Drone raw images of cattle in french grazing areas

The updated data are part of the Horizon Europe project ICAERUS regarding the livestock monitoring use case in a task where the objective is to test, optimize and scale up models regarding animal counting (cattle and sheep). More information here : https://icaerus.eu/ The dataset encompasses around 900 raw .jpeg drone images of grazing areas where cattle graze collected between June and August 2023. Drone used were Mavic 3 Enterprise and Thermal. Data collection is underway, with the aim of collecting images throughout the year and on several farms, to capture variability in animal and background colors and brightness conditions. Image tags (“cattle” vs “no cattle”) are not available for the moment but will be in the next months in next versions of this dataset. There is a strong imbalance between images with “cattle” and image with “no cattle” representative of areas to monitor.  The nadir images were collected during flight planned with DJI Pilot 2 at a constant altitude regarding the take-off position (30 m, 60 m, 100 m).  The data are organized by a first directory by farm where the images were collected and then with one directory by flight planned.  A summary is available in the Table_summary.xls. Name of the directory of each planned flight is defined such as DJI_YYYYMMDDHHMM_XX with the date (YYYYMMDD), the hour in UTC+2 (HHMM), and XX representing a mission number.  .exif data of each images provide many information regarding the drone (GPS position, absolute and relative altitude, gimble information, speed etc.). Further details will be added in the next versions of the dataset. The authors of the dataset are opened to any collaboration regarding animal counting models. For more information, please contact: adrien.lebreton@idele.fr