基于注意力YOLO-V3的合成孔径雷达舰船检测识别一体化方法

传统的合成孔径雷达舰船检测识别需要分两步实现，检测识别精度和效率难以满足实际应用需求。本文结合注意力机制和YOLO-V3网络提出了注意力YOLO-V3网络实现合成孔径雷达舰船检测识别一体化。同时，利用公开的AIR-SARShip-1.0数据集和OpenSARShip数据集构建了大场景舰船检测识别数据集，用于验证目标检测识别性能。实验结果表明，本文提出的注意力YOLO-V3网络可以获得较高的检测识别精度，证明了本文方法的有效性。

SAR Ship Integrated Detection and Recognition Based on Attention YOLO-V3 Network

The traditional SAR ship detection and recognition need to be implemented in two steps, and the detection and recognition accuracy and efficiency have difficulty in meeting the needs of practical applications. Integrating the advantages of attention mechanism and the YOLO-V3 network, an attention YOLO-V3 network is proposed to implement SAR ship integrated detection and recognition. Meanwhile, a large-scene ship detection and recognition dataset is constructed using the public AIR-SARShip-1.0 and Open SAR Ship dataset to verify the target detection and recognition performance. The experimental results show that the proposed attention YOLO-V3 network can achieve high ship detection and recognition accuracy, which proves the effectiveness of this method.