Physical Pointing-Only Grounding on POG
95.6P@0.3Human Study
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Human Study2026.03 | 95.6 | 91.6 | 80.6 | |
| SV-CoTBackbone=Qwen3-vl2026.03 | 69.4 | 63.9 | 55.6 | |
| Qwen3-VL-4BParameters=4B2026.03 | 67.1 | 61.2 | 53.2 | |
| Qwen2.5-VL-7BParameters=7B2026.03 | 67 | 61.1 | 51.5 | |
| Qwen3-VL-8BParameters=8B2026.03 | 65.3 | 59.5 | 51.4 | |
| Qwen2.5-VL-3BParameters=3B2026.03 | 57.5 | 50.1 | 40.7 | |
| GroundingGPT2026.03 | 53.6 | 44.3 | 30.5 | |
| LLaVA-OneVision-4BParameters=4B2026.03 | 52.5 | 29.2 | 6.5 | |
| InternVL-3.5-8BParameters=8B2026.03 | 49.7 | 31.1 | 17.8 | |
| InternVL-3.5-4BParameters=4B2026.03 | 41.6 | 30.5 | 20.3 | |
| Ferret-7BParameters=7B2026.03 | 41.3 | 35.3 | 28.1 | |
| LLaVA-OneVision-8BParameters=8B2026.03 | 40.3 | 18.1 | 2.5 | |
| DeepSeek-VL2-SmallSize=Small2026.03 | 37.5 | 31.8 | 27.4 |