Potato Leaf Disease Detection Using Deep Learning
D. Samant, Rangoli Dhawan, Amit Kumar Mishra, Vaibhav Bora, Manoj Diwakar, Prabhishek Singh
Abstract
India is one of the largest producers of potato crop in the whole world. Potato is responsible for the significant crop production statistics in India. It's contribution is over 27.8 percent of the total crop which is grown in our country. But due to some plant diseases the health of the potato plant gets adversely affected which leads to great economic losses which directly affects the nation's economy as well. Due to illiteracy and poor knowledge of botany and plants farmers used to suffer terrific amount of economic losses. The traditional methods used by farmers were not up to the mark. In earlier times the farmers used to detect if the plant is healthy, or it has some sort of disease. This method is obviously not very productive and also, very cumbersome. So with the help of new technology, we can detect the problem very easily within fraction of seconds and it also provide very optimized results. We have a dataset called Plant village which has more than 2000 images of different leaves. In our project we will give our input image and our model will then predict weather the potato leaf is disease free or not. And if there is any disease then it will figure out whether it is early blight or late blight. We will use Deep learning in our project for optimum result. We will also use various algorithms such as ANN, CNN and apply the one which yield us the most accurate results. It was concluded that CNN gives the outstanding results amongst all with an accuracy of 98.72 percent.