1.Abstract

Road accidents caused by driver distraction have become one of the major concerns in modern transportation systems. Driver negligence such as using mobile phones while driving, talking to passengers, eating, drinking, sleeping, or losing focus can lead to severe accidents, injuries, and loss of human life. In public transportation systems, school buses, taxis, and private vehicles, passenger safety largely depends on the driver’s attentiveness and responsible behavior. Therefore, continuous monitoring of driver activities is essential to reduce accidents and improve road safety. To address this issue, the proposed project presents an AI-Based Driver Distraction Monitoring System that uses Artificial Intelligence, Deep Learning, and Computer Vision techniques to automatically detect distracted driving behaviors in real time.

The main objective of the system is to monitor driver activities continuously and identify dangerous behaviors before accidents occur. The system analyzes driver images and video frames captured through cameras installed inside vehicles and classifies driver activities into multiple categories. A dataset containing ten different driver activity classes was collected from Kaggle for training the deep learning model. The dataset includes activities such as safe driving, texting, talking on the phone, drinking, reaching behind, adjusting controls, hair and makeup activities, and sleeping or drowsiness-related actions. The collected images were preprocessed and augmented to improve model performance under different lighting conditions and viewing angles.

For activity classification, the project uses the ResNet-50 deep learning architecture, which is a powerful Convolutional Neural Network (CNN) model widely used for image classification tasks. ResNet-50 uses residual learning techniques that allow deep neural networks to train efficiently while avoiding problems such as vanishing gradients. The model was trained from scratch using the collected driver activity dataset to accurately recognize distracted driving behaviors. During training, the network learns important visual patterns such as hand movements, facial orientation, eye status, and body posture associated with each activity class.

The proposed system continuously monitors driver behavior through real-time video analysis. Whenever distracted activities such as mobile phone usage, drowsiness, or inattentive behavior are detected, the system can generate alerts or warnings to prevent accidents. This intelligent monitoring mechanism helps improve driver awareness and enhances passenger safety in transportation systems.

To provide an easy-to-use interface, the project uses a Tkinter-based Graphical User Interface (GUI) as the frontend framework. The GUI allows users to upload images, access webcam feeds, and monitor driver activities in real time. The prediction results, including detected driver activity and confidence score, are displayed through the interface for easy understanding and analysis. The GUI-based design makes the system suitable for deployment in vehicles, transport monitoring systems, driving schools, and fleet management applications.

The AI-Based Driver Distraction Monitoring System offers several advantages including real-time monitoring, improved road safety, reduced accident risks, automated driver behavior analysis, and low human intervention. The system can assist transportation authorities, school management systems, and vehicle monitoring organizations in ensuring safe driving practices. It can also be integrated with smart transportation technologies such as IoT devices, CCTV systems, and automated vehicle safety systems.

In conclusion, the proposed Driver Distraction Monitoring System demonstrates the effective use of Artificial Intelligence and Deep Learning in intelligent transportation safety applications. By combining ResNet-50 architecture with a Tkinter-based GUI framework, the system provides an efficient and practical solution for detecting distracted driving behaviors and preventing accidents. Future enhancements may include real-time alarm systems, facial emotion analysis, GPS tracking integration, night vision monitoring, and cloud-based vehicle surveillance systems to further improve driver safety and transportation management.

2. Objectives

1. To detect driver distractions in real time using AI.
2. To classify driver behavior into multiple activity categories.
3. To identify unsafe driving actions such as phone usage and drowsiness.
4. To reduce road accidents caused by human error.
5. To provide a real-time monitoring system using a user-friendly interface.

3. Existing System

• Traditional driver monitoring systems mainly depend on manual observation and basic surveillance methods to monitor driver behavior.

• Some existing systems use simple image processing and sensor-based techniques to detect driver drowsiness and distractions.

• AI-based systems use machine learning and computer vision techniques to identify distracted driving activities such as mobile phone usage and inattentive driving.

• Existing monitoring systems help improve road safety by tracking driver activities and generating alerts during unsafe driving conditions.

4. Proposed System

• The proposed system uses Artificial Intelligence, Deep Learning, and Computer Vision techniques for automatic driver distraction detection.

• The system is trained using a dataset containing ten different driver activity classes collected from Kaggle.

• ResNet-50 architecture is used for accurate driver activity classification and feature extraction.

• The system can detect activities such as mobile phone usage, drowsiness, talking, eating, and inattentive driving.

• A Tkinter-based graphical user interface is developed for real-time monitoring and user interaction.

• The system continuously monitors driver behavior through webcam or surveillance camera feeds and helps prevent accidents by generating alerts during distracted driving situations.

5. Implementation Procedure

1. Collect dataset from Kaggle containing driver activity images.
2. Preprocess dataset (resizing, normalization, augmentation).
3. Train model using ResNet-50 architecture.
4. Save trained model for inference.
5. Capture real-time video using webcam.
6. Predict driver activity frame by frame.
7. Display detected activity using Tkinter GUI.
8. Trigger warning if distraction is detected.

6.Software Requirements

Operating System                     : Windows 10 64-bit

Programming Language           : Python

Front End                                  : Tkinter GUI

Back End                                   : MySQL

Framework                                : TensorFlow / Keras

Libraries : OpenCV, NumPy, Pandas

7.Hardware Requirements

Processor                                     : Intel i3 or Above

RAM                                            : 4 GB Minimum

Hard Disk : 500 GB

Camera : Webcam / CCTV Camera

System Type : 64-bit Computer System

8. Advantages of the Project

1. Reduces road accidents caused by driver distraction
2. Real-time monitoring system
3. Improves passenger safety in public transport
4. Automated detection without human intervention
5. High accuracy using deep learning model
6. Easy-to-use GUI interface
7. Can be integrated into smart vehicles