EEG-Based Driver Fatigue Detection Using Spatio-Temporal Fusion Network With Brain Region Partitioning Strategy

IEEE Transactions on Intelligent Transportation Systems, Journal Year: 2024, Volume and Issue: 25(8), P. 9618 - 9630

Published: July 11, 2024

Language: Английский

---

Machine learning and deep learning techniques for driver fatigue and drowsiness detection: a review

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(3), P. 9441 - 9477

Published: June 19, 2023

Language: Английский

---

Optimizing Road Safety: Advancements in Lightweight YOLOv8 Models and GhostC2f Design for Real-Time Distracted Driving Detection

Sensors, Journal Year: 2023, Volume and Issue: 23(21), P. 8844 - 8844

Published: Oct. 31, 2023

The rapid detection of distracted driving behaviors is crucial for enhancing road safety and preventing traffic accidents. Compared with the traditional methods distracted-driving-behavior detection, YOLOv8 model has been proven to possess powerful capabilities, enabling it perceive global information more swiftly. Currently, successful application GhostConv in edge computing embedded systems further validates advantages lightweight design real-time using large models. Effectively integrating strategies into models reducing their impact on performance become a focal point field based deep learning. Inspired by GhostConv, this paper presents an innovative GhostC2f design, aiming integrate idea linear transformation generate feature maps without additional computation distracted-driving-detection tasks. goal reduce parameters computational load. Additionally, enhancements have made path aggregation network (PAN) amplify multi-level fusion contextual propagation. Furthermore, simple attention mechanisms (SimAMs) are introduced perform self-normalization each map, emphasizing valuable suppressing redundant interference complex backgrounds. Lastly, nine distinct types publicly available SFDDD dataset were expanded 14 categories, nighttime scenarios introduced. results indicate 5.1% improvement accuracy, weight size load reduced 36.7% 34.6%, respectively. During 30 real vehicle tests, accuracy reached 91.9% during daylight 90.3% at night, affirming exceptional proposed assisting when contributing accident-risk reduction.

Language: Английский

---

Driving drowsiness detection using spectral signatures of EEG-based neurophysiology

Frontiers in Physiology, Journal Year: 2023, Volume and Issue: 14

Published: March 30, 2023

Introduction: Drowsy driving is a significant factor causing dire road crashes and casualties around the world. Detecting it earlier more effectively can significantly reduce lethal aftereffects increase safety. As physiological conditions originate from human brain, so neurophysiological signatures in drowsy alert states may be investigated for this purpose. In preface, A passive brain-computer interface (pBCI) scheme using multichannel electroencephalography (EEG) brain signals developed spatially localized accurate detection of drowsiness during tasks. Methods: This pBCI modality acquired electrophysiological patterns 12 healthy subjects prefrontal (PFC), frontal (FC), occipital cortices (OC) brain. Neurological are recorded six EEG channels spread over right left hemispheres PFC, FC, OC sleep-deprived simulated post-hoc analysis, spectral δ, θ, α, β rhythms extracted terms band powers their ratios with temporal correlation complete span experiment. Minimum redundancy maximum relevance, Chi-square, ReliefF feature selection methods used aggregated Z-score based approach global ranking. The attributes classified decision trees, discriminant logistic regression, naïve Bayes, support vector machines, k-nearest neighbors, ensemble classifiers. binary classification results reported confusion matrix-based performance assessment metrics. Results: inter-classifier comparison, optimized model achieved best 85.6% accuracy precision, 89.7% recall, 87.6% F1-score, 80% specificity, 70.3% Matthews coefficient, 70.2% Cohen's kappa score, 91% area under receiver operating characteristic curve 76-ms execution time. inter-channel were obtained at F8 electrode position FC significance all was validated p-value less than 0.05 statistical hypothesis testing methods. Conclusions: proposed has better accomplishment multiple objectives. predictor importance reduced extraction cost computational complexity minimized use conventional machine learning classifiers resulting low-cost hardware software requirements. channel most promising region only single (F8) which reduces physical intrusiveness normal operation. good potential practical applications requiring earlier, accurate, disruptive information biosignals.

Language: Английский

---

Fusion of EEG and Eye Blink Analysis for Detection of Driver Fatigue

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 31, P. 2037 - 2046

Published: Jan. 1, 2023

Objective : The driver fatigue detection using multi-channel electroencephalography (EEG) has been extensively addressed in the literature. However, employment of a single prefrontal EEG channel should be prioritized as it provides users with more comfort. Furthermore, eye blinks from such can analyzed complementary information. Here, we present new method based on simultaneous and analysis an Fp1 channel. xmlns:xlink="http://www.w3.org/1999/xlink">Methods First, moving standard deviation algorithm identifies blink intervals (EBIs) to extract blink-related features. Second, discrete wavelet transform filters EBIs signal. Third, filtered signal is decomposed into sub-bands, various linear nonlinear features are extracted. Finally, prominent selected by neighbourhood components fed classifier discriminate between alert driving. In this paper, two different databases investigated. first one used for parameters' tuning proposed filtering, measures, feature selection. second solely testing robustness tuned parameters. xmlns:xlink="http://www.w3.org/1999/xlink">Main results comparison obtained results both AdaBoost terms sensitivity (90.2% vs. 87.4%), specificity (87.7% 85.5%), accuracy (88.4% 86.8%) indicates reliability detection. xmlns:xlink="http://www.w3.org/1999/xlink">Significance Considering existence commercial headbands, detect real-world scenarios.

Language: Английский

---

Evaluation of a New Lightweight EEG Technology for Translational Applications of Passive Brain-Computer Interfaces

Frontiers in Human Neuroscience, Journal Year: 2022, Volume and Issue: 16

Published: July 14, 2022

Technologies like passive brain-computer interfaces (BCI) can enhance human-machine interaction. Anyhow, there are still shortcomings in terms of easiness use, reliability, and generalizability that prevent passive-BCI from entering real-life situations. The current work aimed to technologically methodologically design a new gel-free system for out-of-the-lab employment. choice the water-based electrodes lightweight headset met need easy-to-wear, comfortable, highly acceptable technology. proposed showed high reliability both laboratory realistic settings, performing not significantly different gold standard based on gel electrodes. In cases, allowed effective discrimination (AUC > 0.9) between low levels workload, vigilance, stress even temporal resolution (<10 s). Finally, has been tested through cross-task calibration. calibrated with data recorded during tasks was able discriminate targeted human factors task reaching AUC values higher than 0.8 at 40 s case vigilance 20 monitoring. These results pave way ecologic use system, where calibration difficult obtain.

Language: Английский

---

An Approach to Hyperparameter Tuning in Transfer Learning for Driver Drowsiness Detection Based on Bayesian Optimization and Random Search

International Journal of Advanced Computer Science and Applications, Journal Year: 2023, Volume and Issue: 14(4)

Published: Jan. 1, 2023

Driver drowsiness is a critical factor in road safety, and developing accurate models for detecting it essential. Transfer learning has been shown to be an effective technique driver detection, as enables leverage large, pre-existing datasets. However, the optimization of hyper-parameters transfer can challenging, involves large search space. The core purpose this research on presenting approach hyperparameter tuning driving fatigue detection based Bayesian Random algorithms. We examine efficiency our publicly available dataset using with MobileNetV2, Xception, VGG19 architectures. explore impact hyperparameters such dropout rate, activation function, number units (the dense nodes), optimizer, rate models' overall performance. Our experiments show that improves performance models, obtaining cutting-edge results all three also compare algorithms terms their ability find optimal indicate more efficient finding than search. study provide insights into importance learning-based different guide selection future studies field. proposed applied other tasks, making valuable contribution field ML.

Language: Английский

---

Facial Feature-Based Drowsiness Detection With Multi-Scale Convolutional Neural Network

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 63417 - 63429

Published: Jan. 1, 2023

Recently, the upsurge in accidents is caused due to driver drowsiness arises lack of sleep, fatigue and other health factors. The leads mortality, loss properties serious conditions. Hence, it necessary prevent by drivers. At present, automated model effective for detection recognition. In this research paper, developed a MCNN (Multi-Scale Convolutional Neural Network) framework classification drowsiness. Initially, YAWDD dataset NTHU-DDD utilized acquiring video sequences about driving. acquired converted into frames keyframe extraction selection. With Dlib library face recognition localization facial points extracted frames. image are pre-processed with Cross Guided Bilateral Filtering followed feature hybrid dual-tree complex wavelet transforms Walsh-Hadamard transform vector frame blocks. optimized Flamingo search algorithm (FSA) integrated deep learning Multiscale convolutional neural network (MCNN). proposed method, based FSA drowsy non-drowsy classified. simulation results illustrated attains an accuracy value around 98.38% exhibits 98.26%. performance approximately 6% higher than conventional state-of-art methods.

Language: Английский

---

EEG and ECG-Based Multi-Sensor Fusion Computing for Real-Time Fatigue Driving Recognition Based on Feedback Mechanism

Sensors, Journal Year: 2023, Volume and Issue: 23(20), P. 8386 - 8386

Published: Oct. 11, 2023

A variety of technologies that could enhance driving safety are being actively explored, with the aim reducing traffic accidents by accurately recognizing driver's state. In this field, three mainstream detection methods have been widely applied, namely visual monitoring, physiological indicator monitoring and vehicle behavior analysis. order to achieve more accurate driver state recognition, we adopted a multi-sensor fusion approach. We monitored signals, electroencephalogram (EEG) signals electrocardiogram (ECG) determine fatigue state, while an in-vehicle camera observed provided information for assessment. addition, outside was used monitor position whether there were any deviations due distraction or fatigue. After series experimental validations, our research results showed approach exhibited good performance recognition. This study provide solid foundation development direction future in-depth recognition research, which is expected further improve road safety.

Language: Английский

---

Efficient Embedded System for Drowsiness Detection Based on EEG Signals: Features Extraction and Hardware Acceleration

Electronics, Journal Year: 2025, Volume and Issue: 14(3), P. 404 - 404

Published: Jan. 21, 2025

Drowsiness detection is crucial for ensuring the safety of individuals engaged in high-risk activities. Numerous studies have explored drowsiness techniques based on EEG signals, but these typically been validated computers, which limits their portability. In this paper, we introduce design and implementation a technique utilizing executed Zynq7020 System Chip (SoC) as part Pynq-Z2 module. This approach more suitable portable applications. We implemented Discrete Wavelet Transform (DWT) feature extraction functions intellectual property (IP) cores, while other run ARM processor Zynq7020.

Language: Английский

---