Two-Stream Spatial–Temporal Graph Convolutional Networks for Driver Drowsiness Detection

Bai, Jing; Yu, Wentao; Xiao, Zhu; Havyarimana, Vincent; Regan, Amelia; Jiang, Hongbo; Jiao, Licheng

doi:10.1109/tcyb.2021.3110813

Cited by 43 publications

(22 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Accuracy (%) HTDBN [5] 84.82 DrowsyNet [4] 86.90 MCNN [3] 90.05 2s-STGCN [1] 92.70 Two-Stream CNN3D [8] 94.46 HGLDD/T=10 HGLDD/T=15 HGLDD/T=20 HGLDD/T=25…”

Section: Methodsmentioning

confidence: 99%

“…2) Computer-vision-based methods: They employ image features, such as facial expression, eye movements, and mouth shapes, to discern whether or not a driver is in a drowsy state. Convolutional neural network (CNN) models using facial characteristics have made significant achievements in this area [1].…”

Section: Introductionmentioning

confidence: 99%

“…A deep LSTM-FCN based drowsiness detection model with facial landmark sequence as input was established by Zu et al (2022) [4], achieving higher accuracy. Bai et al (2022) [1] proposed a model using two-stream graph networks, which constructed a spatial and a temporal graph of facial landmarks, respectively, and eventually achieved a drowsiness detection accuracy of 92.7%.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A hybrid GCN-LSTM model for driver drowsiness detection

Gao,

Duan,

et al. 2023

Fourth International Conference on Signal Processing and Computer Science (SPCS 2023)

View full text Add to dashboard Cite

“…Accuracy (%) HTDBN [5] 84.82 DrowsyNet [4] 86.90 MCNN [3] 90.05 2s-STGCN [1] 92.70 Two-Stream CNN3D [8] 94.46 HGLDD/T=10 HGLDD/T=15 HGLDD/T=20 HGLDD/T=25…”

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A hybrid GCN-LSTM model for driver drowsiness detection

Gao,

Duan,

et al. 2023

Fourth International Conference on Signal Processing and Computer Science (SPCS 2023)

View full text Add to dashboard Cite

“…1) Spatial Feature Aggregation. Graph neural networks (GNNs) typically update node representations by aggregating features of their neighbors iteratively, thereby capturing spatial/structural information across the entire graph [24], [30]. We employ graph isomorphism network (GIN) [31] for spatial representation learning, a GNN variant with high discriminative power in graph classification.…”

Section: A Construction Of Pretrained Source Modelmentioning

confidence: 99%

Unsupervised cross-domain functional MRI adaptation for automated major depressive disorder identification

Fang

Wang

Potter

et al. 2023

Medical Image Analysis

View full text Add to dashboard Cite

“…The proposed PFTL-DDD method is evaluated on the NTHU-DDD and YAWDD benchmark video datasets which are widely used in driver drowsiness detection researches [34][35][36][37][38][39][40][41][42][43]. The NTHU-DDD is an open-source driver drowsiness video dataset collected by the Computer Vision Lab of National Tsing Hua University [7].…”

Section: A Datasetmentioning

confidence: 99%

Privacy-Preserving Federated Transfer Learning for Driver Drowsiness Detection

et al. 2022

View full text Add to dashboard Cite

The drowsiness affects the driver's sensory, cognitive, and psychomotor abilities, which are necessary for safe driving. Drowsiness detection is a critical technique to avoid traffic accidents. The federated learning (FL) can solve the problem of insufficient driver facial data by utilizing different industrial entities' data. However, in the FL system, the privacy information of the drivers might be leaked. In addition, reducing the communication costs and maintaining the model performance is also a challenge in industrial scenarios. In this work, we propose a federated transfer learning method with privacypreserving protocol for driver drowsiness detection, named PFTL-DDD. We use fine-tuning transfer learning on the initial model of the drowsiness detection FL system. Furthermore, a CKKS-based privacypreserving protocol is applied to preserve the drivers' privacy data by encrypting the exchanged parameters. The experimental results show that the PFTL-DDD method is superior in terms of accuracy and efficiency as compared to the conventional federated learning on the NTHU-DDD dataset and YAWDD dataset. The theoretical analysis demonstrates that the proposed transfer learning method can reduce the communication cost of the system, and the CKKS-based security protocol can protect the personal privacy.INDEX TERMS Driver drowsiness detection, transfer learning, federated learning, privacy-preserving.

show abstract

Two-Stream Spatial–Temporal Graph Convolutional Networks for Driver Drowsiness Detection

Cited by 43 publications

References 41 publications

A hybrid GCN-LSTM model for driver drowsiness detection

A hybrid GCN-LSTM model for driver drowsiness detection

Unsupervised cross-domain functional MRI adaptation for automated major depressive disorder identification

Privacy-Preserving Federated Transfer Learning for Driver Drowsiness Detection

Contact Info

Product

Resources

About