Robust Visual Tracking Using the Bidirectional Scale Estimation

Table 2

Per-video CLE (in pixel) on the 32 sequences. “” indicates the best performance, “” indicates the second best ones, and “” indicates the third best ones.

	BSET	DSST	KCF	Struck	SCM	TLD	MTT	LOT	DFT	CXT
biker		68.9	71.2		72.6	66.4	76.8	69.3	74.7
blurCar2			6.8	10.4	125.6		141.8	53.6	152.4	7.2
boy				3.8	51.0	4.5	12.8	66.0	106.3	7.4
car24			4.1	119.7		3.0	2.6	100.0	165.6	3.5
car4			9.9	8.7		12.8	22.3	167.3	61.9	58.1
carScale				36.4	33.4	22.6	87.6	101.2	75.8	24.5
couple	129.9	125.1	47.6		109.6			37.1	108.6	41.8
crossing			2.2	2.8		24.3	57.1	36.7	22.3	23.4
dog1				5.7	7.0			4.6	41.2	4.9
doll			8.4	8.9		6.0	110.3	6.3	59.5	4.7
dudek	13.4	13.3				18.1	14.1	85.1	18.7
fleetface		33.6				41.2	69.1	33.7	68.0	45.1
freeman1			94.9	14.3		39.7	117.8	86.9	10.4	26.8
freeman3		16.6	19.3	16.8		29.3	15.6	40.5	32.6
freeman4	57.4			48.7	37.7	39.2		38.6	57.5	65.6
girl		11.0	11.9			9.8		22.8	24.0	11.0
human5		302.5	175.5		9.3		8.3	90.3	259.1	200.9
human8			3.8	63.8		66.0	76.0	4.0	73.8	67.3
ironman		205.9	194.9	127.7	163.5		215.3		239.7	162.7
liquor	98.4	98.5		91.0	99.2		543.4		221.1	131.8
matrix	83.0		76.4	194.5			75.4	73.5	105.8	151.6
RedTeam			3.8	4.3	4.1	35.5		71.9	50.3	16.8
shaking			112.5	30.7		37.1	97.9	82.6	26.3	129.2
singer1	8.5		12.8	14.5			36.2	141.4	18.8	11.4
skating1				82.9	16.4	145.5	293.3	110.5	174.2	129.8
soccer				71.4	77.6	77.1	84.2	42.2	139.5	89.2
surfer		20.1	8.7	9.0	14.8		37.0	26.1	150.9
toy	8.4			11.4	50.8	8.9	12.8	53.0	31.9
trellis			7.8		7.0	31.1	68.8	47.7	44.9	7.0
walking			4.0	4.6		10.2	3.5		5.9	205.7
walking2			29.0	11.2		44.6	4.0	64.9	29.1	34.7
woman	10.6	9.7	10.1			139.9	137.3	117.1		72.5
Average CLE		34.4		33.9		35.3	77.6	62.0	83.1	55.6