DetectAreasMode

DetectAreasMode enumeration

Determines the type of neural network used for areas detection.

Member name	Description
NONE	Doesn’t detect paragraphs. Better for a simple one-column document without pictures.
DOCUMENT	Detects paragraphs uses NN model for documents. Better for multicolumn document, document with pictures or with other not text objects.
PHOTO	Detects paragraphs uses NN model for photos. Better for image with a lot of pictures and other not text objects.
COMBINE	Detects paragraphs with text and then uses other NN model to detect areas inside of paragraphs. Better for images with complex structure.
TABLE	Detects cells with text. Preferable mode for images with table structure.
CURVED_TEXT	Detects lines and recognizes text on curved images. Preferred mode for photos of book and magazine pages.
TEXT_IN_WILD	A super-powerful neural network specialized in extracting words from low-quality images such as street photos, license plates, passport photos, meter photos, and photos with noisy backgrounds.