video_reid_pytorch
Add reid to your PYTHONPATH, and you can run examples in examples folder.
Result
Here are some results on MARS and iLIDS datasets.
MARS
| Method | rank1 | rank5 | rank10 | mAP |
|---|---|---|---|---|
| cross entropy | 76.11 | 89.55 | 92.63 | 60.11 |
| cross entropy + batch hard triplet | 80.35 | 91.26 | 93.54 | 67.45 |
| cross entropy + adaptive weight triplet | 80.00 | 92.37 | 94.44 | 69.02 |
| quality aware network | 82.17 | 93.03 | 95.30 | 72.24 |
iLIDS
| Method | rank1 | rank5 | rank10 |
|---|---|---|---|
| cross entropy | 45.82 | 74.77 | 83.89 |
| cross entropy + adaptive weight triplet | 59.22 | 86.78 | 92.50 |
| quality aware network | 72.11 | 93.28 | 97.17 |