Shizuoka University
Okabe Lab
日本語

Cartoon reader support system with character face search
and background information presentation functions

Nakamura Toya

Shizuoka University

Makoto Okabe

Shizuoka University

研究のTeaser画像

Image showing the processing flow of the system. From left to right, the system detects character faces from the image, selects the target character, inputs the selection of identification results and the number of volumes read, and finally outputs background information.

Abstract

We have developed a system with character face search and background information presentation functions to assist comic book readers. The proposed system detects and identifies characters in images and provides background information based on the identification results. We used yolov8_animeface for face detection, FaceNet for identification, and GPT-4 for background information generation. Furthermore, the number of volumes read was taken into account to prevent spoilers, and a revision process of the generated information was introduced to improve reliability. The intuitive web interface makes it easy for anyone to use the system.

Paper

Bachelor Thesis

IPSJ SIG-CGVI, 197th conference

Video

BibTeX citation

    @Misc{nakmura.toya.21_bachelor,
        author =    {Nakamura Toya},
        title =     {Cartoon reader support system with character face search and background information presentation functions},
        howpublished = {Bachelor Thesis},
        month =     2,
        year =      2025,
    }