Research on Aircraft Remote Sensing Image Recognition Network Based on Attention Mechanism and TF Learning

Abstract

Effective differentiation of aircraft types using visible remote sensing images is important for providing military combat information as well as civilian aircraft operations. With the emergence of deep learning, remote sensing aircraft image classification has been well solved, and it gets rid of the limitation of traditional image processing methods that require manual feature extraction. However, deep learning requires a large number of samples for training and optimization of the network, and the current publicly available aircraft image database is very limited. In addition, due to the complexity of aircraft target recognition, the important information recognition capability of the existing models cannot meet the task requirements. To address the above problems, this paper proposes the ResNet50 model based on attention mechanism and transfer learning to classify aircraft remote sensing images. Experimental results based on real datasets show that the performance of this method is significantly improved, compared with traditional models and existing convolutional neural network classification models.