Using Image Processing and Deep Learning Techniques Detect and Identify Pomegranate Leaf Diseases

Abstract

Objectives: To detect and identify diseases affecting pomegranate leaves using image processing and deep learning techniques. Method: A dataset of pomegranate leaf images was created with a total of 1844 images and split into 70% for training and 30% for testing. The model was trained using standard parameters like number of filters, activation functions and number of epochs and using Convolutional Neural Network algorithm for improved performance. Evaluation was conducted using standard metrics such as accuracy, precision, recall, and F1 score. Finding:The proposed work obtained the precision values for diseases Bacterial Blight, Fungal Diseases, Viral Diseases and Insect Damage as 98%, 98%, 98% and 97% respectively. Moreover, the classification accuracy obtained for diseases identification is 98.38%. Novelty:The proposed work uses private data set of diseased and healthy pomegranate leaves. Besides this the accuracy obtained is of the best class compared to the existing work in this domain.

Keywords: Pomegranate; Leaf Diseases; Image Processing; Deep Learning Techniques; Dataset Creation; Convolutional Neural Network