Model ZooΒΆ
- Results are presented in the format of <Rank-1 (mAP)>.
- When computing FLOPs, only layers that are used at test time are considered (see
torchreid.utils.compute_model_complexity). - Unless specified otherwise, only
Random2DTranslationandRandomHorizontalFlipare used for data augmentation. - Asterisk (*) means the model is trained from scratch.
| Model | # Param (10^6) | GFLOPs | Loss | Input | market1501 | dukemtmcreid | msmt17 |
|---|---|---|---|---|---|---|---|
| resnet50 | 23.5 | 2.7 | softmax | (256, 128) | 87.9 (70.4) | 78.3 (58.9) | 63.2 (33.9) |
| resnet50_fc512 | 24.6 | 4.1 | softmax | (256, 128) | 90.8 (75.3) | 81.0 (64.0) | 69.6 (38.4) |
| mlfn | 32.5 | 2.8 | softmax | (256, 128) | 90.1 (74.3) | 81.1 (63.2) | 66.4 (37.2) |
| hacnn* | 4.5 | 0.5 | softmax | (160, 64) | 90.9 (75.6) | 80.1 (63.2) | 64.7 (37.2) |
| mobilenetv2_1dot0 | 2.2 | 0.2 | softmax | (256, 128) | 85.6 (67.3) | 74.2 (54.7) | 57.4 (29.3) |
| mobilenetv2_1dot4 | 4.3 | 0.4 | softmax | (256, 128) | 87.0 (68.5) | 76.2 (55.8) | 60.1 (31.5) |