Patient Facial Emotion Recognition and Sentiment Analysis Using Secure Cloud with Hardware Acceleration

Currently, cloud computing has become a de-facto platform for data management and analytics. Meanwhile, data stream processing from smart devices leads to new opportunities and challenges with regards to reliable real-time data analysis. Distributed image understanding across heterogeneous computing platforms, such as smart devices and cloud analytic platforms, poses a great motivation to advance conventional data analytics algorithms to distributed heterogynous mobile cloud computing. The facial emotion recognition system presented in this article has a decoupled architecture where video processing, face detection, and facial emotion recognition inference tasks are run on a local client machine while the intelligence models used for these tasks reside in a cloud-based remote storage. Training of the intelligence models used for these tasks is also conducted in a cloud environment using GPU's. This decoupled design transforms remote GPU resources into a utility-based cloud service, providing cloud-based hardware acceleration to a wide range of distributed facial emotion recognition tasks.