3D Semantic Label Results - ScanNet Benchmark

The 3D semantic labeling task involves predicting a semantic labeling of a 3D scan mesh.

Evaluation and metrics

Our evaluation ranks all methods according to the PASCAL VOC intersection-over-union metric (IoU). IoU = TP/(TP+FP+FN), where TP, FP, and FN are the numbers of true positive, false positive, and false negative pixels, respectively. Predicted labels are evaluated per-vertex over the respective 3D scan mesh; for 3D approaches that operate on other representations like grids or points, the predicted labels should be mapped onto the mesh vertices (e.g., one such example for grid to mesh vertices is provided in the evaluation helpers).

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

Method	avg iou	bathtub	bed	bookshelf	cabinet	chair	counter	curtain	desk	door	floor	otherfurniture	picture	refrigerator	shower curtain	sink	sofa	table	toilet	wall	window

OccuSeg+Semantic	0.764 1	0.758 28	0.796 7	0.839 4	0.746 2	0.907 1	0.562 1	0.850 5	0.680 2	0.672 1	0.978 1	0.610 1	0.335 3	0.777 1	0.819 13	0.847 1	0.830 1	0.691 3	0.972 1	0.885 1	0.727 2

BPNet	0.749 2	0.909 1	0.818 4	0.811 8	0.752 1	0.839 6	0.485 12	0.842 7	0.673 3	0.644 4	0.957 3	0.528 7	0.305 9	0.773 2	0.859 4	0.788 2	0.818 3	0.693 2	0.916 6	0.856 5	0.723 4
Wenbo Hu, Hengshuang Zhao, Li Jiang, Jiaya Jia, Tien-Tsin Wong: Bidirectional Projection Network for Cross Dimension Scene Understanding. CVPR 2021
Virtual MVFusion	0.746 3	0.771 24	0.819 3	0.848 2	0.702 8	0.865 3	0.397 42	0.899 1	0.699 1	0.664 2	0.948 20	0.588 2	0.330 4	0.746 5	0.851 7	0.764 3	0.796 7	0.704 1	0.935 4	0.866 2	0.728 1
Abhijit Kundu, Xiaoqi Yin, Alireza Fathi, David Ross, Brian Brewington, Thomas Funkhouser, Caroline Pantofaru: Virtual Multi-view Fusion for 3D Semantic Segmentation. ECCV 2020
VMNet	0.746 3	0.870 5	0.838 1	0.858 1	0.729 4	0.850 4	0.501 7	0.874 2	0.587 19	0.658 3	0.956 4	0.564 4	0.299 10	0.765 3	0.900 1	0.716 7	0.812 4	0.631 10	0.939 2	0.858 4	0.709 5

MinkowskiNet	0.736 5	0.859 7	0.818 4	0.832 5	0.709 7	0.840 5	0.521 4	0.853 4	0.660 4	0.643 5	0.951 10	0.544 5	0.286 15	0.731 6	0.893 2	0.675 16	0.772 13	0.683 4	0.874 25	0.852 6	0.727 2
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
SparseConvNet	0.725 6	0.647 46	0.821 2	0.846 3	0.721 5	0.869 2	0.533 2	0.754 18	0.603 16	0.614 6	0.955 5	0.572 3	0.325 5	0.710 7	0.870 3	0.724 6	0.823 2	0.628 11	0.934 5	0.865 3	0.683 8

MatchingNet	0.724 7	0.812 16	0.812 6	0.810 9	0.735 3	0.834 8	0.495 10	0.860 3	0.572 24	0.602 10	0.954 6	0.512 9	0.280 16	0.757 4	0.845 10	0.725 5	0.780 10	0.606 18	0.937 3	0.851 7	0.700 7

RFCR	0.702 8	0.889 3	0.745 20	0.813 7	0.672 10	0.818 18	0.493 11	0.815 9	0.623 10	0.610 7	0.947 23	0.470 19	0.249 25	0.594 21	0.848 8	0.705 11	0.779 11	0.646 7	0.892 16	0.823 12	0.611 22

One Thing One Click	0.701 9	0.825 14	0.796 7	0.723 24	0.716 6	0.832 9	0.433 32	0.816 8	0.634 8	0.609 8	0.969 2	0.418 40	0.344 2	0.559 31	0.833 11	0.715 8	0.808 5	0.560 30	0.902 9	0.847 8	0.680 9

JSENet	0.699 10	0.881 4	0.762 15	0.821 6	0.667 11	0.800 28	0.522 3	0.792 11	0.613 11	0.607 9	0.935 42	0.492 13	0.205 36	0.576 27	0.853 6	0.691 12	0.758 16	0.652 6	0.872 28	0.828 10	0.649 14
Zeyu HU, Mingmin Zhen, Xuyang BAI, Hongbo Fu, Chiew-lan Tai: JSENet: Joint Semantic Segmentation and Edge Detection Network for 3D Point Clouds. ECCV 2020
CU-Hybrid Net	0.693 11	0.596 51	0.789 9	0.803 11	0.677 9	0.800 28	0.469 16	0.846 6	0.554 31	0.591 12	0.948 20	0.500 10	0.316 7	0.609 18	0.847 9	0.732 4	0.808 5	0.593 21	0.894 14	0.839 9	0.652 13

FusionNet	0.688 12	0.704 39	0.741 23	0.754 21	0.656 12	0.829 11	0.501 7	0.741 23	0.609 13	0.548 17	0.950 14	0.522 8	0.371 1	0.633 14	0.756 20	0.715 8	0.771 14	0.623 12	0.861 34	0.814 16	0.658 12
Feihu Zhang, Jin Fang, Benjamin Wah, Philip Torr: Deep FusionNet for Point Cloud Semantic Segmentation. ECCV 2020
KP-FCNN	0.684 13	0.847 9	0.758 18	0.784 14	0.647 15	0.814 20	0.473 14	0.772 14	0.605 14	0.594 11	0.935 42	0.450 27	0.181 44	0.587 22	0.805 15	0.690 13	0.785 9	0.614 14	0.882 20	0.819 14	0.632 18
H. Thomas, C. Qi, J. Deschaud, B. Marcotegui, F. Goulette, L. Guibas.: KPConv: Flexible and Deformable Convolution for Point Clouds. ICCV 2019
PointContrast_LA_SEM	0.683 14	0.757 29	0.784 10	0.786 13	0.639 17	0.824 15	0.408 36	0.775 13	0.604 15	0.541 19	0.934 46	0.532 6	0.269 20	0.552 32	0.777 18	0.645 28	0.793 8	0.640 8	0.913 7	0.824 11	0.671 10

ROSMRF3D	0.673 15	0.789 17	0.748 19	0.763 19	0.635 19	0.814 20	0.407 39	0.747 20	0.581 22	0.573 13	0.950 14	0.484 14	0.271 19	0.607 19	0.754 21	0.649 24	0.774 12	0.596 20	0.883 19	0.823 12	0.606 26

SALANet	0.670 16	0.816 15	0.770 14	0.768 17	0.652 14	0.807 25	0.451 20	0.747 20	0.659 5	0.545 18	0.924 49	0.473 18	0.149 54	0.571 29	0.811 14	0.635 31	0.746 19	0.623 12	0.892 16	0.794 27	0.570 37

PointConv	0.666 17	0.781 19	0.759 17	0.699 29	0.644 16	0.822 16	0.475 13	0.779 12	0.564 27	0.504 32	0.953 7	0.428 35	0.203 38	0.586 24	0.754 21	0.661 21	0.753 17	0.588 22	0.902 9	0.813 18	0.642 15
Wenxuan Wu, Zhongang Qi, Li Fuxin: PointConv: Deep Convolutional Networks on 3D Point Clouds. CVPR 2019
PointASNL	0.666 17	0.703 40	0.781 11	0.751 23	0.655 13	0.830 10	0.471 15	0.769 15	0.474 47	0.537 20	0.951 10	0.475 17	0.279 17	0.635 12	0.698 32	0.675 16	0.751 18	0.553 35	0.816 45	0.806 19	0.703 6
Xu Yan, Chaoda Zheng, Zhen Li, Sheng Wang, Shuguang Cui: PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling. CVPR 2020
VACNN++	0.664 19	0.891 2	0.724 29	0.680 36	0.636 18	0.814 20	0.438 31	0.629 46	0.553 32	0.537 20	0.950 14	0.499 11	0.247 26	0.626 15	0.786 17	0.666 19	0.700 32	0.566 28	0.860 35	0.816 15	0.665 11

DCM-Net	0.658 20	0.778 20	0.702 35	0.806 10	0.619 21	0.813 23	0.468 17	0.693 34	0.494 43	0.524 26	0.941 35	0.449 28	0.298 11	0.510 38	0.821 12	0.675 16	0.727 24	0.568 26	0.826 42	0.803 21	0.637 17
Jonas Schult, Francis Engelmann, Theodora Kontogianni, Bastian Leibe: DualConvMesh-Net: Joint Geodesic and Euclidean Convolutions on 3D Meshes. CVPR 2020 [Oral]
HPGCNN	0.656 21	0.698 41	0.743 22	0.650 44	0.564 37	0.820 17	0.505 6	0.758 17	0.631 9	0.479 39	0.945 27	0.480 15	0.226 29	0.572 28	0.774 19	0.690 13	0.735 23	0.614 14	0.853 38	0.776 40	0.597 30
Jisheng Dang, Qingyong Hu, Yulan Guo, Jun Yang: HPGCNN.
SAFNet-seg	0.654 22	0.752 30	0.734 25	0.664 43	0.583 30	0.815 19	0.399 41	0.754 18	0.639 7	0.535 22	0.942 33	0.470 19	0.309 8	0.665 8	0.539 40	0.650 23	0.708 28	0.635 9	0.857 37	0.793 28	0.642 15

RandLA-Net	0.645 23	0.778 20	0.731 26	0.699 29	0.577 32	0.829 11	0.446 24	0.736 24	0.477 46	0.523 28	0.945 27	0.454 25	0.269 20	0.484 45	0.749 24	0.618 35	0.738 21	0.599 19	0.827 41	0.792 30	0.621 20

PointConv-SFPN	0.641 24	0.776 22	0.703 34	0.721 25	0.557 40	0.826 13	0.451 20	0.672 39	0.563 28	0.483 38	0.943 32	0.425 38	0.162 48	0.644 10	0.726 25	0.659 22	0.709 27	0.572 24	0.875 23	0.786 34	0.559 42

MVPNet	0.641 24	0.831 11	0.715 30	0.671 40	0.590 27	0.781 37	0.394 43	0.679 37	0.642 6	0.553 16	0.937 39	0.462 22	0.256 22	0.649 9	0.406 51	0.626 32	0.691 36	0.666 5	0.877 21	0.792 30	0.608 25
Maximilian Jaritz, Jiayuan Gu, Hao Su: Multi-view PointNet for 3D Scene Understanding. GMDL Workshop, ICCV 2019
PointMRNet	0.640 26	0.717 38	0.701 36	0.692 32	0.576 33	0.801 26	0.467 18	0.716 29	0.563 28	0.459 43	0.953 7	0.429 34	0.169 46	0.581 25	0.854 5	0.605 36	0.710 26	0.550 36	0.894 14	0.793 28	0.575 35

FPConv	0.639 27	0.785 18	0.760 16	0.713 28	0.603 24	0.798 30	0.392 44	0.534 52	0.603 16	0.524 26	0.948 20	0.457 23	0.250 24	0.538 35	0.723 26	0.598 40	0.696 35	0.614 14	0.872 28	0.799 22	0.567 39
Yiqun Lin, Zizheng Yan, Haibin Huang, Dong Du, Ligang Liu, Shuguang Cui, Xiaoguang Han: FPConv: Learning Local Flattening for Point Convolution. CVPR 2020
PointSPNet	0.637 28	0.734 33	0.692 41	0.714 27	0.576 33	0.797 31	0.446 24	0.743 22	0.598 18	0.437 47	0.942 33	0.403 42	0.150 53	0.626 15	0.800 16	0.649 24	0.697 34	0.557 33	0.846 39	0.777 39	0.563 40

SConv	0.636 29	0.830 12	0.697 39	0.752 22	0.572 36	0.780 38	0.445 26	0.716 29	0.529 36	0.530 23	0.951 10	0.446 29	0.170 45	0.507 40	0.666 34	0.636 30	0.682 38	0.541 41	0.886 18	0.799 22	0.594 31

Supervoxel-CNN	0.635 30	0.656 44	0.711 31	0.719 26	0.613 22	0.757 44	0.444 28	0.765 16	0.534 35	0.566 14	0.928 47	0.478 16	0.272 18	0.636 11	0.531 42	0.664 20	0.645 46	0.508 46	0.864 33	0.792 30	0.611 22

joint point-based	0.634 31	0.614 49	0.778 12	0.667 42	0.633 20	0.825 14	0.420 33	0.804 10	0.467 49	0.561 15	0.951 10	0.494 12	0.291 12	0.566 30	0.458 46	0.579 45	0.764 15	0.559 32	0.838 40	0.814 16	0.598 29
Hung-Yueh Chiang, Yen-Liang Lin, Yueh-Cheng Liu, Winston H. Hsu: A Unified Point-Based Framework for 3D Segmentation. 3DV 2019
MCCNN	0.633 32	0.866 6	0.731 26	0.771 15	0.576 33	0.809 24	0.410 35	0.684 35	0.497 42	0.491 35	0.949 17	0.466 21	0.105 58	0.581 25	0.646 35	0.620 33	0.680 39	0.542 40	0.817 44	0.795 25	0.618 21
P. Hermosilla, T. Ritschel, P.P. Vazquez, A. Vinacua, T. Ropinski: Monte Carlo Convolution for Learning on Non-Uniformly Sampled Point Clouds. SIGGRAPH Asia 2018
PointMTL	0.632 33	0.731 34	0.688 44	0.675 37	0.591 26	0.784 36	0.444 28	0.565 50	0.610 12	0.492 34	0.949 17	0.456 24	0.254 23	0.587 22	0.706 29	0.599 39	0.665 43	0.612 17	0.868 32	0.791 33	0.579 34

APCF-Net	0.631 34	0.742 31	0.687 46	0.672 38	0.557 40	0.792 33	0.408 36	0.665 41	0.545 33	0.508 30	0.952 9	0.428 35	0.186 42	0.634 13	0.702 30	0.620 33	0.706 29	0.555 34	0.873 27	0.798 24	0.581 33
Haojia, Lin: Adaptive Pyramid Context Fusion for Point Cloud Perception. GRSL
PointNet2-SFPN	0.631 34	0.771 24	0.692 41	0.672 38	0.524 43	0.837 7	0.440 30	0.706 32	0.538 34	0.446 45	0.944 29	0.421 39	0.219 32	0.552 32	0.751 23	0.591 42	0.737 22	0.543 39	0.901 11	0.768 43	0.557 43

3DSM_DMMF	0.631 34	0.626 48	0.745 20	0.801 12	0.607 23	0.751 45	0.506 5	0.729 27	0.565 26	0.491 35	0.866 60	0.434 31	0.197 40	0.595 20	0.630 36	0.709 10	0.705 30	0.560 30	0.875 23	0.740 50	0.491 52

FusionAwareConv	0.630 37	0.604 50	0.741 23	0.766 18	0.590 27	0.747 46	0.501 7	0.734 25	0.503 41	0.527 24	0.919 53	0.454 25	0.323 6	0.550 34	0.420 50	0.678 15	0.688 37	0.544 38	0.896 13	0.795 25	0.627 19
Jiazhao Zhang, Chenyang Zhu, Lintao Zheng, Kai Xu: Fusion-Aware Point Convolution for Online Semantic 3D Scene Segmentation. CVPR 2020
PointMRNet-lite	0.625 38	0.643 47	0.711 31	0.697 31	0.581 31	0.801 26	0.408 36	0.670 40	0.558 30	0.497 33	0.944 29	0.436 30	0.152 52	0.617 17	0.708 28	0.603 37	0.743 20	0.532 44	0.870 31	0.784 35	0.545 44

SIConv	0.625 38	0.830 12	0.694 40	0.757 20	0.563 38	0.772 40	0.448 22	0.647 44	0.520 38	0.509 29	0.949 17	0.431 33	0.191 41	0.496 43	0.614 37	0.647 27	0.672 41	0.535 43	0.876 22	0.783 36	0.571 36

HPEIN	0.618 40	0.729 35	0.668 47	0.647 46	0.597 25	0.766 41	0.414 34	0.680 36	0.520 38	0.525 25	0.946 25	0.432 32	0.215 33	0.493 44	0.599 38	0.638 29	0.617 51	0.570 25	0.897 12	0.806 19	0.605 27
Li Jiang, Hengshuang Zhao, Shu Liu, Xiaoyong Shen, Chi-Wing Fu, Jiaya Jia: Hierarchical Point-Edge Interaction Network for Point Cloud Semantic Segmentation. ICCV 2019
SPH3D-GCN	0.610 41	0.858 8	0.772 13	0.489 57	0.532 42	0.792 33	0.404 40	0.643 45	0.570 25	0.507 31	0.935 42	0.414 41	0.046 63	0.510 38	0.702 30	0.602 38	0.705 30	0.549 37	0.859 36	0.773 41	0.534 46
Huan Lei, Naveed Akhtar, and Ajmal Mian: Spherical Kernel for Efficient Graph Convolution on 3D Point Clouds. TPAMI 2020
AttAN	0.609 42	0.760 27	0.667 48	0.649 45	0.521 44	0.793 32	0.457 19	0.648 43	0.528 37	0.434 49	0.947 23	0.401 43	0.153 51	0.454 46	0.721 27	0.648 26	0.717 25	0.536 42	0.904 8	0.765 44	0.485 53
Gege Zhang, Qinghua Ma, Licheng Jiao, Fang Liu and Qigong Sun: AttAN: Attention Adversarial Networks for 3D Point Cloud Semantic Segmentation. IJCAI2020
LAP-D	0.594 43	0.720 36	0.692 41	0.637 48	0.456 51	0.773 39	0.391 46	0.730 26	0.587 19	0.445 46	0.940 37	0.381 46	0.288 13	0.434 49	0.453 47	0.591 42	0.649 44	0.581 23	0.777 49	0.749 49	0.610 24

DPC	0.592 44	0.720 36	0.700 37	0.602 51	0.480 48	0.762 43	0.380 48	0.713 31	0.585 21	0.437 47	0.940 37	0.369 48	0.288 13	0.434 49	0.509 44	0.590 44	0.639 49	0.567 27	0.772 50	0.755 47	0.592 32
Francis Engelmann, Theodora Kontogianni, Bastian Leibe: Dilated Point Convolutions: On the Receptive Field Size of Point Convolutions on 3D Point Clouds. ICRA 2020
CCRFNet	0.589 45	0.766 26	0.659 51	0.683 34	0.470 50	0.740 48	0.387 47	0.620 47	0.490 44	0.476 40	0.922 51	0.355 51	0.245 27	0.511 37	0.511 43	0.571 46	0.643 47	0.493 49	0.872 28	0.762 45	0.600 28

SegGCN	0.589 45	0.833 10	0.731 26	0.539 55	0.514 45	0.789 35	0.448 22	0.467 54	0.573 23	0.484 37	0.936 40	0.396 44	0.061 62	0.501 41	0.507 45	0.594 41	0.700 32	0.563 29	0.874 25	0.771 42	0.493 51
Huan Lei, Naveed Akhtar, and Ajmal Mian: SegGCN: Efficient 3D Point Cloud Segmentation with Fuzzy Spherical Kernel. CVPR 2020
ROSMRF	0.580 47	0.772 23	0.707 33	0.681 35	0.563 38	0.764 42	0.362 49	0.515 53	0.465 50	0.465 42	0.936 40	0.427 37	0.207 35	0.438 47	0.577 39	0.536 49	0.675 40	0.486 50	0.723 54	0.779 37	0.524 48

TextureNet	0.566 48	0.672 43	0.664 49	0.671 40	0.494 46	0.719 49	0.445 26	0.678 38	0.411 55	0.396 50	0.935 42	0.356 50	0.225 30	0.412 51	0.535 41	0.565 47	0.636 50	0.464 52	0.794 48	0.680 55	0.568 38
Jingwei Huang, Haotian Zhang, Li Yi, Thomas Funkerhouser, Matthias Niessner, Leonidas Guibas: TextureNet: Consistent Local Parametrizations for Learning from High-Resolution Signals on Meshes. CVPR
DVVNet	0.562 49	0.648 45	0.700 37	0.770 16	0.586 29	0.687 53	0.333 51	0.650 42	0.514 40	0.475 41	0.906 57	0.359 49	0.223 31	0.340 54	0.442 49	0.422 57	0.668 42	0.501 47	0.708 55	0.779 37	0.534 46

Pointnet++ & Feature	0.557 50	0.735 32	0.661 50	0.686 33	0.491 47	0.744 47	0.392 44	0.539 51	0.451 51	0.375 53	0.946 25	0.376 47	0.205 36	0.403 52	0.356 53	0.553 48	0.643 47	0.497 48	0.824 43	0.756 46	0.515 49

PanopticFusion-label	0.529 51	0.491 58	0.688 44	0.604 50	0.386 54	0.632 58	0.225 62	0.705 33	0.434 53	0.293 57	0.815 61	0.348 52	0.241 28	0.499 42	0.669 33	0.507 50	0.649 44	0.442 56	0.796 47	0.602 61	0.561 41
Gaku Narita, Takashi Seno, Tomoya Ishikawa, Yohsuke Kaji: PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things. IROS 2019 (to appear)
3DMV, FTSDF	0.501 52	0.558 55	0.608 57	0.424 62	0.478 49	0.690 52	0.246 58	0.586 48	0.468 48	0.450 44	0.911 55	0.394 45	0.160 49	0.438 47	0.212 59	0.432 56	0.541 57	0.475 51	0.742 52	0.727 51	0.477 54

PCNN	0.498 53	0.559 54	0.644 54	0.560 54	0.420 53	0.711 51	0.229 60	0.414 55	0.436 52	0.352 54	0.941 35	0.324 53	0.155 50	0.238 59	0.387 52	0.493 51	0.529 58	0.509 45	0.813 46	0.751 48	0.504 50

3DMV	0.484 54	0.484 59	0.538 60	0.643 47	0.424 52	0.606 61	0.310 52	0.574 49	0.433 54	0.378 52	0.796 62	0.301 54	0.214 34	0.537 36	0.208 60	0.472 55	0.507 61	0.413 59	0.693 56	0.602 61	0.539 45
Angela Dai, Matthias Niessner: 3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation. ECCV'18
PointCNN with RGB	0.458 55	0.577 53	0.611 56	0.356 64	0.321 60	0.715 50	0.299 54	0.376 58	0.328 61	0.319 55	0.944 29	0.285 56	0.164 47	0.216 62	0.229 58	0.484 53	0.545 56	0.456 54	0.755 51	0.709 52	0.475 55
Yangyan Li, Rui Bu, Mingchao Sun, Baoquan Chen: PointCNN. NeurIPS 2018
FCPN	0.447 56	0.679 42	0.604 58	0.578 53	0.380 55	0.682 54	0.291 55	0.106 64	0.483 45	0.258 62	0.920 52	0.258 58	0.025 64	0.231 61	0.325 54	0.480 54	0.560 55	0.463 53	0.725 53	0.666 57	0.231 64
Dario Rethage, Johanna Wald, Jürgen Sturm, Nassir Navab, Federico Tombari: Fully-Convolutional Point Networks for Large-Scale Point Clouds. ECCV 2018
PNET2	0.442 57	0.548 56	0.548 59	0.597 52	0.363 57	0.628 59	0.300 53	0.292 59	0.374 58	0.307 56	0.881 59	0.268 57	0.186 42	0.238 59	0.204 61	0.407 58	0.506 62	0.449 55	0.667 57	0.620 59	0.462 56

SurfaceConvPF	0.442 57	0.505 57	0.622 55	0.380 63	0.342 59	0.654 56	0.227 61	0.397 57	0.367 59	0.276 59	0.924 49	0.240 60	0.198 39	0.359 53	0.262 56	0.366 59	0.581 53	0.435 57	0.640 58	0.668 56	0.398 57
Hao Pan, Shilin Liu, Yang Liu, Xin Tong: Convolutional Neural Networks on 3D Surfaces Using Parallel Frames.
Tangent Convolutions	0.438 59	0.437 62	0.646 53	0.474 59	0.369 56	0.645 57	0.353 50	0.258 61	0.282 63	0.279 58	0.918 54	0.298 55	0.147 55	0.283 56	0.294 55	0.487 52	0.562 54	0.427 58	0.619 59	0.633 58	0.352 59
Maxim Tatarchenko, Jaesik Park, Vladlen Koltun, Qian-Yi Zhou: Tangent convolutions for dense prediction in 3d. CVPR 2018
subcloud_weak	0.411 60	0.479 60	0.650 52	0.475 58	0.285 63	0.519 64	0.087 65	0.725 28	0.396 57	0.386 51	0.621 65	0.250 59	0.117 56	0.338 55	0.443 48	0.188 65	0.594 52	0.369 62	0.377 65	0.616 60	0.306 60

SPLAT Net	0.393 61	0.472 61	0.511 61	0.606 49	0.311 61	0.656 55	0.245 59	0.405 56	0.328 61	0.197 63	0.927 48	0.227 62	0.000 66	0.001 66	0.249 57	0.271 64	0.510 59	0.383 61	0.593 60	0.699 53	0.267 62
Hang Su, Varun Jampani, Deqing Sun, Subhransu Maji, Evangelos Kalogerakis, Ming-Hsuan Yang, Jan Kautz: SPLATNet: Sparse Lattice Networks for Point Cloud Processing. CVPR 2018
ScanNet+FTSDF	0.383 62	0.297 64	0.491 62	0.432 61	0.358 58	0.612 60	0.274 56	0.116 63	0.411 55	0.265 60	0.904 58	0.229 61	0.079 60	0.250 57	0.185 62	0.320 62	0.510 59	0.385 60	0.548 61	0.597 63	0.394 58

PointNet++	0.339 63	0.584 52	0.478 63	0.458 60	0.256 64	0.360 65	0.250 57	0.247 62	0.278 64	0.261 61	0.677 64	0.183 63	0.117 56	0.212 63	0.145 64	0.364 60	0.346 65	0.232 65	0.548 61	0.523 64	0.252 63
Charles R. Qi, Li Yi, Hao Su, Leonidas J. Guibas: pointnet++: deep hierarchical feature learning on point sets in a metric space.
SSC-UNet	0.308 64	0.353 63	0.290 65	0.278 65	0.166 65	0.553 62	0.169 64	0.286 60	0.147 65	0.148 65	0.908 56	0.182 64	0.064 61	0.023 65	0.018 66	0.354 61	0.363 63	0.345 63	0.546 63	0.685 54	0.278 61

ScanNet	0.306 65	0.203 65	0.366 64	0.501 56	0.311 61	0.524 63	0.211 63	0.002 66	0.342 60	0.189 64	0.786 63	0.145 65	0.102 59	0.245 58	0.152 63	0.318 63	0.348 64	0.300 64	0.460 64	0.437 65	0.182 65
Angela Dai, Angel X. Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, Matthias Nießner: ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes. CVPR'17
ERROR	0.054 66	0.000 66	0.041 66	0.172 66	0.030 66	0.062 66	0.001 66	0.035 65	0.004 66	0.051 66	0.143 66	0.019 66	0.003 65	0.041 64	0.050 65	0.003 66	0.054 66	0.018 66	0.005 66	0.264 66	0.082 66

3D Semantic label benchmark