Crop Guard AI

Inspiration

The need to empower farmers with accessible AI technology for early plant disease detection to prevent crop losses and ensure food security.

What it does

A web application that analyzes uploaded leaf images using computer vision to detect and classify diseases across 62 classes spanning 14 major crops, providing instant results and recommendations.

How we built it

Backend: Flask with CNN trained on PlantVillage dataset, integrated YOLOv8 for leaf detection/cropping. Frontend: React/TypeScript with Tailwind CSS and shadcn-ui. Backend services: Supabase for database and authentication.

Challenges we ran into

Training accurate CNN across diverse conditions, integrating YOLOv8 for edge cases, optimizing model for web performance, balancing accuracy with processing speed.

Accomplishments that we're proud of

Developed a functional web app supporting 62 disease classes with high accuracy, automated leaf cropping, intuitive user interface, and scalable architecture.

What we learned

Computer vision applications in agriculture, effective image preprocessing, ML model deployment complexities, modern frontend development, and iterative dataset validation.

What's next for Crop Guard AI

Cloud deployment (AWS/GCP), Grad-CAM heatmaps for explainable AI, real-time video analysis, additional crops/diseases, mobile app, and IoT device integration.