G 0.01161 0.004284 0 4 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.63it/s]
all 10 6 0.99 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
84/299 3.73G 0.01198 0.00355 0 6 640: 100% 6/6 [00:01<00:00, 4.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.12it/s]
all 10 6 0.99 1 0.995 0.731
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
85/299 3.73G 0.01215 0.00367 0 4 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.85it/s]
all 10 6 0.99 1 0.995 0.76
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
86/299 3.73G 0.01369 0.003677 0 5 640: 100% 6/6 [00:01<00:00, 4.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.03it/s]
all 10 6 0.99 1 0.995 0.76
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
87/299 3.73G 0.01265 0.003054 0 2 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.11it/s]
all 10 6 0.99 1 0.995 0.76
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
88/299 3.73G 0.01414 0.004226 0 3 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.34it/s]
all 10 6 0.99 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
89/299 3.73G 0.0115 0.003156 0 4 640: 100% 6/6 [00:01<00:00, 4.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.16it/s]
all 10 6 0.99 1 0.995 0.719
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
90/299 3.73G 0.01124 0.004479 0 7 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.82it/s]
all 10 6 0.99 1 0.995 0.728
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
91/299 3.73G 0.01148 0.004661 0 7 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.70it/s]
all 10 6 0.989 1 0.995 0.728
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
92/299 3.73G 0.01241 0.003232 0 1 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.64it/s]
all 10 6 0.989 1 0.995 0.708
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
93/299 3.73G 0.01164 0.003547 0 2 640: 100% 6/6 [00:01<00:00, 4.57it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.14it/s]
all 10 6 0.99 1 0.995 0.732
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
94/299 3.73G 0.01111 0.003807 0 5 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.79it/s]
all 10 6 0.989 1 0.995 0.744
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
95/299 3.73G 0.01176 0.003925 0 10 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.67it/s]
all 10 6 0.99 1 0.995 0.75
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
96/299 3.73G 0.01101 0.004028 0 8 640: 100% 6/6 [00:01<00:00, 4.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.43it/s]
all 10 6 0.99 1 0.995 0.716
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
97/299 3.73G 0.01114 0.003284 0 1 640: 100% 6/6 [00:01<00:00, 4.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.13it/s]
all 10 6 0.989 1 0.995 0.685
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
98/299 3.73G 0.01092 0.004059 0 5 640: 100% 6/6 [00:01<00:00, 4.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.68it/s]
all 10 6 0.989 1 0.995 0.709
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
99/299 3.73G 0.01145 0.003279 0 2 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.98it/s]
all 10 6 0.99 1 0.995 0.759
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
100/299 3.73G 0.01111 0.00352 0 5 640: 100% 6/6 [00:01<00:00, 5.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.99it/s]
all 10 6 0.989 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
101/299 3.73G 0.01081 0.003193 0 4 640: 100% 6/6 [00:01<00:00, 4.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.75it/s]
all 10 6 0.989 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
102/299 3.73G 0.01125 0.003376 0 3 640: 100% 6/6 [00:01<00:00, 4.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.40it/s]
all 10 6 0.989 1 0.995 0.731
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
103/299 3.73G 0.01082 0.003295 0 5 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.53it/s]
all 10 6 0.989 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
104/299 3.73G 0.01174 0.00343 0 6 640: 100% 6/6 [00:01<00:00, 4.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.56it/s]
all 10 6 0.99 1 0.995 0.712
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
105/299 3.73G 0.01188 0.004028 0 3 640: 100% 6/6 [00:01<00:00, 4.56it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 11.17it/s]
all 10 6 0.989 1 0.995 0.692
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
106/299 3.73G 0.01121 0.003953 0 2 640: 100% 6/6 [00:01<00:00, 5.16it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.51it/s]
all 10 6 0.989 1 0.995 0.755
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
107/299 3.73G 0.01096 0.003464 0 3 640: 100% 6/6 [00:01<00:00, 4.58it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.68it/s]
all 10 6 0.989 1 0.995 0.785
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
108/299 3.73G 0.01102 0.003822 0 4 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.19it/s]
all 10 6 0.988 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
109/299 3.73G 0.01064 0.003349 0 3 640: 100% 6/6 [00:01<00:00, 4.60it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.70it/s]
all 10 6 0.988 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
110/299 3.73G 0.01082 0.003254 0 2 640: 100% 6/6 [00:01<00:00, 4.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.15it/s]
all 10 6 0.988 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
111/299 3.73G 0.01091 0.003575 0 6 640: 100% 6/6 [00:01<00:00, 4.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.24it/s]
all 10 6 0.988 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
112/299 3.73G 0.0109 0.003741 0 10 640: 100% 6/6 [00:01<00:00, 4.51it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.01it/s]
all 10 6 0.988 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
113/299 3.73G 0.01126 0.003164 0 3 640: 100% 6/6 [00:01<00:00, 4.83it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.54it/s]
all 10 6 0.988 1 0.995 0.775
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
114/299 3.73G 0.01012 0.004032 0 6 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.96it/s]
all 10 6 0.988 1 0.995 0.751
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
115/299 3.73G 0.0114 0.003888 0 3 640: 100% 6/6 [00:01<00:00, 4.56it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.03it/s]
all 10 6 0.988 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
116/299 3.73G 0.01009 0.003534 0 4 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.42it/s]
all 10 6 0.989 1 0.995 0.828
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
117/299 3.73G 0.009801 0.003673 0 5 640: 100% 6/6 [00:01<00:00, 4.61it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.99it/s]
all 10 6 0.988 1 0.995 0.8
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
118/299 3.73G 0.01087 0.004118 0 10 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.16it/s]
all 10 6 0.989 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
119/299 3.73G 0.01108 0.003552 0 4 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.34it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
120/299 3.73G 0.01185 0.003734 0 9 640: 100% 6/6 [00:01<00:00, 4.34it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.79it/s]
all 10 6 0.988 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
121/299 3.73G 0.009762 0.003738 0 6 640: 100% 6/6 [00:01<00:00, 4.58it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.02it/s]
all 10 6 0.988 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
122/299 3.73G 0.01246 0.003029 0 6 640: 100% 6/6 [00:01<00:00, 4.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.72it/s]
all 10 6 0.988 1 0.995 0.764
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
123/299 3.73G 0.009114 0.003018 0 6 640: 100% 6/6 [00:01<00:00, 4.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 11.63it/s]
all 10 6 0.988 1 0.995 0.771
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
124/299 3.73G 0.01061 0.003161 0 2 640: 100% 6/6 [00:01<00:00, 4.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.37it/s]
all 10 6 0.988 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
125/299 3.73G 0.01 0.003275 0 5 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.96it/s]
all 10 6 0.987 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
126/299 3.73G 0.01007 0.003525 0 3 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.48it/s]
all 10 6 0.987 1 0.995 0.75
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
127/299 3.73G 0.009394 0.003535 0 5 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.72it/s]
all 10 6 0.987 1 0.995 0.789
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
128/299 3.73G 0.009858 0.003558 0 4 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.69it/s]
all 10 6 0.987 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
129/299 3.73G 0.01052 0.004008 0 8 640: 100% 6/6 [00:01<00:00, 4.46it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 12.99it/s]
all 10 6 0.987 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
130/299 3.73G 0.01051 0.003101 0 5 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.88it/s]
all 10 6 0.987 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
131/299 3.73G 0.01076 0.003768 0 6 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.40it/s]
all 10 6 0.987 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
132/299 3.73G 0.009312 0.00371 0 6 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.38it/s]
all 10 6 0.987 1 0.995 0.819
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
133/299 3.73G 0.009911 0.003673 0 5 640: 100% 6/6 [00:01<00:00, 4.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.70it/s]
all 10 6 0.986 1 0.995 0.819
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
134/299 3.73G 0.009348 0.003809 0 5 640: 100% 6/6 [00:01<00:00, 4.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.72it/s]
all 10 6 0.988 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
135/299 3.73G 0.01018 0.003018 0 1 640: 100% 6/6 [00:01<00:00, 4.55it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.27it/s]
all 10 6 0.985 1 0.995 0.761
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
136/299 3.73G 0.01101 0.003363 0 6 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.33it/s]
all 10 6 0.985 1 0.995 0.782
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
137/299 3.73G 0.01007 0.003793 0 4 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.56it/s]
all 10 6 0.985 1 0.995 0.785
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
138/299 3.73G 0.01045 0.003424 0 5 640: 100% 6/6 [00:01<00:00, 5.13it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.48it/s]
all 10 6 0.984 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
139/299 3.73G 0.009365 0.003881 0 4 640: 100% 6/6 [00:01<00:00, 4.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.04it/s]
all 10 6 0.985 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
140/299 3.73G 0.008698 0.003574 0 3 640: 100% 6/6 [00:01<00:00, 5.11it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.32it/s]
all 10 6 0.985 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
141/299 3.73G 0.01105 0.003861 0 7 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.65it/s]
all 10 6 0.986 1 0.995 0.8
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
142/299 3.73G 0.01067 0.003279 0 4 640: 100% 6/6 [00:01<00:00, 4.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.05it/s]
all 10 6 0.986 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
143/299 3.73G 0.01033 0.003759 0 7 640: 100% 6/6 [00:01<00:00, 4.53it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 11.01it/s]
all 10 6 0.99 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
144/299 3.73G 0.009742 0.003355 0 6 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.15it/s]
all 10 6 0.986 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
145/299 3.73G 0.009744 0.003031 0 2 640: 100% 6/6 [00:01<00:00, 4.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.34it/s]
all 10 6 0.984 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
146/299 3.73G 0.01002 0.003397 0 3 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.74it/s]
all 10 6 0.984 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
147/299 3.73G 0.0115 0.00396 0 5 640: 100% 6/6 [00:01<00:00, 4.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.19it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
148/299 3.73G 0.008925 0.003139 0 1 640: 100% 6/6 [00:01<00:00, 5.13it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.65it/s]
all 10 6 0.985 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
149/299 3.73G 0.00958 0.003417 0 6 640: 100% 6/6 [00:01<00:00, 4.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.32it/s]
all 10 6 0.989 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
150/299 3.73G 0.01075 0.00363 0 4 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.91it/s]
all 10 6 0.989 1 0.995 0.743
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
151/299 3.73G 0.01054 0.003552 0 4 640: 100% 6/6 [00:01<00:00, 4.57it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.44it/s]
all 10 6 0.989 1 0.995 0.775
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
152/299 3.73G 0.01003 0.003568 0 2 640: 100% 6/6 [00:01<00:00, 4.97it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.46it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
153/299 3.73G 0.01126 0.003589 0 7 640: 100% 6/6 [00:01<00:00, 4.60it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.37it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
154/299 3.73G 0.01028 0.003242 0 2 640: 100% 6/6 [00:01<00:00, 4.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.11it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
155/299 3.73G 0.009854 0.003624 0 6 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.63it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
156/299 3.73G 0.007832 0.002851 0 1 640: 100% 6/6 [00:01<00:00, 5.15it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.85it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
157/299 3.73G 0.01012 0.003191 0 4 640: 100% 6/6 [00:01<00:00, 4.50it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.43it/s]
all 10 6 0.989 1 0.995 0.728
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
158/299 3.73G 0.009183 0.003801 0 6 640: 100% 6/6 [00:01<00:00, 4.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.92it/s]
all 10 6 0.989 1 0.995 0.728
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
159/299 3.73G 0.009291 0.003683 0 4 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.58it/s]
all 10 6 0.989 1 0.995 0.768
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
160/299 3.73G 0.009051 0.003014 0 3 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.53it/s]
all 10 6 0.989 1 0.995 0.759
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
161/299 3.73G 0.009207 0.003196 0 4 640: 100% 6/6 [00:01<00:00, 4.57it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.93it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
162/299 3.73G 0.01014 0.003478 0 8 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.05it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
163/299 3.73G 0.009351 0.003349 0 6 640: 100% 6/6 [00:01<00:00, 4.46it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.63it/s]
all 10 6 0.989 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
164/299 3.73G 0.01014 0.00368 0 4 640: 100% 6/6 [00:01<00:00, 4.74it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.77it/s]
all 10 6 0.989 1 0.995 0.755
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
165/299 3.73G 0.009215 0.00335 0 6 640: 100% 6/6 [00:01<00:00, 4.57it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.83it/s]
all 10 6 0.99 1 0.995 0.752
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
166/299 3.73G 0.01063 0.003437 0 4 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.53it/s]
all 10 6 0.99 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
167/299 3.73G 0.008403 0.003524 0 6 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.61it/s]
all 10 6 0.99 1 0.995 0.796
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
168/299 3.73G 0.009336 0.002766 0 2 640: 100% 6/6 [00:01<00:00, 5.18it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.20it/s]
all 10 6 0.99 1 0.995 0.796
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
169/299 3.73G 0.009634 0.002978 0 9 640: 100% 6/6 [00:01<00:00, 4.57it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.97it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
170/299 3.73G 0.009007 0.003022 0 3 640: 100% 6/6 [00:01<00:00, 3.26it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 3.47it/s]
all 10 6 0.99 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
171/299 3.73G 0.009337 0.003999 0 7 640: 100% 6/6 [00:01<00:00, 3.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.27it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
172/299 3.73G 0.008525 0.002994 0 2 640: 100% 6/6 [00:01<00:00, 4.90it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.52it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
173/299 3.73G 0.00911 0.002818 0 9 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.00it/s]
all 10 6 0.99 1 0.995 0.749
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
174/299 3.73G 0.009162 0.003482 0 5 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.09it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
175/299 3.73G 0.009041 0.003926 0 6 640: 100% 6/6 [00:01<00:00, 4.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.54it/s]
all 10 6 0.99 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
176/299 3.73G 0.009077 0.003351 0 2 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.84it/s]
all 10 6 0.99 1 0.995 0.769
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
177/299 3.73G 0.009606 0.003458 0 5 640: 100% 6/6 [00:01<00:00, 4.44it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.04it/s]
all 10 6 0.99 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
178/299 3.73G 0.01039 0.003598 0 5 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.56it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
179/299 3.73G 0.01116 0.003297 0 3 640: 100% 6/6 [00:01<00:00, 4.50it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.42it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
180/299 3.73G 0.009032 0.003572 0 5 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.57it/s]
all 10 6 0.99 1 0.995 0.812
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
181/299 3.73G 0.008482 0.002983 0 2 640: 100% 6/6 [00:01<00:00, 4.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.73it/s]
all 10 6 0.989 1 0.995 0.772
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
182/299 3.73G 0.009409 0.003636 0 4 640: 100% 6/6 [00:01<00:00, 4.56it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.41it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
183/299 3.73G 0.008392 0.003218 0 5 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.67it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
184/299 3.73G 0.008733 0.003173 0 4 640: 100% 6/6 [00:01<00:00, 5.05it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.34it/s]
all 10 6 0.99 1 0.995 0.792
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
185/299 3.73G 0.008969 0.003312 0 7 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.60it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
186/299 3.73G 0.008738 0.003363 0 4 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.08it/s]
all 10 6 0.99 1 0.995 0.796
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
187/299 3.73G 0.009528 0.003781 0 6 640: 100% 6/6 [00:01<00:00, 4.47it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.96it/s]
all 10 6 0.99 1 0.995 0.782
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
188/299 3.73G 0.0107 0.003445 0 2 640: 100% 6/6 [00:01<00:00, 4.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.81it/s]
all 10 6 0.99 1 0.995 0.815
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
189/299 3.73G 0.008911 0.003446 0 9 640: 100% 6/6 [00:01<00:00, 4.55it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.47it/s]
all 10 6 0.99 1 0.995 0.806
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
190/299 3.73G 0.01061 0.00391 0 3 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.07it/s]
all 10 6 0.99 1 0.995 0.788
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
191/299 3.73G 0.007908 0.002835 0 2 640: 100% 6/6 [00:01<00:00, 4.55it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.61it/s]
all 10 6 0.99 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
192/299 3.73G 0.007396 0.002658 0 3 640: 100% 6/6 [00:01<00:00, 4.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.60it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
193/299 3.73G 0.009899 0.004625 0 9 640: 100% 6/6 [00:01<00:00, 4.49it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.06it/s]
all 10 6 0.99 1 0.995 0.807
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
194/299 3.73G 0.008675 0.002876 0 4 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.02it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
195/299 3.73G 0.008164 0.002888 0 4 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.07it/s]
all 10 6 0.99 1 0.995 0.788
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
196/299 3.73G 0.009664 0.003627 0 6 640: 100% 6/6 [00:01<00:00, 5.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.85it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
197/299 3.73G 0.009187 0.002674 0 4 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.01it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
198/299 3.73G 0.008834 0.003182 0 5 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.55it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
199/299 3.73G 0.009005 0.002601 0 3 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.09it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
200/299 3.73G 0.008212 0.003216 0 3 640: 100% 6/6 [00:01<00:00, 5.10it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.40it/s]
all 10 6 0.989 1 0.995 0.831
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
201/299 3.73G 0.009718 0.003333 0 10 640: 100% 6/6 [00:01<00:00, 4.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.10it/s]
all 10 6 0.99 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
202/299 3.73G 0.01367 0.002992 0 4 640: 100% 6/6 [00:01<00:00, 4.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.81it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
203/299 3.73G 0.008764 0.002484 0 4 640: 100% 6/6 [00:01<00:00, 4.78it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.88it/s]
all 10 6 0.99 1 0.995 0.788
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
204/299 3.73G 0.009255 0.003228 0 4 640: 100% 6/6 [00:01<00:00, 4.70it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.57it/s]
all 10 6 0.99 1 0.995 0.788
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
205/299 3.73G 0.00836 0.003504 0 3 640: 100% 6/6 [00:01<00:00, 4.54it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.73it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
206/299 3.73G 0.009353 0.00331 0 3 640: 100% 6/6 [00:01<00:00, 4.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.51it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
207/299 3.73G 0.01096 0.004204 0 7 640: 100% 6/6 [00:01<00:00, 4.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.39it/s]
all 10 6 0.99 1 0.995 0.793
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
208/299 3.73G 0.009436 0.003309 0 6 640: 100% 6/6 [00:01<00:00, 5.01it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.15it/s]
all 10 6 0.99 1 0.995 0.815
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
209/299 3.73G 0.009099 0.003131 0 6 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.17it/s]
all 10 6 0.99 1 0.995 0.793
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
210/299 3.73G 0.0104 0.003654 0 6 640: 100% 6/6 [00:01<00:00, 5.02it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.47it/s]
all 10 6 0.99 1 0.995 0.787
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
211/299 3.73G 0.008379 0.002809 0 5 640: 100% 6/6 [00:01<00:00, 4.49it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.42it/s]
all 10 6 0.99 1 0.995 0.801
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
212/299 3.73G 0.008406 0.003003 0 2 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.41it/s]
all 10 6 0.99 1 0.995 0.809
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
213/299 3.73G 0.008582 0.00332 0 2 640: 100% 6/6 [00:01<00:00, 4.68it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.03it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
214/299 3.73G 0.008906 0.003703 0 3 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.28it/s]
all 10 6 0.99 1 0.995 0.816
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
215/299 3.73G 0.01092 0.003839 0 5 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.75it/s]
all 10 6 0.99 1 0.995 0.836
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
216/299 3.73G 0.008879 0.003542 0 6 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.44it/s]
all 10 6 0.99 1 0.995 0.836
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
217/299 3.73G 0.007919 0.002617 0 5 640: 100% 6/6 [00:01<00:00, 4.49it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.60it/s]
all 10 6 0.99 1 0.995 0.823
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
218/299 3.73G 0.008018 0.002697 0 4 640: 100% 6/6 [00:01<00:00, 5.21it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.14it/s]
all 10 6 0.99 1 0.995 0.823
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
219/299 3.73G 0.01058 0.003174 0 3 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.74it/s]
all 10 6 0.99 1 0.995 0.788
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
220/299 3.73G 0.008325 0.002997 0 4 640: 100% 6/6 [00:01<00:00, 4.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.31it/s]
all 10 6 0.99 1 0.995 0.831
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
221/299 3.73G 0.009034 0.003085 0 8 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.64it/s]
all 10 6 0.99 1 0.995 0.823
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
222/299 3.73G 0.009036 0.00411 0 6 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.40it/s]
all 10 6 0.99 1 0.995 0.803
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
223/299 3.73G 0.008378 0.002915 0 4 640: 100% 6/6 [00:01<00:00, 4.91it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.81it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
224/299 3.73G 0.008377 0.003924 0 11 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.93it/s]
all 10 6 0.99 1 0.995 0.791
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
225/299 3.73G 0.008249 0.003425 0 5 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.23it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
226/299 3.73G 0.009577 0.003274 0 6 640: 100% 6/6 [00:01<00:00, 4.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.26it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
227/299 3.73G 0.008546 0.003127 0 3 640: 100% 6/6 [00:01<00:00, 4.42it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.35it/s]
all 10 6 0.99 1 0.995 0.828
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
228/299 3.73G 0.008031 0.003126 0 4 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.15it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
229/299 3.73G 0.00795 0.003043 0 7 640: 100% 6/6 [00:01<00:00, 4.62it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.81it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
230/299 3.73G 0.008817 0.003123 0 5 640: 100% 6/6 [00:01<00:00, 5.09it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.43it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
231/299 3.73G 0.01064 0.002663 0 5 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.82it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
232/299 3.73G 0.008347 0.00347 0 9 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.16it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
233/299 3.73G 0.009261 0.003589 0 3 640: 100% 6/6 [00:01<00:00, 4.45it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.08it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
234/299 3.73G 0.00976 0.003623 0 4 640: 100% 6/6 [00:01<00:00, 4.93it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.62it/s]
all 10 6 0.99 1 0.995 0.828
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
235/299 3.73G 0.007631 0.002862 0 3 640: 100% 6/6 [00:01<00:00, 4.46it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.24it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
236/299 3.73G 0.0102 0.003178 0 8 640: 100% 6/6 [00:01<00:00, 4.61it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.06it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
237/299 3.73G 0.008814 0.003333 0 6 640: 100% 6/6 [00:01<00:00, 4.51it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.37it/s]
all 10 6 0.99 1 0.995 0.803
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
238/299 3.73G 0.0094 0.003629 0 6 640: 100% 6/6 [00:01<00:00, 5.08it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.11it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
239/299 3.73G 0.008777 0.002798 0 3 640: 100% 6/6 [00:01<00:00, 4.51it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.36it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
240/299 3.73G 0.007499 0.002979 0 3 640: 100% 6/6 [00:01<00:00, 4.97it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.02it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
241/299 3.73G 0.008427 0.003169 0 5 640: 100% 6/6 [00:01<00:00, 4.55it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.57it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
242/299 3.73G 0.007547 0.00297 0 2 640: 100% 6/6 [00:01<00:00, 4.99it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.31it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
243/299 3.73G 0.008311 0.002667 0 5 640: 100% 6/6 [00:01<00:00, 4.81it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.94it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
244/299 3.73G 0.007572 0.003072 0 4 640: 100% 6/6 [00:01<00:00, 5.03it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.13it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
245/299 3.73G 0.008486 0.002789 0 6 640: 100% 6/6 [00:01<00:00, 4.46it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.79it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
246/299 3.73G 0.007749 0.003078 0 3 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.69it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
247/299 3.73G 0.008514 0.003351 0 7 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.99it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
248/299 3.73G 0.007989 0.002891 0 4 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.46it/s]
all 10 6 0.99 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
249/299 3.73G 0.009025 0.002718 0 2 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.96it/s]
all 10 6 0.99 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
250/299 3.73G 0.007318 0.002775 0 4 640: 100% 6/6 [00:01<00:00, 5.06it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.09it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
251/299 3.73G 0.008727 0.003042 0 7 640: 100% 6/6 [00:01<00:00, 4.84it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.52it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
252/299 3.73G 0.007835 0.00311 0 7 640: 100% 6/6 [00:01<00:00, 4.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.36it/s]
all 10 6 0.989 1 0.995 0.808
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
253/299 3.73G 0.01006 0.003175 0 5 640: 100% 6/6 [00:01<00:00, 4.62it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.81it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
254/299 3.73G 0.009437 0.003183 0 7 640: 100% 6/6 [00:01<00:00, 4.95it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.82it/s]
all 10 6 0.989 1 0.995 0.811
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
255/299 3.73G 0.008285 0.003689 0 5 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.32it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
256/299 3.73G 0.009838 0.003891 0 6 640: 100% 6/6 [00:01<00:00, 4.52it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.33it/s]
all 10 6 0.989 1 0.995 0.803
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
257/299 3.73G 0.008853 0.003777 0 6 640: 100% 6/6 [00:01<00:00, 4.56it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.44it/s]
all 10 6 0.99 1 0.995 0.8
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
258/299 3.73G 0.008905 0.003392 0 4 640: 100% 6/6 [00:01<00:00, 4.64it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.53it/s]
all 10 6 0.99 1 0.995 0.8
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
259/299 3.73G 0.007878 0.003151 0 4 640: 100% 6/6 [00:01<00:00, 4.79it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.12it/s]
all 10 6 0.989 1 0.995 0.793
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
260/299 3.73G 0.008253 0.003086 0 2 640: 100% 6/6 [00:01<00:00, 4.87it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.34it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
261/299 3.73G 0.008062 0.002975 0 5 640: 100% 6/6 [00:01<00:00, 4.65it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.79it/s]
all 10 6 0.99 1 0.995 0.778
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
262/299 3.73G 0.007966 0.002766 0 4 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.89it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
263/299 3.73G 0.008914 0.003107 0 2 640: 100% 6/6 [00:01<00:00, 4.53it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.39it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
264/299 3.73G 0.008387 0.003717 0 4 640: 100% 6/6 [00:01<00:00, 4.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.89it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
265/299 3.73G 0.008556 0.003592 0 8 640: 100% 6/6 [00:01<00:00, 4.98it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 5.08it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
266/299 3.73G 0.008762 0.003178 0 8 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.64it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
267/299 3.73G 0.009201 0.003003 0 2 640: 100% 6/6 [00:01<00:00, 4.89it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.35it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
268/299 3.73G 0.007353 0.002541 0 1 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.86it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
269/299 3.73G 0.008942 0.003136 0 4 640: 100% 6/6 [00:01<00:00, 4.59it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.33it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
270/299 3.73G 0.01032 0.00302 0 2 640: 100% 6/6 [00:01<00:00, 4.73it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.73it/s]
all 10 6 0.99 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
271/299 3.73G 0.007931 0.003076 0 5 640: 100% 6/6 [00:01<00:00, 4.45it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.69it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
272/299 3.73G 0.008839 0.002829 0 2 640: 100% 6/6 [00:01<00:00, 3.38it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.83it/s]
all 10 6 0.99 1 0.995 0.773
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
273/299 3.73G 0.008639 0.003088 0 3 640: 100% 6/6 [00:01<00:00, 4.13it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.56it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
274/299 3.73G 0.007453 0.003207 0 3 640: 100% 6/6 [00:01<00:00, 4.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.69it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
275/299 3.73G 0.007968 0.002804 0 3 640: 100% 6/6 [00:01<00:00, 4.92it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.57it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
276/299 3.73G 0.007501 0.00285 0 5 640: 100% 6/6 [00:01<00:00, 4.80it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 10.08it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
277/299 3.73G 0.008524 0.00276 0 4 640: 100% 6/6 [00:01<00:00, 4.76it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.49it/s]
all 10 6 0.989 1 0.995 0.776
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
278/299 3.73G 0.0071 0.00291 0 1 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.68it/s]
all 10 6 0.99 1 0.995 0.776
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
279/299 3.73G 0.008228 0.003286 0 3 640: 100% 6/6 [00:01<00:00, 4.66it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.77it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
280/299 3.73G 0.006748 0.003239 0 4 640: 100% 6/6 [00:01<00:00, 4.88it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.55it/s]
all 10 6 0.989 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
281/299 3.73G 0.007832 0.002889 0 1 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.89it/s]
all 10 6 0.989 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
282/299 3.73G 0.007406 0.002507 0 4 640: 100% 6/6 [00:01<00:00, 4.71it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.70it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
283/299 3.73G 0.008601 0.003283 0 6 640: 100% 6/6 [00:01<00:00, 4.58it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.37it/s]
all 10 6 0.989 1 0.995 0.817
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
284/299 3.73G 0.007784 0.003549 0 9 640: 100% 6/6 [00:01<00:00, 4.77it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.73it/s]
all 10 6 0.989 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
285/299 3.73G 0.009242 0.00284 0 2 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.59it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
286/299 3.73G 0.008363 0.003249 0 5 640: 100% 6/6 [00:01<00:00, 4.94it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.95it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
287/299 3.73G 0.007379 0.003075 0 3 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.40it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
288/299 3.73G 0.007843 0.002769 0 5 640: 100% 6/6 [00:01<00:00, 4.72it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 8.01it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
289/299 3.73G 0.008955 0.002673 0 3 640: 100% 6/6 [00:01<00:00, 4.56it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.74it/s]
all 10 6 0.99 1 0.995 0.803
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
290/299 3.73G 0.008851 0.003321 0 5 640: 100% 6/6 [00:01<00:00, 4.97it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.09it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
291/299 3.73G 0.007907 0.002837 0 3 640: 100% 6/6 [00:01<00:00, 4.67it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.32it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
292/299 3.73G 0.008944 0.003557 0 7 640: 100% 6/6 [00:01<00:00, 4.82it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.84it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
293/299 3.73G 0.008281 0.002684 0 8 640: 100% 6/6 [00:01<00:00, 4.86it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 6.38it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
294/299 3.73G 0.008792 0.00285 0 4 640: 100% 6/6 [00:01<00:00, 4.75it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.73it/s]
all 10 6 0.99 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
295/299 3.73G 0.008708 0.002748 0 2 640: 100% 6/6 [00:01<00:00, 4.63it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.79it/s]
all 10 6 0.99 1 0.995 0.798
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
296/299 3.73G 0.007094 0.00247 0 1 640: 100% 6/6 [00:01<00:00, 4.85it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.17it/s]
all 10 6 0.989 1 0.995 0.795
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
297/299 3.73G 0.007657 0.002593 0 3 640: 100% 6/6 [00:01<00:00, 3.42it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 4.20it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
298/299 3.73G 0.00686 0.00238 0 2 640: 100% 6/6 [00:01<00:00, 4.69it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 7.05it/s]
all 10 6 0.989 1 0.995 0.804
Epoch GPU_mem box_loss obj_loss cls_loss Instances Size
299/299 3.73G 0.007991 0.002985 0 3 640: 100% 6/6 [00:01<00:00, 4.48it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 11.56it/s]
all 10 6 0.989 1 0.995 0.804
300 epochs completed in 0.157 hours.
Optimizer stripped from /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning/weights/last.pt, 14.5MB
Optimizer stripped from /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning/weights/best.pt, 14.5MB
Validating /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning/weights/best.pt...
Fusing layers...
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:00<00:00, 9.47it/s]
all 10 6 0.99 1 0.995 0.836
Results saved to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning
Validation#
Metriken#
Nachfolgend werden die Trainingsmetriken dargestellt. Dabei ist zubeachten, dass der Train-Test-Split nicht optimal ist und die Ergebnisse der Validation entsprechend hinterfragt werden müssten. Anhand der beiden Modelle sollen die Metriken und ihre Wechselwirkungen erklärt werden.
In den folgenden Darstellungen werden je nach Trainings- und Validationsdaten die Metriken für box_loss (Bounding Box Regression Loss), obj_loss (Objectness Loss) und class_loss (Classification Loss) abgebildet. Die Metriken Precision, Recall, mAP50 und mAP50:95 haben einen großen Einfluss auf die Confusion Matrix und werden im nachfolgenden Abschnitt genauer erläutert.
Der Box Loss kann als Mean Squared Error interpretiert werden [Lihi Gur Arie, 2022].
Darunter wird der Grad der Unterscheidung zwischen der so genannten Groundtruth (siehe nachfolgenden Abschnitt) und der predicted Bounding Box verstanden. Umso geringer der Error, desto besser trifft die Prediction auf die Groundtruht zu. [hasty.ai, 2022]
Der Objectness Loss beschreibt die binary cross-entropy. Ein geringerer Loss Wert beschreibt eine hohe Konfidenz zum Vorhandensein eines Objekts. Der Classification Loss dient zur Unterscheidung zwischen den trainierten Objekten. Der class_loss ist Null, weil in dem Beispiel nur ein Label für den B2 Spirit trainiert wurde. Desto geringer dieser Loss, umso besser ist die Unterscheidung zwischen Objekten. [Lihi Gur Arie, 2022]
Modell 2 zeigt hinsichtlich box_loss (MSE) und obj_loss (binary cross-entropy) eine deutliche Verbesserung. Precision, Recall und mAP werden ebenfalls deutlich besser und weniger sprunghaft.
Fig. 6 Eigene Darstellung: Results Modell 1#
Fig. 7 Eigene Darstellung: Results Modell 2#
Confusion Matrix#
Theorie und Praxis#
Nachfolgend werden die zwei Confusion Matrixen für Modell 1 und Modell 2 abgebildet. Die Matrixen sind das Resultat einer automatischen Validierung am Ende des Trainings. Es wird gezeigt, dass das Modell 1 bei der Unterscheidungen zwischen Tarnkappenbomber und Hintergrund (Dreiecke aus Papier, andere Flugzeuge oder tatsächlicher Boden) eine hohe Unsicherheit hat. Für das Modell 2 ergibt sich diese Unsicherheit nicht mehr. Auffällig ist, dass der Background weiterhin nicht als Background klassifiziert wird. Der Grund dafür ist, dass im Kontext von Object Detection die Matrixen anders interpretiert werden müssen.
Für die Zuordnung von True Positive, False Positive, True Negative und False Negativ verwendet die Object Detection die Intersection over Union (IoU), welche maximiert wird. Die Überlappung von Groundtruth und Predicted Box im Verhältnis zur gemeinsamen Fläche ergeben den IoU. Die Groundtruth meint die für das Training bereitgestellte Bounding Box eines gelabelten Objekts. Die predicted Box stellt die Prediction des Modells dar. [Koech, 2020]
D.h. IoU verhält sich invers zum oben erwähnten Box Loss.
\( IoU = {area of overlap\over area of union} \)
Abhängig von einem bestimmten Grad der Überlappung (Threshold) werden die Predictions zugeordnet nach:
True Positive
False Positive
True Negative (nur für Objektklassen gültig)
False Negative
In einer Confusion Matrix für Mulit-Class Klassifikation ergeben sich die True Negatives als Summe aller Zellen, welche für die Predicted Class irrelavant sind [Bharathi, 2022].
Für den Hintergrund bei Object Detection ist dies nicht zutreffend, weil der Hintergrund nicht gelabelt ist. D.h. für False Negative Background wird die Groundtruth nicht erkannt und bei False Positve Background wird ein Objekt ohne Groundtruth (also ein nicht vorhandenes Objekt) erkannt. Daraus ergibt sich, dass die Precision gegen alle Detections (TP+FP) prüft und Recall gegen alle Groundtruths (TP+FN). [Koech, 2020]
Im Fall von YOLOv5 gibt es einen weiteren Sonderfall. Die angezeigten Werte sind auf den Recall normiert und für bessere Interpretation wurden die Zeile und die Spalte für FP Background und FN Background vertauscht. Dies ist eine deutliche Abweichung von üblichen Klassifikationmatrixen. D.h. abweichend von einer üblichen Klassifikationsmatrix verlaufen die False Positive Predictions parallel zur Predicted-Achse. [User rbavery, 2021]
Fig. 8 Eigene Darstellung: Confusion Matrix Modell 1#
Fig. 9 Eigene Darstellung: Confusion Matrix Modell 2#
Einfluss des Konfidenzwerts#
Einen entscheidenen Einfluss auf die Metriken hat auch der Konfidenzwert. Unter der Konfidenz ist die Wahrscheinlichkeit des Objekt-Vorhandenseins zu verstehen.
Bei der Validierung wird der Konfidenz-Defaultwert von 0.001 verwendet. Dies vervollständigt den Inhalt der Confusion Matrix, da alle FP und FN Werte einfließen. Jedoch ist ein so geringer Konfidenzwert nicht zielführend bei der späteren Inference (Anwendung des Modells).
Durch Erhöhen des Konfidenzwerts werden True Positive und False Positive Werte aussortiert und damit Precision und Recall verändert. Dieser Trade-Off zwischen Precision und Recall kann positive und negative Auswirkungen haben.
!python val.py --conf 0.9 --weights /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt --project /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports --name No2_S_Training_Validation_conf_09 --img 640 --data stealth_bomber.yaml
python: can't open file 'C:\dev\Object-Detection-Air-Base-Military-Equipment\val.py': [Errno 2] No such file or directory
!python val.py --conf 0.8 --weights /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt --project /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports --name No2_S_Training_Validation_conf_08 --img 640 --data stealth_bomber.yaml
val: data=/content/yolov5/data/stealth_bomber.yaml, weights=['/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt'], batch_size=32, imgsz=640, conf_thres=0.8, iou_thres=0.6, max_det=300, task=val, device=, workers=8, single_cls=False, augment=False, verbose=False, save_txt=False, save_hybrid=False, save_conf=False, save_json=False, project=/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports, name=No2_S_Training_Validation_conf_05, exist_ok=False, half=False, dnn=False
WARNING ⚠️ confidence threshold 0.8 > 0.001 produces invalid results
YOLOv5 🚀 v6.2-216-g6e544d5 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
Fusing layers...
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
val: Scanning '/content/drive/MyDrive/Exkurs_B2Spirit/labels/validation.cache' images and labels... 6 found, 4 missing, 0 empty, 0 corrupt: 100% 10/10 [00:00<?, ?it/s]
Class Images Instances P R mAP50 mAP50-95: 100% 1/1 [00:03<00:00, 3.34s/it]
all 10 6 1 1 0.995 0.609
Speed: 0.3ms pre-process, 9.2ms inference, 0.7ms NMS per image at shape (32, 3, 640, 640)
Results saved to /content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Training_Validation_conf_05
fig, (ax1,ax2) = plt.subplots(2,2,figsize=(15,15))
fig.tight_layout()
plt.suptitle('Grundmodell - Confusion Matrix per Confidence')
ax1[0].imshow(load_img('/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/confusion_matrix.png'))
ax2[0].imshow(load_img('/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Training_Validation_conf_09/confusion_matrix.png'))
ax1[1].imshow(load_img('/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Training_Validation_conf_08/confusion_matrix.png'))
ax1[0].set_title('Confidence: 0.001')
ax1[1].set_title('Confidence: 0.8')
ax2[0].set_title('Confidence: 0.9')
ax2[1].remove()
Fig. 10 Eigene Darstellung: Confusion Matrixen zu Konfidenzwert#
In den Plots werden Confusion Matrizen des Modells dargestellt. Es wird gezeigt, dass die Steigerung des Konfidenzwert zu mehr False Negative führt. Ergänzend müssen die Modell-Outputs in den jeweiligen Zellen beachtet werden, welche nochmal tabellarisch abgebildet werden.
Confidence |
0.001 |
0.8 |
0.9 |
|---|---|---|---|
Precision |
0.871 |
1 |
1 |
Recall |
1 |
1 |
0.5 |
Beim Sprung von Confidence 0.001 auf 0.8 verändert sich der Recall nicht, stattdessen erhöht sich die Precision. D.h. es werden Predictions (TP sowie FP) unterhalb der 0.8 “aussortiert” - eine positive Auswirkung. Beim Confidence Sprung von 0.8 auf 0.9 nimmt der Recall deutlich ab, es werden weitere True Positives aussortiert - eine negative Auswirkung. D.h. durch setzen eines höheren Konfidenzwerts können bei der Confusion Matrix die Ergebnisse des Modell 2 erreicht werden - ohne die zusätzlichen Tuning-Aktivitäten. Da vermeintliche Fehl-Klassifikation lediglich aussortiert wird ist es nicht zwingend die bessere Vorgehensweise.
Das bisher als Aussortieren bezeichnete Ereignis ist die Non-max suppression, welche auf das YOLOv5 Grid angewendet wird, um mehrfach Detektionen zu vermeiden. Dabei wird nur die höchste Wahrscheinlichkeit des Objekt-Vorhandenseins verwendet und weiter IoU-Werte ausgeblendet. Für mehrfache Objekte innerhalb einer Grid-Zelle wird dies je Anchorbox durchgeführt. Anchorboxen wurden bei der Erläuterung zu YOLOv5 beschrieben. [Yap, 2021]
Generell gilt, dass hohe Konfidenz-Grenzwert den Recall reduziert. Der sehr gute mAP-Wert des obigen Modells verursacht den Trade Off zwischen Precision und Recall erst im hohen Konfidenz-Bereich. Dieser Trade Off kann durch eine PR-Kurve visualisiert werden.
Das folgende Bild aus dem Github Repository zu YOLOv5 verdeutlicht den Zusammenhang zwischen Confidence, Precision und Recall. Zusätzlich wird der mAP als Fläche unter der PR-Kurve gezeigt.
mAP steht für mean Average Precision. Deste höher der Wert, umso besser ist das Modell. Dies wird im Anschluss über manuelle Veränderung des Konfidenzwerts vorgeführt. Die mean Average Precision ist der Durchschnitt aus den Average Precision Werten je Klasse bzw. Label. Es handelt sich um einen Trade-Off zwischen Precision und Recall.
Bzgl. der oben dargestellen Metrik-Grafiken steht die 0.5 in mAP0.5 für einen IoU Threshold von 50%. Der mAP0.5:0.95 beschreibt eine weitere Erhöhung des mAP bis zu 95% - in 5% Schritten.
Fig. 11 Beispiel PR-Kurve und Konfidenz-Threshold von Jocher [2022]#
Modellanwendung / Inference#
Vorbereitung#
Der Konfidenzwert hat auch einen Einfluss auf das Deployment. Für die Inference liegt der YOLOv5 Defaultwert für Confidence bei 0.25.
In den folgenden Codezellen werden zunächst die Modelle mit PyTorch geladen und ein Dictionary erstellt, welches die Resultate zu verschiedenen Internetfunden (Hickam Air Force Base und US Whiteman Air Force)und jeweils unterschiedlichen Zoom enthält.Mit dem Parameter augment = True wird das Eingabebild mit verschiedenen Anpassungen (bspw. Drehungen) untersucht. Dies verbessert die Erkennung auf Kosten der Laufzeit und ist daher nicht für alle Anwendungsfälle angebracht.
Am Ende des Fine Tunings mit kleineren Hyperparemetern zeigt sich eine Verbessung bei Precision und mAP50-95.
model_train = torch.hub.load('ultralytics/yolov5', 'custom', path='/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt')
model_fine = torch.hub.load('ultralytics/yolov5', 'custom', path='/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_FineTuning/weights/best.pt')
/usr/local/lib/python3.7/dist-packages/torch/hub.py:267: UserWarning: You are about to download and run code from an untrusted repository. In a future release, this won't be allowed. To add the repository to your trusted list, change the command to {calling_fn}(..., trust_repo=False) and a command prompt will appear asking for an explicit confirmation of trust, or load(..., trust_repo=True), which will assume that the prompt is to be answered with 'yes'. You can also use load(..., trust_repo='check') which will only prompt for confirmation if the repo is not already trusted. This will eventually be the default behaviour
"You are about to download and run code from an untrusted repository. In a future release, this won't "
Downloading: "https://github.com/ultralytics/yolov5/zipball/master" to /root/.cache/torch/hub/master.zip
INFO:yolov5:YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
INFO:yolov5:Fusing layers...
Fusing layers...
INFO:yolov5:Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
INFO:yolov5:Adding AutoShape...
Adding AutoShape...
Using cache found in /root/.cache/torch/hub/ultralytics_yolov5_master
INFO:yolov5:YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
INFO:yolov5:Fusing layers...
Fusing layers...
INFO:yolov5:Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
INFO:yolov5:Adding AutoShape...
Adding AutoShape...
detect_files = [
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/stealthbomber_testbild_mittel.png',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/stealthbomber_testbild_weit.png',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/stealthbomber_fliegend_nah.jpg',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/stealthbomber_fliegend_weit.jpg',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/Hawaii_Tarnkappenbomber.png',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/Hawaii_Tarnkappenbomber2.png',
'/content/drive/MyDrive/Exkurs_B2Spirit/test_images/Hawaii_Tarnkappenbomber3.png',
]
train_results = dict()
tune_results = dict()
for i in detect_files:
myKey = i.split('/')[-1].split('.')[0]
train_results[myKey] = model_train(i,augment=True)
tune_results[myKey] = model_fine(i,augment=True)
tune_results.keys()
dict_keys(['stealthbomber_testbild_mittel', 'stealthbomber_testbild_weit', 'stealthbomber_fliegend_nah', 'stealthbomber_fliegend_weit', 'Hawaii_Tarnkappenbomber', 'Hawaii_Tarnkappenbomber2', 'Hawaii_Tarnkappenbomber3'])
Hickam Air Force Base - Beispiel I#
Standard Confidence: 0.25#
Im Folgenden die Anwendung auf eine hohe Zoomstufe mit dem Standard-Confidence-Wert von YOLOv5.
Das Grundmodell erkennt ein dreieckiges Feld als Tarnkappenbomber, aber nicht die drei tatsächlichen Tarnkappenbomber im Zentrum des Bildes. Ähnliches gilt für die darauf folgende Ausgabe für das verbesserte Modell. Hier wird jedoch, wie gemäß der Confusion Matrix zu erwarten, das dreieckige Feld nicht mit False Positiv (Type I Error) klassifiziert.
train_results['Hawaii_Tarnkappenbomber'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies, U.S. Geological Survey, USGS
Fig. 12 Output zu Modell 1 mit Conf-Threshold 0.25 und weiten Zoom#
tune_results['Hawaii_Tarnkappenbomber'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies, U.S. Geological Survey, USGS
Fig. 13 Output zu Modell 2 mit Conf-Threshold 0.25 und weiten Zoom#
Erhöhte Confidence: 0.27#
Nachfolgend wird für das Grundmodell die Confidence auf 0.27 erhöht. Wodurch der Type I Error verschwindet.
model_train2 = torch.hub.load('ultralytics/yolov5', 'custom', path='/content/drive/MyDrive/Exkurs_B2Spirit/colab_exports/No2_S_Train/weights/best.pt')
model_train2.conf = 0.27
results = model_train2('/content/drive/MyDrive/Exkurs_B2Spirit/test_images/Hawaii_Tarnkappenbomber.png', augment=True)
results.show()
Using cache found in /root/.cache/torch/hub/ultralytics_yolov5_master
INFO:yolov5:YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
YOLOv5 🚀 2022-10-30 Python-3.7.15 torch-1.12.1+cu113 CUDA:0 (Tesla T4, 15110MiB)
INFO:yolov5:Fusing layers...
Fusing layers...
INFO:yolov5:Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
Model summary: 157 layers, 7012822 parameters, 0 gradients, 15.8 GFLOPs
INFO:yolov5:Adding AutoShape...
Adding AutoShape...
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies, U.S. Geological Survey, USGS
Fig. 14 Output zu Modell 1 mit Conf-Threshold 0.27 und weiten Zoom#
Hickam Air Force Base - Beispiel II#
Abschließend ein Screenshot auf größerer Zoomstufe.
Das Grundmodell ist sich bei der Erkennung des Tarnkappenbombers wesentlich sicherer als das verbesserte Modell. Die Ursache für diesen Effekt im Zusammenhang mit dem oben gezeigten Type I Error liegt darin, dass das Grundmodell die grobe Dreiecksform des Tarnkappenbombers stärker zu berücksichtigen scheint als das verbesserte Modell.
Vor dem Hintergrund der geringen Anzahl an Trainingsdaten ist vorstellbar, dass eine deutliche Erhöhung dieser Anzahl
die False-Positive-Rate des Grundmodells oder
die Confidence des verbesserten Modells für noch unterschiedlichere Flugfelder (Bildhintergrund)
verbessern könnte.
train_results['Hawaii_Tarnkappenbomber2'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies, U.S. Geological Survey, USGS
Fig. 15 Output zu Modell 1 mit nahen Zoom#
tune_results['Hawaii_Tarnkappenbomber2'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies, U.S. Geological Survey, USGS
Fig. 16 Output zu Modell 2 mit nahen Zoom#
US Whiteman Air Force Base#
Zur oberen Erläuterung passen auch die folgenden Ergebnisse bei der Anwendung auf eine vermeintliche Unfallstelle auf der Whitemen Air Force Base.
Das Grundmodell zeigt mehr Confidence für den Tarnkappenbomber, aber auch einen Type I Error für die Grünfläche.
Das verbesserte Modell zeigt weniger Confidence, dafür ohne Type I Error. Im Trainingsverlauf hat es keine Grünflächen in unmittelbarer Nähe des Flugzeugs kennengelernt.
train_results['stealthbomber_testbild_mittel'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies
Fig. 17 Output zu Modell 1 - Whiteman#
tune_results['stealthbomber_testbild_mittel'].show()
Kartendaten Copyright 2022: Google
Bilder Copyright 2022: CNES / Airbus, Maxer Technologies
Fig. 18 Output zu Modell 2 - Whiteman#