Facial Expression Recognition using Multiple Kinect version 2 Cameras

In every presentation, there should be a standard of presenter's performance index to detect whether the audience is interested or not. In this research, the authors propose an application to automate facial expression recognition using multiple Kinect version 2 cameras called Auditoria. The sensor and Kinect FaceBasic API enables real-time audience facial recognition monitoring accurately. First, the depth sensor catch facial emotions such as happy and unhappy. Then, the captured data is processed by Auditoria to be shown in the website later. An interactive result in a form of charts showing the audience's satisfaction level can be accessed through Auditoria's website. This data is particularly useful for educational institution and others to improve speaker's skill. However, there are still some limitation factors regarding optimal range, light interference, and multiple Kinect overlapping that must be further tested to deliver more consistent detection and accurate result.