Object Detection Algorithm of Transmission Lines Based on Improved YOLOv5 Framework

Table 2

Comparative experimental results.

Algorithm	P	R	mAP@.5	mAP@.5:.95	FPS	Inference/ms
Yolov3	0.98	0.94	0.97	0.70	35	27
Yolov4	0.92	0.88	0.96	0.624	26	38
Yolov5s	0.962	0.925	0.947	0.688	269.72	3.7
Yolov5m	0.968	0.949	0.963	0.694	178	5.6
Yolov5l	0.99	0.939	0.964	0.697	126	7.9
Yolov5x	0.967	0.937	0.957	0.689	79	12.6
SSD	0.76	0.71	0.73	0.53	75	13
Faster R-CNN	0.969	0.90	0.967	0.637	11	91
Ours	0.964	0.949	0.969	0.686	258.36	3.9