Surface Representation for Point Clouds

11 May 2022  ·  Haoxi Ran, Jun Liu, Chengjie Wang ·

Most prior work represents the shapes of point clouds by coordinates. However, it is insufficient to describe the local geometry directly. In this paper, we present \textbf{RepSurf} (representative surfaces), a novel representation of point clouds to \textbf{explicitly} depict the very local structure. We explore two variants of RepSurf, Triangular RepSurf and Umbrella RepSurf inspired by triangle meshes and umbrella curvature in computer graphics. We compute the representations of RepSurf by predefined geometric priors after surface reconstruction. RepSurf can be a plug-and-play module for most point cloud models thanks to its free collaboration with irregular points. Based on a simple baseline of PointNet++ (SSG version), Umbrella RepSurf surpasses the previous state-of-the-art by a large margin for classification, segmentation and detection on various benchmarks in terms of performance and efficiency. With an increase of around \textbf{0.008M} number of parameters, \textbf{0.04G} FLOPs, and \textbf{1.12ms} inference time, our method achieves \textbf{94.7\%} (+0.5\%) on ModelNet40, and \textbf{84.6\%} (+1.8\%) on ScanObjectNN for classification, while \textbf{74.3\%} (+0.8\%) mIoU on S3DIS 6-fold, and \textbf{70.0\%} (+1.6\%) mIoU on ScanNet for segmentation. For detection, previous state-of-the-art detector with our RepSurf obtains \textbf{71.2\%} (+2.1\%) mAP$\mathit{_{25}}$, \textbf{54.8\%} (+2.0\%) mAP$\mathit{_{50}}$ on ScanNetV2, and \textbf{64.9\%} (+1.9\%) mAP$\mathit{_{25}}$, \textbf{47.7\%} (+2.5\%) mAP$\mathit{_{50}}$ on SUN RGB-D. Our lightweight Triangular RepSurf performs its excellence on these benchmarks as well. The code is publicly available at \url{https://github.com/hancyran/RepSurf}.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
3D Point Cloud Classification ModelNet40 RepSurf-U Overall Accuracy 94.7 # 2
Number of params 1.48M # 59
FLOPs 0.81G # 1
Semantic Segmentation S3DIS RepSurf-U Mean IoU 74.3 # 2
mAcc 82.6 # 5
oAcc 90.8 # 1
FLOPs 1.04G # 1
Number of params 0.97M # 36
Semantic Segmentation S3DIS Area5 RepSurf-U mIoU 68.9 # 4
oAcc 90.2 # 4
mAcc 76.0 # 3
Number of params 0.97M # 25
FLOPs 1.04G # 1
3D Object Detection ScanNetV2 RepSurf-U mAP@0.25 71.2 # 3
mAP@0.5 54.8 # 4
3D Point Cloud Classification ScanObjectNN RepSurf-U Overall Accuracy 84.6 # 5
Number of params 1.48M # 18
FLOPs 0.81G # 1
3D Point Cloud Classification ScanObjectNN RepSurf-U (2x) Overall Accuracy 86.0 # 2
Number of params 1.48M # 18
FLOPs 2.43G # 1
3D Object Detection SUN-RGBD val RepSurf-U mAP@0.25 64.9 # 1
mAP@0.5 47.7 # 3

Methods


No methods listed for this paper. Add relevant methods here