Attendance Management System

A desktop face recognition system that automatically identifies students from a webcam feed and records their attendance in an Excel file. Built on a full ML pipeline — from dataset collection through model training to real-time inference.

---

What it does

• Admin registers students by capturing ~100 face images per person via webcam
• System trains a face recognition model on those images
• At attendance time, the webcam feed is analysed in real time — faces are detected, identified, and attendance is logged automatically to an Excel file
• Works under varied lighting conditions
• Tkinter GUI for all interactions (dataset creation, training, live recognition)

---

ML Pipeline

DATASET CREATION
  Tkinter GUI → OpenCV webcam → MTCNN detects + crops face → saved per student

MODEL TRAINING
  Saved face images → FaceNet (pretrained CNN) → 128-d embedding vector per face
  All embeddings + labels → SVM classifier trained → model saved to disk

REAL-TIME RECOGNITION
  Webcam frame → MTCNN detects face → FaceNet embeds it
  → SVM predicts identity (+ confidence score)
  → If confidence > threshold: log name + timestamp to Excel

---

Tech Stack

Component	Technology
Language	Python 3
Face detection	MTCNN
Face recognition	FaceNet (TensorFlow / PyTorch)
Classifier	SVM (scikit-learn)
Video input	OpenCV
GUI	Tkinter
Attendance output	openpyxl (Excel)

---

Project Structure

attendance-management-system/
├── main.py                    # Entry point — launches Tkinter app
├── src/
│   ├── face_detection/
│   │   ├── detector.py        # MTCNN face detection + crop/align
│   │   └── embedder.py        # FaceNet embedding generation
│   ├── ui/
│   │   ├── app.py             # Main Tkinter window
│   │   ├── register.py        # Dataset capture screen
│   │   └── attendance.py      # Live recognition screen
│   └── utils/
│       ├── dataset.py         # Image I/O, dataset management
│       ├── trainer.py         # SVM training on embeddings
│       └── excel.py           # Attendance Excel read/write
├── models/                    # MTCNN weights (.npy files)
├── dataset/                   # Captured face images per student
├── attendance/                # Generated Excel files
└── requirements.txt

---

Setup

git clone https://github.com/ajay062002/attendance-management-system.git
cd attendance-management-system
pip install -r requirements.txt
python main.py

---

Usage

1. Register students — enter name, click Capture, look at camera for ~5 seconds
2. Train model — click Train after all students are registered (takes ~30 seconds)
3. Start attendance — click Start Recognition, system logs faces it identifies automatically

---

Requirements

opencv-python
mtcnn
tensorflow
scikit-learn
Pillow
openpyxl
numpy

---

Source

• Web-based interface instead of Tkinter desktop GUI
• Upgrade to TensorFlow 2.x with Keras
• Support for multiple classrooms and subjects
• Email or SMS notification on attendance completion
• Dashboard with attendance analytics and visual reports

---

📄 Research Reference

Based on the paper: Automated Attendance System using CNN — IEEE

Related publication by the author: Sign Language Detection using CNN — IEEE ICSSCS 2023

---

⬇️ Download Pre-trained Model

FaceNet Model — Google Drive

---

👤 Author

Ajay Thota

• GitHub: @ajay062002
• Portfolio: ajaylive.com