A Spatial-Channel Attention-Based Convolutional Neural Network for Remote Sensing Image Classification

This paper proposes a novel model for remote sensing image classification based on CBAM-CNN (Convolutional Block Attention Module-Convolutional Neural Network). CBAM-CNN is a well-known and effective model. However, it is limited when it comes to high-level features due to its shared spatial attention mechanism and narrow sampling range in squeeze-and-excitation module (SEM). We disentangle the shared attention to channel-independent spatial attention and expand the sampling range of SEM. The results show that the proposed approach outperforms other CNN-based models on two large-scale benchmark datasets.