Determination of Citrullus Lanatus “Sweet-16” Ripeness Using Android-Based Application

Abstract

Watermelon is one of the most mouth-watering fruits that people like to eat, especially when it comes to summer-a nondestructive way of determining the ripeness of watermelon considered as a challenge for its customers. This study addresses the problem of identifying between ripe and unripe watermelon using an android mobile to be available remotely. The application of a scientific strategy for determining ripeness is through image processing, which is a more capable, non-destructive, and cost-effective method. Classified samples of Sweet-16 watermelon from the farm and wet market were processed using Open-CV Python and running Tensorflow as the backend for Keras for building and training the CNN classifier. Classification of Sweet-16 watermelon is Unripe and Ripe, and Unknown. The study achieved an overall accuracy of 89.52% regardless of the position of the watermelon as captured.