Utilizing Transfer Learning Approach in Agriculture 4.0 for Banana Leaf Disease Identification

Abstract

This study explores the integration of Convolutional Neural Networks (CNNs), including the ResNet architectures, within the context of Agriculture 4.0, focusing on the prediction of diseases affecting banana leaves. Through an extensive dataset encompassing images of both healthy and diseased banana leaves, such as those afflicted with sigatoka and Xanthomonas, the research underscores the effectiveness of CNNs, specifically Vision Transformer (ViT), ResNet-18, ResNet-34, ResNet50, EfficientNetB0, MobileNetv3 Large and ResNet18 with CBAM in addressing the challenge of disease prediction in banana plants. Leveraging the capabilities of deep learning, this approach offers advanced tools for disease management, yield optimization, and bolstered food security within the agricultural sector. The trained models exhibit promising outcomes in disease prediction, highlighting their potential to support farmers and agricultural professionals in early detection and proactive management practices, thus aligning with the principles of Agriculture 4.0.