Feature Inversion on Pets
52.1SSIMAdjoint Inversion
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Adjoint InversionBackbone=ResNet-502026.04 | 52.1 | 54.3 | |
| UpConvNetBackbone=ResNet-502026.04 | 43.5 | 76.5 | |
| UpConvNetBackbone=ConvNeXt-Base2026.04 | 43.1 | 76 | |
| UpConvNetBackbone=DenseNet-1212026.04 | 41.9 | 77.9 | |
| UpConvNetBackbone=ResNet-182026.04 | 39.6 | 82.6 | |
| Adjoint InversionBackbone=ConvNeXt-Base2026.04 | 37.6 | 71.5 | |
| Adjoint InversionBackbone=ResNet-182026.04 | 35.4 | 72.7 | |
| Adjoint InversionBackbone=DenseNet-1212026.04 | 34.9 | 63.8 | |
| MahendranBackbone=ConvNeXt-Base2026.04 | 13.9 | 87.9 | |
| MahendranBackbone=ResNet-502026.04 | 13.2 | 113.3 | |
| RobustBackbone=ConvNeXt-Base2026.04 | 9.7 | 106 | |
| MahendranBackbone=DenseNet-1212026.04 | 9.6 | 112.2 | |
| MahendranBackbone=ResNet-182026.04 | 7.5 | 125 | |
| RobustBackbone=ResNet-502026.04 | 7.4 | 119.2 | |
| RobustBackbone=DenseNet-1212026.04 | 6.3 | 121.7 | |
| RobustBackbone=ResNet-182026.04 | 4.7 | 124.1 | |
| Vanilla GradientBackbone=ConvNeXt-Base2026.04 | 4.1 | 129.9 | |
| Vanilla GradientBackbone=ResNet-502026.04 | 3.5 | 124.9 | |
| Vanilla GradientBackbone=DenseNet-1212026.04 | 3 | 119.6 | |
| Vanilla GradientBackbone=ResNet-182026.04 | 1.9 | 134.3 |