Classroom Assistant

An AI-powered Classroom Management System that automates student attendance using face recognition and monitors classroom emotions in real-time.

Built with Streamlit, DeepFace, MTCNN, and TensorFlow, this app streamlines attendance tracking, detects impersonation during exams, and provides insights into student engagement through emotion analysis.

Objectives:

• To obtain and preprocess emotion dataset.
• To build and train emotion recognition system based on the collected data.
• To development attendance and emotion management system based on the model in
• To integrate sub-systems developed into a fully functional system.
• To evaluate the system performance using accuracy, precision, recall and Response time.

Features

• Automated Attendance Marking
  Real-time face detection and recognition during class sessions using images streamed from a Raspberry Pi camera.

• Emotion Analysis
  Detects dominant emotions (anger, calm, disgust, fear, happy, sadness, surprise) across the classroom and visualizes percentages with charts.

• Exam Mode with Impersonation Detection
  Restricts access to students meeting 75% attendance threshold and flags unrecognized faces as "Impersonation".

• Student Management
  Enroll/remove students with face images (via upload or local webcam), organized by courses.

• Course Management
  Create new courses and download attendance records as CSV files.

• Raspberry Pi Integration
  Receives live classroom images over a socket connection from a Raspberry Pi.

Screenshots

(Add actual screenshots here once available – recommended: Homepage, Enroll Student, Start Class with emotion chart, Exam Mode)

Tech Stack

• Frontend/UI: Streamlit
• Face Detection: MTCNN
• Face Recognition: DeepFace (Facenet / Facenet512)
• Emotion Recognition: Custom TensorFlow Keras model (Latest FER.keras)
• Image Processing: OpenCV, PIL
• Data Handling: Pandas, JSON
• Networking: Python sockets (for Raspberry Pi image streaming)

Installation & Setup

Prerequisites

• Python 3.9 or higher
• A webcam (for enrollment) or Raspberry Pi with camera (for live capture)
• Pre-trained emotion model: Latest FER.keras (place in project root)

Steps

1. Clone the repository

   git clone https://github.com/yourusername/classroom-assistant.git
   cd classroom-assistant

2. Create a virtual environment (recommended)

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install streamlit tensorflow deepface opencv-python mtcnn pandas matplotlib pillow

4. Folder Structure Setup The app expects:

   student_data/
   ├── attendance/          # CSV files per course
   ├── course list.json     # List of courses
   └── faces/               # Subfolders per course with student face images
   

   Run the app once – it will create default folders and files automatically.

5. (Optional) Raspberry Pi Setup

   • Run a simple server script on your Raspberry Pi to stream images to IP 192.168.43.230:8000.
   • Update raspi_ip in the code if needed.

Run the App

streamlit run main.py

Open your browser at http://localhost:8501.

🎮 Usage Guide

1. Homepage – Create new courses or download attendance CSVs.
2. Enroll Student – Add student details and capture/upload face image.
3. Remove Student – Delete a student and their face data.
4. Start Class – Select course, set duration, and begin real-time attendance + emotion monitoring.
5. Exam Mode – Only eligible students (≥75% attendance) are marked; others flagged.

Notes & Limitations

• Requires good lighting and frontal faces for accurate recognition.
• Raspberry Pi streaming uses a hotspot connection (update IP/port as needed).
• Emotion model is custom-trained; accuracy depends on training data.
• For production use, consider security enhancements (e.g., encrypted connections).

Contributing

Contributions are welcome! Feel free to:

• Open issues for bugs or feature requests
• Submit pull requests with improvements

Please follow standard GitHub flow: fork → branch → PR.

License

This project is licensed under the MIT License – see LICENSE for details.

---

Made with ❤️ for smarter classrooms
If you find this project useful, give it a ⭐ on GitHub!

Contact sundayabraham81@gmail.com for more information and collaboration.