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 50%	bathtub	bed	bookshelf	cabinet	chair	counter	curtain	desk	door	otherfurniture	picture	refrigerator	shower curtain	sink	sofa	table	toilet	window

Volt-SPFormerScanNet	0.827 1	1.000 1	0.981 6	0.975 1	0.801 1	0.940 4	0.426 24	0.693 30	0.752 13	0.762 7	0.800 1	0.804 2	0.855 1	0.959 48	0.745 23	0.879 7	0.806 7	0.997 43	0.710 1
Kadir Yilmaz, Adrian Kruse, Tristan Höfer, Daan de Geus, Bastian Leibe: Volume Transformer: Revisiting Vanilla Transformers for 3D Scene Understanding.
Competitor-MAFT	0.816 2	1.000 1	0.983 4	0.872 12	0.718 6	0.941 3	0.588 5	0.652 42	0.819 3	0.776 3	0.720 7	0.780 7	0.769 12	1.000 1	0.797 11	0.813 32	0.798 9	1.000 1	0.659 5

PointRel	0.816 2	1.000 1	0.971 10	0.908 7	0.743 3	0.923 11	0.573 9	0.714 22	0.695 21	0.734 11	0.747 3	0.725 14	0.809 2	1.000 1	0.814 9	0.899 5	0.820 3	1.000 1	0.610 20
: Relation3D: Enhancing Relation Modeling for Point Cloud Instance Segmentation. CVPR 2025
Spherical Mask(CtoF)	0.812 4	1.000 1	0.973 9	0.852 16	0.718 7	0.917 13	0.574 7	0.677 32	0.748 14	0.729 15	0.715 10	0.795 4	0.809 2	1.000 1	0.831 4	0.854 13	0.787 13	1.000 1	0.638 9

PointComp	0.811 5	0.850 61	0.969 11	0.864 14	0.739 4	0.946 2	0.539 16	0.671 35	0.835 2	0.700 19	0.742 4	0.817 1	0.766 13	1.000 1	0.755 21	0.909 1	0.808 6	1.000 1	0.687 3

EV3D	0.811 5	1.000 1	0.968 12	0.852 16	0.717 8	0.921 12	0.574 8	0.677 32	0.748 14	0.730 14	0.703 16	0.795 4	0.809 2	1.000 1	0.831 4	0.854 13	0.778 17	1.000 1	0.638 10

VDG-Uni3DSeg	0.804 7	1.000 1	0.990 1	0.886 10	0.688 21	0.912 15	0.602 2	0.703 26	0.786 8	0.771 4	0.708 14	0.700 19	0.669 27	0.981 41	0.789 17	0.903 2	0.772 21	1.000 1	0.609 21

SIM3D	0.803 8	1.000 1	0.967 13	0.863 15	0.692 20	0.924 10	0.552 13	0.732 20	0.667 26	0.732 13	0.662 20	0.796 3	0.789 10	1.000 1	0.803 10	0.864 10	0.766 24	1.000 1	0.643 7

OneFormer3D	0.801 9	1.000 1	0.973 8	0.909 6	0.698 16	0.928 8	0.582 6	0.668 38	0.685 22	0.780 2	0.687 18	0.698 23	0.702 16	1.000 1	0.794 13	0.900 4	0.784 15	0.986 56	0.635 11
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
Competitor-SPFormer	0.800 10	1.000 1	0.986 3	0.845 18	0.705 14	0.915 14	0.532 17	0.733 19	0.757 12	0.733 12	0.708 13	0.698 22	0.648 39	0.981 41	0.890 1	0.830 23	0.796 10	0.997 43	0.644 6

InsSSM	0.799 11	1.000 1	0.915 16	0.710 45	0.729 5	0.925 9	0.664 1	0.670 36	0.770 9	0.766 5	0.739 5	0.737 10	0.700 17	1.000 1	0.792 14	0.829 25	0.815 4	0.997 43	0.625 13
Lei Yao, Yi Wang, Moyun Liu, Lap-Pui Chau: SGIFormer: Semantic-guided and Geometric-enhanced Interleaving Transformer for 3D Instance Segmentation. TCSVT, 2024
DCD	0.798 12	1.000 1	0.878 24	0.792 30	0.693 19	0.936 5	0.596 3	0.685 31	0.663 28	0.736 9	0.717 8	0.788 6	0.693 22	1.000 1	0.825 7	0.840 19	0.837 1	1.000 1	0.689 2

TST3D	0.795 13	1.000 1	0.929 15	0.918 5	0.709 11	0.884 24	0.596 4	0.704 25	0.769 10	0.734 10	0.644 25	0.699 21	0.751 14	1.000 1	0.794 12	0.876 9	0.757 27	0.997 43	0.550 37
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
MG-Former	0.791 14	1.000 1	0.980 7	0.837 21	0.626 30	0.897 17	0.543 15	0.759 14	0.800 7	0.766 6	0.659 21	0.769 9	0.697 20	1.000 1	0.791 15	0.707 53	0.791 12	1.000 1	0.610 19

ExtMask3D	0.789 15	1.000 1	0.988 2	0.756 38	0.706 13	0.912 16	0.429 23	0.647 44	0.806 6	0.755 8	0.673 19	0.689 24	0.772 11	1.000 1	0.789 16	0.852 15	0.811 5	1.000 1	0.617 16

UniPerception	0.787 16	1.000 1	0.909 17	0.768 35	0.687 22	0.947 1	0.551 14	0.714 21	0.843 1	0.696 20	0.713 12	0.773 8	0.607 45	0.981 41	0.690 30	0.878 8	0.775 20	1.000 1	0.640 8

Queryformer	0.787 16	1.000 1	0.933 14	0.601 55	0.754 2	0.886 22	0.558 12	0.661 40	0.767 11	0.665 23	0.716 9	0.639 30	0.808 6	1.000 1	0.844 3	0.897 6	0.804 8	1.000 1	0.624 14

MAFT	0.786 18	1.000 1	0.894 22	0.807 25	0.694 18	0.893 20	0.486 19	0.674 34	0.740 16	0.786 1	0.704 15	0.727 13	0.739 15	1.000 1	0.707 28	0.849 17	0.756 28	1.000 1	0.685 4

KmaxOneFormerNet	0.783 19	0.903 59	0.981 5	0.794 29	0.706 12	0.931 7	0.561 11	0.701 27	0.706 19	0.727 16	0.697 17	0.731 12	0.689 24	1.000 1	0.856 2	0.750 44	0.761 26	1.000 1	0.599 25

Mask3D	0.780 20	1.000 1	0.786 48	0.716 43	0.696 17	0.885 23	0.500 18	0.714 22	0.810 5	0.672 22	0.715 10	0.679 25	0.809 2	1.000 1	0.831 4	0.833 22	0.787 13	1.000 1	0.602 23
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
SPFormer	0.770 21	0.903 59	0.903 19	0.806 26	0.609 37	0.886 21	0.568 10	0.815 6	0.705 20	0.711 17	0.655 22	0.652 29	0.685 25	1.000 1	0.789 18	0.809 33	0.776 19	1.000 1	0.583 29
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
SoftGroup++	0.769 22	1.000 1	0.803 41	0.937 2	0.684 23	0.865 26	0.213 40	0.870 2	0.664 27	0.571 30	0.758 2	0.702 18	0.807 7	1.000 1	0.653 36	0.902 3	0.792 11	1.000 1	0.626 12

SoftGroup	0.761 23	1.000 1	0.808 37	0.845 18	0.716 9	0.862 28	0.243 37	0.824 4	0.655 30	0.620 24	0.734 6	0.699 20	0.791 9	0.981 41	0.716 25	0.844 18	0.769 22	1.000 1	0.594 27
Thang Vu, Kookhoi Kim, Tung M. Luu, Xuan Thanh Nguyen, Chang D. Yoo: SoftGroup for 3D Instance Segmentaiton on Point Clouds. CVPR 2022 [Oral]
ISBNet	0.757 24	1.000 1	0.904 18	0.731 41	0.678 24	0.895 18	0.458 21	0.644 46	0.670 25	0.710 18	0.620 30	0.732 11	0.650 29	1.000 1	0.756 20	0.778 36	0.779 16	1.000 1	0.614 17
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
TD3D	0.751 25	1.000 1	0.774 49	0.867 13	0.621 32	0.934 6	0.404 25	0.706 24	0.812 4	0.605 27	0.633 28	0.626 31	0.690 23	1.000 1	0.640 38	0.820 28	0.777 18	1.000 1	0.612 18
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
PBNet	0.747 26	1.000 1	0.818 33	0.837 22	0.713 10	0.844 30	0.457 22	0.647 44	0.711 18	0.614 25	0.617 32	0.657 28	0.650 29	1.000 1	0.692 29	0.822 27	0.765 25	1.000 1	0.595 26
Weiguang Zhao, Yuyao Yan, Chaolong Yang, Jianan Ye, Xi Yang, Kaizhu Huang: Divide and Conquer: 3D Instance Segmentation With Point-Wise Binarization. ICCV 2023
GraphCut	0.732 27	1.000 1	0.788 46	0.724 42	0.642 29	0.859 29	0.248 36	0.787 11	0.618 33	0.596 28	0.653 24	0.722 16	0.583 52	1.000 1	0.766 19	0.861 11	0.825 2	1.000 1	0.504 43

IPCA-Inst	0.731 28	1.000 1	0.788 47	0.884 11	0.698 15	0.788 46	0.252 35	0.760 13	0.646 31	0.511 38	0.637 27	0.665 27	0.804 8	1.000 1	0.644 37	0.778 37	0.747 30	1.000 1	0.561 33

TopoSeg	0.725 29	1.000 1	0.806 40	0.933 3	0.668 26	0.758 51	0.272 34	0.734 18	0.630 32	0.549 34	0.654 23	0.606 32	0.697 21	0.966 47	0.612 42	0.839 20	0.754 29	1.000 1	0.573 30

DKNet	0.718 30	1.000 1	0.814 34	0.782 31	0.619 34	0.872 25	0.224 38	0.751 16	0.569 37	0.677 21	0.585 37	0.724 15	0.633 41	0.981 41	0.515 52	0.819 29	0.736 31	1.000 1	0.617 15
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
SSEC	0.707 31	1.000 1	0.850 26	0.924 4	0.648 27	0.747 54	0.162 42	0.862 3	0.572 36	0.520 36	0.624 29	0.549 35	0.649 38	1.000 1	0.560 47	0.706 54	0.768 23	1.000 1	0.591 28

HAIS	0.699 32	1.000 1	0.849 27	0.820 23	0.675 25	0.808 40	0.279 32	0.757 15	0.465 43	0.517 37	0.596 34	0.559 34	0.600 46	1.000 1	0.654 35	0.767 39	0.676 35	0.994 52	0.560 34
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
SSTNet	0.698 33	1.000 1	0.697 65	0.888 9	0.556 44	0.803 41	0.387 26	0.626 48	0.417 48	0.556 33	0.585 38	0.702 17	0.600 46	1.000 1	0.824 8	0.720 52	0.692 33	1.000 1	0.509 42
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
DualGroup	0.694 34	1.000 1	0.799 43	0.811 24	0.622 31	0.817 35	0.376 27	0.805 9	0.590 35	0.487 42	0.568 41	0.525 39	0.650 29	0.835 60	0.600 43	0.829 24	0.655 38	1.000 1	0.526 39

ODIN - Ins	0.693 35	1.000 1	0.880 23	0.647 50	0.620 33	0.779 48	0.336 29	0.501 63	0.681 23	0.577 29	0.595 35	0.679 26	0.683 26	1.000 1	0.709 27	0.816 31	0.637 42	0.770 72	0.557 35
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
SphereSeg	0.680 36	1.000 1	0.856 25	0.744 39	0.618 35	0.893 19	0.151 43	0.651 43	0.713 17	0.537 35	0.579 40	0.430 49	0.651 28	1.000 1	0.389 63	0.744 47	0.697 32	0.991 54	0.601 24

DANCENET	0.680 36	1.000 1	0.807 38	0.733 40	0.600 38	0.768 50	0.375 28	0.543 56	0.538 38	0.610 26	0.599 33	0.498 40	0.632 43	0.981 41	0.739 24	0.856 12	0.633 45	0.882 67	0.454 52

Box2Mask	0.677 38	1.000 1	0.847 28	0.771 33	0.509 53	0.816 36	0.277 33	0.558 55	0.482 40	0.562 32	0.640 26	0.448 45	0.700 17	1.000 1	0.666 31	0.852 16	0.578 52	0.997 43	0.488 47
Julian Chibane, Francis Engelmann, Tuan Anh Tran, Gerard Pons-Moll: Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes. ECCV 2022
OccuSeg+instance	0.672 39	1.000 1	0.758 57	0.682 47	0.576 42	0.842 31	0.477 20	0.504 62	0.524 39	0.567 31	0.585 39	0.451 44	0.557 54	1.000 1	0.751 22	0.797 34	0.563 55	1.000 1	0.467 51
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
Mask-Group	0.664 40	1.000 1	0.822 32	0.764 37	0.616 36	0.815 37	0.139 47	0.694 29	0.597 34	0.459 46	0.566 42	0.599 33	0.600 46	0.516 70	0.715 26	0.819 30	0.635 43	1.000 1	0.603 22
Min Zhong, Xinghao Chen, Xiaokang Chen, Gang Zeng, Yunhe Wang: MaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation. ICME 2022
INS-Conv-instance	0.657 41	1.000 1	0.760 55	0.667 49	0.581 40	0.863 27	0.323 30	0.655 41	0.477 41	0.473 44	0.549 44	0.432 48	0.650 29	1.000 1	0.655 34	0.738 48	0.585 51	0.944 59	0.472 50

CSC-Pretrained	0.648 42	1.000 1	0.810 35	0.768 34	0.523 51	0.813 38	0.143 46	0.819 5	0.389 51	0.422 55	0.511 48	0.443 46	0.650 29	1.000 1	0.624 40	0.732 49	0.634 44	1.000 1	0.375 59

PE	0.645 43	1.000 1	0.773 51	0.798 28	0.538 46	0.786 47	0.088 55	0.799 10	0.350 55	0.435 53	0.547 45	0.545 36	0.646 40	0.933 50	0.562 46	0.761 42	0.556 60	0.997 43	0.501 45
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
RPGN	0.643 44	1.000 1	0.758 56	0.582 61	0.539 45	0.826 34	0.046 60	0.765 12	0.372 53	0.436 52	0.588 36	0.539 38	0.650 29	1.000 1	0.577 44	0.750 45	0.653 40	0.997 43	0.495 46
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
Dyco3D	0.641 45	1.000 1	0.841 29	0.893 8	0.531 48	0.802 42	0.115 52	0.588 53	0.448 45	0.438 50	0.537 47	0.430 50	0.550 55	0.857 52	0.534 50	0.764 41	0.657 37	0.987 55	0.568 31
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
GICN	0.638 46	1.000 1	0.895 21	0.800 27	0.480 57	0.676 59	0.144 45	0.737 17	0.354 54	0.447 47	0.400 61	0.365 56	0.700 17	1.000 1	0.569 45	0.836 21	0.599 47	1.000 1	0.473 49

PointGroup	0.636 47	1.000 1	0.765 52	0.624 52	0.505 55	0.797 43	0.116 51	0.696 28	0.384 52	0.441 48	0.559 43	0.476 42	0.596 49	1.000 1	0.666 31	0.756 43	0.556 59	0.997 43	0.513 41
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]
DD-UNet+Group	0.635 48	0.667 63	0.797 45	0.714 44	0.562 43	0.774 49	0.146 44	0.810 8	0.429 47	0.476 43	0.546 46	0.399 52	0.633 41	1.000 1	0.632 39	0.722 51	0.609 46	1.000 1	0.514 40
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
Mask3D_evaluation	0.631 49	1.000 1	0.829 31	0.606 54	0.646 28	0.836 32	0.068 56	0.511 60	0.462 44	0.507 39	0.619 31	0.389 54	0.610 44	1.000 1	0.432 58	0.828 26	0.673 36	0.788 71	0.552 36

DENet	0.629 50	1.000 1	0.797 44	0.608 53	0.589 39	0.627 63	0.219 39	0.882 1	0.310 57	0.402 60	0.383 63	0.396 53	0.650 29	1.000 1	0.663 33	0.543 71	0.691 34	1.000 1	0.568 32

3D-MPA	0.611 51	1.000 1	0.833 30	0.765 36	0.526 50	0.756 52	0.136 49	0.588 53	0.470 42	0.438 51	0.432 57	0.358 58	0.650 29	0.857 52	0.429 59	0.765 40	0.557 58	1.000 1	0.430 54
Francis Engelmann, Martin Bokeloh, Alireza Fathi, Bastian Leibe, Matthias Nießner: 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation. CVPR 2020
OSIS	0.605 52	1.000 1	0.801 42	0.599 56	0.535 47	0.728 56	0.286 31	0.436 67	0.679 24	0.491 40	0.433 55	0.256 60	0.404 67	0.857 52	0.620 41	0.724 50	0.510 65	1.000 1	0.539 38

AOIA	0.601 53	1.000 1	0.761 54	0.687 46	0.485 56	0.828 33	0.008 67	0.663 39	0.405 50	0.405 59	0.425 58	0.490 41	0.596 49	0.714 63	0.553 49	0.779 35	0.597 48	0.992 53	0.424 56

PCJC	0.578 54	1.000 1	0.810 36	0.583 60	0.449 60	0.813 39	0.042 61	0.603 51	0.341 56	0.490 41	0.465 52	0.410 51	0.650 29	0.835 60	0.264 69	0.694 58	0.561 56	0.889 64	0.504 44

SSEN	0.575 55	1.000 1	0.761 53	0.473 63	0.477 58	0.795 44	0.066 57	0.529 58	0.658 29	0.460 45	0.461 53	0.380 55	0.331 69	0.859 51	0.401 62	0.692 60	0.653 39	1.000 1	0.348 61
Dongsu Zhang, Junha Chun, Sang Kyun Cha, Young Min Kim: Spatial Semantic Embedding Network: Fast 3D Instance Segmentation with Deep Metric Learning. Arxiv
RWSeg	0.567 56	0.528 73	0.708 64	0.626 51	0.580 41	0.745 55	0.063 58	0.627 47	0.240 61	0.400 61	0.497 49	0.464 43	0.515 56	1.000 1	0.475 54	0.745 46	0.571 53	1.000 1	0.429 55

NeuralBF	0.555 57	0.667 63	0.896 20	0.843 20	0.517 52	0.751 53	0.029 62	0.519 59	0.414 49	0.439 49	0.465 51	0.000 79	0.484 58	0.857 52	0.287 67	0.693 59	0.651 41	1.000 1	0.485 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
MTML	0.549 58	1.000 1	0.807 39	0.588 59	0.327 65	0.647 61	0.004 69	0.815 7	0.180 64	0.418 56	0.364 65	0.182 63	0.445 61	1.000 1	0.442 57	0.688 61	0.571 54	1.000 1	0.396 57
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
ClickSeg_Instance	0.539 59	1.000 1	0.621 68	0.300 66	0.530 49	0.698 57	0.127 50	0.533 57	0.222 62	0.430 54	0.400 60	0.365 56	0.574 53	0.938 49	0.472 55	0.659 63	0.543 61	0.944 59	0.347 62

One_Thing_One_Click	0.529 60	0.667 63	0.718 60	0.777 32	0.399 61	0.683 58	0.000 72	0.669 37	0.138 67	0.391 62	0.374 64	0.539 37	0.360 68	0.641 67	0.556 48	0.774 38	0.593 49	0.997 43	0.251 67
Zhengzhe Liu, Xiaojuan Qi, Chi-Wing Fu: One Thing One Click: A Self-Training Approach for Weakly Supervised 3D Semantic Segmentation. CVPR 2021
Sparse R-CNN	0.515 61	1.000 1	0.538 73	0.282 67	0.468 59	0.790 45	0.173 41	0.345 69	0.429 46	0.413 58	0.484 50	0.176 64	0.595 51	0.591 68	0.522 51	0.668 62	0.476 66	0.986 57	0.327 63

Occipital-SCS	0.512 62	1.000 1	0.716 61	0.509 62	0.506 54	0.611 64	0.092 54	0.602 52	0.177 65	0.346 65	0.383 62	0.165 65	0.442 62	0.850 59	0.386 64	0.618 67	0.543 62	0.889 64	0.389 58

3D-BoNet	0.488 63	1.000 1	0.672 67	0.590 58	0.301 67	0.484 74	0.098 53	0.620 49	0.306 58	0.341 66	0.259 69	0.125 67	0.434 64	0.796 62	0.402 61	0.499 73	0.513 64	0.909 63	0.439 53
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
PanopticFusion-inst	0.478 64	0.667 63	0.712 63	0.595 57	0.259 70	0.550 70	0.000 72	0.613 50	0.175 66	0.250 71	0.434 54	0.437 47	0.411 66	0.857 52	0.485 53	0.591 70	0.267 76	0.944 59	0.359 60
Gaku Narita, Takashi Seno, Tomoya Ishikawa, Yohsuke Kaji: PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things. IROS 2019 (to appear)
SPG_WSIS	0.470 65	0.667 63	0.685 66	0.677 48	0.372 63	0.562 68	0.000 72	0.482 64	0.244 60	0.316 68	0.298 66	0.052 74	0.442 63	0.857 52	0.267 68	0.702 55	0.559 57	1.000 1	0.287 65

SALoss-ResNet	0.459 66	1.000 1	0.737 59	0.159 77	0.259 69	0.587 66	0.138 48	0.475 65	0.217 63	0.416 57	0.408 59	0.128 66	0.315 70	0.714 63	0.411 60	0.536 72	0.590 50	0.873 68	0.304 64
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)
MASC	0.447 67	0.528 73	0.555 71	0.381 64	0.382 62	0.633 62	0.002 70	0.509 61	0.260 59	0.361 64	0.432 56	0.327 59	0.451 60	0.571 69	0.367 65	0.639 65	0.386 67	0.980 58	0.276 66
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
SegGroup_ins	0.445 68	0.667 63	0.773 50	0.185 74	0.317 66	0.656 60	0.000 72	0.407 68	0.134 68	0.381 63	0.267 68	0.217 62	0.476 59	0.714 63	0.452 56	0.629 66	0.514 63	1.000 1	0.222 70
An Tao, Yueqi Duan, Yi Wei, Jiwen Lu, Jie Zhou: SegGroup: Seg-Level Supervision for 3D Instance and Semantic Segmentation. TIP 2022
3D-SIS	0.382 69	1.000 1	0.432 76	0.245 69	0.190 71	0.577 67	0.013 66	0.263 71	0.033 74	0.320 67	0.240 70	0.075 70	0.422 65	0.857 52	0.117 74	0.699 56	0.271 75	0.883 66	0.235 69
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
Hier3D	0.323 70	0.667 63	0.542 72	0.264 68	0.157 74	0.550 69	0.000 72	0.205 74	0.009 76	0.270 70	0.218 71	0.075 70	0.500 57	0.688 66	0.007 80	0.698 57	0.301 72	0.459 77	0.200 71
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
UNet-backbone	0.319 71	0.667 63	0.715 62	0.233 70	0.189 72	0.479 75	0.008 67	0.218 72	0.067 73	0.201 73	0.173 72	0.107 68	0.123 75	0.438 71	0.150 71	0.615 68	0.355 68	0.916 62	0.093 79

R-PointNet	0.306 72	0.500 75	0.405 77	0.311 65	0.348 64	0.589 65	0.054 59	0.068 77	0.126 69	0.283 69	0.290 67	0.028 75	0.219 73	0.214 74	0.331 66	0.396 77	0.275 73	0.821 70	0.245 68

Region-18class	0.284 73	0.250 79	0.751 58	0.228 72	0.270 68	0.521 71	0.000 72	0.468 66	0.008 78	0.205 72	0.127 73	0.000 79	0.068 77	0.070 78	0.262 70	0.652 64	0.323 70	0.740 73	0.173 72

SemRegionNet-20cls	0.250 74	0.333 76	0.613 69	0.229 71	0.163 73	0.493 72	0.000 72	0.304 70	0.107 70	0.147 76	0.100 75	0.052 73	0.231 71	0.119 76	0.039 76	0.445 75	0.325 69	0.654 74	0.141 75

tmp	0.248 75	0.667 63	0.437 75	0.188 73	0.153 75	0.491 73	0.000 72	0.208 73	0.094 72	0.153 75	0.099 76	0.057 72	0.217 74	0.119 76	0.039 76	0.466 74	0.302 71	0.640 75	0.140 76

3D-BEVIS	0.248 75	0.667 63	0.566 70	0.076 78	0.035 80	0.394 78	0.027 64	0.035 79	0.098 71	0.099 78	0.030 79	0.025 76	0.098 76	0.375 73	0.126 73	0.604 69	0.181 78	0.854 69	0.171 73
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Sem_Recon_ins	0.227 77	0.764 62	0.486 74	0.069 79	0.098 77	0.426 77	0.017 65	0.067 78	0.015 75	0.172 74	0.100 74	0.096 69	0.054 79	0.183 75	0.135 72	0.366 78	0.260 77	0.614 76	0.168 74

ASIS	0.199 78	0.333 76	0.253 79	0.167 76	0.140 76	0.438 76	0.000 72	0.177 75	0.008 77	0.121 77	0.069 77	0.004 78	0.231 72	0.429 72	0.036 78	0.445 76	0.273 74	0.333 79	0.119 78

Sgpn_scannet	0.143 79	0.208 80	0.390 78	0.169 75	0.065 78	0.275 79	0.029 63	0.069 76	0.000 79	0.087 79	0.043 78	0.014 77	0.027 80	0.000 79	0.112 75	0.351 79	0.168 79	0.438 78	0.138 77

MaskRCNN 2d->3d Proj	0.058 80	0.333 76	0.002 80	0.000 80	0.053 79	0.002 80	0.002 71	0.021 80	0.000 79	0.045 80	0.024 80	0.238 61	0.065 78	0.000 79	0.014 79	0.107 80	0.020 80	0.110 80	0.006 80

3D Semantic Instance Benchmark