AgriVerseAI is a state-of-the-art, AI-powered agricultural platform designed to bridge the gap between traditional farming and modern technology. Built with React 19, FastAPI, and TensorFlow, it empowers farmers with real-time disease diagnosis, localized weather intelligence, and a direct-to-consumer marketplace.
Traditional agriculture faces significant challenges:
- Delayed Diagnosis: Crop diseases can destroy entire yields before a specialist can identify them.
- Information Gap: Farmers often lack precise, hyperlocal weather and soil data.
- Market Inefficiency: Middlemen take a significant cut, reducing farmer profits.
- Accessibility: High-end agricultural tech is often complex and non-localized.
AgriVerseAI solves these by providing a "Digital Farm Assistant" in every farmer's pocket.
Upload or take a photo of a leaf to receive an instant diagnosis powered by a custom CNN (Convolutional Neural Network).
- Precision: 95%+ accuracy for common crop diseases.
- Actionable Advice: Instant treatment recommendations in multiple languages (English & Kannada).
Real-time weather data tailored to the specific field location.
- Frost & Rain Alerts: Predictive warnings to protect sensitive crops.
- Planting Windows: Data-driven advice on the best time to sow or harvest.
A specialized LLM-powered chatbot trained on agricultural data.
- Voice Support: Speak to Bhoomi naturally.
- 24/7 Support: Expert advice on irrigation, fertilization, and pest control.
A direct-to-buyer platform that eliminates middlemen.
- Farmer Dashboard: Easily list crop yields for sale.
- Buyer Dashboard: Browse and purchase fresh produce directly from the source.
- Secure Transactions: Built-in verification and request tracking.
- Framework: React 19 (Vite)
- Styling: Tailwind CSS + Glassmorphism
- Animations: Framer Motion + GSAP
- State/Auth: Firebase Authentication & Firestore
- Engine: FastAPI (Python 3.10+)
- AI/ML: TensorFlow 2.x + Keras — MobileNetV2 fine-tuned CNN (
app/backend/) - Computer Vision: OpenCV + Pillow
- Database: Firebase Cloud Firestore
Note on language stats: GitHub reports ~87% TypeScript because of the React 19 frontend. The Python ML pipeline — MobileNetV2 fine-tuning, TensorFlow inference, and the FastAPI serving layer — lives entirely in
app/backend/. That is where the model training, CUDA acceleration, and low-latency inference code reside.
AgriVerseAI features a robust triple-tier security system:
- 👩🌾 Farmer (User): Access to AI tools, weather, and the marketplace to sell yields.
- 🏬 Buyer: A curated dashboard to find, track, and purchase fresh crops.
- 🛡️ Admin: Full system oversight, user management, and global alert broadcasting.
graph TD
User((User)) -->|React 19| Frontend[Vite Frontend]
Frontend -->|Firebase Auth| Auth{Authentication}
Auth -->|Verified| App[Main App State]
App -->|Predictions| Backend[FastAPI Backend]
Backend -->|Model Inference| TF[TensorFlow C++ Engine]
App -->|Data| Firestore[(Cloud Firestore)]
App -->|Images| Storage[(Cloud Storage)]
- Node.js 18+
- Python 3.10+
- Firebase Project Account
cd app/frontend
npm install
npm run devcd app/backend
python -m venv venv
source venv/bin/activate # venv\Scripts\activate on Windows
pip install -r requirements.txt
python api.pyCreate a .env file in both frontend and backend using the provided .env.template files.
- Drone Integration: Aerial field scanning for large-scale analysis.
- IoT Sensors: Real-time soil moisture and NPK monitoring.
- Blockchain Ledger: Transparent supply chain tracking for organic produce.
Developed with ❤️ by the AgriVerseAI Team.