High Resolution Remote Sensing Image Captioning Using Group Convolution and Attentional Mechanism

This paper proposes a remote sensing image description generation method based on group convolutional neural network and attention mechanism to address the problem of a single global feature extraction scale in the model for remote sensing image scene description generation. This method mainly applies grouping convolution modules to fuse the features of different network layers and global features at different scales in convolutional neural networks. To overcome the shortcomings of traditional convolutional neural networks in feature extraction, this paper proposes a remote sensing image description generation method based on the inverse bottleneck layer convolutional neural network and attention mechanism. This method significantly increases the global receptive field of convolutional neural networks by introducing inverse convolutional layers, grouped convolutions, and large-scale convolutional kernels. This method utilizes deep separable convolutional layers to further reduce network model parameters and model operations. The experimental results show that our method can generate more accurate text statements for describing remote sensing images.