Self-learning receipt optical character recognition engine

A system for receipt processing using optical character recognition (OCR), comprising:

• a processor of a computer server;
• an optical character recognition (OCR) receipt template selector executing on the processor and configured to:
• select a vendor from a plurality of vendors by matching a portion of a receipt image to one of a plurality of vendor identifiers corresponding to the plurality of vendors, wherein the receipt image corresponds to a receipt generated by one of the plurality of vendors,
• determine a first location match between the portion within the receipt image and a vendor identifier field of an OCR receipt template associated with the vendor, wherein the OCR receipt template comprises a relationship between a plurality of locations on the receipt and a plurality of pre-defined fields of the receipt,
• verify, based on the first location match, that the OCR receipt template is applicable to perform the OCR of the receipt image;
• extract, prior to processing the receipt image based at least on the OCR receipt template, a store location identifier from the receipt image, wherein the store location identifier identifies a store location where the receipt is generated;
• and
• determine that the vendor operates a store at the store location to confirm that the receipt is generated by the vendor,
• wherein verifying that the OCR receipt template is applicable to perform the OCR of the receipt image is further based on confirming that the receipt is generated by the vendor;
• an OCR module configured to:
• generate, in response to the verifying, a textual content of the receipt by processing the receipt image based at least on the OCR receipt template;
• and
• a repository coupled to the processor and configured to store the OCR receipt template.

A non-transitory computer readable medium storing instructions for receipt processing using optical character recognition (OCR), the instructions when executed by a computer processor performing the method steps of:

• detecting, within a receipt image of a receipt, a pre-determined characteristic of the receipt;
• obtaining an OCR receipt template based on the pre-determined characteristic of the receipt, wherein the OCR receipt template specifies a relationship between a plurality of locations on the receipt and a plurality of pre-defined fields of the receipt,
• wherein the pre-determined characteristic of the receipt identifies none of a plurality of vendors associated with a plurality of OCR receipt templates, and
• wherein obtaining the OCR receipt template based on the pre-determined characteristic of the receipt comprises:
• adding the receipt image into a plurality of receipt images to form a training set, wherein each of the plurality of receipt images comprises the pre-determined characteristic of the receipt, and
• generating the OCR receipt template based on a plurality of user inputs with respect to the plurality of receipt images, wherein the plurality of user inputs identifies the vendor;
• extracting, based on at least the OCR receipt template, a vendor attribute from a portion of the receipt image, wherein the portion of the receipt image corresponds to a pre-defined field of the plurality of pre-defined fields of the receipt;
• verifying, based on matching the vendor attribute to a vendor associated with the OCR receipt template, that the receipt is generated by the vendor and that the OCR receipt template is applicable to perform the OCR of the receipt image;
• and
• generating, in response to the verifying, a textual content of the receipt by processing the receipt image based at least on the OCR receipt template, wherein the textual content comprises a plurality of machine-encoded texts.