RivAIrSet: A multitemporal high-resolution dataset of UAV imagery with river water area annotations

The RivAIrSet dataset was collected over a six-year period (2019–2024) along a 3 km reach of the Basento River in the Basilicata region of Southern Italy, along the industrial area of Ferrandina (MT) (40°29′57.2″N, 16°29′59.5″E). Within this section, the river channel varies in width from 90 to 200 m, with a flow regime characterized by peak discharge during autumn–winter and lower flow conditions in summer. Located in the mid-basin area, this reach is prone to recurrent flooding events, resulting in enhanced erosion and sediment transport processes that contribute to significant hydro-geomorphological hazards. RivAIrSet was acquired using Unmanned Aerial Vehicle (UAV) equipped with a nadir-oriented RGB camera following preprogrammed flight paths. The dataset contains 7,630 high-resolution RGB images (.JPG) collected during six survey campaigns under varying weather and hydrological conditions. Each image includes embedded EXIF metadata with GPS coordinates and camera parameters (F-stop, exposure time, ISO, focal length). Corresponding annotation files (.json), generated using the AnyLabeling tool, delineate river water areas through polygon coordinates labeled as “water”. Each image is paired with its respective annotation file, sharing an identical filename (e.g., DJI_11.JPG and DJI_11.json). The dataset is organised into six directories, each named according to the survey year (e.g., 2019, 2020). Within each directory, two subfolders are provided: images, containing the RGB photographs, and labels, containing the corresponding annotation files. Files are numbered sequentially within each survey to enable straightforward integration into a comprehensive dataset spanning all acquisition campaigns. CitationsUsers of RivAIrSet are kindly invited to cite the corresponding publication, as follows: La Salandra, M., Colacicco, R., Dellino, P., & Capolongo, D. (2025). RivAIrSet: A multitemporal high-resolution UAV imagery dataset for machine learning-based river water segmentation. Data in Brief, 112356. https://doi.org/10.1016/j.dib.2025.112356