# Deep Learning for 3D Point Clouds: A Survey

**Authors:** Yulan Guo, Hanyun Wang, Qingyong Hu, Hao Liu, Li Liu, Mohammed Bennamoun  
**arXiv:** [1912.12033](https://arxiv.org/abs/1912.12033)  
**Date:** 2019-12-27  
**Summary page:** https://www.summarizepaper.com/en/paper/1912.12033v1/

## Key points

- The paper provides a comprehensive review of recent progress in deep learning methods for point clouds.
- Point cloud learning has gained significant attention due to its wide range of applications in computer vision, autonomous driving, and robotics.
- Deep learning application to point clouds is still in its early stages due to unique challenges.
- There has been a thriving interest in deep learning on point clouds leading to the proposal of numerous methods.
- The paper focuses on three major tasks: 3D shape classification, 3D object detection and tracking, and 3D point cloud segmentation.
- The authors present a detailed overview of existing techniques and methodologies for each task.
- Comparative results from publicly available datasets are included, providing insights into the performance of different deep learning methods for point clouds.
- The authors offer insightful observations based on their findings and suggest inspiring future research directions.
- The survey serves as a valuable resource for researchers and practitioners interested in deep learning methods for point clouds.

## Summary

The paper "Deep Learning for 3D Point Clouds: A Survey" by Yulan Guo, Hanyun Wang, Qingyong Hu, Hao Liu, Li Liu and Mohammed Bennamoun provides a comprehensive review of recent progress in deep learning methods for point clouds. Point cloud learning has gained significant attention due to its wide range of applications in computer vision, autonomous driving and robotics. While deep learning has been successful in solving various 2D vision problems, its application to point clouds is still in its early stages due to the unique challenges posed by processing point clouds with deep neural networks. In recent years there has been a thriving interest in deep learning on point clouds leading to the proposal of numerous methods to address different problems in this area. To stimulate further research and development this paper focuses on three major tasks: 3D shape classification; 3D object detection and tracking; and 3D point cloud segmentation. The authors present a detailed overview of the existing techniques and methodologies employed for each task. Additionally the paper includes comparative results obtained from several publicly available datasets which provide valuable insights into the performance of different deep learning methods for point clouds. The authors also offer insightful observations based on their findings and suggest inspiring future research directions. Overall this survey serves as a valuable resource for researchers and practitioners interested in deep learning methods for point clouds. It not only summarizes the current state-of-the-art but also highlights areas that require further exploration and improvement.