PPGface

• DOI: 10.1145/3534597
• Publication Date: 2022-07-07T22:50:18Z
• Authors: Seokmin Choi; Yang Gao; Yincheng Jin; Se jun Kim; Jiyang Li; Wenyao Xu; Zhanpeng Jin
• Keywords: Robustness; Facial muscles; Facial expression recognition; Activity Recognition

Recognition of facial expressions has been widely explored to represent people's emotional states. Existing expression recognition systems primarily rely on external cameras which make it less accessible and efficient in many real-life scenarios monitor an individual's a convenient unobtrusive manner. To this end, we propose PPGface, ubiquitous, easy-to-use, user-friendly platform that leverages earable devices with built-in PPG sensor. PPGface understands the through dynamic patterns resulting from muscle movements. With aid accelerometer sensor, can detect recognize user's seven universal relevant body posture unobtrusively. We conducted user study (N=20) using multimodal ResNet evaluate performance showed different 93.5 accuracy 0.93 fl-score. In addition, explore robustness usability our proposed platform, several comprehensive experiments under real-world settings. Overall results work validate great potential be employed future commodity devices.