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 4	0.872 11	0.718 5	0.941 3	0.588 5	0.652 41	0.819 3	0.776 3	0.720 6	0.780 6	0.769 11	1.000 1	0.797 11	0.813 31	0.798 8	1.000 1	0.659 4

PointRel	0.816 1	1.000 1	0.971 9	0.908 6	0.743 2	0.923 10	0.573 9	0.714 22	0.695 20	0.734 10	0.747 2	0.725 13	0.809 1	1.000 1	0.814 9	0.899 5	0.820 3	1.000 1	0.610 19
: Relation3D: Enhancing Relation Modeling for Point Cloud Instance Segmentation. CVPR 2025
Spherical Mask(CtoF)	0.812 3	1.000 1	0.973 8	0.852 15	0.718 6	0.917 12	0.574 7	0.677 31	0.748 13	0.729 14	0.715 9	0.795 3	0.809 1	1.000 1	0.831 4	0.854 12	0.787 12	1.000 1	0.638 8

PointComp	0.811 4	0.850 60	0.969 10	0.864 13	0.739 3	0.946 2	0.539 16	0.671 34	0.835 2	0.700 18	0.742 3	0.817 1	0.766 12	1.000 1	0.755 21	0.909 1	0.808 6	1.000 1	0.687 2

EV3D	0.811 4	1.000 1	0.968 11	0.852 15	0.717 7	0.921 11	0.574 8	0.677 31	0.748 13	0.730 13	0.703 15	0.795 3	0.809 1	1.000 1	0.831 4	0.854 12	0.778 16	1.000 1	0.638 9

VDG-Uni3DSeg	0.804 6	1.000 1	0.990 1	0.886 9	0.688 20	0.912 14	0.602 2	0.703 26	0.786 8	0.771 4	0.708 13	0.700 18	0.669 26	0.981 41	0.789 17	0.903 2	0.772 20	1.000 1	0.609 20

SIM3D	0.803 7	1.000 1	0.967 12	0.863 14	0.692 19	0.924 9	0.552 13	0.732 20	0.667 25	0.732 12	0.662 19	0.796 2	0.789 9	1.000 1	0.803 10	0.864 9	0.766 23	1.000 1	0.643 6

OneFormer3D	0.801 8	1.000 1	0.973 7	0.909 5	0.698 15	0.928 7	0.582 6	0.668 37	0.685 21	0.780 2	0.687 17	0.698 22	0.702 15	1.000 1	0.794 13	0.900 4	0.784 14	0.986 55	0.635 10
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
Competitor-SPFormer	0.800 9	1.000 1	0.986 3	0.845 17	0.705 13	0.915 13	0.532 17	0.733 19	0.757 12	0.733 11	0.708 12	0.698 21	0.648 38	0.981 41	0.890 1	0.830 22	0.796 9	0.997 43	0.644 5

InsSSM	0.799 10	1.000 1	0.915 15	0.710 44	0.729 4	0.925 8	0.664 1	0.670 35	0.770 9	0.766 5	0.739 4	0.737 9	0.700 16	1.000 1	0.792 14	0.829 24	0.815 4	0.997 43	0.625 12
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 11	1.000 1	0.878 23	0.792 29	0.693 18	0.936 4	0.596 3	0.685 30	0.663 27	0.736 8	0.717 7	0.788 5	0.693 21	1.000 1	0.825 7	0.840 18	0.837 1	1.000 1	0.689 1

TST3D	0.795 12	1.000 1	0.929 14	0.918 4	0.709 10	0.884 23	0.596 4	0.704 25	0.769 10	0.734 9	0.644 24	0.699 20	0.751 13	1.000 1	0.794 12	0.876 8	0.757 26	0.997 43	0.550 36
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
MG-Former	0.791 13	1.000 1	0.980 6	0.837 20	0.626 29	0.897 16	0.543 15	0.759 14	0.800 7	0.766 6	0.659 20	0.769 8	0.697 19	1.000 1	0.791 15	0.707 52	0.791 11	1.000 1	0.610 18

ExtMask3D	0.789 14	1.000 1	0.988 2	0.756 37	0.706 12	0.912 15	0.429 23	0.647 43	0.806 6	0.755 7	0.673 18	0.689 23	0.772 10	1.000 1	0.789 16	0.852 14	0.811 5	1.000 1	0.617 15

UniPerception	0.787 15	1.000 1	0.909 16	0.768 34	0.687 21	0.947 1	0.551 14	0.714 21	0.843 1	0.696 19	0.713 11	0.773 7	0.607 44	0.981 41	0.690 29	0.878 7	0.775 19	1.000 1	0.640 7

Queryformer	0.787 15	1.000 1	0.933 13	0.601 54	0.754 1	0.886 21	0.558 12	0.661 39	0.767 11	0.665 22	0.716 8	0.639 29	0.808 5	1.000 1	0.844 3	0.897 6	0.804 7	1.000 1	0.624 13

MAFT	0.786 17	1.000 1	0.894 21	0.807 24	0.694 17	0.893 19	0.486 19	0.674 33	0.740 15	0.786 1	0.704 14	0.727 12	0.739 14	1.000 1	0.707 27	0.849 16	0.756 27	1.000 1	0.685 3

KmaxOneFormerNet	0.783 18	0.903 58	0.981 5	0.794 28	0.706 11	0.931 6	0.561 11	0.701 27	0.706 18	0.727 15	0.697 16	0.731 11	0.689 23	1.000 1	0.856 2	0.750 43	0.761 25	1.000 1	0.599 24

Mask3D	0.780 19	1.000 1	0.786 47	0.716 42	0.696 16	0.885 22	0.500 18	0.714 22	0.810 5	0.672 21	0.715 9	0.679 24	0.809 1	1.000 1	0.831 4	0.833 21	0.787 12	1.000 1	0.602 22
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
SPFormer	0.770 20	0.903 58	0.903 18	0.806 25	0.609 36	0.886 20	0.568 10	0.815 6	0.705 19	0.711 16	0.655 21	0.652 28	0.685 24	1.000 1	0.789 18	0.809 32	0.776 18	1.000 1	0.583 28
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
SoftGroup++	0.769 21	1.000 1	0.803 40	0.937 1	0.684 22	0.865 25	0.213 39	0.870 2	0.664 26	0.571 29	0.758 1	0.702 17	0.807 6	1.000 1	0.653 35	0.902 3	0.792 10	1.000 1	0.626 11

SoftGroup	0.761 22	1.000 1	0.808 36	0.845 17	0.716 8	0.862 27	0.243 36	0.824 4	0.655 29	0.620 23	0.734 5	0.699 19	0.791 8	0.981 41	0.716 24	0.844 17	0.769 21	1.000 1	0.594 26
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 23	1.000 1	0.904 17	0.731 40	0.678 23	0.895 17	0.458 21	0.644 45	0.670 24	0.710 17	0.620 29	0.732 10	0.650 28	1.000 1	0.756 20	0.778 35	0.779 15	1.000 1	0.614 16
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 24	1.000 1	0.774 48	0.867 12	0.621 31	0.934 5	0.404 24	0.706 24	0.812 4	0.605 26	0.633 27	0.626 30	0.690 22	1.000 1	0.640 37	0.820 27	0.777 17	1.000 1	0.612 17
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
PBNet	0.747 25	1.000 1	0.818 32	0.837 21	0.713 9	0.844 29	0.457 22	0.647 43	0.711 17	0.614 24	0.617 31	0.657 27	0.650 28	1.000 1	0.692 28	0.822 26	0.765 24	1.000 1	0.595 25
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 26	1.000 1	0.788 45	0.724 41	0.642 28	0.859 28	0.248 35	0.787 11	0.618 32	0.596 27	0.653 23	0.722 15	0.583 51	1.000 1	0.766 19	0.861 10	0.825 2	1.000 1	0.504 42

IPCA-Inst	0.731 27	1.000 1	0.788 46	0.884 10	0.698 14	0.788 45	0.252 34	0.760 13	0.646 30	0.511 37	0.637 26	0.665 26	0.804 7	1.000 1	0.644 36	0.778 36	0.747 29	1.000 1	0.561 32

TopoSeg	0.725 28	1.000 1	0.806 39	0.933 2	0.668 25	0.758 50	0.272 33	0.734 18	0.630 31	0.549 33	0.654 22	0.606 31	0.697 20	0.966 47	0.612 41	0.839 19	0.754 28	1.000 1	0.573 29

DKNet	0.718 29	1.000 1	0.814 33	0.782 30	0.619 33	0.872 24	0.224 37	0.751 16	0.569 36	0.677 20	0.585 36	0.724 14	0.633 40	0.981 41	0.515 51	0.819 28	0.736 30	1.000 1	0.617 14
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
SSEC	0.707 30	1.000 1	0.850 25	0.924 3	0.648 26	0.747 53	0.162 41	0.862 3	0.572 35	0.520 35	0.624 28	0.549 34	0.649 37	1.000 1	0.560 46	0.706 53	0.768 22	1.000 1	0.591 27

HAIS	0.699 31	1.000 1	0.849 26	0.820 22	0.675 24	0.808 39	0.279 31	0.757 15	0.465 42	0.517 36	0.596 33	0.559 33	0.600 45	1.000 1	0.654 34	0.767 38	0.676 34	0.994 51	0.560 33
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
SSTNet	0.698 32	1.000 1	0.697 64	0.888 8	0.556 43	0.803 40	0.387 25	0.626 47	0.417 47	0.556 32	0.585 37	0.702 16	0.600 45	1.000 1	0.824 8	0.720 51	0.692 32	1.000 1	0.509 41
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
DualGroup	0.694 33	1.000 1	0.799 42	0.811 23	0.622 30	0.817 34	0.376 26	0.805 9	0.590 34	0.487 41	0.568 40	0.525 38	0.650 28	0.835 59	0.600 42	0.829 23	0.655 37	1.000 1	0.526 38

ODIN - Ins	0.693 34	1.000 1	0.880 22	0.647 49	0.620 32	0.779 47	0.336 28	0.501 62	0.681 22	0.577 28	0.595 34	0.679 25	0.683 25	1.000 1	0.709 26	0.816 30	0.637 41	0.770 71	0.557 34
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 35	1.000 1	0.856 24	0.744 38	0.618 34	0.893 18	0.151 42	0.651 42	0.713 16	0.537 34	0.579 39	0.430 48	0.651 27	1.000 1	0.389 62	0.744 46	0.697 31	0.991 53	0.601 23

DANCENET	0.680 35	1.000 1	0.807 37	0.733 39	0.600 37	0.768 49	0.375 27	0.543 55	0.538 37	0.610 25	0.599 32	0.498 39	0.632 42	0.981 41	0.739 23	0.856 11	0.633 44	0.882 66	0.454 51

Box2Mask	0.677 37	1.000 1	0.847 27	0.771 32	0.509 52	0.816 35	0.277 32	0.558 54	0.482 39	0.562 31	0.640 25	0.448 44	0.700 16	1.000 1	0.666 30	0.852 15	0.578 51	0.997 43	0.488 46
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 38	1.000 1	0.758 56	0.682 46	0.576 41	0.842 30	0.477 20	0.504 61	0.524 38	0.567 30	0.585 38	0.451 43	0.557 53	1.000 1	0.751 22	0.797 33	0.563 54	1.000 1	0.467 50
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
Mask-Group	0.664 39	1.000 1	0.822 31	0.764 36	0.616 35	0.815 36	0.139 46	0.694 29	0.597 33	0.459 45	0.566 41	0.599 32	0.600 45	0.516 69	0.715 25	0.819 29	0.635 42	1.000 1	0.603 21
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 40	1.000 1	0.760 54	0.667 48	0.581 39	0.863 26	0.323 29	0.655 40	0.477 40	0.473 43	0.549 43	0.432 47	0.650 28	1.000 1	0.655 33	0.738 47	0.585 50	0.944 58	0.472 49

CSC-Pretrained	0.648 41	1.000 1	0.810 34	0.768 33	0.523 50	0.813 37	0.143 45	0.819 5	0.389 50	0.422 54	0.511 47	0.443 45	0.650 28	1.000 1	0.624 39	0.732 48	0.634 43	1.000 1	0.375 58

PE	0.645 42	1.000 1	0.773 50	0.798 27	0.538 45	0.786 46	0.088 54	0.799 10	0.350 54	0.435 52	0.547 44	0.545 35	0.646 39	0.933 49	0.562 45	0.761 41	0.556 59	0.997 43	0.501 44
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
RPGN	0.643 43	1.000 1	0.758 55	0.582 60	0.539 44	0.826 33	0.046 59	0.765 12	0.372 52	0.436 51	0.588 35	0.539 37	0.650 28	1.000 1	0.577 43	0.750 44	0.653 39	0.997 43	0.495 45
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
Dyco3D	0.641 44	1.000 1	0.841 28	0.893 7	0.531 47	0.802 41	0.115 51	0.588 52	0.448 44	0.438 49	0.537 46	0.430 49	0.550 54	0.857 51	0.534 49	0.764 40	0.657 36	0.987 54	0.568 30
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
GICN	0.638 45	1.000 1	0.895 20	0.800 26	0.480 56	0.676 58	0.144 44	0.737 17	0.354 53	0.447 46	0.400 60	0.365 55	0.700 16	1.000 1	0.569 44	0.836 20	0.599 46	1.000 1	0.473 48

PointGroup	0.636 46	1.000 1	0.765 51	0.624 51	0.505 54	0.797 42	0.116 50	0.696 28	0.384 51	0.441 47	0.559 42	0.476 41	0.596 48	1.000 1	0.666 30	0.756 42	0.556 58	0.997 43	0.513 40
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 47	0.667 62	0.797 44	0.714 43	0.562 42	0.774 48	0.146 43	0.810 8	0.429 46	0.476 42	0.546 45	0.399 51	0.633 40	1.000 1	0.632 38	0.722 50	0.609 45	1.000 1	0.514 39
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 48	1.000 1	0.829 30	0.606 53	0.646 27	0.836 31	0.068 55	0.511 59	0.462 43	0.507 38	0.619 30	0.389 53	0.610 43	1.000 1	0.432 57	0.828 25	0.673 35	0.788 70	0.552 35

DENet	0.629 49	1.000 1	0.797 43	0.608 52	0.589 38	0.627 62	0.219 38	0.882 1	0.310 56	0.402 59	0.383 62	0.396 52	0.650 28	1.000 1	0.663 32	0.543 70	0.691 33	1.000 1	0.568 31

3D-MPA	0.611 50	1.000 1	0.833 29	0.765 35	0.526 49	0.756 51	0.136 48	0.588 52	0.470 41	0.438 50	0.432 56	0.358 57	0.650 28	0.857 51	0.429 58	0.765 39	0.557 57	1.000 1	0.430 53
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 51	1.000 1	0.801 41	0.599 55	0.535 46	0.728 55	0.286 30	0.436 66	0.679 23	0.491 39	0.433 54	0.256 59	0.404 66	0.857 51	0.620 40	0.724 49	0.510 64	1.000 1	0.539 37

AOIA	0.601 52	1.000 1	0.761 53	0.687 45	0.485 55	0.828 32	0.008 66	0.663 38	0.405 49	0.405 58	0.425 57	0.490 40	0.596 48	0.714 62	0.553 48	0.779 34	0.597 47	0.992 52	0.424 55

PCJC	0.578 53	1.000 1	0.810 35	0.583 59	0.449 59	0.813 38	0.042 60	0.603 50	0.341 55	0.490 40	0.465 51	0.410 50	0.650 28	0.835 59	0.264 68	0.694 57	0.561 55	0.889 63	0.504 43

SSEN	0.575 54	1.000 1	0.761 52	0.473 62	0.477 57	0.795 43	0.066 56	0.529 57	0.658 28	0.460 44	0.461 52	0.380 54	0.331 68	0.859 50	0.401 61	0.692 59	0.653 38	1.000 1	0.348 60
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 55	0.528 72	0.708 63	0.626 50	0.580 40	0.745 54	0.063 57	0.627 46	0.240 60	0.400 60	0.497 48	0.464 42	0.515 55	1.000 1	0.475 53	0.745 45	0.571 52	1.000 1	0.429 54

NeuralBF	0.555 56	0.667 62	0.896 19	0.843 19	0.517 51	0.751 52	0.029 61	0.519 58	0.414 48	0.439 48	0.465 50	0.000 78	0.484 57	0.857 51	0.287 66	0.693 58	0.651 40	1.000 1	0.485 47
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 57	1.000 1	0.807 38	0.588 58	0.327 64	0.647 60	0.004 68	0.815 7	0.180 63	0.418 55	0.364 64	0.182 62	0.445 60	1.000 1	0.442 56	0.688 60	0.571 53	1.000 1	0.396 56
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
ClickSeg_Instance	0.539 58	1.000 1	0.621 67	0.300 65	0.530 48	0.698 56	0.127 49	0.533 56	0.222 61	0.430 53	0.400 59	0.365 55	0.574 52	0.938 48	0.472 54	0.659 62	0.543 60	0.944 58	0.347 61

One_Thing_One_Click	0.529 59	0.667 62	0.718 59	0.777 31	0.399 60	0.683 57	0.000 71	0.669 36	0.138 66	0.391 61	0.374 63	0.539 36	0.360 67	0.641 66	0.556 47	0.774 37	0.593 48	0.997 43	0.251 66
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 60	1.000 1	0.538 72	0.282 66	0.468 58	0.790 44	0.173 40	0.345 68	0.429 45	0.413 57	0.484 49	0.176 63	0.595 50	0.591 67	0.522 50	0.668 61	0.476 65	0.986 56	0.327 62

Occipital-SCS	0.512 61	1.000 1	0.716 60	0.509 61	0.506 53	0.611 63	0.092 53	0.602 51	0.177 64	0.346 64	0.383 61	0.165 64	0.442 61	0.850 58	0.386 63	0.618 66	0.543 61	0.889 63	0.389 57

3D-BoNet	0.488 62	1.000 1	0.672 66	0.590 57	0.301 66	0.484 73	0.098 52	0.620 48	0.306 57	0.341 65	0.259 68	0.125 66	0.434 63	0.796 61	0.402 60	0.499 72	0.513 63	0.909 62	0.439 52
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 63	0.667 62	0.712 62	0.595 56	0.259 69	0.550 69	0.000 71	0.613 49	0.175 65	0.250 70	0.434 53	0.437 46	0.411 65	0.857 51	0.485 52	0.591 69	0.267 75	0.944 58	0.359 59
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 64	0.667 62	0.685 65	0.677 47	0.372 62	0.562 67	0.000 71	0.482 63	0.244 59	0.316 67	0.298 65	0.052 73	0.442 62	0.857 51	0.267 67	0.702 54	0.559 56	1.000 1	0.287 64

SALoss-ResNet	0.459 65	1.000 1	0.737 58	0.159 76	0.259 68	0.587 65	0.138 47	0.475 64	0.217 62	0.416 56	0.408 58	0.128 65	0.315 69	0.714 62	0.411 59	0.536 71	0.590 49	0.873 67	0.304 63
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 66	0.528 72	0.555 70	0.381 63	0.382 61	0.633 61	0.002 69	0.509 60	0.260 58	0.361 63	0.432 55	0.327 58	0.451 59	0.571 68	0.367 64	0.639 64	0.386 66	0.980 57	0.276 65
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
SegGroup_ins	0.445 67	0.667 62	0.773 49	0.185 73	0.317 65	0.656 59	0.000 71	0.407 67	0.134 67	0.381 62	0.267 67	0.217 61	0.476 58	0.714 62	0.452 55	0.629 65	0.514 62	1.000 1	0.222 69
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 68	1.000 1	0.432 75	0.245 68	0.190 70	0.577 66	0.013 65	0.263 70	0.033 73	0.320 66	0.240 69	0.075 69	0.422 64	0.857 51	0.117 73	0.699 55	0.271 74	0.883 65	0.235 68
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
Hier3D	0.323 69	0.667 62	0.542 71	0.264 67	0.157 73	0.550 68	0.000 71	0.205 73	0.009 75	0.270 69	0.218 70	0.075 69	0.500 56	0.688 65	0.007 79	0.698 56	0.301 71	0.459 76	0.200 70
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
UNet-backbone	0.319 70	0.667 62	0.715 61	0.233 69	0.189 71	0.479 74	0.008 66	0.218 71	0.067 72	0.201 72	0.173 71	0.107 67	0.123 74	0.438 70	0.150 70	0.615 67	0.355 67	0.916 61	0.093 78

R-PointNet	0.306 71	0.500 74	0.405 76	0.311 64	0.348 63	0.589 64	0.054 58	0.068 76	0.126 68	0.283 68	0.290 66	0.028 74	0.219 72	0.214 73	0.331 65	0.396 76	0.275 72	0.821 69	0.245 67

Region-18class	0.284 72	0.250 78	0.751 57	0.228 71	0.270 67	0.521 70	0.000 71	0.468 65	0.008 77	0.205 71	0.127 72	0.000 78	0.068 76	0.070 77	0.262 69	0.652 63	0.323 69	0.740 72	0.173 71

SemRegionNet-20cls	0.250 73	0.333 75	0.613 68	0.229 70	0.163 72	0.493 71	0.000 71	0.304 69	0.107 69	0.147 75	0.100 74	0.052 72	0.231 70	0.119 75	0.039 75	0.445 74	0.325 68	0.654 73	0.141 74

tmp	0.248 74	0.667 62	0.437 74	0.188 72	0.153 74	0.491 72	0.000 71	0.208 72	0.094 71	0.153 74	0.099 75	0.057 71	0.217 73	0.119 75	0.039 75	0.466 73	0.302 70	0.640 74	0.140 75

3D-BEVIS	0.248 74	0.667 62	0.566 69	0.076 77	0.035 79	0.394 77	0.027 63	0.035 78	0.098 70	0.099 77	0.030 78	0.025 75	0.098 75	0.375 72	0.126 72	0.604 68	0.181 77	0.854 68	0.171 72
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Sem_Recon_ins	0.227 76	0.764 61	0.486 73	0.069 78	0.098 76	0.426 76	0.017 64	0.067 77	0.015 74	0.172 73	0.100 73	0.096 68	0.054 78	0.183 74	0.135 71	0.366 77	0.260 76	0.614 75	0.168 73

ASIS	0.199 77	0.333 75	0.253 78	0.167 75	0.140 75	0.438 75	0.000 71	0.177 74	0.008 76	0.121 76	0.069 76	0.004 77	0.231 71	0.429 71	0.036 77	0.445 75	0.273 73	0.333 78	0.119 77

Sgpn_scannet	0.143 78	0.208 79	0.390 77	0.169 74	0.065 77	0.275 78	0.029 62	0.069 75	0.000 78	0.087 78	0.043 77	0.014 76	0.027 79	0.000 78	0.112 74	0.351 78	0.168 78	0.438 77	0.138 76

MaskRCNN 2d->3d Proj	0.058 79	0.333 75	0.002 79	0.000 79	0.053 78	0.002 79	0.002 70	0.021 79	0.000 78	0.045 79	0.024 79	0.238 60	0.065 77	0.000 78	0.014 78	0.107 79	0.020 79	0.110 79	0.006 79

3D Semantic Instance Benchmark