Presentation Information
[3Yin-A-24]Consideration of automatic graphic recording system
〇Haruhito Zenio1, Koki Saito1, Makoto Enomoto1, Takayuki Hoshino1, Eiji Takahashi1 (1. BIPROGY Inc.)
Keywords:
Graphic Recording,Generative AI,Automation
Using generative AI as an approach to approximate the quality of graphic records produced by human graphic recorders is one feasible option. Recently, with the substantial performance improvements observed in large language models and image generation AI, the feasibility of automated graphic recording has increased significantly. In this study, we define the Listen-Think-Draw process in graphic recording for the purpose of automation using generative AI, and investigate the processing required at each stage as well as the concrete system architecture. We compare the graphic records generated as the final outputs of each processing stage with those drawn by a human graphic recorder. Based on this comparison, we present a preliminary evaluation and discuss the challenges at each stage, thereby reporting findings that provide useful insights toward the realization of an automatic graphic recording system.