Design and realization of learning emotion database

Expression image is depended on analyzing and studying about emotions based on expression recognition technology. However, in-depth research of emotional analysis cannot be supported because of the limited sample size, the shot scene set in constrained environment like laboratory and image with simple labeling expression information. In order to solve these issues, LDA learning emotion database is built while further studying the labeling method of emotion database. As a result, a learning emotion database is formed, which contains 22,708 expressive images that are provided by 144 learners, 1,792 sets of image sequence and 243 video clips. All of expressive images have double labeling information, expression label and emotion label. This database plays a significant role in learning emotion analysis in an academic environment.