Object contour tracking via adaptive data-driven kernel

EURASIP Journal on Advances in Signal Processing

Table 1 Evaluation results of the compared methods on VOT2014 dataset: percentage of correctly tracked frames (score > 0.5)

	Sequence	Pix[28]	DF[27]	HT[32]	SLSM[34]	RPT[29]	Proposed
1	Ball	100	37.31	15.12	100	99.50	100
2	Basketball	41.79	4.00	9.10	37.43	96.55	35.59
3	Bicycle	1.49	98.52	63.43	96.27	83.39	89.3
4	Bolt	10.00	2.57	1.14	2.29	1.43	2.86
5	Car	65.87	39.68	64.68	65.48	100	85.32
6	David	91.69	89.22	72.34	78.57	100	84.94
7	Diving	35.16	27.40	0.46	100	15.98	100
8	Drunk	4.13	18.02	3.14	3.72	100	100
9	Fernando	33.56	62.67	2.05	16.10	65.41	59.93
10	Fish1	1.61	2.29	1.15	6.65	2.29	10.09
11	Fish2	24.19	23.24	5.81	18.06	10.65	9.35
12	Gymnastics	72.46	44.93	9.66	100	42.04	100
13	Hand1	17.43	95.90	100	20.75	21.31	24.59
14	Hand2	19.48	20.60	47.57	48.69	16.85	26.22
15	Jogging	2.28	21.50	80.78	22.15	22.48	100
16	Motocross	6.71	11.59	100	18.29	18.90	15.24
17	Polarbear	100	100	100	100	100	100
18	Skating	9.25	38.00	85.50	53.75	90.00	23.75
19	Sphere	100	9.95	100	100	100	100
20	Sunshade	10.06	50.58	100	68.60	100	100
21	Surfing	98.57	100	100	100	100	100
22	Torus	80.46	20.83	100	100	98.86	100
23	Trellis	87.70	53.08	72.93	39.72	100	36.9
24	Tunnel	1.78	58.55	39.67	25.03	57.73	49.11
25	Woman	17.59	94.47	18.43	88.78	93.80	94.97
	Average	41.3304	44.996	51.7184	56.4132	65.4868	65.9264