React Native OCR - v0.11.0
    Preparing search index...

    Type Alias DocumentModelsConfig

    The document-helper model: one fused PTE that carries page-orientation classification, the UVDoc dewarping grid and, optionally, SLANet table structure. The flags pick which stages run by default; a run can override them.

    type DocumentModelsConfig = {
        dewarp?: boolean;
        dewarpMinVarianceRatio?: number;
        modelPath: string;
        orientation?: boolean;
        orientationMinConfidence?: number;
        table?: TableConfig;
    }
    Index
    dewarp? dewarpMinVarianceRatio? modelPath orientation? orientationMinConfidence? table?
    dewarp?: boolean

    Flatten a physically warped page before OCR. Default false.

    dewarpMinVarianceRatio?: number

    Decline the dewarp when it keeps less than this fraction of the source image variance: a bad grid can collapse a page without clear borders to near-blank, which keeps close to none. Default 0.5; lower is more permissive. A page photographed on a dark background keeps only about a third once the background is cropped away, so set about 0.1 for photos.

    modelPath: string

    Local path or remote URL of the fused helper .pte.

    orientation?: boolean

    Rotate the page upright before OCR. Default false.

    orientationMinConfidence?: number

    Apply the rotation only when the classifier is at least this confident. Default 0.85.

    table?: TableConfig

    Table structure recognition; omit to disable it.

    Settings
    Member Visibility

    On This Page
    Properties