3D Visual Grounding on ScanRefer v1 (test)
58.89Acc@0.25IoUOurs-8B
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Ours-8BRescale factor=1.1, Depth source=GT depth, Category=Specialist MLLM2026.05 | 58.89 | — | — | — | — | |
| Ours-8BRescale factor=1.0, Depth source=GT depth, Category=Specialist MLLM2026.05 | 58.62 | — | — | — | — | |
| Video-3D LLM2026.05 | 58.1 | — | — | — | 51.7 | |
| Ours-8BRescale factor=1.1, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 57.82 | — | — | — | — | |
| Ours-8BRescale factor=0.9, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 57.03 | — | — | — | — | |
| Ours-8BRescale factor=0.9, Depth source=GT depth, Category=Specialist MLLM2026.05 | 57.03 | — | — | — | — | |
| Ours-8BRescale factor=1.0, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 55.17 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=1.0, Category=Specialist MLLM2026.05 | 53.85 | — | — | — | — | |
| GR3D-8BParameters=8B, Backbone=Qwen-2-7B, Input Resolution=4482026.05 | 52 | — | — | — | 46.1 | |
| Qwen3-VL-8B-InstructRescale factor=0.9, Category=Specialist MLLM2026.05 | 51.72 | — | — | — | — | |
| 3D-LLaVA2026.05 | 51.2 | — | — | — | 40.6 | |
| Ours-8BRescale factor=1.2, Depth source=GT depth, Category=Specialist MLLM2026.05 | 51.19 | — | — | — | — | |
| Ours-8BRescale factor=0.8, Depth source=GT depth, Category=Specialist MLLM2026.05 | 50.13 | — | — | — | — | |
| Ours-8BRescale factor=1.2, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 49.6 | — | — | — | — | |
| SPAR2026.05 | 48.8 | — | — | — | 43.1 | |
| Qwen3-VL-8B-InstructRescale factor=1.1, Category=Specialist MLLM2026.05 | 48.54 | — | — | — | — | |
| VG LLM-8BRescale factor=1.0, Category=Specialist MLLM2026.05 | 48.28 | — | — | — | — | |
| Ours-8BRescale factor=0.8, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 48.28 | — | — | — | — | |
| VG LLM-8BRescale factor=1.1, Category=Specialist MLLM2026.05 | 47.75 | — | — | — | — | |
| VG LLM-8BRescale factor=0.9, Category=Specialist MLLM2026.05 | 43.77 | — | — | — | — | |
| Ours-8BRescale factor=1.3, Depth source=GT depth, Category=Specialist MLLM2026.05 | 43.77 | — | — | — | — | |
| Ours-8BRescale factor=1.3, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 43.5 | — | — | — | — | |
| Ours-8BRescale factor=0.7, Depth source=GT depth, Category=Specialist MLLM2026.05 | 42.71 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=0.8, Category=Specialist MLLM2026.05 | 41.64 | — | — | — | — | |
| VG LLM-8BRescale factor=1.2, Category=Specialist MLLM2026.05 | 40.85 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.0, Category=Generalist MLLM2026.05 | 40.58 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=1.2, Category=Specialist MLLM2026.05 | 40.32 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.1, Category=Generalist MLLM2026.05 | 40.05 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.2, Category=Generalist MLLM2026.05 | 38.46 | — | — | — | — | |
| Ours-8BRescale factor=0.7, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 38.2 | — | — | — | — | |
| Qwen3-VL-8BParameters=8B2026.05 | 37.7 | — | — | — | 33.2 | |
| Seed-2.0-Pro-260215Rescale factor=0.9, Category=Generalist MLLM2026.05 | 36.34 | — | — | — | — | |
| Ours-8BRescale factor=1.4, Depth source=GT depth, Category=Specialist MLLM2026.05 | 36.07 | — | — | — | — | |
| VG LLM-8BRescale factor=1.3, Category=Specialist MLLM2026.05 | 35.28 | — | — | — | — | |
| Ours-8BRescale factor=1.4, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 35.28 | — | — | — | — | |
| VG LLM-8BRescale factor=0.8, Category=Specialist MLLM2026.05 | 34.75 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.0, Category=Generalist MLLM2026.05 | 33.42 | — | — | — | — | |
| Ours-8BRescale factor=1.5, Depth source=GT depth, Category=Specialist MLLM2026.05 | 32.89 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.1, Category=Generalist MLLM2026.05 | 32.63 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=1.3, Category=Specialist MLLM2026.05 | 32.36 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.3, Category=Generalist MLLM2026.05 | 31.3 | — | — | — | — | |
| VG LLM-8BRescale factor=1.4, Category=Specialist MLLM2026.05 | 30.77 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=0.8, Category=Generalist MLLM2026.05 | 28.65 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=0.7, Category=Specialist MLLM2026.05 | 28.38 | — | — | — | — | |
| Ours-8BRescale factor=1.5, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 28.12 | — | — | — | — | |
| VG LLM-8BRescale factor=0.7, Category=Specialist MLLM2026.05 | 27.06 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=0.9, Category=Generalist MLLM2026.05 | 26.79 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.2, Category=Generalist MLLM2026.05 | 25.46 | — | — | — | — | |
| Ours-8BRescale factor=0.6, Depth source=GT depth, Category=Specialist MLLM2026.05 | 25.46 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.4, Category=Generalist MLLM2026.05 | 25.2 | — | — | — | — | |
| VG LLM-8BRescale factor=1.5, Category=Specialist MLLM2026.05 | 24.67 | — | — | — | — | |
| Ours-8BRescale factor=0.6, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 24.14 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=1.4, Category=Specialist MLLM2026.05 | 23.61 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=1.5, Category=Generalist MLLM2026.05 | 22.28 | — | — | — | — | |
| VG LLM-8BRescale factor=0.6, Category=Specialist MLLM2026.05 | 22.28 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.3, Category=Generalist MLLM2026.05 | 20.95 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=1.5, Category=Specialist MLLM2026.05 | 20.69 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=0.8, Category=Generalist MLLM2026.05 | 19.89 | — | — | — | — | |
| VG LLM-8BRescale factor=0.5, Category=Specialist MLLM2026.05 | 17.51 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=0.7, Category=Generalist MLLM2026.05 | 17.24 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=0.6, Category=Specialist MLLM2026.05 | 16.98 | — | — | — | — | |
| Ours-8BRescale factor=0.5, Depth source=UniDepthV2, Category=Specialist MLLM2026.05 | 13.26 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.4, Category=Generalist MLLM2026.05 | 12.73 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=0.7, Category=Generalist MLLM2026.05 | 11.67 | — | — | — | — | |
| Ours-8BRescale factor=0.5, Depth source=GT depth, Category=Specialist MLLM2026.05 | 11.41 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=1.5, Category=Generalist MLLM2026.05 | 10.88 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=0.6, Category=Generalist MLLM2026.05 | 9.28 | — | — | — | — | |
| Qwen3-VL-8B-InstructRescale factor=0.5, Category=Specialist MLLM2026.05 | 8.49 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=0.6, Category=Generalist MLLM2026.05 | 7.96 | — | — | — | — | |
| Seed-2.0-Pro-260215Rescale factor=0.5, Category=Generalist MLLM2026.05 | 4.51 | — | — | — | — | |
| Qwen3.6-FlashRescale factor=0.5, Category=Generalist MLLM2026.05 | 3.71 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.1, Category=Generalist MLLM2026.05 | 0.8 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.3, Category=Generalist MLLM2026.05 | 0.8 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=0.5, Category=Generalist MLLM2026.05 | 0.53 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=0.9, Category=Generalist MLLM2026.05 | 0.53 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.0, Category=Generalist MLLM2026.05 | 0.53 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=0.6, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=0.7, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.2, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.4, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=1.5, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| Claude-Sonnet-4.6Rescale factor=0.5, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| Claude-Sonnet-4.6Rescale factor=1.1, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| Claude-Sonnet-4.6Rescale factor=1.5, Category=Generalist MLLM2026.05 | 0.27 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=0.5, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=0.6, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=0.7, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=0.8, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=0.9, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.0, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.1, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.2, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.3, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.4, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| GPT-5.4-2026-03-05Rescale factor=1.5, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| DeepSeek-v4-FlashRescale factor=0.8, Category=Generalist MLLM2026.05 | 0 | — | — | — | — | |
| 3DJCGExtra captioning training data=true2024.10 | — | 60.6 | 31.2 | 37.8 | — | |
| 3DVG-Trans2021.12 | — | 55.15 | 29.33 | 35.12 | — | |
| 3DVG-Trans2024.10 | — | 57.9 | 31 | 37 | — | |
| 3DVG-Trans+2021.12 | — | 57.87 | 31.02 | 37.04 | — |