【摘要】 船舶图像细粒度检测是高分辨遥感图像分析的难题，受船舶尺寸、陆地背景、光照、风浪等因素影响，易降低图像检测的准确性。为克服船舶目标识别的影响因素，针对不同类型和型号的船舶目标检测建起特征提取算法模型，提升最终的识别精度。本文提出一种基于深度学习的船舶图像细粒度检测方法，将深度学习算法应用到高分辨率遥感图像中，借助算法训练得出像素级的识别信息，实现对各类型信息的优化处理，最终生成具备辅助分类子网络功能的优化模型，提升细粒度识别精确度。更多还原

【Abstract】 Fine-grained detection of ship images is a difficult problem in high-resolution remote sensing image analysis.It is affected by factors such as ship size, land background, illumination, wind and waves, which can easily reduce the accuracy of image detection. In order to overcome the influencing factors of ship target recognition, a feature extraction algorithm model should be established for different types and types of ship target detection to improve the final recognition accuracy.This paper proposes a fine-grained detection method of ship images based on deep learning. The deep learning algorithm is applied to high-resolution remote sensing images, and the pixel-level identification information is obtained with the help of algorithm training, and the optimization of various types of information is realized. The optimization model with the function of auxiliary classification sub-network improves the accuracy of fine-grained recognition.更多还原