Measuring the advertisement effect based on emotional responses analysis with crowdsourcing

摘要

In this study, we estimate ad liking and purchase intent in the natural environment by using crowdsourcing and the remote measurement of facial expressions and physiological signals. In recent years the market of online video advertisement has expanded, and demand for measuring advertisement effect is increasing. We collected a significant number of videos of Japanese faces watching video advertisements via the Internet. By combining the multiple features of facial expressions and using support vector machine(SVM), we made it possible to estimate the advertisement effect. In addition, it was possible to improve the accuracy more than using only single-mode features.