Most of text based documents, such as forms, cards, business cars, cheques, books, etc. have four sides. Area of extraction can be detected within the image even though it is shown from various angles, and the result will be flattened and be further analyzed.

The most important thing to improve the accuracy and recognition of OCR is to know where the sentences and text are located in the document. Through artificial neural network, it is possible to detect accurate position of the text or sentences within the image.

The fonts that are written in a document vary greatly depending on the country and language used. The most commonly used fonts are already prelearned for extraction, and additional learning of fonts for different purposes are also possible.

Supports wide range of image quality

Able to process wide range of original document, picture, and scan quality. An image pre-processing algorithm optimized for text identification and extraction is performed to provide text recognition.

Lightweight Module and Fast Processing Speed

The lightweight module and fast processing speed are useful to extract text from large volumes of documents and based on this technology it can be applied and utilized in mobile applications.

Document Text Recognition

ID verification and
non face-to-face identity verification

Automatically sort business cards,
ID type classification