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

Competitor-MAFT	0.816 1	1.000 1	0.983 3	0.872 10	0.718 5	0.941 1	0.588 4	0.652 39	0.819 2	0.776 3	0.720 5	0.780 5	0.769 12	1.000 1	0.797 11	0.813 29	0.798 8	1.000 1	0.659 4

PointRel	0.816 1	1.000 1	0.971 8	0.908 6	0.743 2	0.923 8	0.573 8	0.714 22	0.695 18	0.734 10	0.747 2	0.725 12	0.809 1	1.000 1	0.814 9	0.899 3	0.820 4	1.000 1	0.610 18
: Relation3D (PointRel): Enhancing Relation Modeling for Point Cloud Instance Segmentation.
Spherical Mask(CtoF)	0.812 3	1.000 1	0.973 7	0.852 14	0.718 6	0.917 10	0.574 6	0.677 30	0.748 11	0.729 14	0.715 8	0.795 2	0.809 1	1.000 1	0.831 4	0.854 9	0.787 12	1.000 1	0.638 7

EV3D	0.811 4	1.000 1	0.968 9	0.852 14	0.717 7	0.921 9	0.574 7	0.677 30	0.748 11	0.730 13	0.703 13	0.795 2	0.809 1	1.000 1	0.831 4	0.854 9	0.778 16	1.000 1	0.638 8

SIM3D	0.803 5	1.000 1	0.967 10	0.863 13	0.692 19	0.924 7	0.552 12	0.732 21	0.667 23	0.732 12	0.662 17	0.796 1	0.789 9	1.000 1	0.803 10	0.864 6	0.766 21	1.000 1	0.643 6

OneFormer3D	0.801 6	1.000 1	0.973 6	0.909 5	0.698 15	0.928 5	0.582 5	0.668 35	0.685 19	0.780 2	0.687 15	0.698 20	0.702 15	1.000 1	0.794 13	0.900 2	0.784 14	0.986 53	0.635 9
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
UniPerception	0.800 7	1.000 1	0.930 12	0.872 10	0.727 4	0.862 25	0.454 20	0.764 13	0.820 1	0.746 7	0.706 11	0.750 7	0.772 10	0.926 47	0.764 19	0.818 27	0.826 2	0.997 40	0.660 3

Competitor-SPFormer	0.800 7	1.000 1	0.986 2	0.845 16	0.705 13	0.915 11	0.532 14	0.733 20	0.757 10	0.733 11	0.708 10	0.698 19	0.648 37	0.981 40	0.890 1	0.830 19	0.796 9	0.997 40	0.644 5

InsSSM	0.799 9	1.000 1	0.915 14	0.710 42	0.729 3	0.925 6	0.664 1	0.670 33	0.770 7	0.766 4	0.739 3	0.737 8	0.700 16	1.000 1	0.792 14	0.829 21	0.815 5	0.997 40	0.625 11
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 10	1.000 1	0.878 21	0.792 28	0.693 18	0.936 2	0.596 2	0.685 29	0.663 25	0.736 8	0.717 6	0.788 4	0.693 21	1.000 1	0.825 7	0.840 15	0.837 1	1.000 1	0.689 1

TST3D	0.795 11	1.000 1	0.929 13	0.918 4	0.709 10	0.884 20	0.596 3	0.704 25	0.769 8	0.734 9	0.644 22	0.699 18	0.751 13	1.000 1	0.794 12	0.876 5	0.757 24	0.997 40	0.550 34
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
MG-Former	0.791 12	1.000 1	0.980 5	0.837 19	0.626 27	0.897 13	0.543 13	0.759 15	0.800 6	0.766 5	0.659 18	0.769 6	0.697 19	1.000 1	0.791 15	0.707 50	0.791 11	1.000 1	0.610 17

ExtMask3D	0.789 13	1.000 1	0.988 1	0.756 35	0.706 12	0.912 12	0.429 21	0.647 41	0.806 5	0.755 6	0.673 16	0.689 21	0.772 11	1.000 1	0.789 16	0.852 11	0.811 6	1.000 1	0.617 14

Queryformer	0.787 14	1.000 1	0.933 11	0.601 52	0.754 1	0.886 18	0.558 11	0.661 37	0.767 9	0.665 20	0.716 7	0.639 27	0.808 5	1.000 1	0.844 3	0.897 4	0.804 7	1.000 1	0.624 12

MAFT	0.786 15	1.000 1	0.894 19	0.807 23	0.694 17	0.893 16	0.486 16	0.674 32	0.740 13	0.786 1	0.704 12	0.727 11	0.739 14	1.000 1	0.707 26	0.849 13	0.756 25	1.000 1	0.685 2

KmaxOneFormerNet	0.783 16	0.903 57	0.981 4	0.794 27	0.706 11	0.931 4	0.561 10	0.701 26	0.706 16	0.727 15	0.697 14	0.731 10	0.689 23	1.000 1	0.856 2	0.750 41	0.761 23	1.000 1	0.599 22

Mask3D	0.780 17	1.000 1	0.786 45	0.716 40	0.696 16	0.885 19	0.500 15	0.714 22	0.810 4	0.672 19	0.715 8	0.679 22	0.809 1	1.000 1	0.831 4	0.833 18	0.787 12	1.000 1	0.602 20
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
SPFormer	0.770 18	0.903 57	0.903 16	0.806 24	0.609 34	0.886 17	0.568 9	0.815 6	0.705 17	0.711 16	0.655 19	0.652 26	0.685 24	1.000 1	0.789 17	0.809 30	0.776 18	1.000 1	0.583 26
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
SoftGroup++	0.769 19	1.000 1	0.803 38	0.937 1	0.684 20	0.865 22	0.213 37	0.870 2	0.664 24	0.571 27	0.758 1	0.702 16	0.807 6	1.000 1	0.653 33	0.902 1	0.792 10	1.000 1	0.626 10

SoftGroup	0.761 20	1.000 1	0.808 34	0.845 16	0.716 8	0.862 24	0.243 34	0.824 4	0.655 27	0.620 21	0.734 4	0.699 17	0.791 8	0.981 40	0.716 23	0.844 14	0.769 19	1.000 1	0.594 24
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 21	1.000 1	0.904 15	0.731 38	0.678 21	0.895 14	0.458 18	0.644 43	0.670 22	0.710 17	0.620 27	0.732 9	0.650 27	1.000 1	0.756 20	0.778 33	0.779 15	1.000 1	0.614 15
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 22	1.000 1	0.774 46	0.867 12	0.621 29	0.934 3	0.404 22	0.706 24	0.812 3	0.605 24	0.633 25	0.626 28	0.690 22	1.000 1	0.640 35	0.820 24	0.777 17	1.000 1	0.612 16
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
PBNet	0.747 23	1.000 1	0.818 30	0.837 20	0.713 9	0.844 27	0.457 19	0.647 41	0.711 15	0.614 22	0.617 29	0.657 25	0.650 27	1.000 1	0.692 27	0.822 23	0.765 22	1.000 1	0.595 23
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 24	1.000 1	0.788 43	0.724 39	0.642 26	0.859 26	0.248 33	0.787 11	0.618 30	0.596 25	0.653 21	0.722 14	0.583 49	1.000 1	0.766 18	0.861 7	0.825 3	1.000 1	0.504 40

IPCA-Inst	0.731 25	1.000 1	0.788 44	0.884 9	0.698 14	0.788 43	0.252 32	0.760 14	0.646 28	0.511 35	0.637 24	0.665 24	0.804 7	1.000 1	0.644 34	0.778 34	0.747 27	1.000 1	0.561 30

TopoSeg	0.725 26	1.000 1	0.806 37	0.933 2	0.668 23	0.758 48	0.272 31	0.734 19	0.630 29	0.549 31	0.654 20	0.606 29	0.697 20	0.966 44	0.612 39	0.839 16	0.754 26	1.000 1	0.573 27

DKNet	0.718 27	1.000 1	0.814 31	0.782 29	0.619 31	0.872 21	0.224 35	0.751 17	0.569 34	0.677 18	0.585 34	0.724 13	0.633 39	0.981 40	0.515 49	0.819 25	0.736 28	1.000 1	0.617 13
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
SSEC	0.707 28	1.000 1	0.850 23	0.924 3	0.648 24	0.747 51	0.162 39	0.862 3	0.572 33	0.520 33	0.624 26	0.549 32	0.649 36	1.000 1	0.560 44	0.706 51	0.768 20	1.000 1	0.591 25

HAIS	0.699 29	1.000 1	0.849 24	0.820 21	0.675 22	0.808 37	0.279 29	0.757 16	0.465 40	0.517 34	0.596 31	0.559 31	0.600 43	1.000 1	0.654 32	0.767 36	0.676 32	0.994 49	0.560 31
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
SSTNet	0.698 30	1.000 1	0.697 62	0.888 8	0.556 41	0.803 38	0.387 23	0.626 45	0.417 45	0.556 30	0.585 35	0.702 15	0.600 43	1.000 1	0.824 8	0.720 49	0.692 30	1.000 1	0.509 39
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
DualGroup	0.694 31	1.000 1	0.799 40	0.811 22	0.622 28	0.817 32	0.376 24	0.805 9	0.590 32	0.487 39	0.568 38	0.525 36	0.650 27	0.835 57	0.600 40	0.829 20	0.655 35	1.000 1	0.526 36

ODIN - Ins	0.693 32	1.000 1	0.880 20	0.647 47	0.620 30	0.779 45	0.336 26	0.501 60	0.681 20	0.577 26	0.595 32	0.679 23	0.683 25	1.000 1	0.709 25	0.816 28	0.637 39	0.770 69	0.557 32
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
DANCENET	0.680 33	1.000 1	0.807 35	0.733 37	0.600 35	0.768 47	0.375 25	0.543 53	0.538 35	0.610 23	0.599 30	0.498 37	0.632 41	0.981 40	0.739 22	0.856 8	0.633 42	0.882 64	0.454 49

SphereSeg	0.680 33	1.000 1	0.856 22	0.744 36	0.618 32	0.893 15	0.151 40	0.651 40	0.713 14	0.537 32	0.579 37	0.430 46	0.651 26	1.000 1	0.389 60	0.744 44	0.697 29	0.991 51	0.601 21

Box2Mask	0.677 35	1.000 1	0.847 25	0.771 31	0.509 50	0.816 33	0.277 30	0.558 52	0.482 37	0.562 29	0.640 23	0.448 42	0.700 16	1.000 1	0.666 28	0.852 12	0.578 49	0.997 40	0.488 44
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 36	1.000 1	0.758 54	0.682 44	0.576 39	0.842 28	0.477 17	0.504 59	0.524 36	0.567 28	0.585 36	0.451 41	0.557 51	1.000 1	0.751 21	0.797 31	0.563 52	1.000 1	0.467 48
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
Mask-Group	0.664 37	1.000 1	0.822 29	0.764 34	0.616 33	0.815 34	0.139 44	0.694 28	0.597 31	0.459 43	0.566 39	0.599 30	0.600 43	0.516 67	0.715 24	0.819 26	0.635 40	1.000 1	0.603 19
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 38	1.000 1	0.760 52	0.667 46	0.581 37	0.863 23	0.323 27	0.655 38	0.477 38	0.473 41	0.549 41	0.432 45	0.650 27	1.000 1	0.655 31	0.738 45	0.585 48	0.944 56	0.472 47

CSC-Pretrained	0.648 39	1.000 1	0.810 32	0.768 32	0.523 48	0.813 35	0.143 43	0.819 5	0.389 48	0.422 52	0.511 45	0.443 43	0.650 27	1.000 1	0.624 37	0.732 46	0.634 41	1.000 1	0.375 56

PE	0.645 40	1.000 1	0.773 48	0.798 26	0.538 43	0.786 44	0.088 52	0.799 10	0.350 52	0.435 50	0.547 42	0.545 33	0.646 38	0.933 46	0.562 43	0.761 39	0.556 57	0.997 40	0.501 42
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
RPGN	0.643 41	1.000 1	0.758 53	0.582 58	0.539 42	0.826 31	0.046 57	0.765 12	0.372 50	0.436 49	0.588 33	0.539 35	0.650 27	1.000 1	0.577 41	0.750 42	0.653 37	0.997 40	0.495 43
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
Dyco3D	0.641 42	1.000 1	0.841 26	0.893 7	0.531 45	0.802 39	0.115 49	0.588 50	0.448 42	0.438 47	0.537 44	0.430 47	0.550 52	0.857 49	0.534 47	0.764 38	0.657 34	0.987 52	0.568 28
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
GICN	0.638 43	1.000 1	0.895 18	0.800 25	0.480 54	0.676 56	0.144 42	0.737 18	0.354 51	0.447 44	0.400 58	0.365 53	0.700 16	1.000 1	0.569 42	0.836 17	0.599 44	1.000 1	0.473 46

PointGroup	0.636 44	1.000 1	0.765 49	0.624 49	0.505 52	0.797 40	0.116 48	0.696 27	0.384 49	0.441 45	0.559 40	0.476 39	0.596 46	1.000 1	0.666 28	0.756 40	0.556 56	0.997 40	0.513 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]
DD-UNet+Group	0.635 45	0.667 60	0.797 42	0.714 41	0.562 40	0.774 46	0.146 41	0.810 8	0.429 44	0.476 40	0.546 43	0.399 49	0.633 39	1.000 1	0.632 36	0.722 48	0.609 43	1.000 1	0.514 37
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 46	1.000 1	0.829 28	0.606 51	0.646 25	0.836 29	0.068 53	0.511 57	0.462 41	0.507 36	0.619 28	0.389 51	0.610 42	1.000 1	0.432 55	0.828 22	0.673 33	0.788 68	0.552 33

DENet	0.629 47	1.000 1	0.797 41	0.608 50	0.589 36	0.627 60	0.219 36	0.882 1	0.310 54	0.402 57	0.383 60	0.396 50	0.650 27	1.000 1	0.663 30	0.543 68	0.691 31	1.000 1	0.568 29

3D-MPA	0.611 48	1.000 1	0.833 27	0.765 33	0.526 47	0.756 49	0.136 46	0.588 50	0.470 39	0.438 48	0.432 54	0.358 55	0.650 27	0.857 49	0.429 56	0.765 37	0.557 55	1.000 1	0.430 51
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 49	1.000 1	0.801 39	0.599 53	0.535 44	0.728 53	0.286 28	0.436 64	0.679 21	0.491 37	0.433 52	0.256 57	0.404 64	0.857 49	0.620 38	0.724 47	0.510 62	1.000 1	0.539 35

AOIA	0.601 50	1.000 1	0.761 51	0.687 43	0.485 53	0.828 30	0.008 64	0.663 36	0.405 47	0.405 56	0.425 55	0.490 38	0.596 46	0.714 60	0.553 46	0.779 32	0.597 45	0.992 50	0.424 53

PCJC	0.578 51	1.000 1	0.810 33	0.583 57	0.449 57	0.813 36	0.042 58	0.603 48	0.341 53	0.490 38	0.465 49	0.410 48	0.650 27	0.835 57	0.264 66	0.694 55	0.561 53	0.889 61	0.504 41

SSEN	0.575 52	1.000 1	0.761 50	0.473 60	0.477 55	0.795 41	0.066 54	0.529 55	0.658 26	0.460 42	0.461 50	0.380 52	0.331 66	0.859 48	0.401 59	0.692 57	0.653 36	1.000 1	0.348 58
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 53	0.528 70	0.708 61	0.626 48	0.580 38	0.745 52	0.063 55	0.627 44	0.240 58	0.400 58	0.497 46	0.464 40	0.515 53	1.000 1	0.475 51	0.745 43	0.571 50	1.000 1	0.429 52

NeuralBF	0.555 54	0.667 60	0.896 17	0.843 18	0.517 49	0.751 50	0.029 59	0.519 56	0.414 46	0.439 46	0.465 48	0.000 76	0.484 55	0.857 49	0.287 64	0.693 56	0.651 38	1.000 1	0.485 45
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 55	1.000 1	0.807 36	0.588 56	0.327 62	0.647 58	0.004 66	0.815 7	0.180 61	0.418 53	0.364 62	0.182 60	0.445 58	1.000 1	0.442 54	0.688 58	0.571 51	1.000 1	0.396 54
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
ClickSeg_Instance	0.539 56	1.000 1	0.621 65	0.300 63	0.530 46	0.698 54	0.127 47	0.533 54	0.222 59	0.430 51	0.400 57	0.365 53	0.574 50	0.938 45	0.472 52	0.659 60	0.543 58	0.944 56	0.347 59

One_Thing_One_Click	0.529 57	0.667 60	0.718 57	0.777 30	0.399 58	0.683 55	0.000 69	0.669 34	0.138 64	0.391 59	0.374 61	0.539 34	0.360 65	0.641 64	0.556 45	0.774 35	0.593 46	0.997 40	0.251 64
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 58	1.000 1	0.538 70	0.282 64	0.468 56	0.790 42	0.173 38	0.345 66	0.429 43	0.413 55	0.484 47	0.176 61	0.595 48	0.591 65	0.522 48	0.668 59	0.476 63	0.986 54	0.327 60

Occipital-SCS	0.512 59	1.000 1	0.716 58	0.509 59	0.506 51	0.611 61	0.092 51	0.602 49	0.177 62	0.346 62	0.383 59	0.165 62	0.442 59	0.850 56	0.386 61	0.618 64	0.543 59	0.889 61	0.389 55

3D-BoNet	0.488 60	1.000 1	0.672 64	0.590 55	0.301 64	0.484 71	0.098 50	0.620 46	0.306 55	0.341 63	0.259 66	0.125 64	0.434 61	0.796 59	0.402 58	0.499 70	0.513 61	0.909 60	0.439 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
PanopticFusion-inst	0.478 61	0.667 60	0.712 60	0.595 54	0.259 67	0.550 67	0.000 69	0.613 47	0.175 63	0.250 68	0.434 51	0.437 44	0.411 63	0.857 49	0.485 50	0.591 67	0.267 73	0.944 56	0.359 57
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 62	0.667 60	0.685 63	0.677 45	0.372 60	0.562 65	0.000 69	0.482 61	0.244 57	0.316 65	0.298 63	0.052 71	0.442 60	0.857 49	0.267 65	0.702 52	0.559 54	1.000 1	0.287 62

SALoss-ResNet	0.459 63	1.000 1	0.737 56	0.159 74	0.259 66	0.587 63	0.138 45	0.475 62	0.217 60	0.416 54	0.408 56	0.128 63	0.315 67	0.714 60	0.411 57	0.536 69	0.590 47	0.873 65	0.304 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)
MASC	0.447 64	0.528 70	0.555 68	0.381 61	0.382 59	0.633 59	0.002 67	0.509 58	0.260 56	0.361 61	0.432 53	0.327 56	0.451 57	0.571 66	0.367 62	0.639 62	0.386 64	0.980 55	0.276 63
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
SegGroup_ins	0.445 65	0.667 60	0.773 47	0.185 71	0.317 63	0.656 57	0.000 69	0.407 65	0.134 65	0.381 60	0.267 65	0.217 59	0.476 56	0.714 60	0.452 53	0.629 63	0.514 60	1.000 1	0.222 67
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 66	1.000 1	0.432 73	0.245 66	0.190 68	0.577 64	0.013 63	0.263 68	0.033 71	0.320 64	0.240 67	0.075 67	0.422 62	0.857 49	0.117 71	0.699 53	0.271 72	0.883 63	0.235 66
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
Hier3D	0.323 67	0.667 60	0.542 69	0.264 65	0.157 71	0.550 66	0.000 69	0.205 71	0.009 73	0.270 67	0.218 68	0.075 67	0.500 54	0.688 63	0.007 77	0.698 54	0.301 69	0.459 74	0.200 68
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
UNet-backbone	0.319 68	0.667 60	0.715 59	0.233 67	0.189 69	0.479 72	0.008 64	0.218 69	0.067 70	0.201 70	0.173 69	0.107 65	0.123 72	0.438 68	0.150 68	0.615 65	0.355 65	0.916 59	0.093 76

R-PointNet	0.306 69	0.500 72	0.405 74	0.311 62	0.348 61	0.589 62	0.054 56	0.068 74	0.126 66	0.283 66	0.290 64	0.028 72	0.219 70	0.214 71	0.331 63	0.396 74	0.275 70	0.821 67	0.245 65

Region-18class	0.284 70	0.250 76	0.751 55	0.228 69	0.270 65	0.521 68	0.000 69	0.468 63	0.008 75	0.205 69	0.127 70	0.000 76	0.068 74	0.070 75	0.262 67	0.652 61	0.323 67	0.740 70	0.173 69

SemRegionNet-20cls	0.250 71	0.333 73	0.613 66	0.229 68	0.163 70	0.493 69	0.000 69	0.304 67	0.107 67	0.147 73	0.100 72	0.052 70	0.231 68	0.119 73	0.039 73	0.445 72	0.325 66	0.654 71	0.141 72

tmp	0.248 72	0.667 60	0.437 72	0.188 70	0.153 72	0.491 70	0.000 69	0.208 70	0.094 69	0.153 72	0.099 73	0.057 69	0.217 71	0.119 73	0.039 73	0.466 71	0.302 68	0.640 72	0.140 73

3D-BEVIS	0.248 72	0.667 60	0.566 67	0.076 75	0.035 77	0.394 75	0.027 61	0.035 76	0.098 68	0.099 75	0.030 76	0.025 73	0.098 73	0.375 70	0.126 70	0.604 66	0.181 75	0.854 66	0.171 70
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Sem_Recon_ins	0.227 74	0.764 59	0.486 71	0.069 76	0.098 74	0.426 74	0.017 62	0.067 75	0.015 72	0.172 71	0.100 71	0.096 66	0.054 76	0.183 72	0.135 69	0.366 75	0.260 74	0.614 73	0.168 71

ASIS	0.199 75	0.333 73	0.253 76	0.167 73	0.140 73	0.438 73	0.000 69	0.177 72	0.008 74	0.121 74	0.069 74	0.004 75	0.231 69	0.429 69	0.036 75	0.445 73	0.273 71	0.333 76	0.119 75

Sgpn_scannet	0.143 76	0.208 77	0.390 75	0.169 72	0.065 75	0.275 76	0.029 60	0.069 73	0.000 76	0.087 76	0.043 75	0.014 74	0.027 77	0.000 76	0.112 72	0.351 76	0.168 76	0.438 75	0.138 74

MaskRCNN 2d->3d Proj	0.058 77	0.333 73	0.002 77	0.000 77	0.053 76	0.002 77	0.002 68	0.021 77	0.000 76	0.045 77	0.024 77	0.238 58	0.065 75	0.000 76	0.014 76	0.107 77	0.020 77	0.110 77	0.006 77

3D Semantic Instance Benchmark