MUMUCD: A Multimodal Multiclass Change Detection Dataset

This work introduces the multimodal multiclass change detection (MUMUCD) dataset, which comprises 70 globally distributed georeferenced bitemporal pairs obtained by multiple spaceborne sensors. Acquisitions over heterogeneous terrains (e.g., urban, rural, forests, deserts) for the period 2019–2024 are processed to create the first large-scale curated dataset combining synthetic aperture radar (SAR, Sentinel-1), multispectral (MS, Sentinel-2), and hyperspectral (HS, PRISMA) data with ancillary information. Provided with a resampled image resolution of 10 m and a size of 1536 $\times $ 1536 square pixels, MUMUCD allows to extract several thousands of nonoverlapping patches with size 128 $\times $ 128 square pixels, enabling data-intensive machine learning (ML) applications. Seasonality plays a critical role in the analysis of environmental data, so we carefully selected scenes representing all times of the year and a variety of geographical contexts relevant to key impact sectors, taking also into account the availability of PRISMA acquisitions. While scenes with low cloudiness are prioritized, cloudy pixels are not excluded as different data combinations (e.g., SAR/optical) can be exploited to mitigate the atmospheric effects. Beyond coregistered data, we include surface elevation, land cover (LC), and binary change maps obtained by processing the Dynamic World (DW) dataset, based on the provided MS data. Benchmarking of machine and deep learning (DL) algorithms indicates that the provided labels should be augmented to get the most out of the multiple modalities. MUMUCD is meant to fill a research gap by offering multisensor and task-oriented data, which make it ideal to fine-tune standard and foundational ML models for a variety of tasks, and even unique when these tasks involve CD or HS measurements.