Table 4.

Detect training experiment with different types of YOLOv8 (Bolded data in black indicate the most significant effect in the indicator).

Model	Depth	Width	Parameters (M)	GFLOPs	FPS	p	mAP 0.5:0.95	Size (MB)
YOLOv8n	0.33	0.25	3.2	8.1	188.0	0.985	0.955	6.3
YOLOv8s	0.50	0.50	11.2	28.6	165.9	0.994	0.972	21.5
YOLOv8m	0.67	0.75	25.9	78.9	134.5	0.991	0.971	49.6
YOLOv8l	1.00	1.00	43.7	165.2	109	0.989	0.971	83.6
YOLOv8x	1.00	1.25	68.2	257.8	104.3	0.993	0.972	130.4