We use Intersection over Union (IoU) and Overall Accuracy (OA) as metrics. For more details hover the curser over the symbols or click on a classifier. In order to sort the results differently click on a symbol.
Name | A_IoU | OA | [s] | IoU 1 | IoU 2 | IoU 3 | IoU 4 | IoU 5 | IoU 6 | IoU 7 | IoU 8 | |
1 | ddnet | 0.747 | 0.950 | 1.00 | 0.979 | 0.941 | 0.708 | 0.643 | 0.940 | 0.485 | 0.388 | 0.892 |
Anonymous submission | ||||||||||||
2 | 5354 | 0.763 | 0.948 | 100.00 | 0.967 | 0.927 | 0.700 | 0.625 | 0.942 | 0.508 | 0.494 | 0.938 |
Anonymous submission | ||||||||||||
3 | 3DNet | 0.732 | 0.938 | 400.00 | 0.978 | 0.924 | 0.653 | 0.594 | 0.922 | 0.396 | 0.451 | 0.934 |
Anonymous submission | ||||||||||||
4 | ConvPoint_Keras | 0.777 | 0.950 | 1.00 | 0.959 | 0.900 | 0.791 | 0.705 | 0.963 | 0.433 | 0.561 | 0.907 |
5 | GeomAdapt | 0.752 | 0.947 | 1.00 | 0.963 | 0.897 | 0.709 | 0.672 | 0.960 | 0.457 | 0.458 | 0.903 |
6 | Att_conv | 0.707 | 0.936 | 1.00 | 0.963 | 0.896 | 0.683 | 0.607 | 0.928 | 0.415 | 0.272 | 0.898 |
Attentive Aggregation Networks for Efficient Semantic Segmentation of Large-Scale Point Clouds | ||||||||||||
7 | WreathProdNet | 0.771 | 0.946 | 1.00 | 0.952 | 0.871 | 0.753 | 0.671 | 0.961 | 0.513 | 0.510 | 0.934 |
@misc{wang2020equivariant, title={Equivariant Maps for Hierarchical Structures}, author={Renhao Wang and Marjan Albooyeh and Siamak Ravanbakhsh}, year={2020}, eprint={2006.03627}, archivePrefix={arXiv}, primaryClass={cs.LG} } | ||||||||||||
8 | LightConvPoint. | 0.746 | 0.941 | 1000000.00 | 0.947 | 0.852 | 0.774 | 0.704 | 0.940 | 0.529 | 0.294 | 0.926 |
FKAConv: Feature-Kernel Alignment for Point Cloud Convolution -- https://arxiv.org/abs/2004.04462 implemented in LightConvPoint Framework | ||||||||||||
9 | dp_v4 | 0.752 | 0.938 | 1000.00 | 0.947 | 0.840 | 0.734 | 0.629 | 0.964 | 0.454 | 0.511 | 0.939 |
Anonymous submission | ||||||||||||
10 | conv_pts | 0.765 | 0.934 | 2400.00 | 0.921 | 0.806 | 0.760 | 0.719 | 0.956 | 0.473 | 0.611 | 0.877 |
Generalizing discrete convolutions for unstructured point clouds, A. Boulch, Eurographics 3DOR, 2019 | ||||||||||||
11 | PointGCR | 0.695 | 0.921 | 2.00 | 0.938 | 0.800 | 0.644 | 0.664 | 0.932 | 0.392 | 0.343 | 0.853 |
Global Context Reasoning for Semantic Segmentation of 3D Point Clouds. Submitted to WACV 2020 | ||||||||||||
12 | ILiDAR-C1DC | 0.620 | 0.914 | 604800.00 | 0.918 | 0.798 | 0.576 | 0.672 | 0.938 | 0.294 | 0.209 | 0.557 |
Anonymous submission | ||||||||||||
13 | PointConv_CE | 0.710 | 0.923 | 1.00 | 0.924 | 0.796 | 0.727 | 0.620 | 0.937 | 0.406 | 0.446 | 0.825 |
Semantic Context Encoding for Accurate 3D Point Cloud Segmentation. IEEE Transactions on Multimedia 2020 | ||||||||||||
14 | SnapNet | 0.674 | 0.910 | 0.00 | 0.896 | 0.795 | 0.748 | 0.561 | 0.909 | 0.365 | 0.343 | 0.772 |
Unstructured point cloud semantic labeling using deep segmentation networks. A. Boulch, B. Le Saux and N. Audebert, Eurographics 3DOR 2017 | ||||||||||||
15 | PointNet2_Demo | 0.631 | 0.857 | 10000.00 | 0.819 | 0.781 | 0.643 | 0.517 | 0.759 | 0.364 | 0.437 | 0.726 |
https://github.com/IntelVCL/Open3D-PointNet2-Semantic3D, Yixing Lao | ||||||||||||
16 | SPGraph_ | 0.762 | 0.929 | 10000.00 | 0.915 | 0.756 | 0.783 | 0.717 | 0.944 | 0.568 | 0.529 | 0.884 |
Large-scale Point cloud segmentation with superpoint graphs, Loic Landrieu and Martin Simonovsky, CVPR2018 | ||||||||||||
17 | pointnetpp_sem | 0.521 | 0.825 | 10000.00 | 0.788 | 0.745 | 0.598 | 0.608 | 0.817 | 0.332 | 0.154 | 0.127 |
G. Dekeyser and M. Orhan | ||||||||||||
18 | HarrisNet | 0.623 | 0.881 | 0.00 | 0.818 | 0.737 | 0.742 | 0.625 | 0.927 | 0.283 | 0.178 | 0.671 |
Anonymous submission | ||||||||||||
19 | Wow | 0.720 | 0.906 | 1.00 | 0.864 | 0.703 | 0.695 | 0.680 | 0.969 | 0.434 | 0.523 | 0.895 |
Fast Point Cloud Registration Using Semantic Segmentation - DICTA 2019 | ||||||||||||
20 | super_ss | 0.644 | 0.896 | 1.00 | 0.911 | 0.695 | 0.650 | 0.560 | 0.897 | 0.300 | 0.438 | 0.697 |
@inproceedings{contreras2019edge, title={Edge-Convolution Point Net for Semantic Segmentation of Large-Scale Point Clouds}, author={Contreras, Jhonatan and Denzler, Joachim}, booktitle={IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium}, pages={5236--5239}, year={2019}, organization={IEEE} } | ||||||||||||
21 | TMLC-MS | 0.494 | 0.850 | 38421.00 | 0.911 | 0.695 | 0.328 | 0.216 | 0.876 | 0.259 | 0.113 | 0.553 |
Timo Hackel, Jan D. Wegner, Konrad Schindler: Fast semantic segmentation of 3d point clouds with strongly varying density. ISPRS Annals - ISPRS Congress, Prague, 2016 | ||||||||||||
22 | TML-PC | 0.391 | 0.745 | 0.00 | 0.804 | 0.661 | 0.423 | 0.412 | 0.647 | 0.124 | 0.000 | 0.058 |
Mind the gap: modeling local and global context in (road) networks: Javier Montoya, Jan D. Wegner, Lubor Ladicky, Konrad Schindler. In: German Conference on Pattern Recognition (GCPR), Münster, Germany, 2014 | ||||||||||||
23 | FCNVoxNet | 0.372 | 0.523 | 138929.00 | 0.066 | 0.272 | 0.580 | 0.364 | 0.809 | 0.283 | 0.095 | 0.509 |
Anonymous submission |
@inproceedings{hackel2017isprs,
title={{SEMANTIC3D.NET: A new large-scale point cloud classification benchmark}},
author={Timo Hackel and N. Savinov and L. Ladicky and Jan D. Wegner and K. Schindler and M. Pollefeys},
booktitle={ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences},
year = {2017},
volume = {IV-1-W1},
pages = {91--98}
}