Efficient Classification of Pomegranate Diseases Using Deep Learning Models and Interactive Visualization

Abstract

This work tackles the growing hazard of pomegranate infections by utilizing deep learning for early detection and control. We created a strong classification system by implementing models with TensorFlow, Keras, and NumPy, as well as a user-friendly interface with Python and Streamlit. Five CNN architectures were evaluated: ResNet50, VGG16, DenseNet 161, DenseNet 121, and EfficientNet B0. DenseNet 161 and DenseNet 121 outperformed the others due to improved feature propagation, gradient flow, and accuracy. These models accurately detect complicated disease patterns, allowing for timely intervention to reduce crop losses and increase agricultural production. By incorporating AI-driven solutions into precision agriculture, this study provides farmers with a dependable and practical tool for disease diagnosis and prevention. The findings demonstrate the possibility of deep learning in plant disease management, paving the path for future AI-based advances in precision farming that would ensure global sustainability, economic stability, increased yield, and food security. Future developments could improve model accuracy and broaden the system to include more crops.