Cnn based ripening and maturity monitoring system for musa acuminata (lakatan) using raspberry pi 4/ Kervin D. Ayson, Arnel Jr Y. Hilario, Erwin C. Legaspi, Jamelle Brian C. Martinez, Jayson A. Morados, and Harish Lodrianne A. Saunar.--

Bachelor's thesis

College of Engineering.-- Bachelor of science in electronics engineering: Technological University of the Philippines, 2025.

Includes bibliographic references and index.

Efficient and safe banana ripening methods are urgently needed to meet consumer demand and minimize postharvest losses. This study presents an automated system for accelerating and monitoring the ripening process of Musa acuminata (Lakatan) bananas, emphasizing food quality and safety. The system utilizes Ethephon as a safer alternative to traditional ripening agents and features a custom-designed three-layer acrylic and steel prototype equipped with sensors for humidity, ethylene gas, and temperature, as well as a camera module for image-based maturity assessment. Experimental results showed that Ethephon treatment reduced the average ripening time from 126.48 hours (natural ripening) to 66.36 hours, while maintaining higher humidity (79.35%) and lower temperature (32.15°C) during ripening. Bananas treated with Ethephon exhibited uniform ripeness and improved safety compared to conventional methods. The integration of a Convolutional Neural Network (CNN) enabled automated classification of banana maturity stages, supporting real-time and objective monitoring via a Python-based LCD interface. The system's performance was validated by a local banana distributor and found to be compliant with ISO Standard 3959:1977. Overall, this technology offers a practical, efficient, and safe solution for banana producers, enhancing quality control, reducing waste, and supporting consistent postharvest management.