CSTrack: Rethinking the competition between detection and ReID in Multi-Object Tracking
MOT16-01
Short name:
CSTrack
Detector:
Private
Description:
CSTrack proposes a strong ReID based one-shot MOT framework. It includes a novel cross-correlation network that can effectively impel the separate branches to learn task-dependent representations, and a scale-aware attention network that learns discriminative embeddings to improve the ReID capability. This work also provides an analysis of the weak data association ability in one-shot MOT methods. Our improvements make the data association ability of our one-shot model is comparable to two-stage methods while running more faster.
Reference:
C. Liang, Z. Zhang, Y. Lu, X. Zhou, B. Li. Rethinking the competition between detection and ReID in Multi-Object Tracking. In arXiv:2010.12138 [cs], 2020.
Last submitted:
November 18, 2020 (3 years ago)
Published:
October 21, 2020 at 13:07:50 CET
Submissions:
2
Project page / code:
Open source:
Yes
Hardware:
RTX2080ti
Runtime:
15.8 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| MOT16 | 75.6 | 73.3 | 59.8 | 325 (42.8) | 125 (16.5) | 9,646 | 33,777 | 81.5 | 93.9 | 58.2 | 61.8 | 63.5 | 76.6 | 67.4 | 77.7 | 83.2 | 1.6 | 1,121 (13.8) | 2,450 (30.1) |
Detailed performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| MOT16-01 | 57.0 | 62.9 | 50.8 | 8 | 7 | 112 | 2,604 | 59.3 | 97.1 | 54.1 | 47.7 | 58.4 | 81.0 | 49.8 | 81.6 | 84.2 | 0.2 | 35 | 101 |
| MOT16-03 | 89.2 | 82.0 | 67.6 | 134 | 0 | 4,511 | 6,561 | 93.7 | 95.6 | 63.8 | 72.0 | 68.9 | 79.3 | 78.0 | 79.5 | 83.6 | 3.0 | 263 | 632 |
| MOT16-06 | 61.0 | 64.2 | 51.2 | 77 | 57 | 514 | 3,825 | 66.8 | 93.8 | 51.5 | 51.2 | 59.4 | 72.5 | 55.1 | 77.3 | 83.1 | 0.4 | 164 | 286 |
| MOT16-07 | 65.6 | 62.1 | 49.1 | 23 | 3 | 1,011 | 4,463 | 72.7 | 92.1 | 44.5 | 54.3 | 49.8 | 70.1 | 59.7 | 75.8 | 82.9 | 2.0 | 144 | 337 |
| MOT16-08 | 52.8 | 48.4 | 42.5 | 25 | 2 | 2,324 | 5,288 | 68.4 | 83.1 | 38.6 | 48.1 | 45.9 | 60.6 | 56.3 | 68.5 | 82.1 | 3.7 | 295 | 456 |
| MOT16-12 | 58.7 | 68.1 | 54.9 | 26 | 18 | 635 | 2,742 | 66.9 | 89.7 | 58.6 | 51.6 | 64.1 | 78.5 | 57.1 | 76.5 | 84.9 | 0.7 | 49 | 177 |
| MOT16-14 | 51.3 | 61.2 | 42.9 | 32 | 38 | 539 | 8,294 | 55.1 | 95.0 | 46.2 | 40.3 | 49.8 | 74.1 | 42.7 | 73.6 | 79.5 | 0.7 | 171 | 461 |
Raw data:
