Early Detection of Potato Leaf Diseases using Convolutional Neural Network with Web Application

Potato is most important tuber crop all over the world, around 125 countries are cultivating it. After rice and wheat, potato is the crop that is majorly consumed by a billion of people almost every day in the world. However, the quality and quantity of potato crop is degrading due to various fungal and bacterial diseases. Detecting diseases at an early stage is challenging task due to variations in environmental factors, plant species and plant disease symptoms. Most of the traditional techniques for disease identification are expensive, sample treatment, time consuming and required experienced personnel. In recent research, several machine learning techniques have been developed to recognize the potato leaf diseases. In this work, a multi-layer deep learning model for potato leaf disease detection is developed. In the first layer, the potato leaves features are extracted from the image of the potato plant using the image segmentation technique. At the second level, a new deep learning technique was developed using a convolutional neural network [CNN] to detect fungal and bacterial diseases of potatoes. The leaf disease dataset contains 2250 images collected in real time and from the database. A web-based application is also developed for the young farmers to identify the diseases without any experience so that the precise value of pesticides can be dispensed. The proposed deep learning techniques achieved 99.76% accuracy on potato disease identification.