CONVOLUTIONAL NEURAL NETWORK BASED
DRIVER DROWSINESS DETECTION USING
ARTIFICIAL INTELLIGENT TECHNOLOGY
INTRODUCTION
The growing number of accidents caused by driver fatigue has raised
concerns about road safety. Traditional methods rely on physical monitoring
devices, which may be intrusive. Leveraging Artificial Intelligence (AI),
especially Convolutional Neural Networks (CNNs), provides an efficient and
non-invasive solution. This project integrates AI with hardware components to
create an affordable and portable system for detecting and preventing drowsiness
in real-time.
OBJECTIVE:
The objective of this project is to design and implementation of
Convolutional Neural Network Based Driver Drowsiness Detection Using
Artificial Intelligent Technology.
ABSTRACT
Driver drowsiness is a significant factor contributing to road accidents. This
project presents a real-time driver drowsiness detection system using a Convolutional
Neural Network (CNN) integrated with Arduino, an LCD display, an alarm, and a
motor simulation. The CNN processes live video feed to detect drowsiness by
analyzing eye closure and facial patterns. Upon detection, the system activates an
alarm to alert the driver and stops the motor to simulate halting the vehicle. The
proposed solution enhances road safety using advanced AI techniques in an
embedded system framework.
EXISTING SYSTEM
1. Manual Observation: Relies on self-awareness or external monitoring,
which is inconsistent and prone to human error.
2. Physiological Monitoring: Uses sensors to track heart rate or EEG
signals, but these are expensive and intrusive.
3. Behavioral Monitoring: Employs basic image processing to detect
yawning or head tilts, but lacks accuracy and robustness in varying light
and environments.
PROPOSED SYSTEM
The proposed system uses a CNN model trained to detect signs of
drowsiness by analyzing facial images captured by a camera. The system
components include:
1)AI Module: Runs the CNN model for real-time drowsiness detection.
2)Arduino Microcontroller: Controls the alarm, LCD, and motor actions.
3)LCD Display: Displays driver status (e.g., "Alert" or "Drowsy").
4)Alarm: Activates when drowsiness is detected.
5)Motor: Simulates vehicle shutdown or other preventive actions.
The model processes the input video feed, identifies drowsiness, and
triggers the appropriate response. This system is cost-effective, accurate, and
capable of functioning in diverse environments.
LITERATURE REVIEW / SURVEY
CNN for Traffic Monitoring: Research indicates CNN models achieve high
accuracy in object detection and traffic density estimation. Pre-trained models
like YOLO and MobileNet can be fine-tuned for real-time applications.
IoT in Traffic Systems: Studies reveal that IoT-based platforms improve data
accessibility and decision-making for traffic management.
Existing Implementations: Sensor-based systems lack adaptability, highlighting
the need for AI-based solutions that learn and improve over time.
BLOCK DIAGRAM
Block Diagram Description:
The system comprises:
1.Camera Module: Captures real-time traffic images.
2.Arduino Uno: Interacts with the camera module and ESP8266.
3.ESP8266: Sends traffic data to the IoT platform for visualization and alerts.
4.Cloud IoT Platform: Displays live traffic updates and density levels.
5.Deep Learning Model (Off-System): Processes traffic images for classification.
LCD Display
Input
IOT
Image (camera)
Server
Pre
Processing ARDUINO
Controller
(ATMEGA-328)
Alarm unit
Convolution
Neural Network
AI based Eye Relay With
detection Diriver
algorithm for
drowsiness
detection Motor
application (Braking Unit)
Power supply unit
BLOCK DIAGRAM EXPLANATION
1)Camera Module: Captures real-time video of the driver.
2)CNN Model: Processes the video feed to detect facial features and classify the
state (awake or drowsy).
3)Arduino Microcontroller: Receives the detection results and triggers the
necessary actions.
4)LCD Display: Shows system status (e.g., "Active", "Drowsy").
5)Alarm: Emits a loud sound to wake the driver.
6)Motor: Simulates the vehicle's engine; stops in case of drowsiness.
CIRCUIT DIAGRAM
MICRO
CONTROLLER
ARDUINO UNO
AVR 8-bit RISC architecture
• Available in DIP package
• Up to 20 MHz clock
• 32kB flash memory
• 1 kB SRAM
• 23 programmable I/O channels
• Six 10-bit ADC inputs
• Three timers/counters
• Six PWM outputs
LCD DISPLAY
The most commonly used LCDs found in the market
today are 1 Line, 2 Line or 4 Line LCDs which have only
1 controller and support at most of 80 characters,
whereas LCDs supporting more than 80 characters
Most LCDs with 1 controller has 14 Pins and
LCDs with 2 controller has 16 Pins (two pins are extra in
both for back-light LED connections).
CHARACTER LCD TYPE HD44780 PIN DIAGRAM
LCD DESCRIPTION
Starting with an AC voltage, a steady DC
voltage is obtained by rectifying the AC voltage,
then filtering to a DC level, and finally, regulating
to obtain a desired fixed DC voltage. The
regulation is usually obtained from an IC voltage
regulator unit, which takes a DC voltage and
provides a somewhat lower DC voltage, which
remains the same even if the input DC voltage
varies, or the output load connected to the DC
voltage changes. A block diagram containing the
parts of a typical power supply and the voltage at
various points in the unit is shown in figure.
The AC voltage, typically 120 Vrms, is
connected to a transformer, which steps that AC
voltage down to the level for the desired DC
output. A diode rectifier then provides a full-
wave rectified voltage that is initially filtered by
a simple capacitor filter to produce a DC voltage.
This resulting DC voltage usually has some
ripple or AC voltage variation. A Regulator
circuit can use this DC input to provide a DC
voltage that not only has much less ripple
voltage but also remains the same DC value even
if the input DC voltage varies somewhat, or the
load connected to the output DC voltage
changes.
ADVANTAGES
Non-Intrusive: Monitors drowsiness without physical contact.
Affordable: Utilizes cost-effective hardware components.
Real-Time Detection: Provides immediate feedback.
Portable: Compact and easy to install in vehicles.
Customizable: Adjustable thresholds for different drivers.
APPLICATIONS
Automotive Industry: Enhances safety in commercial and personal vehicles.
Public Transport: Prevents accidents in buses and taxis.
Fleet Management: Ensures safety compliance for logistics companies.
Research: Used as a prototype for further advancements in AI-based safety systems.
HARDWARE REQUIREMENTS
Camera
Personal computer
DC Motor
ARDUINO controller
WIFI Module (ESP8266)
LCD Display
Power supply unit
LED Indicator
Relay with driver
SOFTWARE REQUIREMENTS
EMBEDDED C
ARDUINO-1.8.13
BLYNK MOBILE APPLICATION
Python IDE
REFERENCES
[1] Miaou, “Study of Vehicle Scrap page Rates,” Oak Ridge National
Laboratory, Oak Ridge, TN,, S.P.,April 2012.
[2] Wreggit, S. S., Kim, C. L., and Wierwille, W. W., Fourth Semi-Annual
Research Report”, Research on Vehicle-Based Driver Status
Performance Monitoring”, Blacksburg, VA: Virginia Polytechnic
Institute and State University, ISE Department, January 2013.
[3] Bill Fleming, “New Automotive Electronics Technologies”,
International Conference on Pattern Recognition, pp. 484-
488,December 2012.
TensorFlow or PyTorch for CNN training.
OpenCV for image processing.
blynk or Firebase for IoT integration.
Arduino IDE for microcontroller programming.
THANK YOU
