Multimodal Image-to-Image Translation (RGB+Edge to Depth) on Taskonomy
0.16MAETokenFusion (S)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TokenFusion (S)Architecture=Transformer-based, Scale=Small, Resolution=256x2562022.04 | 0.16 | 0.47 | |
| TokenFusion (Ti)Architecture=Transformer-based, Scale=Tiny, Resolution=256x2562022.04 | 0.21 | 0.57 | |
| CENArchitecture=CNN-based, Resolution=256x2562022.04 | 0.28 | 0.66 | |
| Self-Att.Architecture=CNN-based, Resolution=256x2562022.04 | 0.3 | 0.7 | |
| ConcatArchitecture=CNN-based, Resolution=256x2562022.04 | 0.33 | 0.75 | |
| Concat (S)Architecture=Transformer-based, Scale=Small, Resolution=256x2562022.04 | 0.35 | 0.68 | |
| Concat (Ti)Architecture=Transformer-based, Scale=Tiny, Resolution=256x2562022.04 | 0.4 | 0.83 | |
| Align.Architecture=CNN-based, Resolution=256x2562022.04 | 0.45 | 0.9 |