Improving Plant Disease Identification Using Transfer Learning and Grad-Cam Visualization

Agriculture is vital to human survival, yet plant diseases threaten crop yields and food security. This project presents a deep learning-based approach for early plant disease detection using transfer learning with MobileNet and Grad-CAM for model explainability. The system was trained on a labeled dataset of plant leaf images and achieved promising accuracy, with potential for further improvement through data augmentation and extended training. Grad-CAM heatmaps enhance transparency by highlighting image regions influencing predictions. Results indicate the practical viability of AI-driven, explainable plant disease detection. Future work includes expanding datasets, supporting multi-label classification, and optimizing for mobile deployment, promoting smart and accessible agriculture.