US patent granted to Smart Engines to improve ID document scan speed, accuracy

Smart Engines has patented a method for determining when an analysis by an artificial intelligence algorithm should be discontinued.

This task is important for applications performing text recognition in a video stream, according to the announcement. These kinds of processes will become faster and more accurate thanks to the now-patented technique, Smart Engines says.

One of the images submitted as part of the patent for “Approximate Modeling of Next Combined Result for Stopping Text-Field Recognition in a Video Stream” depicts a person presenting an ID card to a smartphone, and a screenshot of the data from the ID document as captured by optical character recognition (OCR).

The approach is based on predicting the result of analyzing the next frame, and whether the predicted contribution would improve the result of the process. If the predicted next frame would not improve the result, the process ends.

“Since the company’s inception, we have been committed to the principle of Science First and working at the forefront of science while solving business problems. We have received five patents in the USA this year alone, but we have no intention of stopping. We have a lot of other interesting technologies. As soon as we receive the patents, we will tell you about them,” says Smart Engines CEO Vladimir Arlazarov, who holds a PhD in Computer Science.

Smart Engines now holds six U.S. patents in total, including protection for a weighted integration approach that is already included in its document scanning SDK, which was revealed in June.

Article Topics

biometrics  |  digital ID  |  document verification  |  OCR  |  patents  |  Smart Engines