Cloud Removal on Multi-temporal Real-world Cloud-degraded Scenes Sichuan
22.562PSNRPhyVLM-CR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PhyVLM-CRProtocol=T1 input, T2 ground truth, T3 auxiliary reference2026.03 | 22.562 | 79.37 | |
| Vision-Language ModelProtocol=T1 input, T2 ground truth, T3 auxiliary reference2026.03 | 20.89 | 57.06 | |
| Traditional PhysicalProtocol=T1 input, T2 ground truth, T3 auxiliary reference2026.03 | 17.583 | 54.92 | |
| Zero-shot Deep LearningProtocol=T1 input, T2 ground truth, T3 auxiliary reference2026.03 | 12.45 | 26.79 |