Service Scenarios

Content & OCR

Relying on OCR technology, the text content on the picture is intelligently recognized as structured text, which greatly improves the efficiency of information processing. It can be applied to document recognition, bill recognition, custom template recognition, general form text recognition and other scenarios, to recognize text in pictures or copies, and convert them into editable text.

Contact Sales

Application scenarios

1.Small languages multi-scene OCR transliteration

Provide slogans, billboards, menus, commodity packaging, documents and more than 10 kinds of daily scenes, small language image collection and OCR transcription integration services.With the aid of AI, the image content is automatically recognized and converted into text.At present, the difficulties of image acquisition and transcribe in minor languages are as follows: many life scenes are assimilated by English;The writing order of Italian and Portuguese is not fixed and requires human judgment;Small languages contain all kinds of special characters such as A, difficult to transcribe;There are few suppliers of small languages on the market, and the results are uneven.Stardust provides a large number of data resources including bills in small languages, store names, signs, menus, etc., to meet the data needs in various scenarios;The mature multilingual OCR algorithm is applied to automatically identify the position of the lower boundary points of the text to help improve the annotation efficiency and data quality.Standardize the labeling rules, output high-quality structured data, help customers optimize the Spanish language pre-processing algorithm, improve the maturity and accuracy of image and text recognition model.

The words above the door are in English, and in Italy, there are many scenes in English

The position of the lower boundary points should be marked

Niche languages, special fonts

2.OCR transfer of all kinds of bill information

Invoice, fund transfer sheet, bank draft, promissory note, check and other notes information structured extraction.It is more difficult to extract structural information of bills than ordinary OCR transliteration because of the curved font of the seals, unclear ink marks, Chinese and English symbols, overlapping words, chaotic horizontal and vertical typesetting, dim image light and other problems.Stardust adopts polygon OCR algorithm with 99% accuracy to adapt to arc-shaped image recognition scenes.Mature annotation tools help annotators to solve all kinds of "difficult and complicated diseases" with high efficiency. In addition, automatic quality inspection, multi-layer inspection and other layers are checked by the system to output high-quality structured data and give satisfactory answers to customers.

The most outer edge of the seal text, along the arc direction, radian is larger

This is especially true in low-light photos, where the background and the writing at the bottom are easily mixed together

Red and yellow boxes are two different arrangements

3.License plate OCR recognition

License plate information recognition, applied to unmanned, no card vehicle access management, road illegal vehicle capture, remote vehicle certification and other intelligent scenes.Solve the fuzzy problem of license plate information caused by long distance, low pixel, occlusion, dim light and tilted Angle of view, accurately identify license plate number, color and other information, including blue plate, yellow plate, new energy license plate, military license plate, police license plate and embassy license plate, and support the simultaneous recognition of multiple license plates in the same scene.

4.OCR recognition of foreign papers

The mature multilingual OCR recognition algorithm is used to recognize and transfer the title, DOI code, author, abstract and key words of foreign papers.

Professional labeling team

We have set up exclusive data bases in many provinces, and have more than 500 professional annotation teams on the platform. All the annotation personnel on the platform can take up the post only after passing the training and examination, which is highly professional.