TCM: Temporal Consistency Model for Head Detection in Complex Videos

Table 2

Performance evaluation on Casablanca dataset.

Detectors	Baseline CNN	mAP	TCM (mAP)	Improvement (%age)
Faster R-CNN [15]	ZF [42]	0.48	0.54	12.50%
	VGG16 [41]	0.55	0.61	10.90%
	VGGM [41]	0.51	0.55	7.84%
R-CNN [12]	ZF [42]	0.43	0.45	4.65%
	VGG16 [41]	0.53	0.57	7.54%
	VGGM [41]	0.48	0.53	10.41%
YOLO [14]	13-layered architecture	0.31	0.40	20.30%
SSD [13]	VGG16 [41]	0.38	0.42	10.52%
			Average	8.46%