Multiple Object Grounding on Reasoning Segmentation Short query
47.1gIoULISA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LISATraining paradigm=Train LLaVA 1.5-7B, Query source=Original2025.05 | 47.1 | 48.5 | |
| TextRegionQuery source=Interpreted from LLaVA 1.5-7B2025.05 | 28.5 | 30.6 | |
| OVSegQuery source=Interpreted from LLaVA 1.5-7B2025.05 | 24.2 | 18.7 | |
| TridentQuery source=Interpreted from LLaVA 1.5-7B2025.05 | 23 | 24.2 | |
| TextRegionTraining paradigm=Training-free, VLM usage=No, Query source=Original2025.05 | 21.6 | 15.6 | |
| X-DecoderTraining paradigm=Training-based, VLM usage=No, Query source=Original2025.05 | 20.4 | 11.6 | |
| SEEMTraining paradigm=Training-based, VLM usage=No, Query source=Original2025.05 | 20.1 | 11.5 | |
| TridentTraining paradigm=Training-free, VLM usage=No, Query source=Original2025.05 | 19.9 | 12.9 | |
| OVSegTraining paradigm=Training-based, VLM usage=No, Query source=Original2025.05 | 18 | 15.5 | |
| Grounded-SAMTraining paradigm=Training-based, VLM usage=No, Query source=Original2025.05 | 17.8 | 10.8 |