Research on the Application of Remote Sensing Mapping and Deep Learning in Geological Survey

Deep convolutional neural networks have multi-level hierarchical structure and massive parameters, which pose challenges to supervised learning algorithms for surface cover classification. In this paper, we propose a supervised learning method for geological cover classification based on a multi-attention mechanism, which incorporates the global feature attention mechanism on the basis of feature extraction. Experimental results show that the lightweight global feature attention module proposed in this paper achieves good performance in multi-scale surface cover type and boundary information recognition, which can improve the practical application of remote sensing mapping and deep learning in geological survey