GNNMatch: Graph Convolutional Neural Networks for Multi-Object Tracking via Sinkhorn Normalization
ETH-Linthescher
Short name:
GNNMatch
Detector:
Public
Description:
Reference:
I. Papakis, A. Sarkar, A. Karpatne. GCNNMatch: Graph Convolutional Neural Networks for Multi-Object Tracking via Sinkhorn Normalization. In arXiv preprint arXiv:2010.00067, 2020.
Last submitted:
March 07, 2021 (3 years ago)
Published:
March 02, 2021 at 06:22:09 CET
Submissions:
4
Project page / code:
Open source:
Yes
Hardware:
TITAN RTX
Runtime:
0.1 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 | 46.7 | 43.2 | 35.4 | 157 (21.8) | 203 (28.2) | 6,643 | 25,311 | 58.8 | 84.5 | 31.4 | 40.8 | 44.2 | 49.3 | 46.3 | 66.6 | 79.6 | 1.1 | 820 (0.0) | 1,371 (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 | 36.2 | 50.2 | 39.0 | 11 | 3 | 2,518 | 3,368 | 63.8 | 70.2 | 39.6 | 38.7 | 45.4 | 63.4 | 49.8 | 54.8 | 77.6 | 5.0 | 48 | 120 |
| ADL-Rundle-3 | 48.7 | 47.7 | 39.7 | 12 | 8 | 646 | 4,518 | 55.6 | 89.7 | 36.9 | 43.2 | 44.5 | 64.4 | 47.0 | 75.9 | 84.0 | 1.0 | 54 | 61 |
| AVG-TownCentre | 38.9 | 34.3 | 28.3 | 45 | 44 | 1,121 | 2,968 | 58.5 | 78.9 | 22.1 | 37.2 | 40.6 | 31.5 | 43.8 | 59.0 | 76.0 | 2.5 | 281 | 495 |
| ETH-Crossing | 46.4 | 46.9 | 39.0 | 5 | 10 | 74 | 458 | 54.3 | 88.0 | 35.6 | 42.8 | 51.5 | 52.2 | 46.8 | 75.9 | 84.6 | 0.3 | 6 | 11 |
| ETH-Jelmoli | 59.6 | 47.8 | 42.0 | 17 | 12 | 328 | 683 | 73.1 | 85.0 | 34.6 | 51.0 | 66.4 | 39.2 | 59.7 | 69.4 | 82.5 | 0.7 | 14 | 37 |
| ETH-Linthescher | 49.6 | 40.3 | 37.1 | 36 | 98 | 382 | 4,081 | 54.3 | 92.7 | 32.9 | 42.0 | 56.8 | 41.7 | 44.8 | 76.5 | 82.5 | 0.3 | 42 | 88 |
| KITTI-16 | 50.4 | 58.9 | 37.4 | 3 | 1 | 162 | 658 | 61.3 | 86.6 | 35.3 | 39.6 | 45.1 | 51.0 | 44.2 | 62.4 | 74.4 | 0.8 | 24 | 44 |
| KITTI-19 | 47.4 | 47.6 | 34.0 | 9 | 19 | 389 | 2,369 | 55.7 | 88.4 | 30.8 | 37.9 | 44.5 | 44.2 | 41.6 | 66.0 | 76.4 | 0.4 | 51 | 142 |
| PETS09-S2L2 | 48.1 | 31.5 | 24.0 | 5 | 4 | 500 | 4,243 | 56.0 | 91.5 | 14.5 | 40.1 | 19.6 | 33.6 | 43.2 | 70.7 | 78.4 | 1.1 | 263 | 333 |
| TUD-Crossing | 78.5 | 44.8 | 35.9 | 8 | 0 | 30 | 186 | 83.1 | 96.8 | 22.5 | 57.7 | 38.6 | 32.2 | 61.9 | 72.2 | 77.6 | 0.1 | 21 | 18 |
| Venice-1 | 49.9 | 46.1 | 36.9 | 6 | 4 | 493 | 1,779 | 61.0 | 85.0 | 33.8 | 40.6 | 41.6 | 56.6 | 46.5 | 64.7 | 78.9 | 1.1 | 16 | 22 |
Raw data:
