Cross-Task Visual In-Context Learning on Second-tier group Shadow Removal to Reflection Removal
21.79PSNRGemini 2.5 Flash
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 2.5 FlashPrompting Strategy=Fixed2025.11 | 21.79 | 76.1 | 6.24 | |
| T2T-VICL (Gemini 2.5 Flash)Model=Gemini 2.5 Flash, Prompting Strategy=Ours2025.11 | 21.44 | 73.9 | 6.34 | |
| Seedream 4.0Prompting Strategy=Fixed2025.11 | 14.96 | 53.2 | 7.13 | |
| T2T-VICL (Seedream 4.0)Model=Seedream 4.0, Prompting Strategy=Ours2025.11 | 13.1 | 45.9 | 7.17 |