Vision-based machine-mediated mango sorting system using OpenCV and convolutional neural network with tensor flow

• Main Authors: Castillo, Karlos Leo F., Cepria, KevinJeff T., Liwag, Ryan Joshua H., Rapio, Anfernee S.
• Format: text
• Language: English
• Published: Animo Repository 2017
• Subjects: Mango; Neural computers
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/8249
• Institution: De La Salle University

A large part of the Philippine economy depends on the export of fresh produce to other countries. Among these major products is the mango, a yellow-coloured sweet fruit that is prone to bruises and other skin defects and are thus carefully scrutinized and screened by farmers to ensure the quality of their output. The standard of what constitutes an export-quality mango however can vary significantly among farmers throughout the day. The group therefore seeks to develop a system that would aid farmers in examining their mango produce to increase their output and to ensure standardization in the quality of the mangoes they export. The system would be equipped with OpenCV to analyze the images taken from the mango and tensor flow neural network to countercheck the features of the skin of the subject with the store desired characteristics to determine the status of the mango that is being examined by the system. The improvement in the mango selection process would translate into higher quality outputs of farmers and in turn would increase the demand for local products outside of the country, creating a better economy for the Philippines and for Filipinos.