less than 1 minute read

点云:point cloud,是三维点的集合,只需记录物体所在的点的坐标;没有规律性的排列,因此无法套用卷积神经网络;高分辨率下,比较节省内存与运算量,也比较适合表示高分辨率的三维物体;
点云处理概述

1 综述

  1. 3D Object Recognition with Ensemble Learning — A Study of Point Cloud-Based Deep Learning Models
    2019-04-17 Paper
    3D · 分类 · 检测 · 集成学习 · 点云

  2. Deep Learning for 3D Point Clouds: A Survey
    2019-12-27 paper

2 理论

3 技术方向

3.1 GAN

  1. PU-Net: Point Cloud Upsampling Network
    CVPR 2018 Paper | Tensorflow-Offical | Tensorflow-AE-GAN
    点云上采样;

3.2 AE


TOP

附录

A 参考资料

  1. dragonlong. Trending in 3D Vision[EB/OL]. https://github.com/dragonlong/Trending-in-3D-Vision. -/2019-03-28.
  2. timzhang641. 3D-Machine-Learning[EB/OL]. https://github.com/timzhang642/3D-Machine-Learning. -/2019-03-28.
  3. poodar.chu. 点云感知 CVPR 2018 论文总结. https://zhuanlan.zhihu.com/p/41287237. 2018-08-04/2019-03-28
  4. QingyongHu. SoTA-Point-Cloud[EB/OL]. https://github.com/QingyongHu/SoTA-Point-Cloud. 2019/2020-01-22.

Comments