AUTOMATED POULTRY QUALITY ASSURANCE SYSTEM FOR PRECISE INSPECTION AND SORTING

Abstract

This study introduces an automated poultry quality assurance system to address inefficiencies and inconsistencies in traditional inspection methods. Manual inspection processes are labour-intensive, prone to errors, and unsuitable for high-throughput poultry production environments. The proposed system uses real-time image capture via smartphone cameras and YOLOv8 object detection technique used to identify defects such as blood clots and feathers on chicken (Cobb 500) carcasses. Defective chickens are automatically sorted using an Arduino-controlled pneumatic system integrated into the production line. The system’s dataset was prepared through careful annotation and augmentation to enhance detection accuracy. YOLOv8 was trained over 150 epochs, achieving reliable defect classification, with results indicating high precision in separating defective products from quality chickens. The integration of image processing, real-time video analysis, and automated sorting mechanisms significantly reduced human involvement and increased operational efficiency. Limitations, such as false positives in complex scenarios and insufficient dataset diversity, highlight areas for improvement. Future research will focus on optimizing the detection model, expanding the dataset, and improving pneumatic actuation for better sorting accuracy. This work offers a scalable, cost-effective solution for poultry quality assurance, ensuring consistency and reducing wastage. Its implementation in processing plants advances automation and sets a benchmark for integrating computer vision and hardware systems. The developed system achieved an average accuracy of 95.33% in detecting quality chickens and 81.35% in identifying defects across five trials, demonstrating its effectiveness and reliability in sorting.