BLSTM_MTP_O: Online Greedy Data Association using Bilinear LSTM with Multi-track Pooling
MOT16-12
Short name:
BLSTM_MTP_O
Detector:
Public
Description:
Reference:
C. Kim, F. Li, M. Alotaibi, J. Rehg. Discriminative Appearance Modeling with Multi-track Pooling for Real-time Multi-object Tracking. In CVPR (accepted), 2021.
Last submitted:
March 17, 2021 (3 years ago)
Published:
March 17, 2021 at 03:16:37 CET
Submissions:
1
Project page / code:
n/a
Open source:
No
Hardware:
Titan Xp GPU, 3.6GHZ CPU
Runtime:
21.0 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 | 48.3 | 53.5 | 39.7 | 129 (17.0) | 294 (38.7) | 9,792 | 83,707 | 54.1 | 91.0 | 40.4 | 39.3 | 43.2 | 73.2 | 42.0 | 70.6 | 77.7 | 1.7 | 735 (0.0) | 2,349 (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 |
| MOT16-01 | 33.0 | 42.5 | 33.0 | 5 | 8 | 619 | 3,639 | 43.1 | 81.7 | 36.7 | 29.8 | 40.1 | 68.5 | 33.1 | 62.7 | 76.7 | 1.4 | 25 | 84 |
| MOT16-03 | 56.6 | 57.9 | 43.1 | 39 | 21 | 5,585 | 39,478 | 62.2 | 92.1 | 41.3 | 45.3 | 43.9 | 73.5 | 48.4 | 71.6 | 77.6 | 3.7 | 328 | 950 |
| MOT16-06 | 48.0 | 56.0 | 41.7 | 40 | 106 | 348 | 5,573 | 51.7 | 94.5 | 45.8 | 38.1 | 54.8 | 67.2 | 40.1 | 73.2 | 78.5 | 0.3 | 73 | 273 |
| MOT16-07 | 44.7 | 48.9 | 35.2 | 6 | 16 | 484 | 8,451 | 48.2 | 94.2 | 35.4 | 35.2 | 37.2 | 75.1 | 36.9 | 72.2 | 77.7 | 1.0 | 88 | 348 |
| MOT16-08 | 34.0 | 40.6 | 31.3 | 11 | 20 | 830 | 10,116 | 39.6 | 88.9 | 33.5 | 29.4 | 35.5 | 74.3 | 31.3 | 70.3 | 79.0 | 1.3 | 100 | 186 |
| MOT16-12 | 39.1 | 55.4 | 40.3 | 18 | 36 | 900 | 4,112 | 50.4 | 82.3 | 47.1 | 34.7 | 50.0 | 73.9 | 39.0 | 63.6 | 77.6 | 1.0 | 43 | 167 |
| MOT16-14 | 27.3 | 42.0 | 29.2 | 10 | 87 | 1,026 | 12,338 | 33.2 | 85.7 | 35.6 | 24.0 | 37.5 | 72.0 | 25.5 | 65.8 | 76.9 | 1.4 | 78 | 341 |
Raw data:
