Collection and processing of data from poultry farms for anomaly detection and optimization of egg production

Abstract

Recent years have seen numerous efforts focusing on smart poultry farming, with the dual aim of maximizing poultry welfare and farmers’ profits. However, most of these initiatives either introduce high-cost systems beyond the reach of small-scale farms or are tested in controlled, laboratory-like environments that fail to account for the practical challenges of deploying a smart system on a commercial farm. Moreover, there is a notable lack of publicly available datasets containing video and audio data from commercial farms. This work introduces a low-cost pipeline for data collection in commercial poultry farms and presents a dataset of RGB, depth, and audio data collected from a farm in Nicosia. Additionally, it provides an analysis of production data provided by poultry farmers and lays the foundation for a comprehensive system designed to offer insights and warnings to poultry farmers. This system aims to ensure the welfare of the poultry and maximize farm profitability.