3D Instance Segmentation on ScanNet V2 (val)
96.6Average AP50Proposed Interactive 3D Segmentation Method
Evaluation results
| Method | Links | |||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Proposed Interactive 3D Segmentation Methodclicks per object=202022.04 | 96.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.3 | 99.3 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=102022.04 | 94.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.5 | 99 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=102022.04 | 90.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.3 | 97.9 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=202022.04 | 88.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.6 | 97.4 | |
| SegVGGTevaluation_setting=class-agnostic, sampled_scenes=10, down_sampling_rate=40x2026.03 | 87.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 62.9 | 97.4 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=52022.04 | 86 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 61.8 | 95.7 | |
| Spherical MaskMethodology=coarse-to-fine2023.12 | 79.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 62.3 | 88.2 | |
| Spherical MaskArchitecture=Non-transformer-based2025.02 | 79.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 62.3 | 88.2 | |
| BFLArchitecture=Transformer-based2025.02 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 61.7 | 86.5 | |
| SGIFormerVenue=TCSVT 25, Method Category=Query-based2026.02 | 78.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.9 | 86.2 | |
| LaSSMDecoder Type=Hybrid Model, Method Category=Query-based2026.02 | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | 86.1 | |
| MSTA3D2024.11 | 77 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | 85.4 | |
| LaSSMDecoder Type=Query Refinement Decoder, Method Category=Query-based2026.02 | 76.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.6 | 85.2 | |
| Mask-Attention-Free TransformerSurface normal=true2023.09 | 76.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 59.9 | — | |
| OneFormer3DVenue=CVPR 24, Method Category=Query-based, Top Instances during inference=2002026.02 | 76.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.5 | 84 | |
| Mask-Attention-Free Transformer2023.09 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | — | |
| MaftArchitecture=Transformer-based2025.02 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | 84.5 | |
| MAFTVenue=ICCV23, Methodology=Transformer2023.12 | 75.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | 84.5 | |
| MAFT2024.11 | 75.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | 84.5 | |
| HAISclass agnostic=true2022.04 | 75.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.1 | 84.9 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=52022.04 | 75.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.7 | 91.1 | |
| 3IS-ESSSVenue=ICCV23, Methodology=Transformer2023.12 | 75 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.1 | 83.7 | |
| 3IS-ESSSVenue=ICCV 23, Method Category=Query-based2026.02 | 75 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.1 | 83.7 | |
| QueryFormerVenue=ICCV23, Methodology=Transformer2023.12 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.5 | 83.3 | |
| QueryFormer2024.11 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.5 | 83.3 | |
| QueryFormerArchitecture=Transformer-based2025.02 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.5 | 83.3 | |
| SPFormer2022.11 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.3 | 82.9 | |
| SPFormer2023.09 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.3 | — | |
| SPFormer2024.11 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.3 | 82.9 | |
| SPFormerArchitecture=Transformer-based2025.02 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.3 | 82.9 | |
| SPFormerVenue=AAAI 23, Method Category=Query-based2026.02 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.3 | 82.9 | |
| Mask3DRuntime (in ms)=3392022.10 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | — | |
| Mask3DBackbone=Res16UNet34C2023.09 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | — | |
| Mask3DVenue=ICRA23, Methodology=Transformer2023.12 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | 82.9 | |
| Mask3D2024.11 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | 83.5 | |
| Mask3DArchitecture=Transformer-based2025.02 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | 82.9 | |
| Mask3DVenue=ICRA 23, Method Category=Query-based2026.02 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | 82.9 | |
| Proposed Interactive 3D Segmentation Methodclicks per object=32022.04 | 73.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 48.1 | 90.7 | |
| ISBNet2023.03 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.5 | 82.5 | |
| ISBNetVenue=CVPR23, Methodology=kernel2023.12 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.5 | 82.5 | |
| ISBNet2024.11 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.5 | 82.5 | |
| ISBNetArchitecture=Non-transformer-based2025.02 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.5 | 82.5 | |
| ISBNetVenue=CVPR 23, Method Category=Cluster-based2026.02 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.5 | 82.5 | |
| TD3D + Point-GCCSupervision regime=Self-supervised Pre-training, Fine-tuning with pre-training=true2023.05 | 71.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.3 | 81.6 | |
| TD3DVenue=WACV23, Methodology=coarse-to-fine2023.12 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.3 | 81.9 | |
| TD3D2024.11 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.3 | 81.9 | |
| TD3DVenue=WACV 24, Method Category=Proposal-based2026.02 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.3 | 81.9 | |
| TD3DSupervision regime=Self-supervised Pre-training2023.05 | 71.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46.2 | 81.3 | |
| PBNetVenue=ICCV23, Methodology=grouping2023.12 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | 78.9 | |
| PBNet2024.11 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | 78.9 | |
| PBNetVenue=ICCV 23, Method Category=Cluster-based2026.02 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | 78.9 | |
| PBNetInfer. Time(ms)=420.82022.07 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | 78.9 | |
| PBNet2022.07 | 70.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | 78.9 | |
| SoftGroup2022.03 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 78.9 | |
| SoftGroup2022.11 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | 78.9 | |
| SoftGroup2023.03 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | 78.9 | |
| SoftGroupRuntime (in ms)=3452022.10 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | — | |
| SoftGroupSupervision regime=Supervised Only, Extra training dataset (ScanNetV2)=true2023.05 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.8 | 78.9 | |
| SoftGroup2023.09 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | — | |
| SoftGroupVenue=CVPR22, Methodology=grouping2023.12 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | 78.9 | |
| SoftGroup2024.11 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | 78.9 | |
| SoftGroupArchitecture=Non-transformer-based2025.02 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.8 | 78.9 | |
| SoftGroupVenue=CVPR 22, Method Category=Cluster-based2026.02 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.8 | 78.9 | |
| SoftGroup2022.07 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | 78.9 | |
| Di&Co3D2023.03 | 67.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.7 | 77.2 | |
| DKNetInfer. Time(ms)=357.52022.07 | 67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.5 | 77 | |
| DKNet2022.07 | 67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.5 | 77 | |
| DKNetVenue=ECCV22, Methodology=kernel2023.12 | 66.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.8 | 76.9 | |
| DKNetArchitecture=Non-transformer-based2025.02 | 66.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.8 | 76.9 | |
| DKNet2023.03 | 66.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.8 | 76.9 | |
| DKNet2024.11 | 66.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.8 | 76.9 | |
| SSTNet2022.07 | 64.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50 | 73.9 | |
| HAIS2022.03 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.6 | |
| HAIS2022.11 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 44.1 | 75.7 | |
| HAIS2023.03 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAISSupervision regime=Supervised Only, Extra training dataset (ScanNetV2)=true2023.05 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAIS2024.11 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAISArchitecture=Non-transformer-based2025.02 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAISVenue=ICCV 21, Method Category=Cluster-based2026.02 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 74.6 | |
| SSTNet2022.03 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 74 | |
| SSTNet2022.11 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| SSTNet2022.04 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| SSTNet2023.03 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| SSTNetRuntime (in ms)=4282022.10 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | — | |
| SSTNet2023.09 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | — | |
| SSTNetVenue=ICCV21, Methodology=grouping2023.12 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| SSTNetArchitecture=Non-transformer-based2025.02 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| SSTNetVenue=ICCV 21, Method Category=Cluster-based2026.02 | 64.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.4 | 74 | |
| RPGNVenue=ECCV22, Methodology=grouping2023.12 | 64.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HAISclass agnostic=false2022.04 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAISRuntime (in ms)=3392022.10 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | — | |
| HAIS2023.09 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | — | |
| HAISInfer. Time(ms)=206.02022.07 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| HAIS2022.07 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.5 | 75.6 | |
| PointInst3D2023.03 | 63.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.6 | — | |
| PointInst3DVenue=ECCV22, Methodology=kernel2023.12 | 63.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.6 | — | |
| PointInst3D2022.07 | 63.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45.6 | — | |
| MaskGroup2022.10 | 63.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42 | — | |
| MaskGroup2023.09 | 63.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42 | — | |
| MaskGroupArchitecture=Non-transformer-based2025.02 | 63.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42 | 74 |