Apple Leaf Disease Detection using Deep Learning

Abstract

In general, agriculture plays a very important role in contributing to human life on earth. Agriculture acts as the major source of providing food and economic growth of a region and as known plants are affected by several kinds of diseases either by excessive use of chemicals or by bacteria, viruses and fungus. It is important to diagnose plant diseases rightly, since use of wrong chemicals to treat the disease may increase the resistance of the pathogens which affects the plants. Manual diagnosis of diseases that affects the leaves of a plant will delay the process of diagnosis and treatment. Deep Learning frameworks can be used in detection and classification of the diseases. Convolution Neural Network based (CNN) based models are used in detection of apple leaf diseases. VGG16 framework is a CNN based architecture widely used in many deep learning classifications and it is easy to implement. VGG16 is used here for diagnosis and classifying apple leaf diseases. For implementing the framework tools and modules like Kaggle Notebook, Tensorflow, and Keras used. The VGG16 model is applied to the apple leaf disease dataset collected from the Kaggle repository. The proposed model aims in reducing complexity in classifying apple leaf disease using deep learning. The proposed system shows the best validation accuracy of 93.3% on the apple leaf disease dataset. This method outperforms some existing state-of-the-art. The processing time for each image is at an average of 14s. Hence the system proposed can be used by farmers to simplify the apple leaf disease classification process and help in early diagnosis and treatment of the disease.