Classification of Red Lady Papaya and Native Papaya Sweetness Using CNN Method

The Philippines, an archipelagic nation renowned for its tropical fruits, boasts two primary papayas (Carica papaya L.) varieties: the commercially popular Red Lady papaya and the indigenous Native papaya based on UP Diliman Journal publication. Accurately classifying and predicting their sweetness[1], [2], [5], [11], a crucial quality parameter for consumer acceptance and market value, poses a challenge due to destructive and subjective traditional methods. This study presents a novel, non-destructive approach for the sweetness classification of Red Lady and Native papayas using Convolutional Neural Networks (CNNs)[6], [7], [12], [23]. A dataset of papaya images paired Lady with corresponding Brix (°Bx) values, a measure of soluble solids content and sweetness indicator, was acquired Lady from farms across the Philippines [20], [21]. CNN models were trained and optimized on this dataset to differentiate between Red Lady and Native papayas and further classify their sweetness levels (low, medium, high) based on visual cues. The proposed CNN method achieved an accuracy of 94.5% for variety classification and 88.2% for sweetness level ion, significantly exceeding the performance of traditional methods. This research paves the way for rapid, objective, and non-destructive sweetness assessment in the Philippine papaya industry, empowering farmers, distributors, and consumers to make informed decisions and optimize fruit quality throughout the value chain. References were taken from the giant sources of reliable references, namely the MDPI, Scopus, ScienceDirect, and Google Scholar.