Robust Visual Tracking Using the Bidirectional Scale Estimation

Table 1

Per-video overlap precision (OP) (%) 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		26.8	25.4	25.4		31.0	31.0	43.0	25.4
blurCar2			94.7	93.8	14.4		13.3	28.7	17.4
boy					43.9	93.5	48.7	65.0	48.3	49.7
car24		17.3	17.3	17.0					7.2
car4			36.4	39.8			31.1	4.9	25.8	29.9
carScale			44.4	43.3	65.1	43.7	56.7	46.4	44.8
couple	10.7	10.7	24.3	54.3	10.7				8.6
crossing						51.7	22.5	30.8	64.2	34.2
dog1			65.1	65.3	84.7	67.3	78.5		52.1
doll			55.2	68.8		62.4	50.4	87.1	35.0
dudek			97.6		97.6	84.2	92.9	61.8	80.1	92.4
fleetface		66.5		66.6		56.7	54.6	57.9	55.6	64.6
freeman1		35.3	16.3	21.5		21.2	16.0	18.7	17.8
freeman3		31.3	27.8	20.0		58.0	47.8	7.4	33.0
freeman4	16.6		18.4	15.5			18.0	13.4	18.0	18.0
girl	87.4	30.6	74.2			76.4		54.6	25.2	64.2
Human5		24.3	23.6	34.1	41.5		35.1		7.6	28.1
Human8				13.3		13.3	11.7		13.3	10.9
ironman				4.8		6.6	8.4	8.4	3.6	3.0
liquor	40.9	40.9		40.6	32.1		20.1		22.9	21.0
matrix			13.0	12.0		7.0		4.0	6.0	4.0
RedTeam			37.6	39.8	39.1	28.3		10.1	27.0	45.3
shaking			1.4	16.7		40.0	1.1	7.7		10.7
singer1	27.6		27.6	29.9				24.2	27.6	32.2
skating1			36.3	37.0		22.8	13.5	28.0	16.3	12.0
soccer	14.5			15.6		12.2	18.1	21.7	21.9	12.8
surfer		29.0	39.9	15.7	40.7		5.6	37.2	3.7
toy			43.2	49.1	22.1	73.4	53.1	21.4	46.9
trellis			84.0	78.4		47.3	19.3	31.6	51.8	80.8
walking			51.5	56.6		38.3			55.1	21.8
walking2			38.0			34.0		39.0	38.2	39.8
woman	88.3				85.8	16.6	19.8	8.4		20.6
Average OP			47.8	46.8		51.2	41.6	40.0	33.6	49.5