US 12,393,651 B2
Systems and methods for validating and extracting data dynamically from images of identification documents
Shannon Yogerst, New York, NY (US); and Tania Cruz Morales, Washington, DC (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Dec. 15, 2022, as Appl. No. 18/066,514.
Prior Publication US 2024/0202294 A1, Jun. 20, 2024
Int. Cl. H04L 29/00 (2006.01); G06F 21/31 (2013.01); G06V 10/74 (2022.01); G06V 10/82 (2022.01); G06V 30/262 (2022.01)
CPC G06F 21/31 (2013.01) [G06V 10/761 (2022.01); G06V 10/82 (2022.01); G06V 30/274 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A system for dynamic data validation and data extraction from images of identification documents received over computer network communications, the system comprising:
one or more processors; and
one or more non-transitory, computer-readable media comprising instructions that, when executed by the one or more processors, cause operations comprising:
receiving, over a computer network, digital image data comprising a digital image of an identification document;
determining, using an image recognition convolutional neural network model, based on the digital image data, (i) a first image type of the digital image indicating an issuing authority for the identification document and (ii) a first version identifier indicating a design standard for the identification document;
selecting, based on the first image type and the first version identifier, a first ruleset, from an image ruleset database, for processing the digital image data, wherein the first ruleset indicates portions of the digital image expected to include identification data and security features for authenticating the identification document;
determining, based on the first ruleset, a first set of pixel coordinate ranges associated with the portions of the digital image expected to include the identification data and security features;
determining a first authentication status for the digital image indicating that the identification document was not authenticated based on the identification data and the security features being detected outside a first portion of the digital image corresponding to the first set of pixel coordinate ranges;
based on the first authentication status indicating that the identification document was detected outside the first portion of the digital image, determining a second portion of the digital image corresponding to a second set of pixel coordinate ranges within which the identification data and the security features were detected by the image recognition convolutional neural network model;
based on a comparison of the first portion and the second portion, determining a second ruleset for processing the digital image data comprising a third set of pixel coordinate ranges corresponding to an addition of the first set of pixel coordinate ranges and the second set of pixel coordinate ranges;
updating the image ruleset database to include the second ruleset;
determining, based on the second ruleset, a third portion of the digital image corresponding to the third set of pixel coordinate ranges;
determining a second authentication status for the digital image indicating that the identification document has been authenticated based on the identification data and the security features being detected within the third portion; and
generating, for display on a user interface, an indication of the second authentication status for the digital image.