CPC G06V 30/413 (2022.01) [G06N 20/00 (2019.01); G06Q 20/34 (2013.01); G06Q 20/341 (2013.01); G06Q 20/342 (2013.01); G06Q 20/355 (2013.01); G06Q 20/356 (2013.01); G06Q 20/387 (2013.01); G06Q 20/4014 (2013.01); G06V 30/10 (2022.01); G06V 30/224 (2022.01)] | 21 Claims |
1. A computer-implemented method comprising:
generating a card image that represents a card scanned using a mobile device;
identifying a set of characters printed on the card, wherein identifying the set of characters includes applying a machine-learning model to the card image, wherein the machine-learning model is trained using training images that depict previously-scanned cards associated with a plurality of card types, and wherein the machine-learning model enhances identification of characters that were previously limited by a hard coded scheme;
identifying a subset of characters that correspond to an account number associated with a user of the card, wherein identifying the subset of characters includes applying the machine-learning model to the card image to distinguish the subset of characters from the set of characters;
generating a virtual card for the mobile device, wherein the virtual card represents the card, and wherein generating the virtual card includes storing the subset of characters in the mobile device and associating the subset of characters with the virtual card;
performing a particular task by using the virtual card stored in the mobile device;
receiving feedback associated with the subset of characters; and
updating the machine-learning model based on the feedback.
|