Drowsiness_Detection_System

🚀 Built by Mitali under FrameIQ LLP | Updated: August 2025
⚙️ Tech Stack: Flask, OpenCV, MediaPipe, TensorFlow, HTML/CSS

---

🔍 Overview

Drowsiness Detection System is an AI-powered, real-time monitoring solution that detects fatigue signs—like eye closure, yawning, and head nodding—using a hybrid combination of Computer Vision and Deep Learning.
This lightweight system runs locally through a Flask web interface, providing:

• 🎥 Live webcam feed
• 🧠 Eye Aspect Ratio (EAR) + CNN predictions
• 🔔 Sound alerts
• 📜 Automatic event logging
• 🟢 Dynamic status updates in-browser

Perfect for driver monitoring, workplace safety, productivity research, or academic/portfolio showcases.

---

🧠 Key Features

• 🧍 MediaPipe Facial Landmark Detection
• 👁️ Eye State Classification via CNN (MobileNetV2)
• 💤 Drowsiness Detection using EAR + ML logic
• 😮 Yawning & Head Nodding Detection
• 🛜 Real-time Flask UI (No Streamlit required)
• 🔊 Sound Alerts + Timestamped Logging
• 📊 Live Status: Attentive / Drowsy / Yawning / Head Nodding
• ✅ Clean, modular folder structure

---

🗂️ Project Structure

Drowsiness_Detection_System/
├── app.py                          # Flask backend (UI + detection logic)
├── templates/
│   └── index.html                  # Responsive HTML frontend
├── model_training/
│   ├── preprocessing_modeltraining.py
│   ├── eye_state_classifier.h5
│   └── eye_state_classifier.keras
├── test_scripts/
│   └── test_model_prediction.py
├── streamlit_version_archived/     # Previous Streamlit version (archived)
│   ├── app.py
│   └── drowsiness_log.txt
├── drowsiness_log.txt              # Auto-generated event log file
├── requirements.txt                # All dependencies
├── README.md                       # You’re reading it!
└── .gitignore                      # Version control hygiene

---

⚙️ Installation & Usage

# 1. Clone the repository
git clone https://github.com/cvframeiq/Drowsiness_Detection_System.git
cd Drowsiness_Detection_System

# 2. Create a virtual environment
python -m venv cv_env
cv_env\Scripts\activate     # On Windows
# source cv_env/bin/activate   # On Mac/Linux

# 3. Install dependencies
pip install -r requirements.txt

# 4. Launch the app
python app.py

Open your browser at http://localhost:5000 to see the app in action.

---

📊 Model Details

Architecture: MobileNetV2 (Transfer Learning)
Input Size: 224 x 224
Classes: Binary (Open / Closed Eyes)
Dataset: CEW – Closed Eyes in the Wild
Accuracy: ~87% (Training) | ~76% (Validation)

---

🛡️ Security Notes

✅ Webcam access only on local system
✅ No image/video is saved
✅ No external API calls — full offline execution

---

📦 Dataset

Note: Dataset not included due to size. Please create the following folder structure:

dataset/
├── open/
│   └── *.jpg
└── closed/
    └── *.jpg

📌 Archived: Streamlit Version The earlier version using Streamlit has been moved to /streamlit_version_archived for reference.