Improve the reliability over classification of plant diseases using Multi Class - Support Vector Machine

Abstract

Plants have a critical part in the survival of all living things. Incorrect diagnosis of plants illness leads to overuse of pesticides, which has an impact on the kind of harvesting. The classifier is used to reduce losses in agriculture item yields and amount; however, if thorough research is not done in this strategy or categorization, it can have major consequences for crops, affecting item grade and efficiency. Crop disease categorization is crucial for sustainable farming. Physically monitoring and treating plant infections is quite tough. Image processing is employed for the identification of plant infections since it needs a large quantity of effort and a long working period. The goal of this study is to employ Multi Class-Support Vector Machine (MC-SVM) techniques to classify plant diseases. A total of 36 crop features are gathered for 683 examples, and the MC-SVM classification is used to identify 19 sickness types. The plant information collection was subsequently classified using this customized net multilayered perceptron. Its categorization efficiency was 94.1435 percent, with 643 cases properly identified & 40 wrongly identified. The deep training classifier's categorization efficiency was found to be 88.7262 percent. To improve categorization reliability, the properties of the levels of the framework must be improved.