Analysis of Methods of Machine Learning Techniques for Detection and Classification of Brown Spot (Rice) Disease

Abstract

Rice is a major crop and the increased production is pertinent to ensure our food security. Modern techniques and advancements are required in agriculture to enable us to improve productivity and environment-friendliness as well as improve their farming conditions. Automatic disease detection techniques can help in identifying the various rice plant diseases. In agriculture, image processing is commonly used to obtain valuable information about crops. Images are often considered a source of information and data. Machine learning (ML) is a promising modern technique for image processing, and it has been successfully used in various areas such as agriculture. This paper aims to study the various research efforts that are focused on this technology in the field of agriculture. One particular class of ML that is commonly used in this work is convolutional neural networks. Here convolutional neural networks (CNNs) have been developed over the support vector machine (SVM) techniques to identify rice diseases (Brown spot) and measure their accuracy. Their potential applications include the development of screening tools and solutions for agricultural production. This paper presents an overview and research outcomes using SVM and CNN techniques in the study of agricultural problems. This study includes a dataset of 1488 healthy leaves and 523 brown spot leaf data samples. The research outcomes are based on studies and development using SVM and CNN. ML techniques give 82% accuracy using the SVM classification method, while the CNN method gives 95% accuracy.