Pomegranate Leaf Disease Classification using Feature Extraction and Machine Learning

When it comes to agricultural image processing, plant disease detection is considered as one of the most difficult problems to solve. This research study designs a novel and efficient framework for pomegranate disease detection. The image acquisition, scaling, enhancement as well as image segmentation and the extraction of ROIs, are all included. This method makes use of a dataset on pomegranate leaf images that has been divided into two sets; one set for training, and the other set for testing. The discovery of ROI and features takes place mostly during the deployment phase and is primarily accomplished by performing image enhancement and image segmentation process. Afterwards, a supervised learning model and a Support Vector Machine (SVM) model will be utilized in order to classify an image. The proposed framework has been developed by integrating the machine learning technologies. According to the recent research findings, the proposed framework is accurate to the degree of 95.5357% when it comes to differentiating the healthy and unhealthy leaves. The proposed framework allows to perform disease classification in a pomegranate leaf to an accuracy of 96.4286%. The datasets are obtained from Mendeley Data Total: 559 images within which 287 healthy images were identified and 272 diseased images were identified. The proposed dataset is split in an 8:2 ratio.