The manual classification of mangoes has long been a bottleneck in the mango supply chain, characterized by time-consuming efforts and subjective judgment.
Addressing this challenge, the University of the Philippines Cebu (UP Cebu) has launched the “Mango Automated Neural Net Generic Grade Assignor (MANGGA)” project, leveraging artificial intelligence (AI) to automate the labor-intensive task of sorting Carabao mangoes for the fresh export market.
Spearheaded by UP Professor Jonnifer Sinogaya, this two-year initiative is funded by the Philippine Council for Agriculture, Aquatic and Natural Resources Research and Development of the Department of Science and Technology (DOST-PCAARRD).
Collaborating with the Department of Agriculture-Region VII (DA-VII), University of the Philippines Los Baños-Postharvest Horticulture Training and Research Center (UPLB-PHTRC), and the Technological Institute of the Philippines (TIP), the team is developing advanced data acquisition technologies for mangoes.
The project’s systematic approach has yielded an extensive dataset of 10,440 images from various angles and orientations, along with ethylene concentrations from 870 individual mangoes. This comprehensive dataset has been instrumental in training an AI model for sorting Carabao mangoes.
From scratch, the MANGGA team has developed a Convolutional Neural Network (CNN) and an image data acquisition system. Preliminary training of their single-input CNN model achieved an impressive 94% accuracy in determining export suitability based on visual characteristics.
Guided by the Philippine National Standard for quality metrics, the team continues to refine the CNN and Computer Vision System (CVS) for greater efficiency in grading export-quality Carabao mangoes.
Midway through its second year, the project explores innovative preprocessing techniques and assesses multi-input CNN models to enhance precision.
The MANGGA project advocates for the adoption of smart postharvest systems within the local mango industry.
With plans for a conveyor system designed to sort mangoes based on marketability, this initiative aims to revolutionize mango grading, offering improved efficiency and safety for the fresh export market.
