Prediction of Maize Leaf Disease Detection to improve Crop Yield using Machine Learning based Models
Ruthvik Kilaru, Kommisetti Murthy Raju
Abstract
In India, agricultural crops such as rice, wheat, cotton and maize are largely produced, which leads to a rise in the Indian economy, where most Indian people are largely dependent on agriculture. Among the crops, maize is one of the most important crops, because it is the main source for energy in food for humans and yields high productivity all over India. But, the crop yield production of maize is affected due to the diseases in plant leaves. The farmers face the problems in controlling and identifying the plant diseases that affect the quantity and quality of maize crops in high yield production. To avoid this huge loss and increase the maize crop yield productivity, it is essential to identify the diseases at an early stage. Therefore, an automated disease diagnosis system for maize plants is proposed in this research work. There are four stages presented in the work such as pre-processing the input data, segmenting the affected areas of maize leaves, extracting the features and prediction of disease. In this work, supervised Machine Learning (ML) techniques are implemented to predict the diseases of maize plants. YOLO architecture is used for the segmentation process, where Discrete Wavelet Transform (DWT) is used for extracting the features. An input image is taken from Kaggle dataset and experiments are conducted to test the efficiency of ML techniques in terms of accuracy, precision, sensitivity and specificity. The results proved that the Support Vector Machine (SVM) techniques provide better performance than other ML techniques in terms of various parameters for detecting the maize diseases.