Overview

A production-ready attendance system that detects faces from a webcam, recognizes registered users, and logs timestamps to CSV/Excel.

Supports offline usage and works on low-spec laptops.

Features

• Real-time face detection with Haar cascades
• Face recognition using LBPH
• Auto mark in/out with timestamp
• Duplicate prevention within a grace window
• CSV and Excel export
• Admin CLI to register faces

Tech Stack

• Python 3.10+
• OpenCV
• Numpy
• Pandas

Project Structure

attendance/
├─ data/                # Face datasets and trained model
├─ logs/                # Attendance CSV/Excel
├─ register.py          # Register new faces
├─ train.py             # Train recognizer
├─ run.py               # Start attendance app
└─ requirements.txt

Setup

python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Usage

Register at least 30 images per person with unique IDs:

python register.py --person-id 1001 --name "Alice"

Train the model:

python train.py

Run attendance system:

python run.py --export excel

Sample CSV

name,person_id,action,timestamp
Alice,1001,IN,2025-09-01T09:03:11
Alice,1001,OUT,2025-09-01T17:34:22

Notes

• Ensure good lighting for better recognition accuracy.
• Re-train the model after adding/removing users.