3D Object Detection on ScanNet (Fuzzy-F1)
18.93Fuzzy-F1 (IoU=0.5)Ours-8B
Evaluation Results
| Method | Links | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ours-8BMethod Category=Specialist MLLM, Depth source=GT depth2026.05 | 18.93 | — | — | — | — | — | — | — | — | 30.03 | 36.08 | 37.42 | 41.69 | 44.9 | 39.81 | 34.25 | 28.48 | 24.4 | 24.29 | |
| Ours-8BMethod Category=Specialist MLLM, Depth source=UniDepthV22026.05 | 16.97 | — | — | — | — | — | — | — | — | 30.85 | 33.33 | 35.26 | 42.38 | 44.13 | 39.7 | 31.88 | 27.16 | 21.65 | 20.76 | |
| VG LLM-8BMethod Category=Specialist MLLM2026.05 | 13.19 | — | — | — | — | — | — | — | — | 16.77 | 24.3 | 25.12 | 29.67 | 38.49 | 32.78 | 30.4 | 23.08 | 24.07 | 17.16 | |
| Qwen3-VL-8B-InstructMethod Category=Specialist MLLM2026.05 | 10.75 | — | — | — | — | — | — | — | — | 13.83 | 23.99 | 33.15 | 33.29 | 38.2 | 31.18 | 27.17 | 22.04 | 18.52 | 15.05 | |
| Qwen3.6-FlashMethod Category=Generalist MLLM2026.05 | 4.03 | — | — | — | — | — | — | — | — | 9.41 | 10.01 | 19.57 | 18.94 | 25.68 | 19.47 | 16.82 | 19.94 | 17.4 | 17.88 | |
| GPT-5.4-2026-03-05Method Category=Generalist MLLM2026.05 | 2.67 | — | — | — | — | — | — | — | — | 1.11 | 0.81 | 1.96 | 0.83 | 0.63 | 0.41 | 1.37 | 1.31 | 1.34 | 0.59 | |
| Seed-2.0-Pro-260215Method Category=Generalist MLLM2026.05 | 2 | — | — | — | — | — | — | — | — | 5.58 | 7.56 | 14.5 | 17.4 | 24 | 20.84 | 19.23 | 17.2 | 14.28 | 12.54 | |
| Claude-Sonnet-4.6Method Category=Generalist MLLM2026.05 | 0.87 | — | — | — | — | — | — | — | — | 1.5 | 1.35 | 0.79 | 0.53 | 0.94 | 1.64 | 0.78 | 2.44 | 2.7 | 1.84 | |
| Kimi-K2.6Method Category=Generalist MLLM2026.05 | 0.28 | — | — | — | — | — | — | — | — | 0 | 0.06 | 0.08 | 0.08 | 0 | 0.17 | 0.08 | 0.08 | 0 | 0 | |
| DeepSeek-v4-FlashMethod Category=Generalist MLLM2026.05 | 0 | — | — | — | — | — | — | — | — | 0.24 | 0 | 0.11 | 0 | 0.12 | 0 | 0.17 | 0 | 0.42 | 0 | |
| Claude-Opus-4.7Method Category=Generalist MLLM2026.05 | 0 | — | — | — | — | — | — | — | — | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
| OnlineSI2026.01 | — | 33.49 | 34.25 | 32.87 | 16.21 | 27.12 | 71.66 | 43.95 | 61.98 | — | — | — | — | — | — | — | — | — | — | |
| SpatialLM-Finetunefine-tuned=true2026.01 | — | 27.33 | 29 | 28.93 | 13.22 | 24.1 | 47.43 | 36.66 | 56.51 | — | — | — | — | — | — | — | — | — | — | |
| SpatialLM-Ground-Truthground_truth=true2026.01 | — | 61.38 | 64.97 | 61.23 | 30.59 | 60.63 | 90.46 | 71.13 | 94.42 | — | — | — | — | — | — | — | — | — | — | |
| SpatialLM-Ground-Truth-PCDground_truth=point cloud2026.01 | — | 51.82 | 60.5 | 52.54 | 14.64 | 45.11 | 74.82 | 68.55 | 88.03 | — | — | — | — | — | — | — | — | — | — | |
| SpatialLM-Mergemode=merged results2026.01 | — | 22.14 | 32.67 | 22.37 | 11.25 | 19.64 | 35.17 | 31.59 | 30.65 | — | — | — | — | — | — | — | — | — | — | |
| SpatialLM-No-Finetunefine-tuned=false2026.01 | — | 3.83 | 12.91 | 0 | 1.47 | 4.7 | 17.08 | 10.89 | 16.34 | — | — | — | — | — | — | — | — | — | — |