Visual Monitoring of Poultry Farming Based on Deep Learning

Considering that a significant portion of feed ingredients required in Iran's poultry industry are imported—accounting for approximately ۷۰% of production costs—various methods have been developed to enhance the efficiency of these valuable inputs. The application of image processing techniques for monitoring industrial poultry farms can serve as one of the most practical methods to improve performance and optimize productivity, as it can capture and process information that conventional techniques cannot, or that involve temporal, financial, and biological constraints. In the present study, the performance of ۷۰۰ pieces of Aryan broiler chicks (a national strain) was monitored from day ۱ to ۱۰ of age using image capture and YOLO deep learning algorithm. The obtained results indicate that highly valuable information—such as chick count, flock dispersion area, and key graphs including growth rate and flock uniformity curves—can be extracted using this method. Additionally, technical considerations regarding imaging, camera positioning, and other limitations were identified, which will be useful for further research. Although this study was conducted as an initial evaluation of the algorithm response and feasibility of the technique, the results were even more promising than expected, indicating a bright path to further research on artificial intelligence applications in the poultry farming industry.

کلیدواژه ها:

Artificial Intelligence ، Image Processing ، Poultry ، Aryan Broiler ، Food Security