Reasoning Segmentation on ReasonSeg-X (test)
57.9gIoUStAR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| StARBackbone=Qwen 32026.05 | 57.9 | 50.1 | |
| StARBackbone=Qwen 3, Model Components=+ MV2026.05 | 50.6 | 53.1 | |
| StARBackbone=Qwen 2.5, Model Components=+ MV2026.05 | 50.3 | 44.9 | |
| StARBackbone=Qwen 2.52026.05 | 49.2 | 43.6 | |
| GRTO2026.05 | 48.2 | 46.3 | |
| B-GRTO2026.05 | 47.9 | 45.4 | |
| StARBackbone=Qwen 2.5, Training Stage=stage-12026.05 | 47.4 | 40.6 | |
| GRPO2026.05 | 47.3 | 44.2 | |
| GRPOTraining Stage=stage-12026.05 | 44.4 | 41.6 | |
| VisionReasoner2026.05 | 42.2 | 33.8 | |
| SAM 3 AgentBackbone=7B2026.05 | 34.4 | 29.5 | |
| LISA2026.05 | 25.1 | 26 |