Drowsiness Detection System

This project implements a Drowsiness Detection System for drivers using computer vision and embedded electronics. It monitors eye activity through a webcam and alerts the driver via buzzer and LCD if signs of drowsiness are detected. The system uses Python (OpenCV + dlib) for image processing, an Arduino for alerting, and Proteus for circuit simulation. It was also physically implemented on PCB.

---

Table of Contents

• Overview
• How It Works
• Hardware Required
• Software Used
• Folder Structure
• Setup Instructions
• Circuit Connections
• Simulation (Proteus)
• Code Breakdown
• Photos
• License

---

Overview

• Python captures webcam input and uses dlib to detect eyes.
• It calculates Eye Aspect Ratio (EAR) to determine if the driver is sleeping.
• Based on the status, it sends 'a' (sleepy) or 'b' (awake) to Arduino via Serial.
• Arduino receives the data and:
  • Displays status on LCD
  • Triggers a buzzer and LED for alert
• The full setup is simulated in Proteus and implemented physically using PCB.

---

How It Works

1. Laptop webcam continuously monitors the user’s face.
2. Eye landmarks are detected and Eye Aspect Ratio (EAR) is calculated.
3. If EAR drops below threshold for a defined period:
   • Python sends 'a' to Arduino.
   • Arduino activates buzzer, LED and LCD alert.
4. If user is awake:
   • 'b' is sent to Arduino to show “Safe journey”.

---

Hardware Required

Component	Quantity
Arduino Uno	1
16x2 LCD	1
Buzzer	1
Red LED	1
NPN Transistor	1
Resistors	Few
PCB or Breadboard	1
Jumper Wires	As needed
5V/9V Transformer	1 (if powering externally)

---

Software Used

• Python 3.x
• OpenCV
• dlib
• imutils
• Arduino IDE
• Proteus 8 (for simulation)
• Git LFS (for large .dat file)

---

Folder Structure

Drowsiness-Detection/
├── arduino/
│   └── drowsiness_alert.ino
├── python/
│   └── drowsiness_detection.py
├── model/
│   └── shape_predictor_68_face_landmarks.dat
├── images/
│   ├── proteus_schematic.jpg
│   └── pcb_hardware.jpg
├── README.md
└── requirements.txt

---

Setup Instructions

Python

1. Install dependencies:

   pip install -r requirements.txt

2. Clone and set up Git LFS for the .dat file:

   GitHub uses Git LFS (Large File Storage) to store the large .dat model file used by dlib.

   Run the following after cloning the repo:

   git lfs install
   git lfs pull

   This will download the required model:

   model/shape_predictor_68_face_landmarks.dat
   

3. Run the Python script:

   python python/drowsiness_detection.py

Arduino

1. Open drowsiness_alert.ino in Arduino IDE.
2. Connect Arduino via USB.
3. Upload the sketch.
4. Keep the Serial Monitor closed to avoid conflict with Python serial communication.

---

Circuit Connections

Arduino Pin	Connects To	Description
D2–D7	16x2 LCD	Used to display status messages
D8	Buzzer (via transistor)	For audible alert
D9	Red LED	For visual alert
GND	Common ground	Shared with all components
5V	Power supply to LCD and buzzer	Power from Arduino
USB	Laptop	For power + serial communication

• Transistor is used between buzzer and Arduino for current control.
• Resistors are used for base control of transistor and current limiting for LED.

---

Simulation (Proteus)

We used Proteus 8 to simulate the complete circuit.

• LCD shows messages from Arduino
• Buzzer and LED trigger based on serial input
• You can simulate sending 'a' or 'b' using Proteus Serial Terminal

---

Code Breakdown

Python Script (Detailed Explanation)

• Captures webcam feed using OpenCV.
• Detects facial landmarks using the pre-trained .dat model from the dlib library.
• Calculates Eye Aspect Ratio (EAR) using 6 key points around each eye.
• Decision logic based on EAR:
  • EAR < 0.21 → Driver is likely sleeping.
  • 0.21 ≤ EAR ≤ 0.25 → Driver is drowsy.
  • EAR > 0.25 → Driver is awake.
• Based on the EAR, the Python script sends a serial signal:
  • 'a' → Drowsy or sleeping (alerts Arduino)
  • 'b' → Awake (no alert needed)

Arduino Sketch

• Uses LiquidCrystal library to control LCD
• Listens to serial input from Python:
  • 'a' = show alert, trigger buzzer + LED
  • 'b' = show safe message, no alert
• Uses delay intervals for blink and visibility

Model File

• File: shape_predictor_68_face_landmarks.dat
• Used by dlib to locate 68 key facial points
• Pre-trained, required for face/eye tracking

---

Photos

1. PCB Hardware Setup

2. Proteus Simulation

---

License

This project is intended for academic and demonstration purposes. It was originally created by YouTube. Feel free to use, modify, and expand upon it.