GNNMatch: Graph Convolutional Neural Networks for Multi-Object Tracking via Sinkhorn Normalization
MOT16-01
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 02, 2021 (1 year ago)
Published:
July 30, 2020 at 21:33:18 CET
Submissions:
3
Project page / code:
Open source:
Yes
Hardware:
TITAN RTX
Runtime:
0.3 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 | 57.2 | 55.0 | 44.6 | 174 (22.9) | 258 (34.0) | 3,905 | 73,493 | 59.7 | 96.5 | 43.7 | 45.8 | 51.0 | 68.1 | 48.2 | 78.0 | 81.7 | 0.7 | 559 (9.4) | 847 (14.2) |
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 | 44.0 | 41.7 | 36.0 | 6 | 11 | 90 | 3,474 | 45.7 | 97.0 | 37.4 | 34.8 | 44.4 | 64.6 | 36.2 | 76.9 | 80.7 | 0.2 | 19 | 28 |
| MOT16-03 | 68.4 | 63.7 | 50.6 | 61 | 19 | 2,090 | 30,872 | 70.5 | 97.2 | 47.9 | 53.7 | 53.4 | 72.9 | 56.7 | 78.3 | 81.6 | 1.4 | 117 | 230 |
| MOT16-06 | 54.7 | 33.4 | 32.0 | 60 | 80 | 552 | 4,564 | 60.4 | 92.7 | 22.2 | 46.3 | 55.5 | 29.6 | 49.7 | 76.2 | 82.4 | 0.5 | 107 | 130 |
| MOT16-07 | 46.7 | 46.7 | 37.1 | 9 | 14 | 258 | 8,363 | 48.8 | 96.9 | 37.3 | 37.3 | 42.3 | 61.5 | 39.0 | 77.4 | 81.3 | 0.5 | 71 | 140 |
| MOT16-08 | 35.3 | 37.7 | 32.9 | 9 | 24 | 241 | 10,540 | 37.0 | 96.3 | 35.7 | 30.5 | 40.9 | 70.5 | 31.5 | 81.8 | 84.5 | 0.4 | 50 | 67 |
| MOT16-12 | 48.5 | 55.6 | 45.7 | 17 | 39 | 120 | 4,134 | 50.2 | 97.2 | 52.1 | 40.3 | 58.7 | 72.2 | 42.0 | 81.4 | 84.0 | 0.1 | 20 | 30 |
| MOT16-14 | 33.6 | 38.8 | 29.6 | 12 | 71 | 554 | 11,546 | 37.5 | 92.6 | 31.4 | 28.1 | 36.9 | 59.1 | 29.5 | 72.9 | 79.7 | 0.7 | 175 | 222 |
Raw data:
