Depression Recognition using Remote Photoplethysmography from Facial Videos

Depression is a mental illness that may be harmful to an individual's health. The detection of health disorders in the early stages and precise diagnosis are critical avoid social, physiological, or psychological side effects. This work analyzes physiological signals observe if different depressive states have noticeable impact on blood volume pulse (BVP) heart rate variability (HRV) response. Although typically, HRV features calculated from biosignals obtained with contact-based sensors such as wearables, we propose instead novel scheme directly extracts them facial videos, just based visual information, removing need for any device. Our solution pipeline able extract complete remote photoplethysmography (rPPG) fully unsupervised manner. We use these rPPG calculate over 60 statistical, geometrical, further used train several machine learning regressors recognize levels depression. Experiments two benchmark datasets indicate this approach offers comparable results other audiovisual modalities voice expression, potentially complementing them. In addition, achieved proposed method show promising solid performance outperforms hand-engineered methods deep learning-based approaches.