Dataset for creating a red-legged seriema detection model.

This Zenodo dataset contains the data used for creating and evaluating a YOLO26-nano red-legged seriema detection model. It consists of videos of foraging red-legged seriemas (Cariama cristata), which were used for model construction. Additionally, it contains images which were extracted from these videos together with corresponding annotation files (json format). These annotation files are the result of manually drawn bounding boxes around all present seriemas. It also contains converted annotation files (YOLO format) and videos on which the model was tested after construction. This dataset contains the full data workflow from raw data to organized data usable in detection model construction. This dataset has been used in combination with the associated Github repository and was created as part of a workflow demonstrating how small-scale video datasets can be used to develop custom animal detection models for behavioural ecology and conservation applications.