Apollo AI (Stage 2 Submission)

AI-powered crop stress detection platform designed to democratize precision agriculture using standard smartphone/drone imagery.

🚀 Status: Deployed Prototype

This is the Stage 2 submission for the AI500 Hackathon 2025. It demonstrates the core classification engine and user journey.

Key Features

• Frontend: Next.js 15 (App Router), React 19, Tailwind CSS.
• Backend: Python FastAPI (Serverless on Vercel) + ONNX Runtime.
• AI Model: ResNet-18 trained on PlantVillage dataset (70k+ images, 99% accuracy on demo classes).
• Visualization: Interactive 3D Drone Scanner (Three.js).
• Mobile First: Fully responsive glass-morphism UI.

🛠️ Quick Start

1. Install Dependencies (Fixed for React 19 RC):

   npm install

2. Run Development Server:

   npm run dev

3. Open Demo: Visit http://localhost:3000/demo to test the classification engine.

🤖 Telegram Bot

The bot aids farmers with instant analysis and lets them access the web app. Note: The bot uses a polling architecture and must run locally (or on a VPS), even if the website is deployed to Vercel.

1. Setup Keys: Ensure TELEGRAM_BOT_TOKEN and GROQ_API_KEY are in api/.env.
2. Start Bot:

   python bot/main.py

📄 Documentation

• Submission Description - Detailed project narrative and video script.
• Architecture - System design.
• Components - React component structure.