Deep Learning-Based Egg Fertility Classification With Raspberry Pi and Color Sensor Integration
Ratna Aisuwarya, Rian Ferdian, Salman Wafiq · 2025
Egg fertility detection is essential in poultry rearing to enhance hatchery success and prevent contamination by infertile eggs. Candidly is labor-intensive, involves extensive manual labor, and is prone to human error, necessitating automation. The present research foresees a CNN-based system with image processing and the Internet of Things for real-time egg fertility classification. The system consists of a funnel-shaped structure housing a spotlight, camera, and TCS3200 color sensor to determine egg type (village chicken, purebred chicken, or duck) before CNN-based fertility classification. Fertile and infertile eggs are identified using LED and buzzer indicators, with real-time results transmitted via a Telegram bot to notify the user. The CNN model effectively classifies eggs with the extracted spatial feature pattern recognition. Experimental results confirm that the CNN model performs excellent classification accuracy on fertility eggs. The buzzer and LED lights provide real-time feedback. The Telegram bot provides the classification result, allowing users to monitor egg classification remotely.