Development of an Egg Crack Detection System Using Computer Vision to Support Agricultural Development Through Modern Technology
Nawin Kongrugsa, Narumol Chumuang, Todsakorn Chantongpai, Suchat Rakchatying, Shridhar Allagi, Mahasak Ketcham, Thittapom Ganokratanaa
Abstract
This paper presents a design and development an egg crack detection system based on computer vision technology for the export industry. Additionally, it evaluates the system's performance in detecting egg cracks using computer vision techniques. The dataset comprises 400 images of eggs, captured using a SONY SL T -A 77V camera with an ISO setting of 320, an F-number of f/8, and an 18–135 mm f/5.6 lens. The images were taken at a resolution of 2000×3008 pixels under studio lighting conditions of 1000–1500 lumens during daylight hours. All images were stored in. JPG format, with 200 images depicting cracked eggs and 200 images of intact eggs. The images underwent digital image processing, beginning with conversion to grayscale (Gray Scale) and noise reduction using a Gaussian filter. Edge detection techniques were applied to emphasize crack details, followed by segmentation and background removal using Canny Edge Detection to isolate the eggs. To develop the convolutional neural network (CNN) model, the dataset was split into three subsets: 70% for training, 10% for validation, and 20% for testing. The model was trained to classify cracked and intact eggs. Experimental results demonstrated that the model achieved a high detection accuracy of 96.50%. This research can be applied in the food industry to improve quality control and reduce losses caused by cracked eggs.