Training of a target detection model and target detection method, device, equipment and medium

Abstract

The application relates to training and target detection methods, devices, equipment and media of a target detection model. According to historical driving scene data of a vehicle, sample basic simulation point cloud data of the vehicle in a historical scene is simulated. The sample basic simulation point cloud data is processed into high-precision sample high-precision simulation point cloud data through a high-precision point cloud model. The sample high-precision simulation point cloud data and sample real point cloud data are mixed according to a preset ratio to obtain sample mixed point cloud data, and the sample mixed point cloud data is labeled. Based on the sample mixed point cloud data after labeling, a target detection model is trained. The application introduces a high-precision point cloud model to convert the sample basic simulation point cloud data into sample high-precision simulation point cloud data, so as to improve the accuracy of the simulated point cloud data. The mixed sample mixed point cloud data is used to train the target detection model, so as to improve the detection precision of the trained target detection model.

Images