TCM: Temporal Consistency Model for Head Detection in Complex Videos

Table 3

Performance evaluation on BOSS dataset.

Detectors	Baseline CNN	mAP	TCM (mAP)	Improvement (%age)
Faster R-CNN [15]	ZF [42]	0.74	0.79	6.75%
	VGG16 [41]	0.80	0.87	8.75%
	VGGM [41]	0.78	0.84	7.69%
R-CNN [12]	ZF [42]	0.72	0.79	9.72%
	VGG16 [41]	0.75	0.83	10.95%
	VGGM [41]	0.73	0.80	9.58%
YOLO [14]	13-layered architecture	0.72	0.82	13.88%
SSD [13]	VGG16 [41]	0.75	0.81	8.00%
			Average	7.54%