HAM_INTP15: Historical Appearance Matching and Scene Adaptive Detection Filtering (VisBest)
TUD-Crossing
Benchmark:
Short name:
HAM_INTP15
Detector:
Public
Description:
https://github.com/yyc9268/Deep-TAMA
Reference:
Y. Yoon, A. Boragule, Y. Song, K. Yoon, M. Jeon. Online Multi-Object Tracking with Historical Appearance Matching and Scene Adaptive Detection Filtering. In IEEE AVSS, 2018.
Last submitted:
July 15, 2018 (1 year ago)
Published:
July 15, 2018 at 18:08:48 CET
Submissions:
1
Project page / code:
Open source:
Yes
Hardware:
3.7 GHZ, 1 Core, TITAN X
Runtime:
18.7 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| 2D MOT 2015 | 28.6 | 41.4 | 71.1 | 72 (10.0) | 317 (44.0) | 7,485 | 35,910 | 41.6 | 77.3 | 1.3 | 460 (11.1) | 1,038 (25.0) |
Detailed performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| ADL-Rundle-1 | 16.7 | 38.3 | 72.4 | 5 | 13 | 1,949 | 5,770 | 38.0 | 64.5 | 3.9 | 32 | 78 |
| ADL-Rundle-3 | 35.5 | 39.3 | 73.5 | 3 | 15 | 442 | 6,071 | 40.3 | 90.3 | 0.7 | 39 | 56 |
| AVG-TownCentre | 20.9 | 46.5 | 68.8 | 30 | 74 | 1,762 | 3,749 | 47.6 | 65.9 | 3.9 | 145 | 301 |
| ETH-Crossing | 29.0 | 46.3 | 74.0 | 2 | 17 | 28 | 681 | 32.1 | 92.0 | 0.1 | 3 | 3 |
| ETH-Jelmoli | 40.1 | 60.1 | 73.5 | 9 | 15 | 269 | 1,240 | 51.1 | 82.8 | 0.6 | 10 | 33 |
| ETH-Linthescher | 19.8 | 30.6 | 73.6 | 9 | 151 | 189 | 6,959 | 22.1 | 91.3 | 0.2 | 17 | 45 |
| KITTI-16 | 39.5 | 51.6 | 72.6 | 0 | 3 | 103 | 906 | 46.7 | 88.5 | 0.5 | 20 | 42 |
| KITTI-19 | 20.7 | 43.6 | 66.5 | 4 | 17 | 1,311 | 2,879 | 46.1 | 65.3 | 1.2 | 47 | 193 |
| PETS09-S2L2 | 46.0 | 40.9 | 69.4 | 5 | 3 | 704 | 4,385 | 54.5 | 88.2 | 1.6 | 121 | 229 |
| TUD-Crossing | 63.4 | 66.5 | 74.6 | 4 | 3 | 28 | 370 | 66.4 | 96.3 | 0.1 | 5 | 11 |
| Venice-1 | 20.6 | 36.8 | 72.3 | 1 | 6 | 700 | 2,900 | 36.4 | 70.4 | 1.6 | 21 | 47 |
Raw data:
