🎓 Student Attendance System (Face Recognition Based)

This project is a face-recognition-based Student Attendance System built using deep learning and computer vision techniques. It detects and recognizes student faces from video or images and automatically logs attendance.

👨‍💻 This repository contains two separate versions of the system — each in its own branch — using different deep learning approaches.

---

📌 Project Features

• 🎥 Face detection from images or video frames
• 🧠 Deep learning-based face recognition
• 🗂️ Automated attendance logging
• 🧪 Training pipeline for custom student datasets

---

🧠 System Versions

This repo includes two different implementations of the attendance system:

📁 1. cnn-version (Notebook 1)

Custom CNN-based face recognition system

• Uses a basic Convolutional Neural Network.
• Simple model architecture.
• Trained from scratch on student face images.
• Good for learning, prototyping, or smaller datasets.

📁 2. densenet-version (Notebook 2)

Advanced DenseNet121-based system using transfer learning

• Uses DenseNet121, a powerful pre-trained CNN.
• Higher accuracy and faster convergence.
• Ideal for larger, real-world datasets.
• Employs OpenCV for face detection and Keras for model training.

---

🌳 Branch Structure

Branch	Description
main	Project overview + README only
cnn-version	Full project code for custom CNN approach
densenet-version	Full project code using DenseNet121

---

📸 System Flow (Both Versions)

graph TD;
    A[Capture Video or Images] --> B[Extract Frames]
    B --> C[Detect Faces using OpenCV]
    C --> D[Preprocess Faces]
    D --> E[Train CNN / DenseNet Model]
    E --> F[Recognize Faces in New Images]
    F --> G[Mark Attendance in CSV]

Loading

---

📂 Folder Structure (per branch)

Each branch contains its own implementation:

.
├── notebooks
├── requirements.txt
└── README.md

---

🛠 Requirements

Each branch has its own requirements.txt, but common dependencies include:

• Python 3.8+
• TensorFlow / Keras
• OpenCV
• NumPy, Pandas

---

👏 Contributions

Contributions and improvements are welcome! If you'd like to improve the model, fix a bug, or enhance documentation, feel free to fork the repo and submit a pull request.