Sentiment Analysis In Comics With Deep Neural Networks


The project proposes a Cartoon Sentiment Analysis, where information can be perceived through texts and images of its characters. For this purpose, the Mafalda comic developed by the Argentinean caricaturist Quino was selected. Given the breadth of the opinion concept, its focus will be limited to the expressions of Happiness, Anger, and Sadness. Such expressions will also be analyzed in each of the cartoon characters. The drawings are relatively simple compared to more modern black and white comics. The texts are short and straightforward with well-marked feelings and short strips of up to 5 frames, making it an ideal platform for a first approach to the problem. However, it should be noted that it is rich in irony and sarcasm and that understanding a particular feeling depends on the whole strip and not just on a particular frame. The style of humor from Quino makes it necessary to have a combined visual and textual understanding of several frames to be clear about the character's feelings.
The project aims to develop technology associated with identifying complex sentiments, personalized offers, tools for preprocessing and processing of images associated with the Deep Networks studied, and accessible databases that allow the community in the area to continue the development.
External Link


Work with Us

We are always interested in new challenges and looking for new motivated people to work with.

If you are interested in colaborated with us, please contact us.

About Us

GITIA was founded on 2008 in order to promote and apply recent research works on advanced technologies to different of domains. Our main application areas cover collaborative work, decision-helping tools and distributed problem solving.