🎓 Smart Attendance and Emotion Recognition System

This project is a machine learning–based attendance system that detects and identifies students in real time using a webcam.
It automatically marks attendance and logs the detected emotion (happy, neutral, sad, etc.) along with the timestamp in a CSV file.
The system operates only between 9:30 AM and 10:00 AM to simulate a classroom environment.

---

✨ Features

• 🧠 Face Recognition – Identifies registered students using a trained CNN model
• 😊 Emotion Detection – Recognizes facial emotions using the FER dataset
• ⏰ Time Restriction – Works only during a scheduled window (9:30–10:00 AM)
• 🌐 Web App Interface (Flask + OpenCV) – Stream live video and monitor recognition in real time
• 📊 Attendance Logging – Automatically saves student name, emotion, and timestamp in CSV
• 📁 On-Demand Export – Export attendance data via a single button click
• 💾 Lightweight, Modular Design – Clean Python-based code, easy to reproduce

---

🧩 Project Structure

attendance_clean/ ├── web_app/ │ ├── core.py # Handles video feed, face & emotion detection logic │ └── flask_app.py # Flask server + web interface │ ├── scripts/ │ ├── train_faceid.py # Train face identification model │ ├── train_emotion.py # Train emotion classification model │ ├── dataset_audit.py # Inspect dataset distribution │ ├── augment_class.py # Perform image augmentation │ └── attendance_run.py # Optional: run attendance offline (CLI) │ ├── models/ │ ├── faceid_labels.json │ └── emotion_labels.json │ ├── utils.py ├── test_face_pred.py ├── test_emotion.py ├── evaluate_face_model.py ├── inspect_predictions.py ├── .gitignore ├── README.md └── requirements.txt

---

🚀 How to Run

1️⃣ Setup Environment

pip install -r requirements.txt


2️⃣ Run Web App
python web_app/flask_app.py
Then open your browser at:
http://127.0.0.1:5000
Click Start Stream to activate the camera.
The system will detect faces, emotions, and mark attendance automatically.
Click Export Attendance CSV to download the attendance report.

🧠 Model Information
FaceID Model – Trained on custom dataset under data/faceid/
Emotion Model – Trained on FER2013 dataset (fer2013.zip, not uploaded due to size limit)

To recreate models:
python scripts/train_faceid.py --data_dir data/faceid
python scripts/train_emotion.py --data_dir data/emotion
Model binaries (.h5) are excluded from this repository due to GitHub file size limits.

🕒 Time-Based Restriction
Attendance detection is automatically active only between 9:30 AM and 10:00 AM.
Outside this time window, the system won’t log attendance.

📊 Output Example
Name	Emotion	Time
Bushra	Happy	09:35:12 AM
Waniya	Neutral	09:42:05 AM
Zaid	Sad	09:50:33 AM

Exported automatically to:
exports/attendance_YYYYMMDD_HHMMSS.csv


🧩 Tech Stack
Python 3.12
Flask – Web server
OpenCV – Video stream and image processing
TensorFlow / Keras – Model inference
NumPy, Pandas – Data operations