Reference-based Super-Resolution on DRefSR
30.06PSNR (Clean)SSMTF-rec
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SSMTF-recTraining Loss=Reconstruction-only2026.01 | 30.06 | 0.838 | 20.24 | 0.6736 | 9.82 | 0.1644 | 36.55 | 0.9127 | |
| DATSR-recTraining Loss=Reconstruction-only2026.01 | 29.95 | 0.8347 | 19.3 | 0.618 | 10.65 | 0.2167 | 36.35 | 0.9097 | |
| SSMTFTraining Loss=Full2026.01 | 29.54 | 0.8221 | 20.13 | 0.616 | 9.41 | 0.2061 | 36.11 | 0.9066 | |
| MASA-recTraining Loss=Reconstruction-only2026.01 | 29.47 | 0.8213 | 26.68 | 0.7839 | 2.79 | 0.0374 | 37.39 | 0.927 | |
| DATSRTraining Loss=Full2026.01 | 29.37 | 0.8161 | 17.36 | 0.476 | 12.01 | 0.3401 | 36.41 | 0.9093 | |
| TTSR-recTraining Loss=Reconstruction-only2026.01 | 29.28 | 0.8175 | 25 | 0.7207 | 4.28 | 0.0968 | 36.24 | 0.907 | |
| TTSRTraining Loss=Full2026.01 | 28.06 | 0.7886 | 22.94 | 0.5299 | 5.12 | 0.2587 | 35.78 | 0.8982 | |
| MASATraining Loss=Full2026.01 | 27.03 | 0.75 | 20.81 | 0.605 | 6.22 | 0.145 | 37.31 | 0.9237 |