Benchmark:
MOT16 |
Short name:
LP2D_16
Detector:
Public
Description:
The tracker builds a graphical model where nodes are the 2D detections. The solution is then found by computing the minimum cost-flow problem.
Reference:
MOT baseline: Linear programming on 2D image coordinates.
Last submitted:
February 28, 2017 (7 years ago)
Published:
February 28, 2017 at 12:16:56 CET
Submissions:
1
Project page / code:
Open source:
Yes
Hardware:
3 GHz, 1 core
Runtime:
49.3 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 | 35.7 | 34.2 | 27.4 | 66 (8.7) | 385 (50.7) | 5,084 | 111,163 | 39.0 | 93.3 | 26.0 | 29.0 | 27.4 | 73.2 | 30.4 | 72.7 | 79.1 | 0.9 | 915 (23.4) | 1,264 (32.4) |
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 | 1.9 | 30.3 | 24.3 | 2 | 39 | 676 | 5,584 | 12.7 | 54.5 | 32.0 | 18.5 | 34.8 | 69.0 | 19.0 | 71.4 | 77.4 | 1.5 | 12 | 66 |
MOT16-03 | 7.1 | 34.8 | 27.8 | 25 | 79 | 27,412 | 69,492 | 33.5 | 56.1 | 23.6 | 32.9 | 24.4 | 76.7 | 34.5 | 73.6 | 79.2 | 18.3 | 263 | 707 |
MOT16-06 | 38.9 | 46.4 | 35.2 | 21 | 132 | 318 | 6,659 | 42.3 | 93.9 | 37.9 | 33.0 | 42.6 | 66.1 | 35.0 | 70.1 | 77.8 | 0.3 | 73 | 143 |
MOT16-07 | 23.0 | 32.2 | 25.1 | 5 | 33 | 1,806 | 10,695 | 34.5 | 75.7 | 24.8 | 25.6 | 27.2 | 59.7 | 26.8 | 70.0 | 77.5 | 3.6 | 67 | 242 |
MOT16-08 | 7.0 | 29.3 | 24.6 | 6 | 51 | 2,374 | 13,113 | 21.7 | 60.4 | 25.8 | 23.6 | 27.6 | 71.7 | 24.7 | 73.1 | 81.3 | 3.8 | 82 | 143 |
MOT16-12 | 17.5 | 49.5 | 38.2 | 9 | 68 | 1,021 | 5,719 | 31.1 | 71.6 | 46.6 | 31.4 | 49.4 | 73.5 | 33.1 | 73.0 | 80.2 | 1.1 | 100 | 65 |
MOT16-14 | 4.1 | 20.9 | 16.5 | 2 | 270 | 1,685 | 15,994 | 13.5 | 59.6 | 18.5 | 14.7 | 20.2 | 60.5 | 15.2 | 68.9 | 77.7 | 2.2 | 43 | 119 |
Raw data: