Anemia Detection using Deep Learning (ResNet50 + React + Flask)

A non-invasive anemia detection system using conjunctiva and nail bed analysis

🔍 Overview

This project provides a web-based solution for detecting anemia through eye a nd nail images using:

• Deep Learning: ResNet50 models for image classification
• Frontend: TypeScript, React.js, with modern UI/UX
• Backend: Flask CORS API serving predictions

📸 Screenshots

Landing Page

Detection & Diet Recommendation

Personalized Diet Plan

Information Hub

FAQ Section

🛠 Tech Stack

Component	Technology
Frontend	React, TypeScript, TailwindCSS, Vite
Backend	Flask, Python 3.10+
AI Models	TensorFlow/Keras (ResNet50 architecture)
Build Tools	npm, pip, git

🚀 Installation & Setup

Prerequisites

• Python 3.10+
• Node.js 16+
• Git

1. Backend Setup

cd backend
pip install -r requirements.txt

# Set environment variables
export EYE_MODEL_PATH="./path_to/eye_model_ResNet.keras"
export NAIL_MODEL_PATH="./path_to/nail_model_ResNet.keras"

# Run server
python app.py

2. Frontend Setup

cd frontend
npm install
npm run dev

3. Model Setup (Optional)

1. Download datasets from Google Drive
2. Run generators:

python backend/generators/Eye_ResNet50_Generator.py
python backend/generators/Nail_ResNet50_Generator.py

Note: Update paths in generator scripts before running

🌟 Key Features

• Dual-Model Analysis: Combines eye conjunctiva and nail bed assessment
• Responsive UI: Mobile-friendly interface
• Confidence Scoring: Provides prediction confidence percentages
• Developer Friendly: Well-documented code with type hints

🧪 Testing the System

1. Access frontend at http://localhost:5173
2. Upload eye/nail images through the interface
3. View real-time predictions

⚠️ Important Notes

• Run frontend and backend in separate terminals
• Model paths must be correctly configured
• For production, use environment variables for sensitive data

🤝 Contribution

Contributions welcome! Please:

1. Fork the repository
2. Create your feature branch
3. Submit a pull request

📧 Contact

For support or questions:

• Email: syedabdulkareemahmed@gmail.com
• GitHub: @92kareeem

📜 License

MIT License - See LICENSE for details