 Our proposed method is an end-to-end deep learning framework for change detection in remote sensing applications. It uses an improved UNET++ network to fuse global and fine-grained information from image pairs to generate more accurate change maps than previous methods. This method has been tested on VHR satellite image datasets and has demonstrated superior performance compared to other state-of-the-art CD methods. This article was authored by Daifeng Pang, Yongjun Zhang, and Haiyang Guan.