Compositional Zero-Shot Learning on MIT-States (test)
19.9Top-1 AccSymNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SymNetBackbone=ResNet-182020.04 | 19.9 | 28.2 | 33.8 | |
| GenModelBackbone=ResNet-182020.04 | 17.8 | — | — | |
| TAFE-NetBackbone=VGG-162020.04 | 16.4 | 26.4 | 33 | |
| LabelEmbed+Backbone=ResNet-182020.04 | 14.8 | — | — | |
| AttOperatorBackbone=ResNet-182020.04 | 14.2 | 19.6 | 25.1 | |
| Red WineBackbone=ResNet-182020.04 | 13.1 | 21.2 | 27.6 | |
| LabelEmbed (LE)Backbone=ResNet-182020.04 | 11.2 | 17.6 | 22.4 | |
| Visual ProductBackbone=ResNet-182020.04 | 9.8 | 16.1 | 20.6 | |
| LE + RBackbone=ResNet-182020.04 | 9.3 | 16.3 | 20.8 | |
| LEORBackbone=ResNet-182020.04 | 4.5 | 6.2 | 11.8 | |
| AnalogousAttrBackbone=ResNet-182020.04 | 1.4 | — | — |