UAV Multispectral Imagery of Agricultural Fields in Finland

The dataset consists of multispectral aerial imagery collected using a custom-built unmanned aerial vehicle developed by the Agrotechnology Research Group at the University of Helsinki. The UAV was equipped with a MicaSense RedEdge 3 camera, capturing five spectral bands: Blue (475 nm), Green (560 nm), Red (668 nm), Red Edge (717 nm), and Near-Infrared (NIR, 840 nm). Image acquisition took place in 2021 and 2023 over agricultural fields managed by the Viikki Research Farm of the University of Helsinki. The dataset comprises seven flight campaigns conducted during the summer months (May–September), all flown at an altitude of 10 meters. Several crop types are represented, including barley (Hordeum vulgare L.), faba bean (Vicia faba L.), oat (Avena sativa L.), rapeseed (Brassica napus L.), and silage grass (mostly timothy, Phleum pratense L.). All imagery is already radiometrically corrected using photographs of a MicaSense reference target captured after each flight. Calibration was performed with Pix4D software to ensure radiometric consistency across campaigns. The dataset is distributed as 512 × 512 pixel image chips with either four (Green, Red, RedEdge, NIR)) or five spectral channels (Blue, Green, Red, RedEdge, NIR). Images are stored in 8-bit format with per-channel min–max scaling. The data are provided in two separate ZIP archives—one containing four-channel images and the other five-channel images. Each ZIP file contains 1,155 image chips. Apart from the number of channels, the image chips in both ZIP archives are identical, each with a ground sampling distance of 0.7 cm. A flight campaign table is also provided to summarize the data collection details. The four-channel version of this dataset is used in the currently unpublished article “Self-supervised training for high-resolution close-range multispectral remote sensing imagery” by Thomas et al.