4Sight is a real-time, face recognition-based attendance automation system. It uses an ESP32-CAM to stream video footage to a central server, where a Python backend processes faces using computer vision. Attendance records are logged instantly and displayed on a modern React dashboard.
- Real-time Face Recognition: accurate identification using
dlibandface_recognition. - Wireless Streaming: ESP32-CAM streams video over WiFi to the processing server.
- Duplicate Prevention: Smart logic prevents marking the same student multiple times within 60 seconds.
- Live Dashboard: A React + TypeScript web interface to view attendance logs and confidence scores.
- CSV Logging: Automatically maintains an Excel-compatible
attendance.csvfile.
- Hardware: ESP32-CAM (AI-Thinker Model)
- Backend: Python, Flask, OpenCV, Face Recognition lib, Pandas
- Frontend: React (Vite), TypeScript, CSS
- Data: Pickle (Encodings), CSV (Logs)
4Sight/
├── code/
│ ├── recognize.py # Main backend server & recognition loop
│ ├── enroll_faces.py # Script to train/encode new student faces
│ └── convert.py # Utility script
├── data/
│ ├── encodings/ # Stores generated face encodings (.pkl)
│ ├── students.csv # Master list of student details
│ └── attendance.csv # Generated attendance log
├── esp32/
│ └── esp32_stream/ # Arduino sketch for the camera
├── react-app/ # Frontend Dashboard source code
├── static/
│ └── student_images/ # Source images for training
└── requirements.txt # Python dependencies