Optimized Matrix Feature Analysis – Convolutional Neural Network (OMFA-CNN) Model for Potato Leaf Diseases Detection System

Abstract

One of the most often grown crops is the potato. As a main food, potatoes are prioritised for cultivation worldwide. Because potatoes are such a rich source of vitamins and minerals, we can create a robust system for food security. However, a number of illnesses delay the growth of agriculture and harm potato output. Consequently, early disease identification can offer a better answer for effective crop production. In this research work aim is to classify and detect the potato leave (PL) diseases using OMFA-CNN deep learning model. An optimized matrix feature analysis-CNN deep learning model for PL disease detection is implemented. In the first phase, the PLs features are extracted from the potato leave images using K-means clustering image segmentation method. At the last phase, a new OMFA-CNN model are proposed using CNN to classify virus, and bacterial diseases of PLs, The PL disease dataset consists 2351 images gathered in real-time and from the Kaggle (PlantVillage) dataset. The implemented OMFA-CNN model attained 99.3 % precision and 99 % recall on potato disease detection. The implemented method is also compared with MASK RCNN,SVM and other models and attained significantly high precision and recall.