Region Change Detection Model for Remote Sensing Images Based on U-net

Abstract

This paper presents a region-based convolutional neural network based on the U-net model and region change detection method, aiming to address the shortcomings of traditional remote sensing change detection methods such as poor feature characterization and long processing time. By employing a region-based object detection method and data augmentation strategy, a more generalized network model is obtained, and attempts are made to further improve training efficiency using U-net++. Experimental results demonstrate that the U-net and its derived models exhibit faster convergence and higher training efficiency, proving their effectiveness. The research methods and experimental results of this paper have important theoretical and practical significance for remote sensing image change detection.