Cassava Leaf Disease Detection Using Convolutional Neural Networks

Abstract

Cassava is a plant that is widely found in Indonesia with various benefits. One of the benefits of cassava is as a substitute for rice. According to data from the Indonesian Central Statistics Agency in 2015, cassava production in Indonesia was 21,801,415 tons a year. Lampung Province is the largest producer of cassava in Indonesia. In 2016, its production decreased due to disease attacking the cassava plant. One of the deep learning methods currently being developed is Convolutional Neural Network (CNN). This network is built with the assumption that the input used is an image. This technique can make the image learning function more efficient to implement. Therefore, this study will take advantage of the advantages of CNN, namely being able to classify an object intended for image data so that the CNN model will be used as an introduction to the four types of healthy cassava leaf and cassava leaf diseases that can be found in Indonesia. By using the Tensorflow library, the results of model trials and evaluations of cassava leaf images show an accuracy of 0.8538 for training and 0.7496 for data validation. So it can be concluded that the implementation of Deep Learning with the Convolutional Neural Network (CNN) method can detect cassava leaf disease images.