Artificial Neural Network (ANN) With Chan-Vese(CV) Algorithm-Based Plant Disease Detection And Classification

A country’s economy is influenced by a wide range of factors, with the agriculture, service, and food processing sectors. In our country, agriculture is one of the most significant economic sectors. The agriculture sector, however, has several difficulties, such as the different climatic conditions in other regions of our country that give rise to various infectious diseases in different plant species, resulting in a significant drop in crop productivity and income generation. Early and accurate plant disease diagnosis is crucial in improving the foundation for harvest formation. The traditional method of plant disease detection through eye observation is unsuccessful since it necessitates knowledge. Machine learning (ML) techniques are being utilized to identify and categorize plant ailments to address this issue. In this work, plant diseases can be classified using machine learning methods. These involve classifier training, classification, and stages like dataset construction, image loading, pre-processing, segmentation, and feature extraction. An artificial neural network is initially used to distinguish and find leaves in a challenging environment. The Chan-Vese (CV) technique states that symptoms can be found in images that have been segmented depending on the outcomes of the ANN algorithm. A transfer learning model trained on a dataset of ill leaves on an essential background is then given the segmented leaves. The model is also checked for rust, black rot, and bacterial plaque conditions. The findings demonstrate that the methodology is more accurate—98.22%—than the previous approach, minimizing the adverse effects of illness on agricultural productivity and promoting long-term agrarian growth. Utilizing data from PlantVillage, the model is trained and validated. The dataset includes images of plant diseases afflicting pepper, tomato, and potato plants. Fifteen different plant disease categories are used to train and evaluate the classifier. The results of the trials show that the proposed ANNCVA deep learning model can categorize the numerous varieties of plant leaves accurately and effectively.