Cassava Disease Categorization Using Deep Neural Networks

Abstract

In this research, we are using convolutional neural networks (CNNs), machine learning algorithms and transfer learning. In many tropical and subtropical climates, cassava is a staple food crop that gives smallholder farmers a significant source of income and nourishment. The crop is vulnerable to a wide range of illnesses, which can significantly reduce yields and have a negative influence on the economy and food security. In this research, we suggest a deep neural network method for automatically classifying cassava illnesses. Prior to pre-processing the pictures to extract pertinent features, we first gathered a dataset of cassava leaf images that had been labelled with disease names. The GUI will be user friendly in that it will just take pictures of cassava leaves and then some tools of machine learning and deep learning will be used, then it will predict the disease type and show disease. There are five disease is “Cassava Bacterial Blight”, “Cassava Brown Streak Disease”, “Cassava Green Mottle”, “Cassava Mosaic Disease” and “Healthy”. Our findings shown that the suggested approach can classify cassava illnesses accurately, with a total classification accuracy of 92.6%. This strategy may aid in the early detection and treatment of cassava diseases, reducing yield losses and enhancing food security in the afflicted areas. Keywords: convolutional neural network; neural network; cassava leaf disease detection, Transfer learning