Potato Leaf Diseases Detection Using Convolutional Neural Networks

Abstract

We know that the root of the potato tree is used as a food in many countries of the world. It is called potato. Scientific name is “Solanum Tuberosum”. It is from the Solanaceae family. The Solanaceae family prefers cool weather to grow. The best is between 15 to 18 degrees and required PH for soil is 5.5 to 6.0. More temperature can make an effect for growing in tropical Africa, potatoes are cultivated in high lands about 1500 to 3500 meters above sea level. The year “2008” was celebrated as “international potato year” which was organized by a United States organization dated 18th October, 2007. The sector of agriculture has been now a key backbone to Bangladesh’s economy. Bangladesh has a large population who take potato as a side meal with “Rice”. As potato cultivation is huge it has several diseases. The most common is late blight and early blight. It is quite tough to detect potato’s disease from sightseeing. That's why we used different kinds of sensors for detecting those diseases. Detecting those diseases through AI we can find an easy solution. And also it will help our farmer to take necessary steps in time. And they will be able to grow as their target. To grow healthy crops, we need to find out the problem, to maintain the health of the food. And by using technology now it is very simple. Here we decided to use convolutional neural networks. It is a class of deep neural networks, in deep learning, which is applied commonly to analyzing visual imagery. We will use the image of potato leaves to analyze if it is healthy or not. In this segmentation approach and utilization of support vector machines demonstrate disease classification over 2152 images with an accuracy level of 98.29%. Thus, our proposed approach presents a path toward automatic disease diagnosis of plants on a great scale.