3D Semantic Instance Results - ScanNet Benchmark

The 3D semantic instance prediction task involves detecting and segmenting the object in an 3D scan mesh.

Evaluation and metrics

Our evaluation ranks all methods according to the average precision for each class. We report the mean average precision AP at overlap 0.25 (AP 25%), overlap 0.5 (AP 50%), and over overlaps in the range [0.5:0.95:0.05] (AP). Note that multiple predictions of the same ground truth instance are penalized as false positives.

This table lists the benchmark results for the 3D semantic instance scenario.

Method	avg ap 25%	bathtub	bed	bookshelf	cabinet	chair	counter	curtain	desk	door	otherfurniture	picture	refrigerator	shower curtain	sink	sofa	table	toilet	window

Box2Mask	0.803 31	1.000 1	0.962 32	0.874 13	0.707 52	0.887 31	0.686 27	0.598 55	0.961 1	0.715 30	0.694 29	0.469 52	0.700 21	1.000 1	0.912 19	0.902 8	0.753 37	0.997 39	0.637 43
Julian Chibane, Francis Engelmann, Tuan Anh Tran, Gerard Pons-Moll: Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes. ECCV 2022
SphereSeg	0.835 23	1.000 1	0.963 31	0.891 9	0.794 26	0.954 7	0.822 7	0.710 42	0.961 2	0.721 28	0.693 30	0.530 47	0.653 29	1.000 1	0.867 38	0.857 38	0.859 20	0.991 50	0.771 21

PointRel	0.901 1	1.000 1	0.978 23	0.928 3	0.879 1	0.962 5	0.882 4	0.749 36	0.947 3	0.912 2	0.802 3	0.753 17	0.820 2	1.000 1	0.984 4	0.919 5	0.894 3	1.000 1	0.815 14
: Relation3D (PointRel): Enhancing Relation Modeling for Point Cloud Instance Segmentation.
MG-Former	0.887 4	1.000 1	0.991 14	0.837 25	0.801 23	0.935 18	0.887 3	0.857 10	0.946 4	0.891 10	0.748 17	0.805 5	0.739 16	1.000 1	0.993 2	0.809 57	0.876 14	1.000 1	0.842 3

TD3D	0.875 12	1.000 1	0.976 24	0.877 12	0.783 29	0.970 2	0.889 2	0.828 19	0.945 5	0.803 22	0.713 24	0.720 24	0.709 19	1.000 1	0.936 14	0.934 3	0.873 15	1.000 1	0.791 18
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
InsSSM	0.883 6	1.000 1	0.996 6	0.800 38	0.865 4	0.960 6	0.808 10	0.852 15	0.940 6	0.899 8	0.785 4	0.810 3	0.700 21	1.000 1	0.912 19	0.851 42	0.895 2	0.997 39	0.827 8
Lei Yao, Yi Wang, Moyun Liu, Lap-Pui Chau: SGIFormer: Semantic-guided and Geometric-enhanced Interleaving Transformer for 3D Instance Segmentation. TCSVT, 2024
EV3D	0.877 11	1.000 1	0.996 8	0.873 14	0.854 8	0.950 10	0.691 25	0.783 30	0.926 7	0.889 13	0.754 14	0.794 10	0.820 2	1.000 1	0.912 19	0.900 9	0.860 19	1.000 1	0.779 20

Spherical Mask(CtoF)	0.875 12	1.000 1	0.991 15	0.873 14	0.850 9	0.946 12	0.691 25	0.752 35	0.926 7	0.889 12	0.759 11	0.794 9	0.820 2	1.000 1	0.912 19	0.900 9	0.878 11	1.000 1	0.769 22

Mask3D	0.870 16	1.000 1	0.985 17	0.782 46	0.818 19	0.938 16	0.760 16	0.749 36	0.923 9	0.877 15	0.760 10	0.785 12	0.820 2	1.000 1	0.912 19	0.864 35	0.878 11	0.983 53	0.825 9
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
TST3D	0.879 9	1.000 1	0.994 9	0.921 5	0.807 22	0.939 15	0.771 15	0.887 7	0.923 10	0.862 17	0.722 22	0.768 14	0.756 13	1.000 1	0.910 29	0.904 7	0.836 26	0.999 38	0.824 10
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
Queryformer	0.874 14	1.000 1	0.978 22	0.809 36	0.876 2	0.936 17	0.702 22	0.716 41	0.920 11	0.875 16	0.766 9	0.772 13	0.818 6	1.000 1	0.995 1	0.916 6	0.892 4	1.000 1	0.767 23

OneFormer3D	0.896 2	1.000 1	1.000 1	0.913 6	0.858 6	0.951 9	0.786 14	0.837 18	0.916 12	0.908 4	0.778 8	0.803 6	0.750 14	1.000 1	0.976 6	0.926 4	0.882 7	0.995 47	0.849 1
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
SoftGroup	0.865 18	1.000 1	0.969 27	0.860 20	0.860 5	0.913 22	0.558 37	0.899 4	0.911 13	0.760 23	0.828 1	0.736 21	0.802 8	0.981 44	0.919 18	0.875 23	0.877 13	1.000 1	0.820 12
Thang Vu, Kookhoi Kim, Tung M. Luu, Xuan Thanh Nguyen, Chang D. Yoo: SoftGroup for 3D Instance Segmentaiton on Point Clouds. CVPR 2022 [Oral]
Competitor-SPFormer	0.881 8	1.000 1	1.000 1	0.845 23	0.854 7	0.962 4	0.714 21	0.857 11	0.904 14	0.902 6	0.782 7	0.789 11	0.662 27	1.000 1	0.988 3	0.874 25	0.886 6	0.997 39	0.847 2

Competitor-MAFT	0.896 2	1.000 1	1.000 1	0.872 16	0.847 11	0.967 3	0.955 1	0.778 32	0.901 15	0.919 1	0.784 5	0.812 1	0.770 12	1.000 1	0.949 8	0.865 33	0.868 17	1.000 1	0.840 4

PointGroup	0.778 40	1.000 1	0.900 52	0.798 40	0.715 50	0.863 36	0.493 47	0.706 43	0.895 16	0.569 56	0.701 26	0.576 38	0.639 42	1.000 1	0.880 35	0.851 41	0.719 45	0.997 39	0.709 34
Li Jiang, Hengshuang Zhao, Shaoshuai Shi, Shu Liu, Chi-Wing Fu, Jiaya Jia: PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation. CVPR 2020 [oral]
Mask3D_evaluation	0.843 22	1.000 1	0.955 33	0.847 22	0.795 25	0.932 19	0.750 18	0.780 31	0.891 17	0.818 19	0.737 19	0.633 34	0.703 20	1.000 1	0.902 31	0.870 28	0.820 27	0.941 61	0.805 15

SIM3D	0.878 10	1.000 1	0.972 25	0.863 19	0.817 20	0.952 8	0.821 8	0.783 29	0.890 18	0.902 7	0.735 20	0.797 7	0.799 9	1.000 1	0.931 16	0.893 13	0.853 21	1.000 1	0.792 17

Mask-Group	0.792 33	1.000 1	0.968 29	0.812 32	0.766 34	0.864 35	0.460 49	0.815 23	0.888 19	0.598 49	0.651 36	0.639 30	0.600 47	0.918 50	0.941 9	0.896 12	0.721 44	1.000 1	0.723 31
Min Zhong, Xinghao Chen, Xiaokang Chen, Gang Zeng, Yunhe Wang: MaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation. ICME 2022
IPCA-Inst	0.851 20	1.000 1	0.968 28	0.884 11	0.842 13	0.862 39	0.693 24	0.812 24	0.888 20	0.677 36	0.783 6	0.698 25	0.807 7	1.000 1	0.911 26	0.865 34	0.865 18	1.000 1	0.757 26

SoftGroup++	0.874 14	1.000 1	0.972 26	0.947 1	0.839 14	0.898 26	0.556 40	0.913 2	0.881 21	0.756 24	0.828 2	0.748 19	0.821 1	1.000 1	0.937 13	0.937 1	0.887 5	1.000 1	0.821 11

PBNet	0.825 27	1.000 1	0.963 30	0.837 27	0.843 12	0.865 34	0.822 6	0.647 50	0.878 22	0.733 26	0.639 39	0.683 27	0.650 30	1.000 1	0.853 40	0.870 29	0.820 28	1.000 1	0.744 28
W.Zhao, Y.Yan, C.Yang, J.Ye,X.Yang,K.Huang: Divide and Conquer: 3D Instance Segmentation With Point-Wise Binarization. ICCV 2023
ExtMask3D	0.867 17	1.000 1	1.000 1	0.756 53	0.816 21	0.940 14	0.795 12	0.760 34	0.862 23	0.888 14	0.739 18	0.763 15	0.774 11	1.000 1	0.929 17	0.878 21	0.879 9	1.000 1	0.819 13

GraphCut	0.832 25	1.000 1	0.922 48	0.724 57	0.798 24	0.902 25	0.701 23	0.856 13	0.859 24	0.715 29	0.706 25	0.748 18	0.640 41	1.000 1	0.934 15	0.862 36	0.880 8	1.000 1	0.729 29

SPFormer	0.851 20	1.000 1	0.994 10	0.806 37	0.774 31	0.942 13	0.637 29	0.849 16	0.859 25	0.889 11	0.720 23	0.730 22	0.665 26	1.000 1	0.911 26	0.868 32	0.873 16	1.000 1	0.796 16
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
SALoss-ResNet	0.695 54	1.000 1	0.855 60	0.579 68	0.589 63	0.735 62	0.484 48	0.588 56	0.856 26	0.634 42	0.571 50	0.298 60	0.500 61	1.000 1	0.824 45	0.818 52	0.702 48	0.935 64	0.545 57
Zhidong Liang, Ming Yang, Hao Li, Chunxiang Wang: 3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud Segmentation. IEEE Robotics and Automation Letters (IROS2020)
ISBNet	0.835 23	1.000 1	0.950 34	0.731 55	0.819 17	0.918 20	0.790 13	0.740 38	0.851 27	0.831 18	0.661 32	0.742 20	0.650 30	1.000 1	0.937 12	0.814 56	0.836 25	1.000 1	0.765 24
Tuan Duc Ngo, Binh-Son Hua, Khoi Nguyen: ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution. CVPR 2023
MAFT	0.860 19	1.000 1	0.990 16	0.810 35	0.829 15	0.949 11	0.809 9	0.688 47	0.836 28	0.904 5	0.751 15	0.796 8	0.741 15	1.000 1	0.864 39	0.848 44	0.837 24	1.000 1	0.828 7

UniPerception	0.884 5	1.000 1	0.979 20	0.872 16	0.869 3	0.892 27	0.806 11	0.890 6	0.835 29	0.892 9	0.755 13	0.811 2	0.779 10	0.955 47	0.951 7	0.876 22	0.914 1	0.997 39	0.840 5

RPGN	0.806 30	1.000 1	0.992 12	0.789 41	0.723 48	0.891 28	0.650 28	0.810 25	0.832 30	0.665 39	0.699 28	0.658 28	0.700 21	1.000 1	0.881 33	0.832 48	0.774 30	0.997 39	0.613 49
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
OSIS	0.725 50	1.000 1	0.885 58	0.653 63	0.657 58	0.801 52	0.576 36	0.695 45	0.828 31	0.698 32	0.534 57	0.457 54	0.500 61	0.857 53	0.831 44	0.841 46	0.627 60	1.000 1	0.619 46

KmaxOneFormerNet	0.883 6	1.000 1	1.000 1	0.798 39	0.848 10	0.971 1	0.853 5	0.903 3	0.827 32	0.910 3	0.748 16	0.809 4	0.724 18	1.000 1	0.980 5	0.855 39	0.844 23	1.000 1	0.832 6

HAIS	0.803 31	1.000 1	0.994 10	0.820 31	0.759 36	0.855 40	0.554 41	0.882 8	0.827 33	0.615 45	0.676 31	0.638 31	0.646 39	1.000 1	0.912 19	0.797 62	0.767 31	0.994 48	0.726 30
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
TopoSeg	0.832 25	1.000 1	0.981 19	0.933 2	0.819 18	0.826 48	0.524 46	0.841 17	0.811 34	0.681 35	0.759 12	0.687 26	0.727 17	0.981 44	0.911 26	0.883 17	0.853 22	1.000 1	0.756 27

GICN	0.788 36	1.000 1	0.978 21	0.867 18	0.781 30	0.833 45	0.527 45	0.824 20	0.806 35	0.549 58	0.596 46	0.551 40	0.700 21	1.000 1	0.853 40	0.935 2	0.733 41	1.000 1	0.651 40

DKNet	0.815 29	1.000 1	0.930 40	0.844 24	0.765 35	0.915 21	0.534 44	0.805 26	0.805 36	0.807 21	0.654 33	0.763 16	0.650 30	1.000 1	0.794 52	0.881 18	0.766 32	1.000 1	0.758 25
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
DENet	0.786 37	1.000 1	0.929 41	0.736 54	0.750 42	0.720 64	0.755 17	0.934 1	0.794 37	0.590 51	0.561 52	0.537 45	0.650 30	1.000 1	0.882 32	0.804 60	0.789 29	1.000 1	0.719 32

SSEC	0.820 28	1.000 1	0.983 18	0.924 4	0.826 16	0.817 51	0.415 55	0.899 5	0.793 38	0.673 37	0.731 21	0.636 32	0.653 28	1.000 1	0.939 11	0.804 59	0.878 10	1.000 1	0.780 19

PE	0.776 41	1.000 1	0.900 53	0.860 20	0.728 47	0.869 32	0.400 56	0.857 12	0.774 39	0.568 57	0.701 27	0.602 36	0.646 39	0.933 49	0.843 43	0.890 14	0.691 53	0.997 39	0.709 33
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
DualGroup	0.782 39	1.000 1	0.927 42	0.811 33	0.772 32	0.853 41	0.631 31	0.805 26	0.773 40	0.613 46	0.611 44	0.610 35	0.650 30	0.835 61	0.881 33	0.879 20	0.750 39	1.000 1	0.675 38

NeuralBF	0.718 52	1.000 1	0.945 35	0.901 7	0.754 38	0.817 50	0.460 49	0.700 44	0.772 41	0.688 33	0.568 51	0.000 74	0.500 61	0.981 44	0.606 65	0.872 26	0.740 40	1.000 1	0.614 48
Weiwei Sun, Daniel Rebain, Renjie Liao, Vladimir Tankovich, Soroosh Yazdani, Kwang Moo Yi, Andrea Tagliasacchi: NeuralBF: Neural Bilateral Filtering for Top-down Instance Segmentation on Point Clouds. WACV 2023
PanopticFusion-inst	0.693 55	1.000 1	0.852 61	0.655 62	0.616 60	0.788 54	0.334 58	0.763 33	0.771 42	0.457 68	0.555 54	0.652 29	0.518 58	0.857 53	0.765 55	0.732 68	0.631 58	0.944 57	0.577 54
Gaku Narita, Takashi Seno, Tomoya Ishikawa, Yohsuke Kaji: PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things. IROS 2019 (to appear)
CSC-Pretrained	0.791 34	1.000 1	0.996 6	0.829 30	0.767 33	0.889 30	0.600 32	0.819 22	0.770 43	0.594 50	0.620 43	0.541 44	0.700 21	1.000 1	0.941 9	0.889 15	0.763 33	1.000 1	0.526 59

SegGroup_ins	0.637 62	1.000 1	0.923 47	0.593 67	0.561 65	0.746 60	0.143 72	0.504 64	0.766 44	0.485 66	0.442 64	0.372 58	0.530 57	0.714 62	0.815 48	0.775 64	0.673 56	1.000 1	0.431 65
An Tao, Yueqi Duan, Yi Wei, Jiwen Lu, Jie Zhou: SegGroup: Seg-Level Supervision for 3D Instance and Semantic Segmentation. TIP 2022
DANCENET	0.786 37	1.000 1	0.936 37	0.783 44	0.737 45	0.852 42	0.742 19	0.647 50	0.765 45	0.811 20	0.624 42	0.579 37	0.632 44	1.000 1	0.909 30	0.898 11	0.696 49	0.944 57	0.601 52

SSEN	0.724 51	1.000 1	0.926 43	0.781 47	0.661 56	0.845 43	0.596 35	0.529 62	0.764 46	0.653 40	0.489 63	0.461 53	0.500 61	0.859 52	0.765 55	0.872 27	0.761 35	1.000 1	0.577 53
Dongsu Zhang, Junha Chun, Sang Kyun Cha, Young Min Kim: Spatial Semantic Embedding Network: Fast 3D Instance Segmentation with Deep Metric Learning. Arxiv
Dyco3D	0.761 45	1.000 1	0.935 38	0.893 8	0.752 41	0.863 37	0.600 32	0.588 56	0.742 47	0.641 41	0.633 40	0.546 42	0.550 54	0.857 53	0.789 54	0.853 40	0.762 34	0.987 51	0.699 35
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
OccuSeg+instance	0.742 46	1.000 1	0.923 45	0.785 42	0.745 43	0.867 33	0.557 38	0.578 59	0.729 48	0.670 38	0.644 38	0.488 50	0.577 53	1.000 1	0.794 52	0.830 49	0.620 62	1.000 1	0.550 55
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
3D-BoNet	0.687 57	1.000 1	0.887 57	0.836 28	0.587 64	0.643 71	0.550 42	0.620 52	0.724 49	0.522 63	0.501 61	0.243 63	0.512 59	1.000 1	0.751 57	0.807 58	0.661 57	0.909 66	0.612 50
Bo Yang, Jianan Wang, Ronald Clark, Qingyong Hu, Sen Wang, Andrew Markham, Niki Trigoni: Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds. NeurIPS 2019 Spotlight
AOIA	0.767 42	1.000 1	0.937 36	0.810 34	0.740 44	0.906 23	0.550 42	0.800 28	0.706 50	0.577 55	0.624 41	0.544 43	0.596 52	0.857 53	0.879 37	0.880 19	0.750 38	0.992 49	0.658 39

SPG_WSIS	0.678 60	1.000 1	0.880 59	0.836 28	0.701 53	0.727 63	0.273 65	0.607 54	0.706 51	0.541 61	0.515 60	0.174 66	0.600 47	0.857 53	0.716 59	0.846 45	0.711 46	1.000 1	0.506 60

DD-UNet+Group	0.764 43	1.000 1	0.897 55	0.837 26	0.753 39	0.830 47	0.459 51	0.824 20	0.699 52	0.629 43	0.653 34	0.438 55	0.650 30	1.000 1	0.880 35	0.858 37	0.690 54	1.000 1	0.650 41
H. Liu, R. Liu, K. Yang, J. Zhang, K. Peng, R. Stiefelhagen: HIDA: Towards Holistic Indoor Understanding for the Visually Impaired via Semantic Instance Segmentation with a Wearable Solid-State LiDAR Sensor. ICCVW 2021
3D-MPA	0.737 48	1.000 1	0.933 39	0.785 42	0.794 27	0.831 46	0.279 63	0.588 56	0.695 53	0.616 44	0.559 53	0.556 39	0.650 30	1.000 1	0.809 49	0.875 24	0.696 50	1.000 1	0.608 51
Francis Engelmann, Martin Bokeloh, Alireza Fathi, Bastian Leibe, Matthias Nießner: 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation. CVPR 2020
Sparse R-CNN	0.714 53	1.000 1	0.926 44	0.694 58	0.699 54	0.890 29	0.636 30	0.516 63	0.693 54	0.743 25	0.588 48	0.369 59	0.601 46	0.594 67	0.800 50	0.886 16	0.676 55	0.986 52	0.546 56

RWSeg	0.739 47	1.000 1	0.899 54	0.759 51	0.753 40	0.823 49	0.282 61	0.691 46	0.658 55	0.582 54	0.594 47	0.547 41	0.628 45	1.000 1	0.795 51	0.868 31	0.728 43	1.000 1	0.692 36

Occipital-SCS	0.688 56	1.000 1	0.913 49	0.730 56	0.737 46	0.743 61	0.442 52	0.855 14	0.655 56	0.546 59	0.546 55	0.263 62	0.508 60	0.889 51	0.568 66	0.771 65	0.705 47	0.889 67	0.625 45

INS-Conv-instance	0.762 44	1.000 1	0.923 45	0.765 49	0.785 28	0.905 24	0.600 32	0.655 49	0.646 57	0.683 34	0.647 37	0.530 46	0.650 30	1.000 1	0.824 45	0.830 49	0.693 52	0.944 57	0.644 42

SSTNet	0.789 35	1.000 1	0.840 62	0.888 10	0.717 49	0.835 44	0.717 20	0.684 48	0.627 58	0.724 27	0.652 35	0.727 23	0.600 47	1.000 1	0.912 19	0.822 51	0.757 36	1.000 1	0.691 37
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
R-PointNet	0.544 66	0.500 73	0.655 72	0.661 61	0.663 55	0.765 57	0.432 53	0.214 72	0.612 59	0.584 53	0.499 62	0.204 65	0.286 71	0.429 70	0.655 62	0.650 73	0.539 63	0.950 56	0.499 62

PCJC	0.684 59	1.000 1	0.895 56	0.757 52	0.659 57	0.862 38	0.189 70	0.739 39	0.606 60	0.712 31	0.581 49	0.515 49	0.650 30	0.857 53	0.357 71	0.785 63	0.631 59	0.889 67	0.635 44

MTML	0.731 49	1.000 1	0.992 12	0.779 48	0.609 61	0.746 59	0.308 60	0.867 9	0.601 61	0.607 47	0.539 56	0.519 48	0.550 54	1.000 1	0.824 45	0.869 30	0.729 42	1.000 1	0.616 47
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
ASIS	0.422 72	0.333 74	0.707 70	0.676 60	0.401 72	0.650 70	0.350 57	0.177 73	0.594 62	0.376 71	0.202 72	0.077 72	0.404 66	0.571 68	0.197 75	0.674 72	0.447 71	0.500 74	0.260 74

ClickSeg_Instance	0.685 58	1.000 1	0.818 64	0.600 66	0.715 51	0.795 53	0.557 38	0.533 61	0.591 63	0.601 48	0.519 59	0.429 57	0.638 43	0.938 48	0.706 60	0.817 54	0.624 61	0.944 57	0.502 61

MASC	0.615 63	0.711 70	0.802 65	0.540 69	0.757 37	0.777 55	0.029 73	0.577 60	0.588 64	0.521 64	0.600 45	0.436 56	0.534 56	0.697 63	0.616 64	0.838 47	0.526 64	0.980 54	0.534 58
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
One_Thing_One_Click	0.675 61	1.000 1	0.823 63	0.782 45	0.621 59	0.766 56	0.211 67	0.736 40	0.560 65	0.586 52	0.522 58	0.636 33	0.453 65	0.641 65	0.853 40	0.850 43	0.694 51	0.997 39	0.411 66
Zhengzhe Liu, Xiaojuan Qi, Chi-Wing Fu: One Thing One Click: A Self-Training Approach for Weakly Supervised 3D Semantic Segmentation. CVPR 2021
UNet-backbone	0.605 64	1.000 1	0.909 50	0.764 50	0.603 62	0.704 65	0.415 54	0.301 69	0.548 66	0.461 67	0.394 65	0.267 61	0.386 67	0.857 53	0.649 63	0.817 53	0.504 66	0.959 55	0.356 69

SemRegionNet-20cls	0.470 71	1.000 1	0.727 67	0.447 72	0.481 69	0.678 68	0.024 74	0.380 67	0.518 67	0.440 69	0.339 67	0.128 68	0.350 68	0.429 70	0.212 74	0.711 70	0.465 69	0.833 71	0.290 73

tmp	0.474 70	1.000 1	0.727 67	0.433 73	0.481 70	0.673 69	0.022 75	0.380 67	0.517 68	0.436 70	0.338 68	0.128 68	0.343 69	0.429 70	0.291 73	0.728 69	0.473 67	0.833 71	0.300 72

3D-SIS	0.558 65	1.000 1	0.773 66	0.614 65	0.503 68	0.691 67	0.200 68	0.412 65	0.498 69	0.546 60	0.311 70	0.103 70	0.600 47	0.857 53	0.382 68	0.799 61	0.445 72	0.938 63	0.371 67
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
Hier3D	0.540 67	1.000 1	0.727 67	0.626 64	0.467 71	0.693 66	0.200 68	0.412 65	0.480 70	0.528 62	0.318 69	0.077 73	0.600 47	0.688 64	0.382 68	0.768 66	0.472 68	0.941 61	0.350 70
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
Sgpn_scannet	0.390 74	0.556 72	0.636 73	0.493 70	0.353 73	0.539 74	0.271 66	0.160 74	0.450 71	0.359 73	0.178 73	0.146 67	0.250 73	0.143 74	0.347 72	0.698 71	0.436 73	0.667 73	0.331 71

3D-BEVIS	0.401 73	0.667 71	0.687 71	0.419 74	0.137 75	0.587 73	0.188 71	0.235 70	0.359 72	0.211 74	0.093 75	0.080 71	0.311 70	0.571 68	0.382 68	0.754 67	0.300 74	0.874 69	0.357 68
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Sem_Recon_ins	0.484 69	0.764 69	0.608 74	0.470 71	0.521 67	0.637 72	0.311 59	0.218 71	0.348 73	0.365 72	0.223 71	0.222 64	0.258 72	0.629 66	0.734 58	0.596 74	0.509 65	0.858 70	0.444 64

Region-18class	0.497 68	0.250 75	0.902 51	0.689 59	0.540 66	0.747 58	0.276 64	0.610 53	0.268 74	0.489 65	0.348 66	0.000 74	0.243 74	0.220 73	0.663 61	0.814 55	0.459 70	0.928 65	0.496 63

MaskRCNN 2d->3d Proj	0.261 75	0.903 68	0.081 75	0.008 75	0.233 74	0.175 75	0.280 62	0.106 75	0.150 75	0.203 75	0.175 74	0.480 51	0.218 75	0.143 74	0.542 67	0.404 75	0.153 75	0.393 75	0.049 75

3D Semantic Instance Benchmark