Intelligent Ship Identity Identification and Verification with Multisource Fusion

The multi-source ship identity identification and verification system is developed for ship monitoring at marine pollutant reception station. Ship declaration information, AIS data and camara monitoring images are used together for ship identification and ID verification. The ship's MMSI is obtained from its declaration information, and the position data of the ship is extracted from the AIS data associated with the MMSI to decide the ship's arriving at the pollutant reception station. Ship identity verification is achieved by means of monitoring camara system, where the visual target detection box associated with the ship is produced with the improved YOLOv5 and the cross-match to the AIS-data-mapped calibration frame is performed to fulfil ship identity verification. Tests are conducted with the data published by SeaShips, which shows that the average output accuracy of the presented design reaches 80.83 %, a 3.14 % improvement. With the help of TensorRT, the Inference speed of the model is raised by 73 % and the frame rate reaches 64.18 fps, sufficient for marine pollutant reception station operation.