TCM: Temporal Consistency Model for Head Detection in Complex Videos

Table 4

Performance evaluation on PAMELA dataset.

Detectors	Baseline CNN	mAP	TCM (mAP)	Improvement (%age)
Faster R-CNN [15]	ZF [42]	0.63	0.67	6.34%
	VGG16 [41]	0.67	0.71	5.97%
	VGGM [41]	0.65	0.69	6.25%
R-CNN [12]	ZF [42]	0.52	0.55	5.76%
	VGG16 [41]	0.56	0.59	5.35%
	VGGM [41]	0.54	0.59	8.25%
YOLO [14]	13-layered architecture	0.59	0.63	6.77%
SSD [13]	VGG16 [41]	0.62	0.64	3.22%
			Average	4.79%