DSST-RFNet: A Differential Spatial-Spectral-Temporal Residual Feature Fusion Deep Learning Change Detection Network for Large-Area Mangrove Gain and Recovery Monitoring using Bi-Temporal Sentinel-2 Multispectral Data

This dataset contains large-area mangrove gain and recovery maps derived from bi-temporal Sentinel-2 imagery using the proposed DSST-RFNet deep learning change detection framework. The products include a mangrove gain map for the southwest coast of Pakistan (2019-2024) and a mangrove recovery map for Everglades National Park, Florida (post-Hurricane Irma, 2017/2018-2024). Training and testing datasets and the source code are available upon reasonable request.