Spatial VQA on 3DSRBench Level 1-3
64.34AccuracyHiSpatial-3B-RGB
Evaluation Results
| Method | Links | |
|---|---|---|
| HiSpatial-3B-RGBInput=RGB2026.03 | 64.34 | |
| HiSpatial-3BInput=RGB-XYZ2026.03 | 63.81 | |
| InternVL3.5-14BInput=RGB2026.03 | 57.32 | |
| Qwen-3-VL-30B-A3BInput=RGB2026.03 | 55.7 | |
| SpatialBot-3BInput=RGB-D2026.03 | 54.67 | |
| Qwen-3-VL-8BInput=RGB2026.03 | 52.8 | |
| InternVL3.5-8BInput=RGB2026.03 | 51.58 | |
| RoboRefer-8B-SFTInput=RGB-D2026.03 | 51.24 | |
| Gemini-2.5-ProInput=RGB2026.03 | 48.47 | |
| SpatialRGPT-8BInput=RGB-D2026.03 | 48.4 | |
| Claude-3.7-SonnetInput=RGB2026.03 | 48.2 | |
| SpaceLLaVA-13BInput=RGB2026.03 | 46.55 | |
| GPT-4oInput=RGB2026.03 | 44.2 | |
| PaliGemma2-3BInput=RGB2026.03 | 8.15 |