Point Cloud Oversegmentation with Graph-Structured Deep Metric Learning

CVPR 2019  ·  Loic Landrieu, Mohamed Boussaha ·

We propose a new supervized learning framework for oversegmenting 3D point clouds into superpoints. We cast this problem as learning deep embeddings of the local geometry and radiometry of 3D points, such that the border of objects presents high contrasts. The embeddings are computed using a lightweight neural network operating on the points' local neighborhood. Finally, we formulate point cloud oversegmentation as a graph partition problem with respect to the learned embeddings. This new approach allows us to set a new state-of-the-art in point cloud oversegmentation by a significant margin, on a dense indoor dataset (S3DIS) and a sparse outdoor one (vKITTI). Our best solution requires over five times fewer superpoints to reach similar performance than previously published methods on S3DIS. Furthermore, we show that our framework can be used to improve superpoint-based semantic segmentation algorithms, setting a new state-of-the-art for this task as well.

PDF Abstract CVPR 2019 PDF CVPR 2019 Abstract

Datasets


Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Semantic Segmentation S3DIS SSP+SPG Mean IoU 68.4 # 28
mAcc 78.3 # 21
oAcc 87.9 # 24
Number of params 0.290M # 38
Params (M) 0.29 # 15
Semantic Segmentation S3DIS Area5 SSP+SPG mIoU 61.7 # 45
oAcc 87.9 # 28
mAcc 68.2 # 33
Number of params 290K # 2

Methods


No methods listed for this paper. Add relevant methods here