Agriculture cyber-physical systems are becoming increasingly significant in improving crop quality and output while employing the least amount of farmland possible. Many agricultural components and methods have been automated to produce faster and superior-quality items. As a result, several methods and techniques to help prevent or reduce plant diseases have been created. Imaging analysis tools and gas sensors are increasingly being included in innovative cyber agribusiness for plant disease monitoring. This chapter develops intelligent cyber-physical system-based plant disease detection for successfully preventing and managing plant diseases and decision-making. To extract the important features linked to the sick region, a novel multi-statistical feature extraction technique has been developed. The proposed methodology is implemented using a Raspberry Pi board running Stretch OS. Metrics like detection and classification accuracy are used to assess the efficacy of the multi-statistical feature extraction methodology on the plant disease detection system.
TopIntroduction
Crop diseases detection is critical in agriculture because workers must continuously check whether the crop harvested is adequate. Taking an critical since it can create serious plant issues that influence quality of the product, volume, and performance. Plant diseases generate disease epidemics on a regular basis, resulting in widespread death and a large economic effect. Computer vision techniques may be used to do image-based automated inspection. Manual identification takes longer, requires greater precision, and can only be done in small amounts. This technique can identify early plant infections, and infection management measures may be utilized to handle pest issues while decreasing dangers to the environment and public people. Because visual monitoring plants for illnesses is less accurate and time-consuming, machine learning automates diagnosing diseases based on features extracted from photos. Image acquisition and extraction of feature are carried out at the field to be watched. Experts classify the monitoring station. Finally, the farmers are advised on how to solve the situation via text messages or a mobile app. The farm's camera sensor captures images of the plant leaves. The idea is then further processed to identify and segment the affected leaf section.
The agricultural specialist extracts features from the segmented section and sends them to him for study. The farmer could appropriately apply pesticides to the affected areas if he had precise knowledge of where the illness had spread, resulting in economic and environmental benefits. Pre-processing and segmentation are critical components of a vision-based disease diagnosis system. Attribute extraction from either the split image is required for effective classification. The process of grouping or splitting an image into various portions is known as image segmentation Khirade (2015). Different techniques can be used to segment an image, ranging from simple to complex segmentation procedures.
To extract features, color, texture, and area are used. Experts use classifiers to categorize data. The researchers Singh (2016) suggested unique image segmentation techniques for detecting as well as categorizing crop leaves illnesses automatically. The images were segmented using the genetic algorithm and classified using.
The author Dhakate (2015) suggested a method based on artificial neural networks for identifying diseases in pomegranate plants. The diseases that will be demonstrated. With a 90% accuracy rate, the proposed method produced satisfactory results. The author Aasha Nandhini (2018) presents a proposed method to detect and identify illnesses in crop using compressed sensing (CS). For several plants, the plan was tested using simulation and experimental analyses. The data were classified using SVM, and the results showed 98.5 percent accuracy.