Disease Detection in Cactus (Beles) via the Use of Machine Learning: A Proposed Technique

Machine learning is a crucial technology that may assist individuals in various fields (agriculture, healthcare, household, transportation, etc.) and stages of life. Machine learning improves performance precision (prediction). It utilises numerous data formats (picture, video, audio, and text) for varied applications and purposes. Our effort has centred on the early detection of cactus diseases to prevent the quantitative and qualitative decline of crop yield. To do this, we have utilised photos of diseased and healthy cacti. Using the imadjust, guided filter, and K-means clustering approaches, the images were improved, noises were removed, and the images were segmented to generate a better model. After executing each technique and evaluating their performance, these picture preparation techniques were picked from a large pool of options. As part of the model creation process, feature extraction approaches (Color histogram, Bag of features, and GLCM) were used to extract colour, bag of features, and texture features, respectively. After testing the model with these features, a bag of features was determined to be the best for generating a better model, hence they were chosen as the model's features. Our proposed machine learning model will be developed utilising a bag of features and linear SVM. We will use VGG16 to clear and enhanced the image quality. Other machine learning techniques will tried to train and evaluate the model for disease detection, however linear SVM will demonstrate the highest performance (97.2%) as we analyze and predict it from the previous work (average).