MSFECGAN-OpticalReconstruction-Dataset

The MSFECGAN-OpticalReconstruction-Dataset is a remote sensing dataset designed to support research on optical image reconstruction using multi-source data. It is developed in conjunction with the MSF-ECGAN (Multi-Source Fusion Enhanced Conditional GAN) model, which aims to generate high-quality optical time series from auxiliary sources such as SAR imagery and previous optical observations. The dataset contains temporally-aligned, co-registered multi-source inputs and their corresponding reconstructed optical outputs at 10-meter resolution and 8-day intervals, matching Sentinel-2 specifications. It is intended to facilitate the development and benchmarking of deep learning models for cloud-robust optical image reconstruction, spatiotemporal information fusion, and Earth observation applications.