Reasoning Instance Segmentation on Inst2Seg
35.7Overall mAPSAM3-Agent_Qwen2.5-VL-7B
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SAM3-Agent_Qwen2.5-VL-7BModel Architecture Category=Multi-round Agent Pipeline, Fine-tuned=false2026.05 | 35.7 | 63.2 | 47.8 | 65.1 | 30.4 | 47.9 | 90.1 | |
| InstructSAM-2BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 31.5 | 60.4 | 52.6 | 66.8 | 22.2 | 44 | 74.3 | |
| SAM3-Agent_Qwen3-VL-2BModel Architecture Category=Multi-round Agent Pipeline, Fine-tuned=false2026.05 | 29.7 | 58.8 | 43.2 | 62.1 | 24.4 | 43.2 | 81.3 | |
| SAM3-Agent_Qwen2.5-VL-3BModel Architecture Category=Multi-round Agent Pipeline, Fine-tuned=false2026.05 | 23.2 | 48.7 | 33.9 | 47.6 | 18.8 | 33.1 | 86.2 | |
| X-SAM-3.8BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 11 | 36.6 | 33.4 | 51.9 | 2 | 30 | 0 | |
| SA2VA-8BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 9.4 | 53.9 | 35.5 | 65.2 | 1.4 | 46.3 | 32.4 | |
| SA2VA-4BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 8.2 | 52.6 | 30.1 | 59.2 | 1.1 | 40.2 | 56.9 | |
| PixelLM-7BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 4.6 | 27.2 | 13.9 | 36.7 | 0.8 | 22.9 | 5.1 | |
| LISA++-7BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 2.2 | 25.8 | 7.7 | 34.1 | 0.5 | 24.9 | 0.2 | |
| LISA-7BModel Architecture Category=End-to-end Model, Fine-tuned=false2026.05 | 1.9 | 28.6 | 8.2 | 37.7 | 0.2 | 27.7 | 0.2 |