Skip to Main content Skip to Navigation

Learning new representations for 3D point cloud semantic segmentation

Abstract : In the recent years, new technologies have allowed the acquisition of large and precise 3D scenes as point clouds. They have opened up new applications like self-driving vehicles or infrastructure monitoring that rely on efficient large scale point cloud processing. Convolutional deep learning methods cannot be directly used with point clouds. In the case of images, convolutional filters brought the ability to learn new representations, which were previously hand-crafted in older computer vision methods. Following the same line of thought, we present in this thesis a study of hand-crafted representations previously used for point cloud processing. We propose several contributions, to serve as basis for the design of a new convolutional representation for point cloud processing. They include a new definition of multiscale radius neighborhood, a comparison with multiscale k-nearest neighbors, a new active learning strategy, the semantic segmentation of large scale point clouds, and a study of the influence of density in multiscale representations. Following these contributions, we introduce the Kernel Point Convolution (KPConv), which uses radius neighborhoods and a set of kernel points to play the role of the kernel pixels in image convolution. Our convolutional networks outperform state-of-the-art semantic segmentation approaches in almost any situation. In addition to these strong results, we designed KPConv with a great flexibility and a deformable version. To conclude our argumentation, we propose several insights on the representations that our method is able to learn.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, January 28, 2020 - 5:06:10 PM
Last modification on : Wednesday, November 17, 2021 - 12:31:08 PM
Long-term archiving on: : Wednesday, April 29, 2020 - 4:44:03 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02458455, version 1


Hugues Thomas. Learning new representations for 3D point cloud semantic segmentation. Machine Learning [cs.LG]. Université Paris sciences et lettres, 2019. English. ⟨NNT : 2019PSLEM048⟩. ⟨tel-02458455⟩



Record views


Files downloads