https://github.com/adityabichhave/facial-reaction-password

AI-powered facial reaction authentication system using MediaPipe + DTW + Anti-Spoofing.
https://github.com/adityabichhave/facial-reaction-password

ai-authentication anti-spoofing biometric-authentication cybersecurity dtw facial-recognition mediapipe react

Last synced: about 2 months ago
JSON representation

AI-powered facial reaction authentication system using MediaPipe + DTW + Anti-Spoofing.

• Host: GitHub
• URL: https://github.com/adityabichhave/facial-reaction-password
• Owner: adityabichhave
• License: mit
• Created: 2025-12-04T07:35:02.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2025-12-04T07:52:49.000Z (7 months ago)
• Last Synced: 2025-12-07T14:48:50.694Z (6 months ago)
• Topics: ai-authentication, anti-spoofing, biometric-authentication, cybersecurity, dtw, facial-recognition, mediapipe, react
• Language: JavaScript
• Homepage:
• Size: 2.7 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          


  



🧠 Facial Reaction Password (FRP)

A next-generation biometric authentication system that identifies users using facial micro-reactions, blinking behavior, and natural movement patterns.

Unlike traditional passwords or face-recognition systems, FRP records your reaction sequence (blink → smile → slight head movement, etc.) and uses AI to verify it.

It works using any standard webcam — no specialized hardware needed.

🚀 Features

🔹 Real-Time Facial Landmark Tracking

Powered by MediaPipe Face Landmarker, detecting 468+ facial points with high precision and low latency.

🔹 Reaction-Based Password (Instead of Text Passwords)

Users perform a natural facial sequence which becomes the unique password template.

🔹 AI Matching Using Dynamic Time Warping (DTW)

DTW compares two sets of motion-based landmark sequences to determine user identity.

🔹 Anti-Spoofing System

Detects and blocks fake login attempts using:

Motion variance

Blink detection

Micro-movement consistency

Landmark noise analysis

🔹 Secure Local Storage

User facial templates are stored as numerical vectors, not images.

Supports:

LocalStorage (frontend-only mode)

Express backend storage (optional)

🔹 Beautiful VisionOS-Inspired UI

Premium holographic design featuring:

Floating glass panels

Sleek motion indicators

Real-time metrics

Minimalistic control panel

🛠️ Tech Stack

Frontend

React (Vite)

TailwindCSS

MediaPipe Tasks Vision

Canvas API

Backend

Node.js + Express

Local JSON template storage

(Upgradable to MongoDB / Firebase easily)

Other Tools

Dynamic Time Warping (DTW)

LocalStorage caching

Web Share API

📸 Demo Preview

Replace screenshot.png with your actual screenshot

![FRP Screenshot](./screenshot.png)

📦 Folder Structure

facial-reaction-password/

│

├── backend/

│   ├── server.js

│   ├── package.json

│

├── frontend/

│   ├── public/

│   ├── src/

│   │   ├── components/

│   │   │   ├── FacialReactionPassword.jsx

│   │   │   ├── FacialReactionPasswordVision.jsx

│   │   ├── App.jsx

│   │   ├── main.jsx

│   ├── package.json

│

├── README.md

└── .gitignore

⚙️ Installation & Setup

1️⃣ Clone Repository

git clone https://github.com/adityabichhave/facial-reaction-password.git

cd facial-reaction-password

🎨 Frontend Setup (React + Vite)

cd frontend

npm install

npm run dev

Frontend starts at:

👉 http://localhost:5173

🧩 Backend Setup (Node.js + Express)

cd backend

npm install

npm run dev

Backend runs at:

👉 http://localhost:5001

🔐 How Authentication Works

FRP records 468+ facial landmarks across multiple frames

Normalizes vectors around the face center

Creates a unique reaction pattern

Runs DTW to compare sequences

Applies anti-spoofing checks

Computes similarity score

Accepts or rejects login based on threshold

🧪 API Endpoints (Backend Optional)

POST /api/enroll

Stores a user’s facial reaction template.

GET /api/template/:username

Retrieves stored template for login matching.

🔒 Security Notes

✔ No raw video or images are stored

✔ Templates contain only floating-point vectors

✔ Works completely offline

✔ Users control their own data

✔ Backend storage is optional

🌍 Real-World Applications

Passwordless authentication

Secure workstation login

High-security lab access

Personal computer unlocking

Research in biometrics & AI

Gesture-based UI systems

Human-computer interaction studies

🤝 Contributing

Pull requests are welcome!

Open issues for:

UI/UX improvements

Performance tuning

Model optimization

Security enhancements

📜 License

This project is licensed under the MIT License — free to use and modify.

👨‍💻 Developed By

Aditya Kumar Bichhave

B.Tech CSE | Cyber Security Enthusiast | Full-Stack Developer