Unsupervised Conditional Image Generation on Oxford102-Flowers Coarse-Grained
14.63FIDOurs (Repar.)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ours (Repar.)Backbone=DDPM2026.06 | 14.63 | 3.58 | 0.3045 | 0.5147 | |
| Ours (Noise)Backbone=DDPM2026.06 | 16.17 | 3.81 | 0.3045 | 0.5147 | |
| SG-DM+DINOBackbone=DDPM2026.06 | 26.18 | 3.57 | 0.2294 | 0.4177 | |
| SG-DM+MSNBackbone=DDPM2026.06 | 28.85 | 3.58 | 0.2001 | 0.3412 | |
| SG-DM+SimCLRBackbone=DDPM2026.06 | 33.86 | 3.44 | 0.0761 | 0.1282 | |
| ClusterGANBackbone=BigGAN2026.06 | 39.81 | 3.5 | 0.0902 | 0.3606 | |
| Self-Cond GANBackbone=BigGAN2026.06 | 45.88 | 2.71 | 0.1409 | 0.3228 | |
| MIC-GANsBackbone=BigGAN2026.06 | 65.84 | 2.34 | 0.0564 | 0.2587 | |
| IC-GANBackbone=BigGAN2026.06 | 72.48 | 2.82 | — | — |