Segmentation on MIMOSeg Perspective II held-in (test)
66.5mIoUMIMO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MIMOMulti-modal Input Aligner=true, LLM=Vicuna-7B2025.10 | 66.5 | 27.9 | 0.406 | |
| MIMO(w/o Aligner)Multi-modal Input Aligner=false, LLM=Vicuna-7B2025.10 | 62.2 | 18.5 | 0.314 | |
| SAM-hZero-shot=true, Backbone=ViT-H2025.10 | 57.1 | 13.2 | — | |
| SAM-bZero-shot=true, Backbone=ViT-B2025.10 | 56.8 | 12.5 | — | |
| GLaMMFine-tuned=true2025.10 | 49.6 | 13.7 | 0.279 | |
| SAM-MedZero-shot=true2025.10 | 49.4 | 9.7 | — |