Target Tracking Based on Adaptive Multilayer Convolutional Feature Decision Fusion

Faling Chen; Qinghai Ding; Haibo Luo; Bin Hui; Zheng Chang; Yunpeng Liu

doi:10.3788/AOS202040.2315002

Acta Optica Sinica, Volume. 40, Issue 23, 2315002(2020)

Target Tracking Based on Adaptive Multilayer Convolutional Feature Decision Fusion

Faling Chen^1,2,3,4,5, Qinghai Ding^1,6, Haibo Luo^1,2,4,5、*, Bin Hui^1,2,4,5, Zheng Chang^1,2,4,5, and Yunpeng Liu^1,2,4,5

Author Affiliations

¹Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang, Liaoning 110016, China

²Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences, Shenyang, Liaoning 110169, China

³University of Chinese Academy of Sciences, Beijing 100049, China

⁴Key Laboratory of Opto-Electronic Information Processing, Chinese Academy of Sciences, Shenyang, Liaoning 110016, China

⁵Liaoning Key Laboratory of Image Understanding and Computer Vision, Shenyang, Liaoning 110016, China

⁶Space Star Technology Co., Ltd., Beijing 100086, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(15)

Fig. 1. Visualization results of convolutional features in different layers on VGG-Net-19. (a) Input image; (b) feature in Conv1-2 layer; (c) feature in Conv2-2 layer; (d) feature in Conv3-4 layer; (e) feature in Conv4-4 layer; (f) feature in Conv5-4 layer

Download full size

Fig. 2. Procedure diagram of proposed algorithm

Download full size

Fig. 3.

Flow chart of proposed algorithm

Download full size

Fig. 4. Distance precision and success rate of proposed algorithm and other weak trackers. (a) Distance precision; (b) success rate

Download full size

Fig. 5. Distance precision and success rate of different algorithms under scale variation sequences. (a) Distance precision; (b) success rate

Download full size

Fig. 6. Distance precision and success rate of different algorithms under all sequences. (a) Distance precision; (b) success rate

Download full size

Fig. 7. Tracking results of different algorithms on Doll sequence

Download full size

Fig. 8. Tracking results of different algorithms on Carscale sequence

Download full size

Fig. 9. Tracking results of different algorithms on Freeman3 sequence

Download full size

Fig. 10. Tracking results of different algorithms on Skating1 sequence

Download full size

Fig. 11. Tracking results of different algorithms on Matrix sequence

Download full size

Fig. 12. Tracking results of different algorithms on Liquor sequence

Download full size

Table 1. DP scores of the top twelve algorithms under eleven challenging attributes

View table

Table 1. DP scores of the top twelve algorithms under eleven challenging attributes

Algorithm	IV	DEF	SV	OCC	MB	FM	IPR	OPR	OV	BC	LR
Proposed algorithm	0.868	0.876	0.906	0.908	$\bar{0.828}$	0.837	0.881	0.897	$\bar{0.766}$	0.886	0.847
HDT	$\bar{0.845}$	0.884	$\bar{0.866}$	$\bar{0.874}$	0.840	$\bar{0.782}$	$\bar{0.869}$	$\bar{0.871}$	0.679	$\bar{0.871}$	$\bar{0.846}$
FCNT	0.830	0.917	0.830	0.797	0.789	0.767	0.811	0.831	0.741	0.799	0.765
MEEM	0.778	0.859	0.808	0.814	0.740	0.757	0.809	0.853	0.730	0.808	0.494
DLSSVM	0.754	$\bar{0.892}$	0.761	0.811	0.749	0.731	0.807	0.840	0.697	0.818	0.553
SiamFC	0.709	0.744	0.796	0.802	0.700	0.721	0.744	0.788	0.777	0.732	0.659
TGPR	0.709	0.812	0.648	0.738	0.662	0.611	0.724	0.750	0.509	0.730	0.399
DSST	0.730	0.636	0.738	0.692	0.544	0.513	0.768	0.725	0.511	0.694	0.497
KCF	0.657	0.698	0.648	0.695	0.571	0.534	0.691	0.678	0.590	0.676	0.387
Struck	0.558	0.521	0.639	0.564	0.551	0.604	0.617	0.597	0.539	0.585	0.545
SCM	0.594	0.586	0.672	0.640	0.339	0.333	0.597	0.618	0.429	0.578	0.305
TLD	0.537	0.512	0.606	0.563	0.518	0.551	0.584	0.596	0.576	0.428	0.349

Table 2. OP scores of the top twelve algorithms under eleven challenging attributes

View table

Table 2. OP scores of the top twelve algorithms under eleven challenging attributes

Algorithm	IV	DEF	SV	OCC	MB	FM	IPR	OPR	OV	BC	LR
Proposed algorithm	0.758	$\bar{0.786}$	0.801	0.839	0.748	0.768	0.769	0.793	0.742	0.774	0.738
HDT	0.662	0.765	0.587	0.739	0.748	0.705	0.715	0.716	0.685	$\bar{0.769}$	0.636
FCNT	$\bar{0.755}$	0.813	0.669	0.706	0.724	0.718	0.688	0.724	$\bar{0.748}$	0.721	0.620
MEEM	0.667	0.719	0.599	0.694	0.721	$\bar{0.726}$	0.643	0.695	0.741	0.741	0.472
DLSSVM	0.648	0.778	0.567	0.723	$\bar{0.730}$	0.693	0.690	0.714	0.712	0.745	0.503
SiamFC	0.679	0.705	$\bar{0.771}$	$\bar{0.769}$	0.666	0.699	$\bar{0.725}$	$\bar{0.756}$	0.794	0.705	$\bar{0.651}$
TGPR	0.629	0.770	0.509	0.665	0.655	0.590	0.645	0.661	0.527	0.688	0.410
DSST	0.681	0.610	0.640	0.632	0.528	0.503	0.679	0.632	0.512	0.627	0.497
KCF	0.543	0.628	0.474	0.580	0.561	0.523	0.613	0.579	0.610	0.630	0.355
Struck	0.491	0.473	0.471	0.493	0.518	0.567	0.528	0.506	0.550	0.545	0.410
SCM	0.568	0.565	0.635	0.599	0.339	0.335	0.560	0.575	0.449	0.550	0.308
TLD	0.460	0.456	0.494	0.468	0.482	0.473	0.476	0.497	0.516	0.388	0.327

Table 3. Average tracking speed of different algorithms on all video sequences
View table
Table 3. Average tracking speed of different algorithms on all video sequences
Parameter Proposed algorithm HDT FCNT DSST MEEM TGPR
Code format M+C M+C M M+C M+C C
Average FPS 11.6 10 3 32.2 10 3

Tools

Get Citation

Copy Citation Text

Faling Chen, Qinghai Ding, Haibo Luo, Bin Hui, Zheng Chang, Yunpeng Liu. Target Tracking Based on Adaptive Multilayer Convolutional Feature Decision Fusion[J]. Acta Optica Sinica, 2020, 40(23): 2315002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Machine Vision

Received: Aug. 7, 2020

Accepted: Aug. 31, 2020

Published Online: Nov. 23, 2020

The Author Email: Haibo Luo (luohb@sia.cn)

DOI:10.3788/AOS202040.2315002

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. DP scores of the top twelve algorithms under eleven challenging attributes

Table 1. DP scores of the top twelve algorithms under eleven challenging attributes

Table 2. OP scores of the top twelve algorithms under eleven challenging attributes

Table 2. OP scores of the top twelve algorithms under eleven challenging attributes

Table 3. Average tracking speed of different algorithms on all video sequences

Table 3. Average tracking speed of different algorithms on all video sequences