Sleep Detector Program

Real-Time Driver Drowsiness Detection System

A real-time computer vision system that detects driver drowsiness using facial landmark analysis and triggers an audible alert to help prevent fatigue-related accidents.

---

Overview

Driver fatigue is a major cause of road accidents.
Drive Alarm monitors the driver's eye state using a webcam and calculates the Eye Aspect Ratio (EAR) to determine drowsiness. If the eyes remain closed for a defined duration, an alarm is triggered.

This project demonstrates:

• Real-time computer vision
• Facial landmark detection (dlib)
• Geometric eye state analysis
• Alert-based safety system implementation

---

Features

• Real-time face and eye detection
• Eye Aspect Ratio (EAR) calculation
• Frame-based drowsiness threshold detection
• Audible alert system
• Live EAR display on video feed

---

Tech Stack

• Python
• OpenCV
• dlib
• NumPy
• SciPy
• Pygame

---

Project Structure

Drive_alarm/
│
├── project 2.py
├── project 3.py
├── requirements.txt
├── .gitignore
└── assets/
    ├── beep.mp3
    └── shape_predictor_68_face_landmarks.dat (not included in repo)

---

Installation

Clone the Repository

git clone https://github.com/AdityaSharma-Git3207/Sleep-Detector-Program.git
cd Sleep-Detector-Program

---

Create Virtual Environment (Recommended)

python -m venv venv
venv\Scripts\activate

---

Install Dependencies

pip install -r requirements.txt

Note: dlib installation may require Visual Studio Build Tools on Windows.

---

Required Model File

This project requires the pre-trained dlib facial landmark predictor model.

Due to its large size (~100MB), it is not included in this repository.

Download Steps:

1. Download from:
   http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2

2. Extract the .bz2 file

3. Place the extracted file inside:

assets/

Final structure should look like:

assets/shape_predictor_68_face_landmarks.dat

---

Running the Application

python "project 2.py"

Press q to quit the application.

---

How It Works

The system calculates the Eye Aspect Ratio (EAR) using facial landmarks.

Detection Logic

• Eyes are considered closed when:

EAR < 0.25

• If eyes remain closed for:

20 consecutive frames

The alarm is triggered.

---

Git Ignore Policy

The following files are intentionally excluded:

• venv/ → Virtual environment directory
• assets/shape_predictor_68_face_landmarks.dat → Large binary model file

This keeps the repository lightweight and maintainable.

---

Future Improvements

• Blink rate monitoring
• Head pose estimation
• Performance optimization
• GUI interface
• Mobile device integration

---

Author

Aditya Sharma
Open Source Contributor | Computer Vision Enthusiast