Deep learning network based on vehicle-mounted 4D radar point cloud

• Main Authors: ZHENG CHENGXIN, KE WENXIONG, XI GUANGRONG, LI XIAOLIU, FU CHAOWEI, LI YOUZHI, ZHANG ZHONGZE, WEI WEIWEI, YIN JIEJUN
• Format: Patent
• Language: Chinese; English
• Published: 07.03.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention provides a deep learning network based on a vehicle-mounted 4D radar point cloud, and the method comprises the following steps: S1, collecting scene data to form a 4D radar point cloud data set, and dividing the 4D radar point cloud data set into a training set, a verification set and a test set; s2, dividing scene point cloud data in the training set into a plurality of space regions, and learning point cloud distribution characteristics in different space regions through a density sensing PointRCNN network; s3, global semantic features of points are learned point by point, a foreground point is segmented from an input point cloud after a spatial region is searched and traversed, and a 3D proposal box is generated to estimate the position and size of a vehicle target; s4, learning local features through local coordinate transformation, and optimizing the position and direction of the 3D proposal box, so as to generate a 3D bounding box for detecting the vehicle target; and S5, setting a network