基于全卷积神经网络的船舶检测和船牌识别系统

船舶检测与识别对于港口智能监控，实现港口资源的有效管理具有重要意义。由于复杂的船舶轮廓、船牌位置不固定、船牌文本类型复杂多样和船牌文字个数不确定等因素，使得船舶的检测和识别非常具有挑战性。本文提出一种基于全卷积神经网络的船舶检测与识别方法：SDR-FCN。SDR-FCN利用本文提出的船舶检测算法SDNet进行船舶检测定位，然后利用本文提出的船牌文本检测算法PDNet进行船牌文字检测，最后利用具备在线自适应性的分类器OA-Classifier进行船牌分类识别。OA-Classifier综合了AIS（船舶自动识别系统）反馈的信息，提高了分类器的识别精度。实际SDR-FCN部署运行表明，它能够以较高的精度可靠地工作，满足实际应用。

A Ship Detection and Plate Recognition System Based on FCN

Ship detection and recognition are important for smart monitoring of ships in order to manage port resources effectively. 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 an efficient method based on fully convolutional neural network for ship detection and recognition named SDR-FCN. SDR-FCN, which uses a tiny fully convolutional neural network named SDNet to locate ships, then detects text of plate with PDNet designed in this paper, at last, recognizes the plate with an online adaptive classifier named OA-Classifier. The recognition accuracy of the classifier is improved with integrating the AIS (Automatic Identification System) information. The actual SDR-FCN deployment demonstrates that it can work reliably with a high accuracy for satisfying practical usages.