Deep Learning Based Automatic Charging Identification and Positioning Method for Electric Vehicle

Electric vehicle charging identification and positioning is critically important to achieving automatic charging. In terms of the problem for electric vehicles, a dual recognition method based on deep learning proposed. The divided into two parts: global localization local localization. specific implementation process, collected pictures attitude are classified labeled. It trained with improved YOLOv4 network model corresponding detection obtained. contour extracted by BiSeNet semantic segmentation algorithm. minimum external rectangle used vehicle. Based location relationship between port vehicle, rough information equipment moves vicinity port, camera near gun collects port. detected Hough circle, KM algorithm feature matching, homography matrix solve attitude. results show that proposed in this paper can accurately locate accuracy connection reach 80%. provides an effective way problems vehicles has strong engineering practical value.