yolov8训练过程打印结果中的指标的含义

码农小乐

于 2025-02-17 17:12:50 发布

阅读量794

点赞数 7

文章标签： python 人工智能深度学习 YOLO

本文链接：https://blog.youkuaiyun.com/wanglele2010/article/details/145687136

版权

class：代表模型检测的类别名称；
Images:代表验证集图片总数；
Instances:代表每个类别目标所标注的总数；
P:代表精确率Precision=TP / (TP+FP), 在预测是Positive所有结果中，预测正确的比重
R:召回率recall=TP / (TP+FN), 在真实值为Positive的所有结果中，预测正确的比重
mAP50:表示IOU阈值大于0.5的平均精确度（Mean Average Precision, mAP）
mAP50-95:表示在不同IoU阈值（从0.5到0.95，步长0.05）（0.5、0.55、0.6、0.65、0.7、0.75、0.8、0.85、0.9、0.95）上的平均mAP

epochs=100, time=None, patience=100, batch=8, imgsz=640, save=True, save_period=-1, cache=False, device=0, workers=4, project=runs/train, name=exp, exist_ok=False, save_dir=runs\train\exp

Starting training for 100 epochs...

Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
1/100 2.55G 3.86 3.943 3.635 16 640: 100%|██████████| 196/196 [00:39<00:00, 4.99it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:23<00:00, 4.14it/s]
all 1564 9023 0.699 0.0383 0.0413 0.0148
0%| | 0/196 [00:00<?, ?it/s]
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
2/100 2.54G 2.921 2.99 2.887 26 640: 100%|██████████| 196/196 [00:37<00:00, 5.17it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:14<00:00, 6.63it/s]
all 1564 9023 0.334 0.161 0.133 0.0628
0%| | 0/196 [00:00<?, ?it/s]
。。。。。。

Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
97/100 2.51G 0.6249 0.3714 0.9784 27 640: 100%|██████████| 196/196 [00:37<00:00, 5.17it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:14<00:00, 6.95it/s]
all 1564 9023 0.966 0.969 0.991 0.859

Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
98/100 2.51G 0.6257 0.3692 0.9659 34 640: 100%|██████████| 196/196 [00:37<00:00, 5.19it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:13<00:00, 7.01it/s]
all 1564 9023 0.962 0.97 0.991 0.855

Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
99/100 2.52G 0.6219 0.3562 0.9713 35 640: 100%|██████████| 196/196 [00:37<00:00, 5.19it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:13<00:00, 7.01it/s]
all 1564 9023 0.964 0.97 0.991 0.857
0%| | 0/196 [00:00<?, ?it/s]
Epoch GPU_mem box_loss cls_loss dfl_loss Instances Size
100/100 2.52G 0.6328 0.3653 0.978 10 640: 100%|██████████| 196/196 [00:37<00:00, 5.20it/s]
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:13<00:00, 7.01it/s]
all 1564 9023 0.965 0.969 0.991 0.863

100 epochs completed in 1.516 hours.
Optimizer stripped from runs\train\exp\weights\last.pt, 22.5MB
Optimizer stripped from runs\train\exp\weights\best.pt, 22.5MB

Validating runs\train\exp\weights\best.pt...
WARNING ⚠️ validating an untrained model YAML will result in 0 mAP.
YOLOv8s summary (fused): 168 layers, 11,126,358 parameters, 0 gradients, 28.4 GFLOPs
Class Images Instances Box(P R mAP50 mAP50-95): 100%|██████████| 98/98 [00:14<00:00, 6.69it/s]
all 1564 9023 0.963 0.969 0.991 0.868
LightCir 764 4414 0.955 0.973 0.99 0.89
DarkCir 1146 4609 0.97 0.966 0.991 0.845
Speed: 0.2ms preprocess, 6.6ms inference, 0.0ms loss, 0.5ms postprocess per image
Results saved to runs\train\exp

Process finished with exit code 0