By: Arne Boon, Guido Huijser, Roy van Rooijen
This project explores the application of recognizing facial expressions for decision-making processes in interactive technology systems.
The proliferation of the Internet, and more specifically of online channels such as YouTube, has caused an exponential rise in the amount of digital media available today. Navigating this huge database demands increasingly sophisticated filtering and query techniques, based on the somewhat behaviorist approach of ranking data on the amount views and 'likes' it receives from its users.
Most people would agree that their opinion is not as clear-cut as 'like' versus 'do not like', and the fact that rating is done post-viewing allows for contrived instead of natural appreciation. The SmileTracker application developed in this study aims to accurately gauge the users' emotive response to digital media, specified to humorous videos. Using an interface with video playback and webcam capturing functionalities, the application detects the user’s emotive response and dynamically composes a ranking of the most fun-appreciated videos.
This project is the first step towards a more developed goal, which aims non-linear storytelling steered by facial expressions of the viewer.