An intelligent attendance system that uses Face Recognition (YOLOv8 + FaceNet) to automatically detect and mark attendance in real time.
This project eliminates manual attendance by using computer vision and deep learning.
Users register with their face, admins approve them, and attendance is marked automatically using real-time face recognition.
- Register with personal details and face capture
- Check approval status (Pending / Approved / Rejected)
- Mark attendance using webcam
- Automatic check-in and check-out system
- Secure admin login
- Approve / reject users
- View attendance records
- Dashboard with analytics
- Email notifications
- Face detection using YOLOv8
- Face recognition using FaceNet embeddings
- Multi-frame embedding averaging for better accuracy
| Category | Technology |
|---|---|
| Frontend | Streamlit |
| Backend | Python |
| Database | SQL Server (pyodbc) |
| Computer Vision | OpenCV, YOLOv8 |
| Deep Learning | FaceNet (facenet-pytorch) |
| Data Processing | NumPy, Pandas |
AI-Face-Recognition-Attendance-System/ โ โโโ admin_app/ # Admin panel (approval, analytics, dashboard) โโโ user_app/ # User portal (registration, attendance) โโโ screenshots/ # Project screenshots โโโ .gitignore โโโ README.md
git clone https://github.com/maniac-24/AI-Face-Recognition-Attendance-System.git
cd AI-Face-Recognition-Attendance-System
cd admin_app pip install -r requirements.txt
cd ../user_app pip install -r requirements.txt
Create file:
.streamlit/secrets.toml
[email] sender = "your_email@gmail.com " password = "your_app_password"
cd admin_app streamlit run app.py
cd user_app streamlit run app.py
User registers โ Admin approves โ User marks attendance โ System records check-in/out
- Only approved users can mark attendance
- Ensure camera permissions are enabled
- Good lighting improves recognition accuracy
- Database must be configured correctly
- Cloud deployment (AWS / Streamlit Cloud)
- Mobile support
- Role-based access system
- Live notifications
- Multi-face tracking
Prashanth M
๐ง prashanthmadival64@gmail.com
If you like this project, consider giving it a โญ on GitHub!
If you like this project, consider giving it a โญ on GitHub!






