Generalized Compositional Zero-Shot Learning on MIT-States (val)
4.3AUC (1)SymNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SymNetBackbone=ResNet-182020.04 | 4.3 | 9.8 | 14.8 | — | |
| TMNBackbone=ResNet-182020.04 | 3.5 | 8.1 | 12.4 | — | |
| GenModelBackbone=ResNet-182020.04 | 3.1 | 6.9 | 10.5 | — | |
| LabelEmbed+Backbone=ResNet-182020.04 | 3 | 7.6 | 12.2 | — | |
| Red WineBackbone=ResNet-182020.04 | 2.9 | 7.3 | 11.8 | — | |
| AttrOperatorBackbone=ResNet-182020.04 | 2.5 | 6.2 | 10.1 | — | |
| AttOpFeature Extraction Strategy=fixed2021.02 | — | — | — | 2.5 | |
| CGEFeature Extraction Strategy=end-to-end2021.02 | — | — | — | 8.6 | |
| CGEffFeature Extraction Strategy=fixed Resnet182021.02 | — | — | — | 6.8 | |
| LE+Feature Extraction Strategy=fixed2021.02 | — | — | — | 3 | |
| SymNetFeature Extraction Strategy=fixed2021.02 | — | — | — | 4.3 | |
| TMNFeature Extraction Strategy=fixed2021.02 | — | — | — | 3.5 |