YOLOv5-Based Vehicle Detection Method for High-Resolution UAV Images

Table 1

Comparison of different algorithms.

Model	mAP@0.5	mAP@0.5:0.95	Precision	Recall	Average speed (ms)
FRCNN	0.494	0.243	0.525	0.480	96
Cascade RCNN	0.586	0.330	0.592	0.538	117
YOLOv5(s)	0.479	0.261	0.795	0.442	46
FRCNN+AC	0.784	0.488	0.715	0.655	341
Cascade RCNN+AC	0.815	0.540	0.742	0.657	788
YOLOv5(s)+AC	0.896	0.624	0.919	0.825	212