Vision system for detection of white root disease infection based on capacitance properties

Abstract

This paper presents the findings of Visions System performance for the detection of White root disease infection based on capacitance properties. A number of 100 latex samples representing healthy and white root infected rubber tree is tested for its capacitance value using Prototype Console Unit (PCU) developed. An optimized model for ANN using Levenberg Marquardt was designed. It is found that the hidden layer size of neuron 2 gave the best optimized ANN model with 77% sensitivity, 88% specificity, 82.5% accuracy, and uses 5 numbers of connections. A vision system based on this optimized model is developed and has the performance of 78.34% total accuracy.