Table 3.

The best YOLO model is based on the training and validation datasets.

Model	Peak Accuracy Epoch	Training Hours(hr)	Threshold	Precision(P)	Recall(R)	F1-Score	mAP50
YOLOv7x	654	65.12hr	0.2	0.95	0.94	0.94	0.97
YOLOv7-W6	675	60.65hr	0.2	0.93	0.938	0.92	0.97
YOLOv7-D6	600	93.29hr	0.2	0.95	0.94	0.94	0.98
YOLOv8s	137	5.799hr	0.7	0.94	0.95	0.94	0.98
YOLOv8n	407	11.71hr	0.7	0.946	0.953	0.94	0.986
YOLOv8m	247	20.127hr	0.7	0.949	0.952	0.95	0.986
YOLOv8l	186	21.345hr	0.7	0.945	0.956	0.95	0.986
YOLOv8x	220	44.239hr	0.7	0.942	0.957	0.94	0.984