Fair: A simple baseline for multi-object tracking
TUD-Crossing
Benchmark:
MOT16 | MOT20 | MOT17 | 2D MOT 2015 |
Short name:
Fair
Detector:
Private
Description:
n/a
Reference:
Y. Zhang, C. Wang, X. Wang, W. Zeng, W. Liu. FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking. In arXiv preprint arXiv:2004.01888, 2020.
Last submitted:
August 02, 2020 (5 months ago)
Published:
May 16, 2020 at 05:52:27 CET
Submissions:
3
Project page / code:
Open source:
Yes
Hardware:
RTX2080ti
Runtime:
30.5 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| 2D MOT 2015 | 60.6 | 64.7 | 76.5 | 343 (47.6) | 79 (11.0) | 7,854 | 15,785 | 74.3 | 85.3 | 1.4 | 591 (8.0) | 1,731 (23.3) |
Detailed performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| ADL-Rundle-1 | 41.3 | 56.5 | 74.1 | 11 | 3 | 2,271 | 3,131 | 66.4 | 73.1 | 4.5 | 57 | 221 |
| ADL-Rundle-3 | 53.7 | 56.4 | 80.0 | 18 | 3 | 1,152 | 3,483 | 65.7 | 85.3 | 1.8 | 71 | 172 |
| AVG-TownCentre | 58.9 | 73.0 | 70.6 | 110 | 22 | 1,163 | 1,680 | 76.5 | 82.5 | 2.6 | 98 | 312 |
| ETH-Crossing | 76.7 | 76.0 | 81.8 | 10 | 2 | 31 | 197 | 80.4 | 96.3 | 0.1 | 6 | 23 |
| ETH-Jelmoli | 68.7 | 77.5 | 81.4 | 34 | 1 | 456 | 325 | 87.2 | 82.9 | 1.0 | 12 | 59 |
| ETH-Linthescher | 70.7 | 71.8 | 81.1 | 74 | 39 | 525 | 1,986 | 77.8 | 93.0 | 0.4 | 109 | 196 |
| KITTI-16 | 54.7 | 68.4 | 74.2 | 7 | 1 | 314 | 432 | 74.6 | 80.2 | 1.5 | 25 | 56 |
| KITTI-19 | 55.9 | 70.2 | 72.4 | 31 | 2 | 1,062 | 1,238 | 76.8 | 79.4 | 1.0 | 58 | 202 |
| PETS09-S2L2 | 77.1 | 58.0 | 75.4 | 27 | 1 | 500 | 1,582 | 83.6 | 94.2 | 1.1 | 121 | 392 |
| TUD-Crossing | 88.8 | 82.5 | 76.9 | 13 | 0 | 50 | 64 | 94.2 | 95.4 | 0.2 | 9 | 21 |
| Venice-1 | 55.7 | 64.0 | 78.7 | 8 | 5 | 330 | 1,667 | 63.5 | 89.8 | 0.7 | 25 | 77 |
Raw data:
