List your company

3D Referring Expression Segmentation on ScanRefer

62Accuracy @ 0.25Video-3D LLM†
19.56830.58441.652.616Mar 6, 2026Mar 13, 2026Mar 21, 2026Mar 28, 2026Apr 5, 2026Apr 12, 2026Apr 20, 2026

Evaluation results

MethodLinks
Video-3D LLM†Video sequence length=32, 3D Encoder=Additional 3D encoder2026.046255.1—
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓60%2026.0461.354.6—
HCF-RES2026.0360.955.750.5
PC-CrossDiffVenue=/, FPS=4.07, batchsize=1, GPU=NVIDIA 30902026.0360.4152.5246.39
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓80%2026.0460.153.5—
IPDNVenue=AAAI20252026.0359.954.449.5
MCLNVenue=ECCV20242026.0358.750.744.7
MCLNVenue=ECCV’24, FPS=4.81, batchsize=1, GPU=NVIDIA 30902026.0358.750.744.72
Video-3D LLMVideo sequence length=322026.0458.151.7—
MDINVenue=ACMMM20242026.035853.148.3
Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓90%2026.045851.8—
Reason3DVenue=3DV20252026.0357.941.942
VG LLMVideo sequence length=322026.0457.650.9—
RefMask3DVenue=ACMMM20242026.0355.949.244.9
RefMask3DVenue=ACMMM’24, FPS=1.65, batchsize=1, GPU=NVIDIA 30902026.0355.8749.2444.86
3D-STMNVenue=AAAI20242026.0354.639.839.5
3D-STMNVenue=AAAI’24, FPS=4.14, batchsize=1, GPU=NVIDIA 30902026.0354.639.839.5
LLaVA-3D2026.0454.142.4—
M3DRef-CLIP2026.0451.944.7—
X-RefSeg3DVenue=AAAI’24, FPS=2.03, batchsize=1, GPU=NVIDIA 30902026.0340.3333.7729.94
InstanceReferVenue=ICCV20212026.0340.233.530.6
TGNNVenue=AAAI’21, FPS=2.85, batchsize=1, GPU=NVIDIA 30902026.0337.531.427.8
ScanRefer2026.0437.324.3—
3D-LLM (BLIP2-FlanT5)2026.0430.3——
3D-LLM (Flamingo)2026.0421.2——
SegPointVenue=ECCV20242026.03——41.7
SegPointVenue=ECCV’24, batchsize=1, GPU=NVIDIA 30902026.03——41.7