Facial Recognition Attendance System

A Tkinter desktop application that enrols people from a webcam, trains an OpenCV LBPH face recogniser, and marks attendance to a CSV when it recognises a face. Student/staff records are kept in MySQL.

What it does

main.py opens a full-screen dashboard with eight buttons, each backed by its own module:

Button	Module	What it does
User Details	UserDetailsButton.py	CRUD over a userdetails MySQL table (department, course, year, semester, ID, name, gender, DOB, email, phone, address, teacher). A "Take Photo Sample" radio option captures face crops from the webcam into a local data/ directory, named user.<id>.<n>.jpg
Detect Face	face_recog.py	Opens the webcam, detects faces with a Haar cascade, runs the trained LBPH recogniser, looks the ID up in MySQL for name and department, and draws the label on the frame
Attendance	attendance.py	A table view of attendance rows with Import CSV / Export CSV buttons (Tkinter filedialog)
Help	help.py	Support/contact form backed by MySQL
Train Face	train.py	Reads every image in data/, trains cv2.face.LBPHFaceRecognizer_create(), writes classifier.xml
Photos	—	os.startfile("data") — opens the dataset folder in the file manager
Developer	developer.py	An "about the developer" panel plus a complaints table
Exit	—	Confirm-and-quit

Attendance logic (face_recog.py): the LBPH prediction distance is converted to a confidence score with confidence = int(100 * (1 - predict / 300)), and a row is appended to harold.csv when confidence is above 86. Below that the face is drawn as "Unknown Face". The CSV columns are name, department, id, time, date, and the literal status Present.

Tech stack

• Python 3, Tkinter (GUI), Pillow (image loading for the UI)
• OpenCV with opencv-contrib — required, because cv2.face.LBPHFaceRecognizer_create() lives in the contrib modules and is absent from the plain opencv-python wheel
• NumPy
• MySQL via mysql-connector-python (and pymysql, imported in face_recog.py but unused)

Setup

There is no requirements.txt. From the imports:

python3 -m venv venv && source venv/bin/activate
pip install opencv-contrib-python numpy pillow mysql-connector-python pymysql

1. MySQL

You need a local MySQL server with a database named facialrecognition containing a userdetails table (columns referenced in the code include id, name, dep, plus the rest of the enrolment fields). TODO: verify — no schema file or CREATE TABLE statement is committed, so the table definition has to be reconstructed from the INSERT/SELECT statements in UserDetailsButton.py.

Connection details are hardcoded in five separate files (UserDetailsButton.py, developer.py, face_recog.py, help.py) as host="localhost", user="root", passwd=<hardcoded>, database="facialrecognition". You must edit each one — see the security note below.

2. Run

python main.py

The app must be started from the repository root: main.py, train.py and face_recog.py all open image and cascade files by relative path (Christine1.jpg, haarcascade_frontalface_default.xml, data, classifier.xml), so any other working directory raises FileNotFoundError.

3. Enrol and train

The data/ directory is not committed and does not exist in a fresh clone. train.py calls os.listdir("data") unguarded, so "Train Face" crashes until you have enrolled at least one person via User Details → Take Photo Sample.

A pre-trained classifier.xml (10 MB) is committed, so "Detect Face" may appear to work immediately — but it was trained on the original author's dataset and its numeric IDs will not match rows in your userdetails table.

Platform note

main.py uses os.startfile("data") for the Photos button. os.startfile is Windows-only; on Linux and macOS that button raises AttributeError. The rest of the app is cross-platform.

Status

Working prototype, Windows-oriented. Last commit June 2024. The enrolment → train → recognise → CSV loop is implemented. Hardcoded credentials, hardcoded relative paths, no schema file and no requirements.txt make it awkward to set up from a clean machine.

Licence

Proprietary software — all rights reserved. Copyright © 2026 Joel Harold Onyango.

This repository is not open source. The full terms are in LICENSE; in summary, you may not copy, redistribute, modify, sublicense, publish, re-host or commercially exploit this software, in whole or in part, without the prior written permission of the copyright holder. Access to this repository does not grant any licence beyond reading it.