Empirical Study on Rice Plant Disease Detection Using Machine Learning Architectures

Rice is deliberate as a major source of food among the rural population and also it is considered and the second most cereal crop cultivated over the world. Over the past decades, rice crops are crucially admitted as one of the powerful energy streams for the production of resources. Rice plant diseases are considered as a raising factor behind the agricultural, economic and communal loss in the upcoming development of the agricultural field. Identifying disease from the images of the plant is one of the interesting research areas in computer and agriculture field. This paper presents a survey of various image processing techniques using machine-learning approaches to identify region and classes of rice plant diseases based on images of diseases affected rice plants. Especially those techniques identify Leaf blast, Brown spot, sheath blight and bacterial blight disease of the plant on basis of the leaf characteristics. Further it analyses the noise reduction and segmentation algorithm to reduce the interference of complex background with the detection of target blade in the image by charactering it on basis of texture, color and shape. Feature identified on basis of the supervised and unsupervised learning algorithms to screen the seedling on the feature identified but it has some limitation. Towards effective identification, deep learning architecture based Convolution Neural Network has been outlined which can be effectively employed to determine the different sizes of the dataset on rice plant diseases.