A Deep Learning Approach of Vehicle Multitarget Detection from Traffic Video

Table 2

Statistics of different model results for synchronous flow sample training.

frame sample	YOLO 9000			YOLO-voc			YOLO v3			YOLO-vocRV
	free flow	225	221	0.9%	225	223	0.4%	225	-	-	225	223	0.4%
synchronous flow	547	500	5.2%	547	501	4.4%	547	-	-	547	524	2.1%
blocking flow	1050	940	5.5%	1050	925	6.3%	1050	-	-	1050	921	6.5%