AgerAmicus

Inspiration

Agriculture faces increasing challenges from climate change, pest outbreaks, and yield uncertainty. Many farmers lack access to timely insights about crop health or tailored recommendations. AgerAmicus was born from the idea of building an AI-powered farming assistant that bridges this gap through accessible technology.

What it does

AgerAmicus is a multimodal AI assistant designed to support farmers and agricultural researchers. Our MVP includes: 📸 Image Classifier for detecting crop diseases from leaf images (e.g., Tomato Early Blight, Potato Healthy) 💬 NLP Chatbot that answers agricultural queries using Azure OpenAI (GPT-35 Turbo) 🎯 Simple backend API built with FastAPI 📂 Hosted on GitHub with clean modular structure and .env support

How we built it

Trained a custom CNN model on the PlantVillage dataset using TensorFlow Built a FastAPI backend with two endpoints: /analyze-image for image-based disease detection /chat for chatbot conversation using Azure OpenAI Managed secrets and config with .env and python-dotenv Used Git, VS Code, and Azure Cognitive Services to deploy and test

Challenges we ran into

Azure OpenAI setup required careful attention to deployment names and API versions Faced Git repository corruption due to OneDrive conflicts during final submission Debugged FastAPI Swagger docs not updating due to caching Had to deprioritize frontend due to time limits

Accomplishments that we're proud of

Successfully integrated two major AI features into a single working backend Deployed and tested GPT-powered chat with agricultural context Created a clean, scalable architecture for future feature expansion Learned real-world debugging, deployment, and Azure integration under time pressure

What we learned

Deep integration of Azure OpenAI with Python (non-OpenAI.com endpoints) Managing environment variables securely with .env FastAPI structure and documentation via Swagger UI Handling Git edge cases like object corruption and forced reset The value of building small, testable MVPs early

What's next for AgerAmicus

📊 Add yield prediction and fertilizer recommendation engine 🌍 Integrate weather forecasting via Azure Maps API 💾 Store chat/image history in Azure Blob Storage 🌱 Partner with local agri-institutes to pilot the assistant with real farmers

Built With

• fastapi
• html
• javascript
• python
• tensorflow