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

PointComp	0.897 2	1.000 1	0.998 6	0.864 20	0.869 3	0.969 3	0.830 8	0.783 33	0.905 15	0.894 10	0.791 4	0.834 1	0.769 14	1.000 1	0.982 5	0.920 5	0.868 20	1.000 1	0.872 2

DKNet	0.815 33	1.000 1	0.930 44	0.844 27	0.765 39	0.915 24	0.534 48	0.805 28	0.805 39	0.807 25	0.654 37	0.763 20	0.650 33	1.000 1	0.794 56	0.881 21	0.766 36	1.000 1	0.758 28
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
TopoSeg	0.832 29	1.000 1	0.981 21	0.933 2	0.819 22	0.826 52	0.524 50	0.841 19	0.811 37	0.681 39	0.759 14	0.687 30	0.727 19	0.981 48	0.911 30	0.883 20	0.853 25	1.000 1	0.756 30

SoftGroup++	0.874 17	1.000 1	0.972 28	0.947 1	0.839 15	0.898 29	0.556 44	0.913 2	0.881 23	0.756 28	0.828 2	0.748 23	0.821 1	1.000 1	0.937 15	0.937 1	0.887 6	1.000 1	0.821 13

SPFormer	0.851 23	1.000 1	0.994 11	0.806 41	0.774 35	0.942 16	0.637 33	0.849 18	0.859 27	0.889 13	0.720 26	0.730 26	0.665 29	1.000 1	0.911 30	0.868 36	0.873 17	1.000 1	0.796 19
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
SSEC	0.820 32	1.000 1	0.983 20	0.924 4	0.826 19	0.817 55	0.415 59	0.899 5	0.793 42	0.673 41	0.731 24	0.636 36	0.653 31	1.000 1	0.939 13	0.804 63	0.878 11	1.000 1	0.780 22

SegGroup_ins	0.637 66	1.000 1	0.923 51	0.593 71	0.561 69	0.746 64	0.143 76	0.504 68	0.766 48	0.485 70	0.442 68	0.372 62	0.530 61	0.714 66	0.815 52	0.775 68	0.673 60	1.000 1	0.431 69
An Tao, Yueqi Duan, Yi Wei, Jiwen Lu, Jie Zhou: SegGroup: Seg-Level Supervision for 3D Instance and Semantic Segmentation. TIP 2022
SPG_WSIS	0.678 64	1.000 1	0.880 63	0.836 31	0.701 57	0.727 67	0.273 69	0.607 58	0.706 55	0.541 65	0.515 64	0.174 70	0.600 51	0.857 57	0.716 63	0.846 49	0.711 50	1.000 1	0.506 64

PBNet	0.825 31	1.000 1	0.963 32	0.837 30	0.843 13	0.865 38	0.822 9	0.647 54	0.878 24	0.733 30	0.639 43	0.683 31	0.650 33	1.000 1	0.853 44	0.870 33	0.820 32	1.000 1	0.744 31
Weiguang Zhao, Yuyao Yan, Chaolong Yang, Jianan Ye, Xi Yang, Kaizhu Huang: Divide and Conquer: 3D Instance Segmentation With Point-Wise Binarization. ICCV 2023
DCD	0.885 6	1.000 1	0.933 43	0.856 24	0.832 16	0.959 8	0.930 2	0.858 11	0.802 40	0.859 20	0.767 10	0.796 11	0.709 22	1.000 1	0.971 8	0.871 31	0.904 2	1.000 1	0.874 1

SoftGroup	0.865 21	1.000 1	0.969 29	0.860 22	0.860 6	0.913 25	0.558 41	0.899 4	0.911 14	0.760 27	0.828 1	0.736 25	0.802 8	0.981 48	0.919 20	0.875 26	0.877 14	1.000 1	0.820 15
Thang Vu, Kookhoi Kim, Tung M. Luu, Xuan Thanh Nguyen, Chang D. Yoo: SoftGroup for 3D Instance Segmentaiton on Point Clouds. CVPR 2022 [Oral]
IPCA-Inst	0.851 23	1.000 1	0.968 30	0.884 12	0.842 14	0.862 43	0.693 28	0.812 26	0.888 22	0.677 40	0.783 7	0.698 29	0.807 7	1.000 1	0.911 30	0.865 38	0.865 21	1.000 1	0.757 29

DualGroup	0.782 43	1.000 1	0.927 46	0.811 37	0.772 36	0.853 45	0.631 35	0.805 28	0.773 44	0.613 50	0.611 48	0.610 39	0.650 33	0.835 65	0.881 37	0.879 23	0.750 43	1.000 1	0.675 42

DD-UNet+Group	0.764 47	1.000 1	0.897 59	0.837 29	0.753 43	0.830 51	0.459 55	0.824 22	0.699 56	0.629 47	0.653 38	0.438 59	0.650 33	1.000 1	0.880 39	0.858 41	0.690 58	1.000 1	0.650 45
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
DENet	0.786 41	1.000 1	0.929 45	0.736 58	0.750 46	0.720 68	0.755 21	0.934 1	0.794 41	0.590 55	0.561 56	0.537 49	0.650 33	1.000 1	0.882 36	0.804 64	0.789 33	1.000 1	0.719 35

SSTNet	0.789 39	1.000 1	0.840 66	0.888 11	0.717 53	0.835 48	0.717 24	0.684 52	0.627 62	0.724 31	0.652 39	0.727 27	0.600 51	1.000 1	0.912 22	0.822 55	0.757 40	1.000 1	0.691 41
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
Mask-Group	0.792 37	1.000 1	0.968 31	0.812 36	0.766 38	0.864 39	0.460 53	0.815 25	0.888 21	0.598 53	0.651 40	0.639 34	0.600 51	0.918 54	0.941 11	0.896 14	0.721 48	1.000 1	0.723 34
Min Zhong, Xinghao Chen, Xiaokang Chen, Gang Zeng, Yunhe Wang: MaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation. ICME 2022
GICN	0.788 40	1.000 1	0.978 23	0.867 19	0.781 34	0.833 49	0.527 49	0.824 22	0.806 38	0.549 62	0.596 50	0.551 44	0.700 24	1.000 1	0.853 44	0.935 2	0.733 45	1.000 1	0.651 44

SSEN	0.724 55	1.000 1	0.926 47	0.781 51	0.661 60	0.845 47	0.596 39	0.529 66	0.764 50	0.653 44	0.489 67	0.461 57	0.500 65	0.859 56	0.765 59	0.872 30	0.761 39	1.000 1	0.577 57
Dongsu Zhang, Junha Chun, Sang Kyun Cha, Young Min Kim: Spatial Semantic Embedding Network: Fast 3D Instance Segmentation with Deep Metric Learning. Arxiv
OccuSeg+instance	0.742 50	1.000 1	0.923 49	0.785 46	0.745 47	0.867 37	0.557 42	0.578 63	0.729 52	0.670 42	0.644 42	0.488 54	0.577 57	1.000 1	0.794 56	0.830 53	0.620 66	1.000 1	0.550 59
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
CSC-Pretrained	0.791 38	1.000 1	0.996 7	0.829 34	0.767 37	0.889 33	0.600 36	0.819 24	0.770 47	0.594 54	0.620 47	0.541 48	0.700 24	1.000 1	0.941 11	0.889 18	0.763 37	1.000 1	0.526 63

MTML	0.731 53	1.000 1	0.992 13	0.779 52	0.609 65	0.746 63	0.308 64	0.867 10	0.601 65	0.607 51	0.539 60	0.519 52	0.550 58	1.000 1	0.824 49	0.869 34	0.729 46	1.000 1	0.616 51
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
3D-MPA	0.737 52	1.000 1	0.933 42	0.785 46	0.794 31	0.831 50	0.279 67	0.588 60	0.695 57	0.616 48	0.559 57	0.556 43	0.650 33	1.000 1	0.809 53	0.875 27	0.696 54	1.000 1	0.608 55
Francis Engelmann, Martin Bokeloh, Alireza Fathi, Bastian Leibe, Matthias Nießner: 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation. CVPR 2020
GraphCut	0.832 29	1.000 1	0.922 52	0.724 61	0.798 28	0.902 28	0.701 27	0.856 15	0.859 26	0.715 33	0.706 28	0.748 22	0.640 44	1.000 1	0.934 17	0.862 40	0.880 9	1.000 1	0.729 32

RWSeg	0.739 51	1.000 1	0.899 58	0.759 55	0.753 44	0.823 53	0.282 65	0.691 50	0.658 59	0.582 58	0.594 51	0.547 45	0.628 48	1.000 1	0.795 55	0.868 35	0.728 47	1.000 1	0.692 40

NeuralBF	0.718 56	1.000 1	0.945 38	0.901 7	0.754 42	0.817 54	0.460 53	0.700 48	0.772 45	0.688 37	0.568 55	0.000 78	0.500 65	0.981 48	0.606 69	0.872 29	0.740 44	1.000 1	0.614 52
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
Queryformer	0.874 17	1.000 1	0.978 24	0.809 40	0.876 2	0.936 20	0.702 26	0.716 45	0.920 12	0.875 18	0.766 11	0.772 17	0.818 6	1.000 1	0.995 1	0.916 7	0.892 5	1.000 1	0.767 26

TD3D	0.875 15	1.000 1	0.976 26	0.877 13	0.783 33	0.970 2	0.889 3	0.828 21	0.945 5	0.803 26	0.713 27	0.720 28	0.709 21	1.000 1	0.936 16	0.934 3	0.873 16	1.000 1	0.791 21
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
EV3D	0.877 14	1.000 1	0.996 9	0.873 15	0.854 9	0.950 13	0.691 29	0.783 32	0.926 8	0.889 15	0.754 17	0.794 13	0.820 2	1.000 1	0.912 22	0.900 11	0.860 22	1.000 1	0.779 23

ExtMask3D	0.867 20	1.000 1	1.000 1	0.756 57	0.816 25	0.940 17	0.795 15	0.760 38	0.862 25	0.888 16	0.739 21	0.763 19	0.774 11	1.000 1	0.929 19	0.878 24	0.879 10	1.000 1	0.819 16

MG-Former	0.887 5	1.000 1	0.991 15	0.837 28	0.801 27	0.935 21	0.887 4	0.857 12	0.946 4	0.891 12	0.748 20	0.805 6	0.739 18	1.000 1	0.993 2	0.809 61	0.876 15	1.000 1	0.842 5

SIM3D	0.878 13	1.000 1	0.972 27	0.863 21	0.817 24	0.952 10	0.821 11	0.783 31	0.890 20	0.902 8	0.735 23	0.797 9	0.799 9	1.000 1	0.931 18	0.893 15	0.853 24	1.000 1	0.792 20

Spherical Mask(CtoF)	0.875 15	1.000 1	0.991 16	0.873 15	0.850 10	0.946 15	0.691 29	0.752 39	0.926 8	0.889 14	0.759 13	0.794 12	0.820 2	1.000 1	0.912 22	0.900 11	0.878 12	1.000 1	0.769 25

KmaxOneFormerNet	0.883 8	1.000 1	1.000 1	0.798 43	0.848 11	0.971 1	0.853 7	0.903 3	0.827 34	0.910 3	0.748 19	0.809 5	0.724 20	1.000 1	0.980 6	0.855 43	0.844 26	1.000 1	0.832 8

PointRel	0.901 1	1.000 1	0.978 25	0.928 3	0.879 1	0.962 6	0.882 5	0.749 40	0.947 3	0.912 2	0.802 3	0.753 21	0.820 2	1.000 1	0.984 4	0.919 6	0.894 4	1.000 1	0.815 17
: Relation3D: Enhancing Relation Modeling for Point Cloud Instance Segmentation. CVPR 2025
MAFT	0.860 22	1.000 1	0.990 17	0.810 39	0.829 17	0.949 14	0.809 12	0.688 51	0.836 30	0.904 6	0.751 18	0.796 10	0.741 17	1.000 1	0.864 43	0.848 48	0.837 27	1.000 1	0.828 9

OSIS	0.725 54	1.000 1	0.885 62	0.653 67	0.657 62	0.801 56	0.576 40	0.695 49	0.828 33	0.698 36	0.534 61	0.457 58	0.500 65	0.857 57	0.831 48	0.841 50	0.627 64	1.000 1	0.619 50

ISBNet	0.835 27	1.000 1	0.950 37	0.731 59	0.819 21	0.918 23	0.790 16	0.740 42	0.851 29	0.831 21	0.661 36	0.742 24	0.650 33	1.000 1	0.937 14	0.814 60	0.836 28	1.000 1	0.765 27
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
Competitor-MAFT	0.896 3	1.000 1	1.000 1	0.872 17	0.847 12	0.967 4	0.955 1	0.778 35	0.901 17	0.919 1	0.784 6	0.812 2	0.770 13	1.000 1	0.949 10	0.865 37	0.868 19	1.000 1	0.840 6

TST3D	0.879 12	1.000 1	0.994 10	0.921 5	0.807 26	0.939 18	0.771 18	0.887 8	0.923 11	0.862 19	0.722 25	0.768 18	0.756 15	1.000 1	0.910 33	0.904 8	0.836 29	0.999 40	0.824 12
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
VDG-Uni3DSeg	0.880 11	1.000 1	0.990 17	0.889 10	0.823 20	0.952 11	0.764 19	0.893 6	0.941 6	0.907 5	0.756 15	0.781 16	0.628 48	1.000 1	0.918 21	0.903 9	0.872 18	0.999 40	0.821 14

PointGroup	0.778 44	1.000 1	0.900 56	0.798 44	0.715 54	0.863 40	0.493 51	0.706 47	0.895 18	0.569 60	0.701 29	0.576 42	0.639 45	1.000 1	0.880 39	0.851 45	0.719 49	0.997 42	0.709 38
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]
PE	0.776 45	1.000 1	0.900 57	0.860 22	0.728 51	0.869 36	0.400 60	0.857 14	0.774 43	0.568 61	0.701 30	0.602 40	0.646 42	0.933 53	0.843 47	0.890 17	0.691 57	0.997 42	0.709 37
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
Competitor-SPFormer	0.881 10	1.000 1	1.000 1	0.845 26	0.854 8	0.962 5	0.714 25	0.857 13	0.904 16	0.902 7	0.782 8	0.789 14	0.662 30	1.000 1	0.988 3	0.874 28	0.886 7	0.997 42	0.847 4

RPGN	0.806 34	1.000 1	0.992 13	0.789 45	0.723 52	0.891 31	0.650 32	0.810 27	0.832 32	0.665 43	0.699 31	0.658 32	0.700 24	1.000 1	0.881 37	0.832 52	0.774 34	0.997 42	0.613 53
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
InsSSM	0.883 8	1.000 1	0.996 7	0.800 42	0.865 5	0.960 7	0.808 13	0.852 17	0.940 7	0.899 9	0.785 5	0.810 4	0.700 24	1.000 1	0.912 22	0.851 46	0.895 3	0.997 42	0.827 10
Lei Yao, Yi Wang, Moyun Liu, Lap-Pui Chau: SGIFormer: Semantic-guided and Geometric-enhanced Interleaving Transformer for 3D Instance Segmentation. TCSVT, 2024
UniPerception	0.884 7	1.000 1	0.979 22	0.872 17	0.869 4	0.892 30	0.806 14	0.890 7	0.835 31	0.892 11	0.755 16	0.811 3	0.779 10	0.955 51	0.951 9	0.876 25	0.914 1	0.997 42	0.840 7

Box2Mask	0.803 35	1.000 1	0.962 34	0.874 14	0.707 56	0.887 34	0.686 31	0.598 59	0.961 1	0.715 34	0.694 32	0.469 56	0.700 24	1.000 1	0.912 22	0.902 10	0.753 41	0.997 42	0.637 47
Julian Chibane, Francis Engelmann, Tuan Anh Tran, Gerard Pons-Moll: Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes. ECCV 2022
One_Thing_One_Click	0.675 65	1.000 1	0.823 67	0.782 49	0.621 63	0.766 60	0.211 71	0.736 44	0.560 69	0.586 56	0.522 62	0.636 37	0.453 69	0.641 69	0.853 44	0.850 47	0.694 55	0.997 42	0.411 70
Zhengzhe Liu, Xiaojuan Qi, Chi-Wing Fu: One Thing One Click: A Self-Training Approach for Weakly Supervised 3D Semantic Segmentation. CVPR 2021
OneFormer3D	0.896 3	1.000 1	1.000 1	0.913 6	0.858 7	0.951 12	0.786 17	0.837 20	0.916 13	0.908 4	0.778 9	0.803 7	0.750 16	1.000 1	0.976 7	0.926 4	0.882 8	0.995 50	0.849 3
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
HAIS	0.803 35	1.000 1	0.994 11	0.820 35	0.759 40	0.855 44	0.554 45	0.882 9	0.827 35	0.615 49	0.676 35	0.638 35	0.646 42	1.000 1	0.912 22	0.797 66	0.767 35	0.994 51	0.726 33
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
AOIA	0.767 46	1.000 1	0.937 39	0.810 38	0.740 48	0.906 26	0.550 46	0.800 30	0.706 54	0.577 59	0.624 45	0.544 47	0.596 56	0.857 57	0.879 41	0.880 22	0.750 42	0.992 52	0.658 43

SphereSeg	0.835 27	1.000 1	0.963 33	0.891 9	0.794 30	0.954 9	0.822 10	0.710 46	0.961 2	0.721 32	0.693 33	0.530 51	0.653 32	1.000 1	0.867 42	0.857 42	0.859 23	0.991 53	0.771 24

Dyco3D	0.761 49	1.000 1	0.935 41	0.893 8	0.752 45	0.863 41	0.600 36	0.588 60	0.742 51	0.641 45	0.633 44	0.546 46	0.550 58	0.857 57	0.789 58	0.853 44	0.762 38	0.987 54	0.699 39
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
Sparse R-CNN	0.714 57	1.000 1	0.926 48	0.694 62	0.699 58	0.890 32	0.636 34	0.516 67	0.693 58	0.743 29	0.588 52	0.369 63	0.601 50	0.594 71	0.800 54	0.886 19	0.676 59	0.986 55	0.546 60

Mask3D	0.870 19	1.000 1	0.985 19	0.782 50	0.818 23	0.938 19	0.760 20	0.749 40	0.923 10	0.877 17	0.760 12	0.785 15	0.820 2	1.000 1	0.912 22	0.864 39	0.878 12	0.983 56	0.825 11
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
MASC	0.615 67	0.711 74	0.802 69	0.540 73	0.757 41	0.777 59	0.029 77	0.577 64	0.588 68	0.521 68	0.600 49	0.436 60	0.534 60	0.697 67	0.616 68	0.838 51	0.526 68	0.980 57	0.534 62
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
UNet-backbone	0.605 68	1.000 1	0.909 54	0.764 54	0.603 66	0.704 69	0.415 58	0.301 73	0.548 70	0.461 71	0.394 69	0.267 65	0.386 71	0.857 57	0.649 67	0.817 57	0.504 70	0.959 58	0.356 73

R-PointNet	0.544 70	0.500 77	0.655 76	0.661 65	0.663 59	0.765 61	0.432 57	0.214 76	0.612 63	0.584 57	0.499 66	0.204 69	0.286 75	0.429 74	0.655 66	0.650 77	0.539 67	0.950 59	0.499 66

ClickSeg_Instance	0.685 62	1.000 1	0.818 68	0.600 70	0.715 55	0.795 57	0.557 42	0.533 65	0.591 67	0.601 52	0.519 63	0.429 61	0.638 46	0.938 52	0.706 64	0.817 58	0.624 65	0.944 60	0.502 65

INS-Conv-instance	0.762 48	1.000 1	0.923 49	0.765 53	0.785 32	0.905 27	0.600 36	0.655 53	0.646 61	0.683 38	0.647 41	0.530 50	0.650 33	1.000 1	0.824 49	0.830 53	0.693 56	0.944 60	0.644 46

DANCENET	0.786 41	1.000 1	0.936 40	0.783 48	0.737 49	0.852 46	0.742 23	0.647 54	0.765 49	0.811 24	0.624 46	0.579 41	0.632 47	1.000 1	0.909 34	0.898 13	0.696 53	0.944 60	0.601 56

PanopticFusion-inst	0.693 59	1.000 1	0.852 65	0.655 66	0.616 64	0.788 58	0.334 62	0.763 37	0.771 46	0.457 72	0.555 58	0.652 33	0.518 62	0.857 57	0.765 59	0.732 72	0.631 62	0.944 60	0.577 58
Gaku Narita, Takashi Seno, Tomoya Ishikawa, Yohsuke Kaji: PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things. IROS 2019 (to appear)
Mask3D_evaluation	0.843 26	1.000 1	0.955 35	0.847 25	0.795 29	0.932 22	0.750 22	0.780 34	0.891 19	0.818 22	0.737 22	0.633 38	0.703 23	1.000 1	0.902 35	0.870 32	0.820 31	0.941 64	0.805 18

Hier3D	0.540 71	1.000 1	0.727 71	0.626 68	0.467 75	0.693 70	0.200 72	0.412 69	0.480 74	0.528 66	0.318 73	0.077 77	0.600 51	0.688 68	0.382 72	0.768 70	0.472 72	0.941 64	0.350 74
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
3D-SIS	0.558 69	1.000 1	0.773 70	0.614 69	0.503 72	0.691 71	0.200 72	0.412 69	0.498 73	0.546 64	0.311 74	0.103 74	0.600 51	0.857 57	0.382 72	0.799 65	0.445 76	0.938 66	0.371 71
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
SALoss-ResNet	0.695 58	1.000 1	0.855 64	0.579 72	0.589 67	0.735 66	0.484 52	0.588 60	0.856 28	0.634 46	0.571 54	0.298 64	0.500 65	1.000 1	0.824 49	0.818 56	0.702 52	0.935 67	0.545 61
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)
Region-18class	0.497 72	0.250 79	0.902 55	0.689 63	0.540 70	0.747 62	0.276 68	0.610 57	0.268 78	0.489 69	0.348 70	0.000 78	0.243 78	0.220 77	0.663 65	0.814 59	0.459 74	0.928 68	0.496 67

3D-BoNet	0.687 61	1.000 1	0.887 61	0.836 31	0.587 68	0.643 75	0.550 46	0.620 56	0.724 53	0.522 67	0.501 65	0.243 67	0.512 63	1.000 1	0.751 61	0.807 62	0.661 61	0.909 69	0.612 54
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
Occipital-SCS	0.688 60	1.000 1	0.913 53	0.730 60	0.737 50	0.743 65	0.442 56	0.855 16	0.655 60	0.546 63	0.546 59	0.263 66	0.508 64	0.889 55	0.568 70	0.771 69	0.705 51	0.889 70	0.625 49

PCJC	0.684 63	1.000 1	0.895 60	0.757 56	0.659 61	0.862 42	0.189 74	0.739 43	0.606 64	0.712 35	0.581 53	0.515 53	0.650 33	0.857 57	0.357 75	0.785 67	0.631 63	0.889 70	0.635 48

ODIN - Ins	0.847 25	1.000 1	0.951 36	0.834 33	0.828 18	0.875 35	0.871 6	0.767 36	0.821 36	0.816 23	0.690 34	0.800 8	0.771 12	1.000 1	0.912 22	0.891 16	0.821 30	0.886 72	0.713 36
Ayush Jain, Pushkal Katara, Nikolaos Gkanatsios, Adam W. Harley, Gabriel Sarch, Kriti Aggarwal, Vishrav Chaudhary, Katerina Fragkiadaki: ODIN: A Single Model for 2D and 3D Segmentation. CVPR 2024
3D-BEVIS	0.401 77	0.667 75	0.687 75	0.419 78	0.137 79	0.587 77	0.188 75	0.235 74	0.359 76	0.211 78	0.093 79	0.080 75	0.311 74	0.571 72	0.382 72	0.754 71	0.300 78	0.874 73	0.357 72
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Sem_Recon_ins	0.484 73	0.764 73	0.608 78	0.470 75	0.521 71	0.637 76	0.311 63	0.218 75	0.348 77	0.365 76	0.223 75	0.222 68	0.258 76	0.629 70	0.734 62	0.596 78	0.509 69	0.858 74	0.444 68

SemRegionNet-20cls	0.470 75	1.000 1	0.727 71	0.447 76	0.481 73	0.678 72	0.024 78	0.380 71	0.518 71	0.440 73	0.339 71	0.128 72	0.350 72	0.429 74	0.212 78	0.711 74	0.465 73	0.833 75	0.290 77

tmp	0.474 74	1.000 1	0.727 71	0.433 77	0.481 74	0.673 73	0.022 79	0.380 71	0.517 72	0.436 74	0.338 72	0.128 72	0.343 73	0.429 74	0.291 77	0.728 73	0.473 71	0.833 75	0.300 76

Sgpn_scannet	0.390 78	0.556 76	0.636 77	0.493 74	0.353 77	0.539 78	0.271 70	0.160 78	0.450 75	0.359 77	0.178 77	0.146 71	0.250 77	0.143 78	0.347 76	0.698 75	0.436 77	0.667 77	0.331 75

ASIS	0.422 76	0.333 78	0.707 74	0.676 64	0.401 76	0.650 74	0.350 61	0.177 77	0.594 66	0.376 75	0.202 76	0.077 76	0.404 70	0.571 72	0.197 79	0.674 76	0.447 75	0.500 78	0.260 78

MaskRCNN 2d->3d Proj	0.261 79	0.903 72	0.081 79	0.008 79	0.233 78	0.175 79	0.280 66	0.106 79	0.150 79	0.203 79	0.175 78	0.480 55	0.218 79	0.143 78	0.542 71	0.404 79	0.153 79	0.393 79	0.049 79

3D Semantic Instance Benchmark