ANALYSIS OF GRAPE LEAF DISEASE BY USING DEEP CONVOLUTIONAL NEURAL NETWORK

Abstract

Smart agriculture is a strategy for restructuring and reorganizing agricultural systems to ensure food security in the face of emerging climate change challenges. Diseases cause problems on agricultural development and yield and they're generally tough to control. It is necessary to have a precise diagnosis of the grape leaf diseases and preventative measures before time. In order to diagnose grape leaf diseases, this research suggests a novel recognition method that is based on enhanced convolutional neural networks. Firstly; addressed the grape leaf disease types into four categories such as Esca, black rot, Leaf Blight, and healthy which cause loss for grape industry every year. A large dataset of labeled images is collected and prepared for training. The images are typically pre-processed to enhance their features and remove any noise or artifacts that might interfere with the CNN's ability to recognize patterns. Data collection, data pre-processing, and image categorization are the three main phases of the study's approach. Secondly; Images are classified and mapped to their respective disease categories on the basis of three features namely, color and texture. Extensive experiments performed on MATLAB using CNN model AlexNet. The CNN training process used learning rate 0.0001 which produced better results and obtained better accuracy. Overall, an accurate diagnosis of grape leaf diseases and the implementation of effective preventative measures will help to reduce the impact of diseases on agricultural development and yield. This will help to ensure a sustainable and profitable grape production industry for farmers and communities.