CORTEXA
← Browse

Jiajun Song

1 paper indexed

arxivcs.CV2026-07-13

MonkeyOCRv2: A Visual-Text Foundation Model for Document AI

Yuliang Liu, Zhang Li, Ziyang Zhang, Shuo Zhang, Qiang Liu, Jiajun Song, et al.

Mainstream visual encoders are pretrained on natural images and cannot be effectively applied to document images without document-oriented adaptation, as dense text and fine-grained character strokes demand character-level visual perception. We present MonkeyOCRv2, a visual-text…

View free PDFSource page