- 🐶 means that the model is initialized with imagenet pretrained weights.
- Results are presented in the format of Rank1 (mAP).
- Classification layer is ignored when computing the model size.
- Unless specified otherwise, the following data augmentation techniques are used: (1) Random2DTranslation, and (2) RandomHorizontalFlip.
- Click the highlighted results to download the model weights and the training scripts.
| Model | # param (M) | Loss | Input | market1501 | dukemtmcreid | msmt17 |
|---|---|---|---|---|---|---|
| resnet50🐶 | 23.5 | xent | (256, 128) | 87.9 (70.4) | 78.3 (58.9) | 63.2 (33.9) |
| resnet50_fc512🐶 | 24.6 | xent | (256, 128) | 90.8 (75.3) | 81.0 (64.0) | 69.6 (38.4) |
| densenet121_fc512🐶 | 7.5 | xent | (256, 128) | 87.8 (68.0) | 79.7 (58.8) | 67.6 (35.0) |
| se_resnet50_fc512🐶 | 27.1 | xent | (256, 128) | 91.9 (75.9) | 81.5 (63.7) | 71.1 (39.8) |
| squeezenet1_0_fc512🐶 | 1.0 | xent | (256, 128) | 79.3 (52.2) | 66.6 (42.6) | 44.1 (17.1) |
| resnet50mid🐶 | 27.7 | xent | (256, 128) | 90.2 (76.0) | 81.6 (64.0) | 69.0 (38.0) |
| mlfn🐶 | 32.5 | xent | (256, 128) | 90.1 (74.3) | 81.1 (63.2) | 66.4 (37.2) |
| hacnn | 3.7 | xent | (256, 128) | 90.9 (75.6) | 80.1 (63.2) | 64.7 (37.2) |