A document layout analysis model from PaddleOCR, detecting regions like text, table, and figure in documents.