Combining a Deep Convolutional Neural Network with Transfer Learning for Ship Classification
Hui Zhou, Na Chu, Liu ZhenYu · 2019
The classification of ship targets is essential for maritime search and rescue, fishing vessel monitoring, and other maritime supervision. Aiming at the problems of low recognition accuracy due to insufficient image resolution of the ship, a high-resolution remote sensing image ship target recognition method based on deep convolutional neural network is proposed. First VGG19 network model is used as the basic network, and then the transfer learning technology is applied to initialize parameters. The model is trained by ship target data set. Finally we use Adam method to optimize this model. The experimental results show that the final recognition accuracy is 95.8%, and it is proved higher accuracy compared with other classification methods.