ITM: Individualized Tubelet Model
ADL-Rundle-1
Benchmark:
MOT20 | MOT17 | MOT16 | 2D MOT 2015 |
Short name:
ITM
Detector:
Public
Description:
Reference:
Pivot Correlation Network with Individualized Tubelets for Efficient Multi-object Tracking
Last submitted:
November 30, 2020 (1 month ago)
Published:
November 12, 2020 at 03:34:16 CET
Submissions:
3
Project page / code:
n/a
Open source:
No
Hardware:
2080Ti
Runtime:
1.2 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| 2D MOT 2015 | 50.1 | 49.8 | 76.1 | 286 (39.7) | 95 (13.2) | 7,638 | 21,311 | 65.3 | 84.0 | 1.3 | 1,730 (26.5) | 1,891 (29.0) |
Detailed performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| ADL-Rundle-1 | 38.8 | 50.7 | 74.2 | 11 | 4 | 2,103 | 3,533 | 62.0 | 73.3 | 4.2 | 60 | 162 |
| ADL-Rundle-3 | 44.2 | 45.8 | 81.9 | 12 | 4 | 1,330 | 4,258 | 58.1 | 81.6 | 2.1 | 89 | 123 |
| AVG-TownCentre | 56.6 | 51.2 | 70.8 | 135 | 11 | 1,020 | 1,273 | 82.2 | 85.2 | 2.3 | 806 | 399 |
| ETH-Crossing | 67.7 | 68.6 | 83.5 | 8 | 6 | 105 | 202 | 79.9 | 88.4 | 0.5 | 17 | 30 |
| ETH-Jelmoli | 58.5 | 66.2 | 80.3 | 16 | 13 | 286 | 749 | 70.5 | 86.2 | 0.7 | 19 | 45 |
| ETH-Linthescher | 59.1 | 60.2 | 79.8 | 74 | 35 | 1,294 | 2,098 | 76.5 | 84.1 | 1.1 | 260 | 249 |
| KITTI-16 | 52.8 | 62.3 | 70.7 | 3 | 1 | 83 | 701 | 58.8 | 92.3 | 0.4 | 19 | 56 |
| KITTI-19 | 47.4 | 56.3 | 71.8 | 9 | 16 | 450 | 2,284 | 57.3 | 87.2 | 0.4 | 74 | 159 |
| PETS09-S2L2 | 49.4 | 30.2 | 75.0 | 4 | 2 | 488 | 4,053 | 58.0 | 92.0 | 1.1 | 335 | 600 |
| TUD-Crossing | 77.2 | 50.5 | 72.7 | 9 | 0 | 29 | 195 | 82.3 | 96.9 | 0.1 | 27 | 33 |
| Venice-1 | 46.5 | 44.3 | 75.0 | 5 | 3 | 450 | 1,965 | 56.9 | 85.2 | 1.0 | 24 | 35 |
Raw data:
