An optimized machine learning framework for crop disease detection

The management of crops from the early to mature stage contains nutrient deficiency, monitoring plant disease, controlling irrigation, and controlling the use of pesticides and fertilizers. Moreover, lack of immunity and climate changes cause the crops and minimize the growth of agriculture due to crop disease. The identification and detection of crop diseases is the most challenging task due to less detection accuracy, overfitting, and error rate. So this research work designed a novel Krill Herd based Random Forest (KHbRF) for the accurate detection of crop disease, enhancing the performance of detection accuracy by using an optimized fitness function. The krill herd fitness function is updated to the classification layer for effective crop disease detection. Furthermore, development involves preprocessing, segmentation, feature extraction, and classification. The developed framework is implemented in the python tool, and the plant villa image dataset is tested and trained in the system. After that preprocessing removes errors and feature extraction extracts the texture features from the crop. At last, the classification layer detects the crop disease present in the dataset using the fitness of the krill herd. Additionally, attained results of the developed framework are compared with other state-of-the-art techniques in terms of detection accuracy, sensitivity, F-measure, and error.