Interactive Image Segmentation on ThinObject-5K
77.95IoULitePGE-SAM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LitePGE-SAMFine-tuning strategy=Full model2026.06 | 77.95 | 85.5 | — | |
| LitePGE-SAMFine-tuning strategy=Feature Enhancer2026.06 | 75.66 | 83.82 | — | |
| LiteMedSAMFine-tuning strategy=Full Mask Decoder2026.06 | 72.11 | 81.23 | — | |
| LiteMedSAMFine-tuning strategy=Output Token2026.06 | 59 | 70.69 | — | |
| LiteMedSAM2026.06 | 25.38 | 34.75 | — | |
| LitePGE-SAMFine-tuning strategy=Feature Enhancer, Segmentation Loss=False2026.06 | 18.96 | 27.55 | — |