CDMD: An Efficient Crop Disease Detection and Pesticide Recommendation System Using Mobile Vision and Deep Learning

The traditional agriculture systems are less suitable for better crop productivity. The use of modern technologies such as mobile vision and deep learning, can help in solving some of the farming related problems for improving the crop growth. In this paper, we primarily focus on developing a user-friendly system for crop disease detection and pesticide recommendation, called CDMD to improve crop productivity. We employ a pre-trained deep learning model Visual Geometry Group (VGG)-13 for learning the features of various diseased and non-diseased images. CDMD also recommends the appropriate pesticide based on the type of disease detected. CDMD provides an average accuracy $$\sim $$ 95.12% with loss 0.1607 for 24 different diseases of Plant Village dataset which seems to be promising. A user-friendly android app has also been developed as a prototype of the proposed system. It currently supports both English and Hindi languages as a user interface.