Vehicle Tracking Algorithm Based on Scale Search

Fig. 3. Tracking renderings of partial video sequences. (a) BlurCar1-OTB; (b) BlurCar2-OTB; (c) Car1-OTB; (d) Car7-UAV; (e) Car24-OTB; (f) CarDark-OTB

Download full size

Table 1. Overall performance of each algorithm
View table
Table 1. Overall performance of each algorithm
Algorithm CSK KCF CN CT DSST RPT SAMF Ours
Scale estimation N N N N Y Y Y Y
DP 0.636 0.865 0.088 0.398 0.890 0.859 0.888 0.968
AUC 0.477 0.580 0.006 0.231 0.603 0.656 0.746 0.778
FPS 240.69 155.70 99.00 49.69 23.50 2.88 10.15 15.65

Table 2. Performance comparison of fixed weight update and adaptive weight update
View table
Table 2. Performance comparison of fixed weight update and adaptive weight update
Impactfactor Update withfixed weight Update withadaptive weight
DP AUC DP AUC
SV 0.929 0.638 0.964 0.731
IV 0.899 0.659 0.972 0.752
BC 0.856 0.604 0.973 0.744
MB 0.879 0.721 0.965 0.783
LR 0.927 0.586 0.976 0.651

Table 3. Performance of each algorithm under different factors

View table

Table 3. Performance of each algorithm under different factors

Algorithm	SV		IV		BC		MB		LR
Algorithm	DP	AUC	DP	AUC	DP	AUC	DP	AUC	DP	AUC
Ours	0.964	0.731	0.972	0.752	0.973	0.744	0.965	0.783	0.976	0.651
SAMF	0.883	0.697	0.874	0.688	0.848	0.622	0.854	0.743	0.897	0.683
RPT	0.890	0.618	0.888	0.649	0.837	0.646	0.806	0.651	0.906	0.601
DSST	0.886	0.516	0.870	0.552	0.863	0.578	0.836	0.697	0.876	0.497
CT	0.370	0.234	0.352	0.228	0.265	0.184	0.286	0.248	0.424	0.231
CN	0.052	0.002	0.103	0.002	0.152	0.002	0.033	0.009	0.052	0.002
KCF	0.849	0.504	0.838	0.523	0.820	0.539	0.818	0.688	0.834	0.485
CSK	0.605	0.433	0.639	0.479	0.650	0.533	0.549	0.497	0.607	0.401