Text recognition method, text recognition model training method and device

Abstract

This application relates to a text recognition method, a text recognition model training method, and an apparatus. The text recognition method includes: inputting an image to be recognized into at least two predetermined text recognition models to obtain corresponding first recognition results and first confidence levels; when the first recognition results corresponding to the at least two text recognition models are inconsistent, determining a second confidence level for each of the text recognition models based on the first confidence level, a first evaluation index, and a second evaluation index; wherein the first evaluation index and the second evaluation index are determined based on the evaluation values ​​of the recognition results of each of the text recognition models on at least two evaluation dimensions for the image to be recognized and preset sample images; and determining a second recognition result for the image to be recognized based on each second confidence level. This solves the problem of low recognition accuracy of existing text recognition models for real-world images in related technologies.

Images