🖐️ PalmAuthAI – AI-Based Palm Recognition System

PalmAuthAI is a real-time biometric authentication system that recognizes users based on palm landmarks and texture analysis using Machine Learning & Computer Vision.

🚀 Built With:
✅ MediaPipe – Extracts 21 key palm landmarks
✅ HOG (Histogram of Oriented Gradients) – Captures palm texture
✅ SVM (Support Vector Machine) – Trained to classify palm features
✅ SQLite – Stores user feature data for authentication
✅ Flask API – Allows authentication through HTTP requests

---

📌 Project Structure

PalmAuthAI/               
│── data/                       # Stores SQLite database
│   ├── palm_auth.db            # Database file storing user features
│── models/                     # Trained ML model & mappings
│   ├── palm_recognition_model.pkl  # Trained SVM model
│── src/                        # Source code for feature extraction, training, & API
│   ├── db_setup.py             # Sets up SQLite database
│   ├── store_features.py       # Captures palm features & saves to database
│   ├── train_model.py          # Trains SVM model using stored features
│   ├── authenticate.py         # Performs real-time authentication
│   ├── api.py                  # Flask API for authentication
│── requirements.txt            # Python dependencies
│── README.md                   # Documentation

---

📌 Installation & Setup

1️⃣ Install Dependencies

pip install -r requirements.txt

2️⃣ Set Up the Database

Run this once to initialize SQLite:

python src/db_setup.py

---

📌 Data Collection & Training

3️⃣ Store Palm Features in Database

Run this script to capture palm features (press 's' to save, 'q' to quit):

python src/store_features.py

🔹 What happens?
✅ Opens webcam → detects palm → extracts features → saves to database

---

4️⃣ Train the Machine Learning Model

Once features are stored, train the model:

python src/train_model.py

🔹 What happens?
✅ Loads features from SQLite → Trains an SVM model → Saves the trained model

---

📌 Authentication Methods

5️⃣ Real-Time Authentication (Using Webcam)

Run the following script to authenticate users in real time:

python src/authenticate.py

🔹 What happens?
✅ Captures palm → Extracts features → Predicts user → Checks DB → Authenticates

---

6️⃣ Authentication via API (External Apps/Web Services)

Start the Flask API server:

python src/api.py

🔹 API Endpoint:

• URL: http://localhost:5000/authenticate
• Method: POST
• Body: Upload an image (image key)

🔹 Example API Request (Using cURL)

curl -X POST -F "image=@test_palm.jpg" http://localhost:5000/authenticate

🔹 Expected API Responses

Scenario	Response
✅ Successful Authentication	{"success": true, "user_id": 1, "message": "Authentication Successful"}
❌ Palm Not Detected	{"success": false, "message": "No palm detected!"}
❌ Unknown User (Low Confidence)	{"success": false, "message": "Authentication Failed: Unknown User"}
❌ User Not Found in DB	{"success": false, "message": "Authentication Failed: User Not Found"}

---

📌 Viewing Database Entries

To manually inspect the database via SQLite shell:

sqlite3 data/palm_auth.db

Run SQL queries like:

SELECT * FROM users;

---

📌 Next Steps & Future Improvements

✅ Improve Model Training with More Data
✅ Fine-Tune SVM Parameters for Higher Accuracy
✅ Add a User Registration API for Seamless Enrollment
✅ Support Cloud Database for Multi-Device Authentication

---

🎯 Conclusion

PalmAuthAI is a powerful proof-of-concept biometric system. With further improvements, it can be used in secure access control, identity verification, and AI-driven authentication. 🚀

💡 Want to contribute or improve accuracy? Let’s discuss next steps!