Image-filter generation system, image-filter generation device, learning device, and learning method

An image filter generation device for generating image filters usable for image processing to be performed before performance of OCR on article image data that is data of an image of an article captured by an imaging component, the OCR being optical character recognition, the image filter generation device comprising:

• training data acquiring circuitry to acquire training data including (i) the article image data acquired in advance, (ii) image-filter-related data indicating a combination of a plurality of image filters used for image processing of the article image data and a value of a parameter for each of the plurality of image filters, and (iii) OCR score data indicating a score of character recognition output through the OCR when image processing is performed on the article image data using the image filters based on the image-filter-related data;
• trained model generating circuitry to generate a trained model indicating a relationship between the article image data, the image-filter-related data, and the OCR score data through machine learning using the training data;
• and
• article image data acquiring circuitry to acquire the article image data as a target of the OCR;
• inference result data generating circuitry to generate, by inputting the article image data as the target of the OCR into the trained model, inference result data indicating a combination of a plurality of image filters usable for image processing of the article image data as the target of the OCR and a value of a parameter for each of the plurality of image filters;
• image filter generating circuitry to generate the image filters based on the inference result data;
• and
• image filter outputting circuitry to output the image filters.

A learning device for learning a relationship between (i) image filters usable for image processing to be performed before performance of OCR on article image data that is data of an image of an article captured by an imaging component and (ii) the article image data, the OCR being optical character recognition, the learning device comprising:

• training data acquiring circuitry to acquire training data including (i) the article image data acquired in advance, (ii) image-filter-related data indicating a combination of a plurality of image filters used for image processing of the article image data and a value of a parameter for each of the plurality of image filters, and (iii) OCR score data indicating a score of character recognition output through the OCR when image processing is performed on the article image data using the image filters based on the image-filter-related data;
• trained model generating circuitry to generate a trained model indicating a relationship between the article image data, the image-filter-related data, and the OCR score data through machine learning using the training data;
• and
• trained model outputting circuitry to output the trained model.

A learning method for learning a relationship between (i) image filters usable for image processing to be performed before performance of OCR on article image data that is data of an image of an article captured by an imaging component and (ii) the article image data, the OCR being optical character recognition, the learning method comprising:

• acquiring, by a computer, training data including (i) the article image data acquired in advance, (ii) image-filter-related data indicating a combination of a plurality of image filters used for image processing of the article image data and a value of a parameter for each of the plurality of image filters, and (iii) OCR score data indicating a score of character recognition output through the OCR when image processing is performed on the article image data using the image filters based on the image-filter-related data;
• generating, by the computer, a trained model indicating a relationship between the article image data, the image-filter-related data, and the OCR score data through machine learning using the training data;
• and
• outputting the trained model by the computer.