A Convolutional Learning Model for Multicrop Plant Disease Detection

Crop cultivation is very important in agriculture. 70% of the people are in the agriculture field. Generally, it is estimated that various kinds of pests like insects, weeds, animals, and diseases cause crop yield losses of 20–40%. This research work looks into the possibility of developing a part of a holistic system which works for disease detection of multiple varieties of plants, not just one type of plant. Our proposed system predicts the crop variety and the leaf disease as well. Such a system shall help farmers who cultivate crops in large land mass and cultivate multiple varieties of crops at the same time. For the purpose of identifying leaf diseases in a variety of plant species, we have suggested a two-stage deep learning model. Convolutional neural networks were used for developing the system to provide the best results. For evaluation of the proposed model, we have used a dataset comprising of a collection of 0.125 million images of ten different plants and thirty-seven different diseases. The dataset contains the images of healthy leaves and disease infected leaves in order to train the model for discretion. The suggested system makes predictions about the types of plants and plant diseases, as well as offers advice on how to treat the ailments. Our CNN-based deep learning model achieved an accuracy of 96.3% in predicting the plant variety and the disease.