List your company

Multi-object 3D Visual Grounding on Multi3DRefer

60.1F1@0.25Video-3D LLM†
7.16420.90734.6548.393Dec 13, 2023May 4, 2024Sep 24, 2024Feb 14, 2025Jul 7, 2025Nov 27, 2025Apr 20, 2026

Evaluation results

MethodLinks
Video-3D LLM†Video sequence length=32, 3D Encoder=Additional 3D encoder2026.0460.154.6
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓60%2026.045953.7
DVTIEPruning Ratio=35%2026.0458.6253.61
DVTIEPruning Ratio=65%2026.0458.4753.88
Fast3DPruning Ratio=65%2026.0458.4653.86
Fast3DPruning Ratio=35%2026.0458.3353.52
Baseline (Chat-Scene)Pruning Ratio=0%2026.0458.1453.24
Video-3D LLM (Uniform)3D Generalist=true, sampling strategy=uniform sampling2024.115852.7
Video-3D LLMVideo sequence length=322026.045852.7
Video-3D LLM (MC)3D Generalist=true, sampling strategy=maximum coverage2024.1157.952.4
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓80%2026.0457.552.5
FastVPruning Ratio=35%2026.0457.3952.69
Chat-SceneModel Category=LLM-based Models2023.1257.152.4
ChatScene3D Generalist=true2024.1157.152.4
Fast3DPruning Ratio=90%2026.0455.651.43
DVTIEPruning Ratio=90%2026.0455.5751.44
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓90%2026.0453.949.3
ToMePruning Ratio=35%2026.0451.9547.7
Grounded 3D-LLM3D Generalist=true2024.1145.240.6
Chat-3D v23D Generalist=true2024.1145.141.6
M3DRef-CLIPModel Category=Expert Models2023.1242.838.4
M3DRef-CLIP3D Generalist=false2024.1142.8—
M3DRef-CLIP2026.0442.8—
Random PruningPruning Ratio=35%2026.0440.8536.62
FastVPruning Ratio=65%2026.0433.6131.44
ToMePruning Ratio=65%2026.0430.9928.98
Random PruningPruning Ratio=65%2026.0419.9218.04
ToMePruning Ratio=90%2026.049.639.28
Random PruningPruning Ratio=90%2026.049.599.25
FastVPruning Ratio=90%2026.049.28.91
PQ3D3D Generalist=true2024.11—50.1