A speaker must have the ability to read the feeling of the audience. In order to provide the speaker an immediate feedback from the audience's reaction, we propose an emotion-reading system, which is based on the analysis of the facial emotion of the audience, to let the speaker decide on what to say in the next couple of minutes.
The emotion-reading system detects the faces appear in the given input, and analyzes the emotion of it with nural networks. Based on the results, it then calculates a score that reflects the response of the people.
By the emotion score, it may serve as an indicator of the response of the audience
Source | |
---|---|
MTCNN | https://github.com/kpzhang93/MTCNN_face_detection_alignment |
Videos | https://www.pond5.com/ |