Detection and Classification of Cracked Eggs

Eggshell crack detection and species classification are critical for ensuring quality in egg production, yet traditional manual inspections are inefficient and prone to errors. This study addresses these challenges by proposing an automated system leveraging traditional image processing techniques over deep learning for cost-effective, real-time application. Methods include Gaussian blur, Canny edge detection, and Hough Circle Transform for crack detection, alongside morphological operations and feature extraction (size, shape, texture) for classifying chicken versus duck eggs. Results demonstrate a crack detection accuracy of 90.5%–98% with processing times under 0.7 seconds on a 2.5 GHz CPU using OpenCV 4.5, outperforming YOLO in resource-limited settings. Species classification achieves over 98% accuracy, enhancing quality control in Industry 4.0 settings. In conclusion, this system offers a scalable, efficient alternative to manual methods, with potential for broader adoption despite limitations in variable lighting conditions.