ANN Diagnostic System for Various Grades of Yellow Flesh Watermelon based on the Visible light and NIR properties

• Published in: 2018 4th International Conference on Electrical, Electronics and System Engineering (ICEESE) pp. 70 - 75
• Main Authors: Abdullah, Noor Ezan, Korlina Madzhi, Nina, Azamudin Abg Yahya, Abg Mohd, Afhzan A. Rahim, Azraa, Rosli, Anis Diyana
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-11-2018
• Subjects: Adaptive optics; ANN; Classification algorithms; Integrated optics; Neurons; optical; Optical sensors; Sensitivity; SPSS; Training; watermelon
• DOI: 10.1109/ICEESE.2018.8703498

There are various traditional methods to identify the quality of the watermelon such as ripeness, grades and others. Amongst of them were from destructively technique and may need the knowledge from skillful person. The aim of this study is to develop an intelligent system that able to classify the grades of ripe yellow flesh watermelon using Artificial Neural Network (ANN) as the classifier system. This intelligent system is generated using MATLAB through three selected training algorithms which are Levenberg-Marquardt, Scaled Conjugate Gradient and Resilient Backpropagation. The classifying technique is made based on the optical properties (VIS/NIR) for yellow watermelons. A high percentage of accuracy had been achieved in classifying the grades of the yellow watermelon via Levenberg-Marquardt training algorithm. It can produce optimum and better output despite its lower number of connections by having a 86.7% sensitivity and 80% accuracy.