Enhancing Stress Detection: A Comprehensive Approach through rPPG Analysis and Deep Learning Techniques.

Area of Science:

• Computer Science
• Biomedical Engineering
• Psychology

Background:

• Stress significantly impacts modern health and society, necessitating effective monitoring solutions.
• Current stress detection methods often rely on wearable devices, limiting accessibility and comfort.
• Accurate stress detection is crucial for understanding its health and social consequences.

Purpose of the Study:

• To propose an efficient deep learning approach for remote stress detection using facial videos.
• To develop novel Hybrid Deep Learning (DL) networks for stress detection based on remote photoplethysmography (rPPG).
• To enhance stress detection accuracy and computational efficiency compared to existing methods.

Main Methods:

• Utilized hybrid deep learning models including Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and 1D Convolutional Neural Network (1D-CNN).
• Employed remote photoplethysmography (rPPG) signals extracted from facial videos for stress detection.
• Incorporated hyperparameter optimization and data augmentation techniques to improve model performance.

Main Results:

• Achieved up to 95.83% accuracy in stress detection on the UBFC-Phys dataset.
• Demonstrated substantial improvements in both accuracy and efficiency of stress detection.
• Maintained excellent computational efficiency throughout the experiments.

Conclusions:

• The proposed Hybrid DL models are effective for rPPG-based remote stress detection.
• This approach offers a promising non-invasive alternative to wearable-based stress monitoring.
• The findings highlight the potential of deep learning in addressing the societal challenge of stress.