Reasoning Segmentation on In-house OR video dataset 1.0 (test)
75.8cIoU (Semantic)ORDIRS-Agent
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ORDIRS-AgentLLM Backbone=GPT-4o2025.03 | 75.8 | 78.74 | 77.25 | 76.85 | 82.48 | 79.55 | 89.05 | |
| LISA-7B-SAM1LLM Backbone Size=7B, SAM Version=SAM1, Fine-tuned=true2025.03 | 69.68 | 69 | 68.3 | 64.62 | 65.38 | 64.56 | 1.52 | |
| LISA-13B-SAM1LLM Backbone Size=13B, SAM Version=SAM1, Fine-tuned=true2025.03 | 53.32 | 64.84 | 59.38 | 54.33 | 65.25 | 58.19 | 1.85 | |
| LISA-7B-SAM2LLM Backbone Size=7B, SAM Version=SAM2, Fine-tuned=true2025.03 | 51.14 | 51.42 | 52.81 | 43.84 | 46.99 | 47.05 | 1.42 | |
| LISA-13B-SAM2LLM Backbone Size=13B, SAM Version=SAM2, Fine-tuned=true2025.03 | 49.06 | 49.45 | 51.21 | 39.22 | 42 | 42.69 | 2.09 | |
| LISA-13B-SAM1LLM Backbone Size=13B, SAM Version=SAM1, Fine-tuned=false2025.03 | 28.15 | 65.35 | 42.08 | 34.19 | 68.85 | 50.75 | 2.03 | |
| LISA-7B-SAM1LLM Backbone Size=7B, SAM Version=SAM1, Fine-tuned=false2025.03 | 25.17 | 59.93 | 39.85 | 34.94 | 63.77 | 48.97 | 1.51 | |
| V*2025.03 | 3.09 | 3.35 | 3.26 | 2.93 | 3.27 | 3.15 | 22.62 |