Abstract:
This paper explores the use of deep learning algorithms in identifying diseases affecting
Bangladesh jujube leaves, with the goal of improving crop yields and yields’
management. A dataset comprising 2400 images was curated, encompassing four target
attributes: These are the diseases such as Jujube Sun Burn, Jujube Anthracnose, Jujube
Fresh, and Jujube Brown Spot. Applying the steps of data augmentation techniques, the
dataset was preprocessed to make the model more effective. Deep learning models such
as Xception, VGG19, InceptionV3, MobileNetV2 and a Custom CNN was used. Hence,
MobileNetV2 was identified as the model with the highest accuracy of 98.75% higher
than other models for disease classification. As a result, this study has proved that
transfer learning produces high accuracy and efficient models for detecting diseases to
implement the models practically for disease management in agricultural fields for
perpetuating sustainable farming practices and food security in Bangladesh.
