ConNextFair: FairMOT with ConvNeXt-Tiny backbone
TUD-Crossing
Short name:
ConNextFair
Detector:
Private
Description:
We use FairMOT as baseline, and ConvNeXt-Tiny as the backbone network to get a new model.
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 International Journal of Computer Vision, 2021.
Last submitted:
April 20, 2022 (1 year ago)
Published:
April 20, 2022 at 17:19:44 CET
Submissions:
1
Project page / code:
n/a
Open source:
No
Hardware:
1.5 GHZ, 1 Core
Runtime:
3.9 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| MOT15 | 43.2 | 52.9 | 37.7 | 100 (13.9) | 216 (30.0) | 2,420 | 31,732 | 48.4 | 92.5 | 41.0 | 35.0 | 44.7 | 73.8 | 37.1 | 70.9 | 77.8 | 0.4 | 748 (0.0) | 2,221 (0.0) |
Detailed performance:
| Sequence | MOTA | IDF1 | HOTA | MT | ML | FP | FN | Rcll | Prcn | AssA | DetA | AssRe | AssPr | DetRe | DetPr | LocA | FAF | ID Sw. | Frag |
| ADL-Rundle-1 | 32.6 | 49.5 | 34.6 | 4 | 9 | 612 | 5,617 | 39.6 | 85.8 | 43.6 | 27.7 | 46.0 | 75.3 | 30.0 | 64.8 | 76.7 | 1.2 | 40 | 361 |
| ADL-Rundle-3 | 44.0 | 50.9 | 35.2 | 5 | 11 | 271 | 5,350 | 47.4 | 94.7 | 36.9 | 33.9 | 38.8 | 73.8 | 35.5 | 70.9 | 77.2 | 0.4 | 72 | 185 |
| AVG-TownCentre | 31.0 | 45.7 | 29.9 | 7 | 81 | 315 | 4,371 | 38.9 | 89.8 | 32.8 | 27.6 | 35.0 | 71.1 | 28.9 | 66.7 | 73.9 | 0.7 | 246 | 458 |
| ETH-Crossing | 60.2 | 70.9 | 48.9 | 3 | 8 | 29 | 363 | 63.8 | 95.7 | 50.5 | 47.5 | 54.0 | 76.9 | 50.4 | 75.5 | 80.4 | 0.1 | 7 | 34 |
| ETH-Jelmoli | 58.7 | 65.8 | 48.2 | 14 | 13 | 137 | 895 | 64.7 | 92.3 | 48.4 | 48.1 | 53.3 | 75.3 | 52.2 | 74.4 | 81.8 | 0.3 | 16 | 81 |
| ETH-Linthescher | 61.4 | 65.4 | 48.8 | 42 | 68 | 212 | 3,132 | 64.9 | 96.5 | 48.6 | 49.1 | 55.4 | 73.1 | 51.8 | 77.0 | 80.9 | 0.2 | 104 | 226 |
| KITTI-16 | 52.4 | 70.0 | 43.5 | 2 | 1 | 85 | 710 | 58.3 | 92.1 | 46.7 | 40.6 | 49.3 | 72.0 | 43.4 | 68.5 | 75.8 | 0.4 | 15 | 70 |
| KITTI-19 | 46.7 | 62.7 | 41.0 | 8 | 9 | 396 | 2,412 | 54.9 | 88.1 | 44.4 | 38.4 | 48.7 | 70.8 | 41.6 | 66.7 | 75.8 | 0.4 | 39 | 180 |
| PETS09-S2L2 | 32.5 | 30.6 | 21.2 | 1 | 10 | 150 | 6,173 | 36.0 | 95.9 | 17.4 | 26.1 | 18.2 | 72.6 | 27.0 | 71.8 | 76.8 | 0.3 | 189 | 475 |
| TUD-Crossing | 81.8 | 72.8 | 54.9 | 9 | 0 | 5 | 185 | 83.2 | 99.5 | 49.9 | 60.5 | 56.2 | 73.0 | 64.0 | 76.5 | 80.1 | 0.0 | 11 | 29 |
| Venice-1 | 39.9 | 52.8 | 39.2 | 5 | 6 | 208 | 2,524 | 44.7 | 90.7 | 47.0 | 32.8 | 51.0 | 73.8 | 34.8 | 70.7 | 78.3 | 0.5 | 9 | 122 |
Raw data:
