Neural Network-based Optical Character Recognition

A system comprising:

• one or more processors;
• and
• a non-transitory computer readable medium storing a plurality of instructions, which when executed, cause the one or more processors to:
• generating a set of bounding boxes based on reshaped image data portions that correspond to image data of a first source image;
• merging, according to one or more human judgement heuristics, any intersecting bounding boxes into a merged bounding box to generate a first set of merged bounding boxes indicative of first image data portions that likely portray one or more words in the first source image;
• executing a comparison between the one or more of the merged bounding boxes that correspond to the first source image and one or more ground truth merged bounding boxes;
• determining a quality of the one or more human judgement heuristics based on generating a matching score resulting from the comparison;
• and
• determining an update to the one or more human judgement heuristics.

A computer program product comprising a non-transitory computer-readable medium having a computer-readable program code embodied therein to be executed by one or more processors, the program code including instructions for:

• generating a set of bounding boxes based on reshaped image data portions that correspond to image data of a first source image;
• merging, according to one or more human judgement heuristics, any intersecting bounding boxes into a merged bounding box to generate a first set of merged bounding boxes indicative of first image data portions that likely portray one or more words in the first source image;
• executing a comparison between the one or more of the merged bounding boxes that correspond to the first source image and one or more ground truth merged bounding boxes;
• determining a quality of the one or more human judgement heuristics based on generating a matching score resulting from the comparison;
• and
• determining an update to the one or more human judgement heuristics.