点云上的卷积神经网络及其应用,深蓝学院,Introduction of the course Convolutional Neural Networks (CNNs) have led to a revolution in the recognition of raster images. However, many data, especially 3D data, come naturally in the form of point clouds where raster-based convolution operations are not readily available to be used. In this tutorial we will discuss several recent work that make it possible to build a convolutional network or similar operations on point clouds.Besides,certain operations that are computationally expensive using raster-based CNNs,such as cost volume convolution,could potentially be significantly cheaper to compute using point could convolution layers,this leads to better performance on tasks such as scene flow computation. We will also briefly touch upon the topic of object detection on point clouds as well as the visualization of point cloud classifiers. 适用人群：1、学习三维视觉、感知方向的同学 2、从事机器人、三维视觉领域工作的工程师 3、三维点云方向的科研人员