Laser & Optoelectronics Progress, Volume. 56, Issue 12, 121501(2019)
Multiple Object Tracking Based on Kernelized Correlation Filter
Figures & Tables(9)
Fig. 2. Object similarity judgment. (a) 4th frame; (b) 21st frame; (c) template-to-template change
Fig. 5. Effects of different parameters. (a) MOTA; (b) MOTP; (c) FAF; (d) MT; (e) ML; (f) FP; (g) FN; (h) IDS; (i) FM
Table 1. Selectable parameter range of algorithm
View tableTable 1. Selectable parameter range of algorithm
Parameter K lostK sameK scoreRange 2,4,6,8 10,20,30,40 0.6,0.7,0.8,0.9
Table 2. Performance of proposed approach on 2DMOT2015 sequences
View tableTable 2. Performance of proposed approach on 2DMOT2015 sequences
Tracker MOTA MOTP FAF MT /% ML /% FP FN IDS FM Hz DP-NMS 14.5 70.8 2.3 6.0 40.8 13171 34814 4537 3090 444.8 LDCT 4.7 71.7 2.4 11.4 32.5 14066 32156 12348 2918 20.7 TC_ODAL 15.1 70.5 2.2 3.2 55.9 12970 38538 637 1716 1.7 CEM 19.3 70.7 2.5 8.5 46.5 14180 34591 813 1023 1.1 Proposed 29.2 71.1 0.9 10.0 34.5 5369 36590 1513 2859 30.0
Table 3. Performance of proposed approach on MOT16 sequences
View tableTable 3. Performance of proposed approach on MOT16 sequences
Tracker MOTA MOTP FAF MT /% ML /% FP FN IDS FM Hz DP-NMS 26.2 76.3 0.6 4.1 67.5 3689 130557 365 638 5.9 CEM 33.2 75.8 1.2 7.8 54.4 6837 114322 642 731 0.3 EAGS16 47.4 75.9 1.4 17.3 42.7 8369 86931 575 913 197.3 SMOT 29.7 75.2 2.9 5.3 47.7 17426 107552 3108 4483 0.2 MAP-RF 64.8 79.7 - 37.8 22.3 7780 57081 645 - 30.0 Proposed 55.8 76.5 1.2 22.8 33.3 7046 71601 1856 3114 15.0
Tools
Get Citation
Copy Citation Text
Huan Liu, Chungeng Li, Jubai An, Guo Wei, Junli Ren. Multiple Object Tracking Based on Kernelized Correlation Filter[J]. Laser & Optoelectronics Progress, 2019, 56(12): 121501
Paper Information
Category: Machine Vision
Received: Dec. 4, 2018
Accepted: Jan. 14, 2019
Published Online: Jun. 13, 2019
The Author Email: Chungeng Li (li_chungeng@dlmu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20