Library Structure

        graph LR
    manuscript[manuscript]

    manuscript --> Pipeline[Pipeline]
    manuscript --> data[data]
    manuscript --> detectors[detectors]
    manuscript --> layouts[layouts]
    manuscript --> recognizers[recognizers]
    manuscript --> correctors[correctors]
    manuscript --> utils[utils]
    manuscript --> api[api]

    Pipeline --> p1[predict]
    Pipeline --> p2[get_text]
    Pipeline --> p3[last_detection_page]
    Pipeline --> p4[last_layout_page]
    Pipeline --> p5[last_recognition_page]
    Pipeline --> p6[last_correction_page]

    detectors --> EAST[EAST]
    detectors --> YOLO[YOLO]
    layouts --> SimpleSorting[SimpleSorting]
    recognizers --> TRBA[TRBA]
    correctors --> CharLM[CharLM]

    utils --> organize_page[organize_page wrapper]
    

Overview

• Pipeline orchestrates OCR stages.

• detectors provide detection models such as YOLO and EAST.

• layouts provide ordering/grouping models (for example SimpleSorting).

• recognizers provide text recognition models.

• correctors provide text post-processing models.

• utils.organize_page remains as a compatibility wrapper over layout logic.