Backend service system of ship detection and plate recognition system based on deep learning

Acquiring the identity information of the ships passing through the canal and the sea automatically by intelligent equipment is helpful for the water transportation department to strengthen the management and is meaningful for the construction of the urban intelligent traffic management system and national defense security. However, this is challenging due to complex ship profile, ship license background and object occlusion, variations of ship license plate locations and text types. This paper proposes a backend service system for ship detection and plate recognition based on Yolov4 and PaddleOCR, and produces a dataset containing 54059 pictures. By setting up a Socket server, the system receives the image information sent by the front-end, uses the YOLOV4 target detection algorithm to identify the ship and locate the ship plate, and simultaneously uses the PaddleOCR character detection and recognition algorithm to locate and identify the characters in the image. After frame synchronization and data fusion of target detection result and character detection and recognition result, the result is transmitted to the front-end through Socket. The system can be deployed on Win10, Linux and embedded system, working reliably with high precision to meet the practical application.