ML Based Disease Identification and Grain Classification of Rice

Abstract

In general agriculture is one of the main sources of income for the farmers and Indian economy greatly
depends on agriculture growth and development for better production. Almost three fourths of the world require
rice production for their survival and this is cultivated almost all over the world, mostly in Asian countries.
However, the farmers have been facing with some continuous challenges for centuries, such as different diseases
of rice. If those diseases are not identified in the early stages, there will be a huge loss for the farmers as well as
human beings who wish to consume that product. If the plant disease is identified in the early stages, it will be
very helpful for the agriculture specialist or farmers to boost up the crop by taking necessary preventive steps
and try to increase the profit. Normally the plant experts or specialist try to find out the plant or leaf illnesses
based on external symptoms examination, but sometimes this may not give accurate reports. As we all know that
the structure of rice plant diseases and insects is very minute and hence it is complex task to predict the diseases
and different species in the plants and try to take necessary steps by spreading the pesticides or insecticides for
the plants.In order to overcome all these problems, we try to design an application which can able to identify the
plant disease from the affected part of crop image and then find out remedies for that disease. At present, it is
very interesting to design the deep intricate neural network (CNN) is the latest image recognition solution. Here
we try to gather several infected rice plant images and apply ML Algorithm and CNN model to identify the
disease name and also find out the necessary preventive measures for that plant. Nevertheless, manual detection
of disease costs a large amount of time and labor, so it is inevitably prudent to have an automated system to
detect disease. To solve the above problem, we are developing a Machine Learning model using a CNN
algorithm to detect the rice crop disease using the image and provide a suitable remedy. By conducting various
experiments on our proposed model, we achieved a classification accuracy of 97.17% and 99.45% when
applied to the test dataset. These remedies give information on pesticide use to control the disease. As an
extension for the current application we try to try to find out the characteristics of rice grain and try to classify
the grain name based on shape, size and color. Finally we try to conclude that proposed dataset was trained with
a range of different machine learning algorithms and achieved an accuracy of 91.30% on Decision Tree
Classifier.