Vehicle Orientation Detection Using CNN

Abstract

Vehicle orientation detection is a challenging task because the orientations of vehicles can vary in a wide rangein captured images. The existing methods for oriented vehicle detection require too much computation time to beapplied to a real-time system. We propose Rotate YOLO, which has a set of anchor boxes with multiple scales,ratios, and angles to predict bounding boxes. For estimating the orientation angle, we applied angle-related IoUwith CIoU loss to solve the underivable problem from the calculation of SkewIoU. Evaluation results on threepublic datasets DLR Munich, VEDAI and UCAS-AOD demonstrate the efficiency of our approach