Emotion based classification of natural images
Michela Dellagiacoma, Pamela Zontone, Giulia Boato, Liliana Albertazzi · 2011
Images convey opinions and emotional messages in the communication process. With the increasing use of images in various scenarios, the area of opinion mining and sentiment analysis has recently received a huge burst of interest. In particular, in the context of social web the ability of identify different emotions in images might help providing diversification of results, thus proposing different viewpoints to users. In this paper we analyze which are the features (e.g., colors, texture) that are more strictly related to the emotional content of a picture, thus allowing a classification connected with the emotion conveyed by images. We present the results on a set of natural images in order to reduce as much as possible the interaction with content semantics.