MOSAIC-SEN2-CC: A Multispectral Change Captioning Dataset

📢 This dataset is from "Introducing MOSAIC-SEN2-CC: A Multispectral Dataset and Adaptation Framework for Remote Sensing Change Captioning" paper published in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (JSTARS), 2025.
📄 DOI: 10.1109/JSTARS.2025.3615113

Although the state-of-the-art methods focus on predicting captions from RGB image pairs, change captioning in multispectral images has not been investigated yet. For this purpose, we created a new MOSAIC-SEN2-CC dataset, which contains 5232 pairs of multispectral (MS) images captured from Sentinel-2 satellites and 26 160 change captions over a 12-month period. Our dataset consists of a total of eight categories, namely Wildfire (WF), Flood (FL), Wetland (WET), Green Field (GF), Glacier (GL), Urban (UR), Agriculture (AG), along with a No-Change (NO) category.

For more information, please visit: https://github.com/ChangeCapsInRS/MOSAIC-SEN2-CC