The PAGE frame links automatic reading of the plates and human verification. Three studies cover 3,929 American comics and 87 volumes of Japanese manga.
PAGE divides the scans into boxes and text, associates them with bibliographic information and then applies a vision-language model guided by an annotation grid. Samples are compared with human annotations to control the analyses.
The authors study the representations of political power, the visual form of the pages and the distribution of speech between characters. Their corpus includes American comics from 1938 to 1956 and Japanese manga. They also envisage a search by visual content and better access to archives for blind or visually impaired people; these uses are perspectives.
Preprint filed on arXiv on October 9, 2026; results are those reported by the authors and remain to be independently confirmed.
Read the full article on the original source
Go to the source →