Multi-label Classification on NUS-WIDE 81 unseen labels (test)
0.603mAPMKT
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MKT2022.07 | 0.603 | 0.626 | 0.557 | 0.565 | 0.702 | 0.413 | 0.856 | |
| BiAM2022.07 | 0.478 | 0.596 | 0.534 | — | — | — | — | |
| GAN-MLZSL2022.07 | 0.467 | 0.593 | 0.531 | 0.535 | 0.665 | 0.394 | 0.816 | |
| DualCoOp2022.07 | 0.436 | 0.413 | 0.387 | — | — | — | — | |
| MKTPre-training=CLIP2022.07 | 0.425 | 0.341 | 0.315 | — | — | — | — | |
| ADDSInput resolution=3362022.07 | 0.39 | 0.37 | 0.393 | — | — | — | — | |
| MKTPre-training=IM-1K2022.07 | 0.376 | 0.341 | 0.311 | — | — | — | — | |
| ADDSInput resolution=2242022.07 | 0.366 | 0.342 | 0.366 | — | — | — | — | |
| LESA2022.07 | 0.315 | 0.58 | 0.52 | 0.523 | 0.651 | 0.386 | 0.8 | |
| CNN-RNN2022.07 | 0.283 | 0.552 | 0.508 | 0.499 | 0.617 | 0.377 | 0.781 | |
| FastTag2022.07 | 0.224 | 0.538 | 0.486 | 0.486 | 0.604 | 0.36 | 0.746 | |
| Logistic2022.07 | 0.216 | 0.511 | 0.461 | 0.461 | 0.573 | 0.342 | 0.708 | |
| nCLIPZero-shot=true, Backbone=ViT-B/16, Pre-trained=IT35M2022.10 | 0.161 | 0.356 | 0.166 | — | — | — | — | |
| xCLIPZero-shot=true, Backbone=ViT-B/16, Pre-trained=IT35M2022.10 | 0.153 | 0.352 | 0.168 | — | — | — | — | |
| CLIPZero-shot=true, Backbone=ViT-B/16, Pre-trained=IT35M2022.10 | 0.151 | 0.333 | 0.16 | — | — | — | — | |
| WARP2022.07 | 0.031 | 0.544 | 0.494 | 0.491 | 0.61 | 0.366 | 0.759 | |
| WSABIE2022.07 | 0.031 | 0.538 | 0.492 | 0.485 | 0.604 | 0.365 | 0.756 |