Laser point cloud road 3D target detection based on deep learning

摘要

In order to solve the problem that the two-dimensional image road target detection cannot provide spatial coordinate information, it is proposed to use lidar to scan the target to achieve the effect of road target detection. The ground point cloud is filtered by multi-plane ground point fitting on the original data. In this paper, we use a density-based three-dimensional clustering method to obtain a target candidate set, and use the deep learning method to identify the candidate set, and finally obtain the detection result. The detection method in this paper uses a combination of three-dimensional clustering and deep learning to overcome the shortcomings of difficult to judge categories in point cloud detection, and realizes a 3D target detection method based on lidar in road scenes.