Automated Inspection System for Crop Disease Detection

This project aims to advance agricultural practices through the development of a sophisticated plant disease detection system, focusing on critical crops such as tomatoes, potatoes, and peppers. Utilizing InceptionV3 as the base model for image detection/classification, our approach integrates high-level feature extraction with enhancements inspired by VGG and ResNet architectures for improved accuracy in identifying plant diseases. The system employs a novel architecture that freezes the base model layers, incorporates convolutional layers for feature processing, utilizes batch normalization for training stability, and introduces simulated ResNet-style skip connections to overcome the vanishing gradient problem. A dense prediction layer finalizes the classification task, catering to the dynamic nature of agricultural environments. Additionally, this project explores the integration of blockchain technology to secure the web interface, ensuring data integrity and transparency in the dissemination of disease detection results. This dual-faceted approach not only enhances the efficiency of disease identification but also establishes a reliable and secure platform for agricultural data exchange, setting a new standard for technological applications in sustainable farming. The predictive model represents a significant advancement in the agricultural domain, offering the potential to improve crop yield and sustainability through early and precise disease detection. By enabling timely and accurate interventions, the model contributes to healthier crops and reduced losses, aligning with sustainable farming practices.