AMMUNet: Multi-Scale Attention Map Merging for Remote Sensing Image Segmentation

arXiv:2404.13408v1 Announce Type: new
Abstract: The advancement of deep learning has driven notable progress in remote sensing semantic segmentation. Attention mechanisms, while enabling global modeling and utilizing contextual information, face challenges of high computational costs and require window-based operations that weaken capturing long-range dependencies, hindering their effectiveness for remote sensing image processing. In this letter, we propose AMMUNet, a UNet-based framework that employs multi-scale attention map merging, comprising two key innovations: the granular multi-head self-attention (GMSA) module and the attention …

arxiv attention cs.cv image map merging scale segmentation sensing type