An information identification method, device, equipment and readable storage medium

CN116246276BActive Publication Date: 2026-07-07ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
Filing Date
2022-12-22
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing technologies struggle to quickly and effectively extract service provider compliance information from images, particularly in identifying the correspondence between structured information and words within the images, leading to inefficient compliance audits.

Method used

By generating target images of target word pairs as training samples, the recognition model learns the correspondence between target words, identifies characters in the image and their coordinate information, and trains the model to minimize the predicted correspondence and label differences, thereby improving the accuracy and efficiency of extracting structured information from images.

Benefits of technology

It enables the rapid and efficient extraction of structured information from images, improving the efficiency and accuracy of compliance audits, especially in identifying the correspondence between words in images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116246276B_ABST
    Figure CN116246276B_ABST
Patent Text Reader

Abstract

The specification discloses an information recognition method, device, equipment and readable storage medium. A target image is generated as a training sample according to at least a target word pair formed by target words having a corresponding relationship, and a corresponding relationship between the target words in the target word pair is taken as a first label of the training sample. By inputting each character contained in the target words and coordinate information of each character in the target image into a recognition model, a predicted corresponding relationship between each undetermined word is obtained, and a difference between the predicted corresponding relationship and the first label is minimized as a training target to train the recognition model. It can be seen that the target image is generated as the training sample based on the target word pair, which solves the problem of insufficient training samples. The target words used to generate the training sample have a corresponding relationship, so that the recognition model can output a corresponding relationship between words contained in an image, improve the efficiency of extracting structured information from the image, and improve the security of private information.
Need to check novelty before this filing date? Find Prior Art