Deep learning-based point cloud generation method and device, electronic equipment and system

• Main Author: ZHANG SHANGDI
• Format: Patent
• Language: Chinese; English
• Published: 09.04.2024
• Subjects: ANALOGOUS ARRANGEMENTS USING OTHER WAVES; CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION ORRERADIATION OF RADIO WAVES; MEASURING; PHYSICS; RADIO DIRECTION-FINDING; RADIO NAVIGATION; TESTING

The invention provides a deep learning-based point cloud generation method, device, electronic equipment and system, and the method comprises the steps: carrying out the feature extraction of the ADC original data of a millimeter wave radar, and obtaining a first 3D radar feature; according to the first 3D radar features, generating a dense point cloud by using a trained deep learning model; wherein in the training process of the deep learning model, data supervision is carried out by using acquired data of high-precision 3D acquisition equipment. According to the method, the quality of the generated dense point cloud can be improved, and data support is provided for improving the accuracy of downstream tasks. 本申请提供一种基于深度学习的点云生成方法、装置、电子设备及系统，该方法包括：对毫米波雷达的ADC原始数据进行特征提取，得到第一3D雷达特征；依据所述第一3D雷达特征，利用训练好的深度学习模型，生成稠密点云；其中，所述深度学习模型的训练过程中，利用高精度3D采集设备的采集数据进行数据监督。该方法可以提高所生成的稠密点云的质量，为提高下游任务的准确性提供了数据支持。