Driver Distraction Detection and Alert System Using Convolutional Neural Networks

Abstract

The Driver Distraction and Alert System Using Convolutional Neural Networks provide a real-time machine learning and webcam-based monitoring system for driver distraction detection. The system uses a sophisticated Convolutional Neural Network (CNN) architecture trained on the State Form of Distracted Driver Detection dataset, which is made possible by utilizing the OpenCV, TensorFlow, and Pandas packages. With improvements like dropout layers and Xavier weight initialization, the model efficiently classifies ten driver attention states through preprocessing stages like image scaling and normalization. The results of the experiment show how well the suggested method works to identify distracted driving behaviors; a revised model on the Kaggle competition platform achieved a competitive public score of 2.67118.

Introduction

I. INTRODUCTION

In Driver Distraction And Alert System Using Convolutional Neural Network, a camera technique to real-time driver attention monitoring is presented. To identify and address driver distraction, it makes use of Pandas, TensorFlow, and OpenCV. A CNN model is created to precisely categorize ten driving behaviors by utilizing the state form of distracted driver detection dataset. The method improves real-time distraction detection using data preparation and model optimization techniques like weight initialization and dropout layers. Using a multi-pronged strategy, the system makes use of TensorFlow for smooth model generation, OpenCV for reliable image processing, and Pandas for effective data manipulation. By means of methodical preprocessing and deliberate model improvements, like weight initialization and dropout layers, it attains increased precision in recognising various driving behaviours in real-time situations. This combination of technologies highlights a thorough approach to reducing driver distraction and also improving traffic safety.

II. LITERATURE REVIEW

III. METHODOLOGY

The Driver Distraction Detection and Alert System is a cutting-edge solution designed to enhance road safety by effectively monitoring and alerting drivers about potential distractions. By integrating various technologies, it provides comprehensive analysis using a single camera system. Utilizing Python with OpenCV and Media-pipe libraries, the system processes real-time video feed from a single camera. Initially, the system detects the driver's face, employing advanced face tracking algorithms to ensure accurate monitoring. Once the face is detected, the system proceeds to extract key facial landmarks, categorizing them into three general regions: the eye region, mouth, and head. This step enables the system to gather essential data regarding the driver's attention state, including indicators like tiredness, drowsiness, eye closure, gaze deviation, and head pose deviations.Next, attention estimation of the drivers is performed using the extracted facial features. By analyzing the collected data, the system can accurately assess the driver's level of distraction and promptly issue alerts when necessary. Moreover, the system incorporates a Convolutional Neural Networks(CNN) architecture to further enhance its capabilities. By preprocessing the captured images to include various driver postures such as eating, texting on a cell phone, or conversing with others, the system ensures comprehensive coverage of potential distractions. The pre-trained Convolutional model embedded within the system is then employed to identify distracted behaviors exhibited by the driver. This includes activities like eating, texting, talking to others, or engaging in other distracting behaviors. Upon detection, the system promptly alerts the driver, promoting immediate corrective action.By combining both facial analysis and behavior recognition techniques into a single-camera system, the proposed solution offers real-time monitoring and alerting mechanisms, thereby significantly contributing to enhanced road safety through proactive detection of driver distraction behaviors. Additionally, the system continuously adapts and learns from real-world data, refining its algorithms to enhance accuracy and adaptability to diverse driving scenarios . Its seamless integration with the existing vehicle systems ensures easy implementation and wide spread adoption, making it a valuable tool in mitigating risks associated to drivers distraction on the roads.

IV. RESULTS AND DISCUSSION

The proposed driver’s distraction detection system utilizes a refined CNN architecture trained on the state form of distracted driver detection data set. Through techniques such as dropout regularization, Xavier weight initialization, and zero-mean normalization, the model achieved improved performance, with a public score of 2.67118. By prepossessing the images, including re-sizing, normalization, and channel utilization, the system effectively handles input data. Implemented in Python using OpenCV, TensorFlow, and Pandas, the system demonstrates real-time monitoring of driver attention states, enabling classification into 10 distinct classes such as texting, talking on phone, and safe driving.If driver is texting the system detects that the driver has been texting towards right, texting left and if the driver has been looking behind then the system detects that the person is reaching behind and even if the driver is speaking to passenger outside the car environment then the system detects that the the driver is speaking to passenger all the above stated driver distracted activity is identified by the system and alerts the driver with beep sound. Furthermore, the system uses a blend of deep learning(DL) and computer vision methods to precisely recognize and categories different types of driver distracted behaviors in real time. By utilizing the capability in Convolutional Neural Network(CNN), the model is able to detect distracted driving behaviors with robustness. It does this by effectively learning discriminating features with raw visual data. By reducing the danger associated with driver distraction, the technology seeks to improve road safety through constant monitoring and immediate alerts.

V. ACKNOWLEDGEMENT

The satisfactions that accompany the successful completion of project on "Driver Distraction Detection and Alert System Using Convolutional Neural Networks” would be incomplete without the mention of people who made it possible, whose noble gesture, affection, guidance, encouragement and support crowned my efforts with success. It is my privilege to express my gratitude and respect to all those who inspired us in the completion of this work.

We are extremely grateful to our Guide Mrs.Rachel Evangeline Christian  Asst Prof of Department of AIML for noble gesture, support, co-ordination and valuable suggestions given in completing the work. We also thank Dr. BM Vidyavathi, H.O.D Department of AIML, for her coordination and valuable suggestions given in completing the work.

Conclusion

Driver Distraction and Alert System, which offers a comprehensive solution for real-time driver attention monitoring and distraction detection. The system accomplishes accurate categorization of 10 different driving behaviours by utilising a combination of cutting-edge technologies such as TensorFlow, OpenCV, and Pandas in conjunction with a well-designed convolutional neural networks model that was trained on the state form of distracted driver detection dataset. By implementing rigorous data preparation and model optimisation strategies including weight initialization and dropout layers, the system can more successfully identify and handle driver attention incidents in real-world situations. This all-encompassing strategy emphasises how important it is to use state-of-the-art instruments and techniques that reduce the dangers which comes with distracted driving, ultimately resulting in increased traffic safety.

References

[1] Baheti, Bhakti & Gajre, Suhas & Talbar, Sanjay. (2024). Detection of Distracted Driver Using Convolutional Neural Network. 1145-11456. 10.1109/CVPRW.2018.00150. [2] Deepthi M. Pisharody, Binu P. Chacko, K.P. Mohamed Basheer, Driver distraction detection using machine learning techniques, Materials Today: Proceedings, Volume 58, Part 1, 2022, Pages 251-255, ISSN 2214-7853, https://doi.org/10.1016/j.matpr.2022.02.108. [3] Shumayla Yaqoob, Giacomo Morabito, Salvatore Cafiso, Giuseppina Pappalardo, Ata Ullah, \"AI-Driven Driver Behavior Assessment Through Vehicle and Health Monitoring for Safe Driving—A Survey\", IEEE Access, vol.12, pp.48044-48056, 2024.https://ieeexplore.ieee.org/document/9405644. [4] Eva Michelaraki Accident Analysis & Prevention Volume 192, November2023, 107241, https://www.sciencedirect.com/science/article/abs/pii/S0001457523002889. [5] Jamal Raiyn, Galia Weidls, \"Improving Autonomous Vehicle Reasoning with Non-Monotonic Logic: Advancing Safety and Performance in Complex Environments\", 2023 IEEE International Smart Cities Conference (ISC2), pp.1-8, 2023. https://ieeexplore.ieee.org/abstract/document/10041146/citations#citation. [6] A. Kashevnik, R. Shchedrin, C. Kaiser and A. Stocker, \"Driver Distraction Detection Methods: A Literature Review and Framework,\" in IEEE Access, vol. 9, pp. 60063-60076, 2021, doi: 10.1109/ACCESS.2021.3073599.

Copyright

Copyright © 2024 Ms. R Sai Ashritha, Mr. Girish V, Ms. K. S. Meghana Bai, Ms. Meenakshi K, Mrs. Rachel Evangeline Christian . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.