Crop Disease Detection Accelerated by GPU

Abstract

Agriculture is one of the vital occupations in the world. A major problem that a farmer is facing is the plants getting affected by diseases. For preventing the loss in the yield, it is very much important to detect the disease in the crop. Monitoring the diseased crop manually becomes very time-consuming and difficult because if the farm is large the workload for the farmer becomes more and sometimes it cannot be accurate as it is done manually. If the disease is nonnative, many times farmers are not aware of it. Hence, with the help of technologies like image processing, crop disease can be detected manually. Here, the chapter deals with the same to detect crop disease. It consists of image acquisition in which data collection is done, then the image is preprocessed, segmentation of the image is done, then features are extracted after this the disease is classified with the help of some classifier. In image preprocessing the RGB photo which is captured by the camera is converted into a grayscale image for accuracy of the result. Image Segmentation the image is partitioned into a range of pixels with admiration to their depth levels. In feature extraction, we can use different techniques, such as gray level co-occurrence matrix, local binary pattern, color space, etc. Classification is used to differentiate between different diseases, it mainly uses SVM, ANN, and SAS classifiers. This chapter presents a study and comparison of robotic applications for agriculture which mainly focuses on AI-based solutions for crop disease detection.