Rotation Aware 3D Point Cloud Vehicle Detection

3D point cloud object detection is an important task of environment perception in autonomous driving. However, the point cloud data collected by LiDAR is limited in angle. In order to solve this problem, we propose a Rotation Aware Detection Network. We apply an augmentation method of angle transformation in the original scenes to obtain the corresponding derivative scenes. We propose an auxiliary network to learn the difference between the two scenes we designed. Our method can effectively enhance the detection capability of the network without additional inference cost. Our approach can improve the performance in the car detection task on the KITTI dataset.