Manga Understanding

Manga understanding research focuses on developing computational methods to analyze and interpret the complex visual and textual information within manga comics. Current efforts concentrate on tasks such as automatic transcription generation (including speaker diarization), benchmarking large multimodal models for various understanding tasks (e.g., emotion recognition, narrative comprehension), and improving image processing techniques like colorization and screentone generation. These advancements are significant for improving accessibility for visually impaired individuals, enabling new forms of manga analysis, and potentially revolutionizing manga creation tools.