A comprehensive study of plant pest and disease detection using different computer vision techniques

There are a vast variety of crops, land characteristics, fertilizers, and thereby different ranges and extents of the diseases, which need an ensemble method to detect and cure these crop diseases. The research fraternity is striving more to get a better solution for curing the crop disease which will thereby increase the yield of the crop production. With the blessings of machine vision and its supportive devices, a farmer in any region may get information in the early stage of the disease and can save his/her crop before it spreads further. For a huge farm, it is difficult for the farmers to analyze the crop at each and every place manually in time. Creating an ad hoctype sensor-based network, which monitors the soil condition, atmospheric condition, and other features of the crop from different parts of the farm, will solve the uncertain troubles faced by the farmers. Using this type of cluster-head network, the computation can be done at the master-driven side using high-speed computation. An ensemble approach is required to be developed that results in cumulative decision to classify in between the healthy and infected crop. The identification will help farmers to detect disease from its symptoms to take preventive measures. This chapter provides details of various techniques for classification of diseases in plants and the fundamental theory of detecting pests and diseases through various machine learning algorithms, the current technology available in the market. The authors have majorly focused on studies of fundamentals, current trends, and future scopes of disease detection in crops using various machine vision techniques. The chapter also showcases the future scope of machine vision in the agriculture industry.