PIO, A Large-Scale Dataset for Broiler Chicken Detection under Real Poultry Farming Conditions DATASET v.1

This dataset, named PIO, contains 1,435 high-resolution images and over 327,289 manually annotated instances of broiler chickens. The images were captured in a commercial farm and a prototype poultry house located in Veraguas Province, Panama, covering the entire life cycle of the birds, from early stages to full maturity. The dataset includes variations in population density, lighting conditions, and environmental features, making it a robust and representative resource. Annotations were performed in YOLO format to ensure high-quality object detection labels. Additionally, a folder with original videos from the poultry houses is provided, allowing the research community to continue labeling, expand the dataset, or explore new computer vision tasks. This dataset is designed for computer vision applications in poultry monitoring, animal welfare assessment, automated counting, and precision livestock farming, offering a comprehensive resource for the development and evaluation of deep learning models in poultry production systems. For comprehensive technical documentation on the dataset's collection methodology, annotation procedures, and experimental framework, please consult our associated paper in Scientific Data: https://doi.org/10.1038/s41597-026-07114-5.