3D Visual Question Answering on SQA3D
58.6EM@1Video-3D LLM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Video-3D LLMModality=RGB-D2025.11 | 58.6 | — | |
| CAPrunerBackbone=Llama-3-8B2026.04 | 56.3 | — | |
| LLaVA-3DModality=RGB-D2025.11 | 55.6 | — | |
| QuatRoPE2026.04 | 55.2 | — | |
| 3DGraphLLMSize=8B2026.04 | 55.2 | — | |
| Chat-Scene-7BSize=7B2026.04 | 54.6 | — | |
| Scene-LLM2026.04 | 54.2 | — | |
| CAPrunerBackbone=Llama-3.2-1B2026.04 | 52.8 | — | |
| 3DGraphLLMSize=1B2026.04 | 52.6 | — | |
| UniScene3DInput=CPM2026.04 | 52.5 | 92.3 | |
| POMA-3D_llmModality=PM, Grid inputs=H x W x 3, LoRA-tuned=true2025.11 | 51.6 | — | |
| POMA-3DInput=PM2026.04 | 51.1 | 91.2 | |
| POMA-3D_specModality=PM2025.11 | 51.1 | 91.2 | |
| LEOModality=C-PC2025.11 | 50 | — | |
| SceneVerseInput=PC2026.04 | 49.9 | 85 | |
| SceneVerseModality=C-PC2025.11 | 49.9 | 85 | |
| FG-CLIPInput=I2026.04 | 49.5 | 89.7 | |
| FG-CLIPModality=PM2025.11 | 49.5 | 89.7 | |
| 3D-VisTAInput=PC2026.04 | 48.5 | 85.6 | |
| 3D-ViSTAModality=C-PC2025.11 | 48.5 | 85.6 | |
| 3D-VisTA2026.04 | 48.5 | — | |
| SQA3D-LLaMALLM transformer=LLaMA (frozen)2023.10 | 48.09 | 89.03 | |
| Qwen2.5-VL-7BModality=RGB2025.11 | 47.8 | — | |
| LLaVA-OV-7BModality=RGB2025.11 | 47.7 | — | |
| SplatTalkModality=RGB(GS), Grid inputs=H x W x 3, LoRA-tuned=true2025.11 | 47.6 | — | |
| SQA3D2023.10 | 47.2 | 86.82 | |
| ScanQAModality=C-PC2025.11 | 47.2 | — | |
| PQ3D2026.04 | 47.1 | — | |
| SQA3DModality=C-PC2025.11 | 46.6 | — | |
| ScanQA2023.10 | 46.58 | 85.97 |