CBCS - Comic Book Cover Synopsis: Generating Synopsis of a Comic Book with Unsupervised Abstractive Dialogue

Abstract

Document analysis which aims at understanding the semantics of the document is an active field of research in today’s world. Growth of graphical novels such as comics is magnificent, by incorporating digitized and digital-born comics. This enables a machine intelligence to recognize the key elements present in a comic book story. Analyzing comic story book is complex as they contains text drawings, balloons embedded in a comic page and these elements are considered as key components of a comic The transition in digital era has made the people to have a different approach in reading out the books through comics and with the world moving towards automation, the proposed work incorporated deep learning approaches to analyze the key components contained in comics. This work aimed to design a model to automatically learn to extract the comic specific components and further identification of text has been done using pre-trained language recognition models. The main objective of the proposed work is to have quick and crisp understanding of a comic story. Hence, the proposed work designed to generate the unsupervised abstractive dialogues representing the whole story without losing the essence of it. The proposed work has attained an improved performance over traditional method for generating the abstractive summarized story and state-of-the-art method for comic component detection.