Method, apparatus, readable medium and electronic device of key-value matching

A non-transitory_computer-readable medium having a computer program stored thereon, the program, when executed by a processing apparatus, implementing acts comprising:

• obtaining an image to be detected, the image to be detected comprising at least one attribute data and at least one attribute value data;
• and
• inputting the image to be detected into a predetermined key-value matching model, to cause the predetermined key-value matching model to output a matching relationship between the attribute data and the attribute value data;
• wherein the predetermined key-value matching model comprises a semantic segmentation submodel and an image matching submodel coupled with the semantic segmentation submodel;
• wherein the semantic segmentation submodel is configured to obtain a text coverage region, a text center region, an attribute data region, and an attribute value data region in the image to be detected, determine a target attribute data region based on the text coverage region, the text center region, and the attribute data region, and determine a target attribute value data region based on the text coverage region, the text center region, and the attribute value data region;
• and
• wherein the image matching submodel is configured to determine the matching relationship between the attribute data and the attribute value data in the image to be detected based on the target attribute data region and the target attribute value data region.

An electronic device, comprising:

• a storage apparatus having a computer program stored thereon;
• and
• a processing apparatus for executing the computer program in the storage apparatus to implement acts comprising:
• obtaining an image to be detected, the image to be detected comprising at least one attribute data and at least one attribute value data;
• and
• inputting the image to be detected into a predetermined key-value matching model, to cause the predetermined key-value matching model to output a matching relationship between the attribute data and the attribute value data;
• wherein the predetermined key-value matching model comprises a semantic segmentation submodel and an image matching submodel coupled with the semantic segmentation submodel;
• wherein the semantic segmentation submodel is configured to obtain a text coverage region, a text center region, an attribute data region, and an attribute value data region in the image to be detected, determine a target attribute data region based on the text coverage region, the text center region, and the attribute data region, and determine a target attribute value data region based on the text coverage region, the text center region, and the attribute value data region;
• and
• wherein the image matching submodel is configured to determine the matching relationship between the attribute data and the attribute value data in the image to be detected based on the target attribute data region and the target attribute value data region.