Attention-Based Convolutional and Recurrent Neural Networks for Driving Behavior Recognition Using Smartphone Sensor Data

Zhang, Jun; Zhang, Wu; Fang, Li; Luo, Jianfei; Ren, Tingting; Hu, Song; Li, Wenjing; Li, Wei

doi:10.1109/access.2019.2932434

Cited by 34 publications

(19 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, there were 10 drivers whose faces were occluded or far away from the camera, and the expression data could not be well recognized, thus a total of 12 drivers were finally employed. Considering the long test time and high sampling frequency, a dataset with the test results from 12 drivers should already be sufficient [ 32 , 33 , 34 ]. Due to the specificity of this occupation, the 12 drivers were all male, with an average age of 36 years old, and they all had three or more years’ driving experience.…”

Section: Data Collection and Pre-processingmentioning

confidence: 99%

Real-Time Driving Behavior Identification Based on Multi-Source Data Fusion

Xie

Chen

et al. 2021

IJERPH

View full text Add to dashboard Cite

Real-time driving behavior identification has a wide range of applications in monitoring driver states and predicting driving risks. In contrast to the traditional approaches that were mostly based on a single data source with poor identification capabilities, this paper innovatively integrates driver expression into driving behavior identification. First, 12-day online car-hailing driving data were collected in a non-intrusive manner. Then, with vehicle kinematic data and driver expression data as inputs, a stacked Long Short-Term Memory (S-LSTM) network was constructed to identify five kinds of driving behaviors, namely, lane keeping, acceleration, deceleration, turning, and lane changing. The Artificial Neural Network (ANN) and XGBoost algorithms were also employed as a comparison. Additionally, ten sliding time windows of different lengths were introduced to generate driving behavior identification samples. The results show that, using all sources of data yields better results than using the kinematic data only, with the average F1 value improved by 0.041, while the S-LSTM algorithm is better than the ANN and XGBoost algorithms. Furthermore, the optimal time window length is 3.5 s, with an average F1 of 0.877. This study provides an effective method for real-time driving behavior identification, and thereby supports the driving pattern analysis and Advanced Driving Assistance System.

show abstract

Section: Data Collection and Pre-processingmentioning

confidence: 99%

Real-Time Driving Behavior Identification Based on Multi-Source Data Fusion

Xie

Chen

et al. 2021

IJERPH

View full text Add to dashboard Cite

show abstract

“…Besides that, portable sensors still face some technical problems, such as battery life, sensor size, water resistance and wearing comfort, thereby greatly limiting the wide application of this technology. To alleviate these problems, some researchers use smartphones to assist in daily activity monitoring [16].…”

Section: ) Sensor-based Activity Monitoringmentioning

confidence: 99%

Graph-Based Semi-Supervised Learning for Activity Labeling in Health Smart Home

Wang

Sun

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Health Smart Home (HSH) is an important part of smart city. This technology provides a new kind of remote medical treatment, and can effectively alleviate the shortage of medical resources caused by aging population and help elderly people live at home more safely and independently. Activity recognition is the core of Health Smart Home. However, constructing activity recognition models usually requires a large amount of labeled data, which imposes a heavy burden on manual labeling. In this paper, the authors propose an activity labeling approach based on a graph-based semi-supervised learning algorithm. This approach can divide the raw sensor event sequence without any label information into appropriate segments. Consecutive sensor events that occurred in a same activity are grouped into a same segment. In addition, this approach requires only a small number of manually labeled segments to complete the labeling of the remaining large number of unlabeled segments, thereby greatly reducing the burden of manual labeling. After that, all the labeled data can be further used for activity recognition in smart homes. Finally, a series of comprehensive experiments are conducted on freely available data sets to validate the effectiveness of the proposed activity labeling approach.

show abstract

“…To alleviate these problems, smart phones are used for daily activity monitoring. 13 Compared with portable sensors, non-intrusive sensors do not impose any burden on users. They are usually low-cost and can be deployed at different positions in a smart home to record locations of users at any time.…”

Section: Activity Monitoring Technologymentioning

confidence: 99%

Genetic algorithm–optimized support vector machine for real-time activity recognition in health smart home

Wang

Sun

et al. 2020

International Journal of Distributed Sensor Networks

View full text Add to dashboard Cite

Health smart home, as a typical application of Internet of things, provides a new solution for remote medical treatment. It can effectively relieve pressure from shortage of medical resources caused by aging population and help elderly people live at home more independently and safely. Activity recognition is the core of health smart home. This technology aims to recognize the activity patterns of users from a series of observations on the user’ actions and the environmental conditions, so as to avoid distress situations as much as possible. However, most of the existing researches focus on offline activity recognition, but not good at online real-time activity recognition. Besides, the feature representation techniques used for offline activity recognition are generally not suitable for online scenarios. In this article, the authors propose a real-time online activity recognition approach based on the genetic algorithm–optimized support vector machine classifier. In order to support online real-time activity recognition, a new sliding window-based feature representation technique enhanced by mutual information between sensors is devised. In addition, the genetic algorithm is used to automatically select optimal hyperparameters for the support vector machine model, thereby reducing the recognition inaccuracy caused by manual tuning of hyperparameters. Finally, a series of comprehensive experiments are conducted on freely available data sets to validate the effectiveness of the proposed approach.

show abstract

Attention-Based Convolutional and Recurrent Neural Networks for Driving Behavior Recognition Using Smartphone Sensor Data

Cited by 34 publications

References 39 publications

Real-Time Driving Behavior Identification Based on Multi-Source Data Fusion

Real-Time Driving Behavior Identification Based on Multi-Source Data Fusion

Graph-Based Semi-Supervised Learning for Activity Labeling in Health Smart Home

Genetic algorithm–optimized support vector machine for real-time activity recognition in health smart home

Contact Info

Product

Resources

About