Crop Disease Detection Using Convolutional Neural Network Algorithm

Abstract

It is essential to detect diseases early to produce crops that are healthy and this can be achieved by analyzing the leaf of the diseased plant. The primary focus of this paper is to detect plant diseases and give recommendations using the images of the diseased plant. Most disease symptoms are reflected on the leaves of plants such as Apple, corn (maize), blueberry, cherry, rice etc. Unlike earlier approaches, in this approach we process the images captured in the field using normal or a mobile phone camera by an untrained person who aims to find the disease affecting his crop. The field images gathered are processed by leaf segmentation. The goal of this project is to increase the accuracy of earlier approaches to give better results at detecting plant diseases. Using local statistical features, first classifier segments leaf from the background, Then using Convolution Neural network algorithm another classifier is trained to detect disease .Varying filters are applied to identify the plant disease. The developed algorithm is a generalized as it can be applied for any disease. The range of diseases includes common rust, cercospora leaf spots etc.