DP_NMS: Dynamic Programming
TUD-Crossing
Benchmark:
2D MOT 2015 | MOT20 |
Short name:
DP_NMS
Detector:
Public
Description:
A dynamic programming (DP) approximation to the successive shortest path (SSP) solution of a network flow framework.
Parameters:
c_en=18.2882
c_ex=16.8287
c_ij=0.42473
betta=0.24366
max_it=Inf
thr_cost=9.0096
Reference:
H. Pirsiavash, D. Ramanan, C. Fowlkes. Globally-Optimal Greedy Algorithms for Tracking a Variable Number of Objects. In CVPR, 2011.
Last submitted:
September 23, 2017 (3 years ago)
Published:
November 01, 2014 at 03:09:33 CET
Submissions:
3
Project page / code:
Open source:
Yes
Hardware:
2.6 GHz, 16 Cores
Runtime:
444.8 Hz
Benchmark performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| 2D MOT 2015 | 14.5 | 19.7 | 70.8 | 43 (6.0) | 294 (40.8) | 13,171 | 34,814 | 43.3 | 66.9 | 2.3 | 4,537 (104.7) | 3,090 (71.3) |
Detailed performance:
| Sequence | MOTA | IDF1 | MOTP | MT | ML | FP | FN | Recall | Precision | FAF | ID Sw. | Frag |
| ADL-Rundle-1 | -2.2 | 20.3 | 69.9 | 6 | 8 | 4,213 | 5,058 | 45.6 | 50.2 | 8.4 | 241 | 336 |
| ADL-Rundle-3 | 23.7 | 27.0 | 72.0 | 3 | 9 | 2,193 | 5,407 | 46.8 | 68.5 | 3.5 | 158 | 189 |
| AVG-TownCentre | 13.4 | 15.0 | 68.8 | 8 | 93 | 1,206 | 4,682 | 34.5 | 67.2 | 2.7 | 299 | 414 |
| ETH-Crossing | 21.1 | 22.3 | 75.5 | 0 | 15 | 27 | 757 | 24.5 | 90.1 | 0.1 | 7 | 16 |
| ETH-Jelmoli | 34.8 | 31.0 | 73.3 | 8 | 13 | 314 | 1,280 | 49.5 | 80.0 | 0.7 | 59 | 86 |
| ETH-Linthescher | 12.4 | 15.5 | 74.7 | 3 | 157 | 164 | 7,609 | 14.8 | 89.0 | 0.1 | 49 | 102 |
| KITTI-16 | 26.3 | 19.4 | 68.5 | 0 | 2 | 290 | 896 | 47.3 | 73.5 | 1.4 | 67 | 116 |
| KITTI-19 | 17.7 | 18.2 | 68.3 | 4 | 16 | 1,388 | 2,818 | 47.3 | 64.5 | 1.3 | 191 | 242 |
| PETS09-S2L2 | 38.3 | 15.7 | 71.6 | 4 | 6 | 1,016 | 4,611 | 52.2 | 83.2 | 2.3 | 320 | 417 |
| TUD-Crossing | 57.2 | 33.6 | 71.7 | 4 | 2 | 81 | 348 | 68.4 | 90.3 | 0.4 | 43 | 49 |
| Venice-1 | 12.7 | 12.6 | 71.7 | 0 | 8 | 686 | 3,240 | 29.0 | 65.9 | 1.5 | 56 | 114 |
Raw data:
