Reasoning Segmentation on LISA
67.24AccuracyRECAP
Evaluation results
| Method | Links | |
|---|---|---|
| RECAPBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 67.24 | |
| PropMixBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 66.8 | |
| UniformBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 65.18 | |
| Qwen2.5-VL-7Bvariant=Base model2025.10 | 65.13 | |
| LwFBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 65.08 | |
| CoresetBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 64.82 | |
| VLAA-Thinker-7B2025.10 | 63.14 | |
| Reasoning-onlyBackbone=Qwen2.5-VL-7B, Training=finetuned2025.10 | 57.58 | |
| Vision-R1-7B2025.10 | 47.3 | |
| OpenVLThinker-7B2025.10 | 42.73 |