Doors Close Status Detection for Box Car Based on Target Detection

Box car door close status detection is a necessary check before train departure. The closing state of the door can be reflected by measuring the longitudinal opening distance of the door and the rotation angle of the door lock. Aiming at the problems of low accuracy and incomplete check points of traditional detection algorithms based on image processing, this paper proposes the keypoint detection on the door and the rotating object detection on the lock based on deep learning. Experiment shows that the proposed methods have high detection accuracy and can accurately reflect the door close status.