Date of Award

Spring 2022

Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Ahamed, Sheikh Iqbal


Stress has a significant impact on the physical and mental health of an individual and is a growing concern for society, especially during the COVID-19 pandemic. Facial video-based stress evaluation from non-invasive cameras has proven to be a significantly more efficient method to evaluate stress in comparison to approaches that use questionnaires or wearable sensors. Plenty of classification models have been built for stress detection. However, most do not consider individual differences. Also, the results for such models are limited by a uni-dimensional definition of stress levels lacking a comprehensive quantitative definition of stress. The dissertation focuses on building a framework that utilizes the multilevel video frame representations from deep learning and the remote photoplethysmography signals extracted from the facial videos for stress assessment. The fusion model takes the inputs of a baseline video and a target video of the subject. The physiological features such as heart rate and heart rate variability are used with the initial stress scores generated from deep learning are used to predict the stress scores in cognitive anxiety, somatic anxiety, and self-confidence. To generate stress scores with better accuracy, the signal extraction method is improved by introducing the CWT-SNR method that uses the signal-to-noise ratio to assist the adaptive bandpass filtering in the post-processing of the signals. A study on phase space reconstruction features is performed and the results show the potential for additional accuracy improvement for the heart rate variability detection. To select the best deep learning architecture, multiple deep learning architectures are tested to build the deep learning model. Support Vector Regression is used to generate the output stress score results. Testing with the data from the UBFC-Phys dataset, the fusion model shows a strong correlation between ground truth and the predicted results.