Human-Object Interaction on VG-HOI Annotated-box
21.65mAPAnyHOI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AnyHOIBackbone=LLaVA OV 7B2026.04 | 21.65 | 64.91 | |
| AnyHOIBackbone=InternVL2 8B2026.04 | 19.83 | 62.37 | |
| AnyHOIBackbone=Qwen2-VL 7B2026.04 | 19.61 | 63.76 | |
| AnyHOIBackbone=Idefics2 8B2026.04 | 19.6 | 59.65 | |
| AnyHOIBackbone=CogVLM2 19B2026.04 | 18.2 | 60.25 | |
| AnyHOIBackbone=InternVL2 4B2026.04 | 17.6 | 58.73 | |
| AnyHOIBackbone=Phi3V 4B2026.04 | 17.37 | 54.35 | |
| AnyHOIBackbone=InstructBLIP 7B2026.04 | 15.71 | 56.37 | |
| AnyHOIBackbone=InternVL2 2B2026.04 | 15.34 | 50.5 | |
| AnyHOIBackbone=InstructBLIP 13B2026.04 | 15.2 | 57.14 | |
| AnyHOIBackbone=Qwen2-VL 2B2026.04 | 14.16 | 47.64 | |
| AnyHOIBackbone=LLaVA OV 0.5B2026.04 | 12.76 | 42.7 | |
| Idefics2 8B2026.04 | 12.54 | 39.96 | |
| LLaVA OV 7B2026.04 | 10.54 | 33.53 | |
| InternVL2 8B2026.04 | 10.36 | 39.14 | |
| CLIPLabel Source=VG2026.04 | 9.8 | 55.83 | |
| Qwen2-VL 7B2026.04 | 9.46 | 38.32 | |
| DHDLabel Source=WN2026.04 | 8.86 | 50.36 | |
| DHDLabel Source=VG2026.04 | 8.73 | 52.85 | |
| InternVL2 4B2026.04 | 7.87 | 33.22 | |
| InternVL2 2B2026.04 | 6.29 | 25.3 | |
| Qwen2-VL 2B2026.04 | 4.56 | 17.68 | |
| CLIPLabel Source=WN2026.04 | 2.95 | 39.75 | |
| InstructBLIP 7B2026.04 | 2.37 | 9.03 | |
| CogVLM2 19B2026.04 | 2.28 | 7.53 | |
| InstructBLIP 13B2026.04 | 1.87 | 8.54 | |
| Phi3V 4B2026.04 | 1.78 | 5.57 | |
| LLaVA OV 0.5B2026.04 | 1.21 | 4.41 | |
| Random2026.04 | 0.3 | 35.89 |