Driver distraction detection using capsule network
- PDF / 1,788,149 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 23 Downloads / 267 Views
(0123456789().,-volV)(0123456789(). ,- volV)
ORIGINAL ARTICLE
Driver distraction detection using capsule network Deepak Kumar Jain1 • Rachna Jain2 • Xiangyuan Lan3 • Yash Upadhyay2 • Anuj Thareja2 Received: 4 January 2020 / Accepted: 24 September 2020 Ó Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract With the onset of the new technological age, the distractions caused due to handheld devices have been a major cause of traffic accidents as they affect the decision-making capabilities of the driver and give them less time to react to difficult situations. Often drivers try to multitask which reduces their reaction time leading to accidents which could have been easily avoided if they had been attentive. As such problems are related to the driver’s negligence toward safety, a possible solution is to monitor driver’s behavior and notify if they are distracted. We propose a CapsNet-based approach for detecting the distracted driver which is a novel approach. The proposed method scores perform well on the real-world environment inputs when compared to other famous methods used for the same. Our proposed methods get high scores for all the most commonly used metrics for classification. On the testing set, the proposed method gets an accuracy of 0.90, 0.92 as precision score, 0.90 as recall score and 0.91 as F-measure. Keywords Driver distraction CapsNet Dynamic routing Posture classification Abbreviations CNN Convolutional neural network ANN Artificial neural network MSE Mean squared error SVM Support vector machine RNN Recurrent neural network IMU Inertial measurement unit LSTM Long short-term memory ELM Extreme learning machine KNN K-nearest neighbors
& Rachna Jain [email protected] 1
Key Laboratory of Intelligent Air-Ground Cooperative Control for Universities in Chongqing, College of Automation, Chongqing University of Posts and Telecommunications, Chongqing, China
2
Department of Computer Science and Engineering, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
3
Department of Computer Science, Hong Kong Bapist University, Kowloon Tong, China
1 Introduction According to National Safety Council, more than 390,000 injuries occur each year just from texting and driving; this number would surely go up if we consider other causes of distractions. In addition to distractions, drivers are also fatigued, and in a state of sleepiness, as per the research conducted by Sleep Research Center (UK), distracted driving was the cause for 20% of the accidents. Many studies conducted on traffic accidents have concluded that driver’s not paying attention or being sleeping is a major cause of road accidents. Many systems have been developed which try to reduce such accidents, but still we don’t have a solution which can be used uniformly in all the vehicles. In this paper, we will study what shortcomings do the existing methods have and what improvement can be made. As the root cause of such problems lies with the driver, the best way to combat distractions is to develop sy
Data Loading...
