Text recognition model training method and device, text recognition method and device and equipment

Abstract

The invention provides a text recognition model training method and device, a text recognition method and device, and equipment. The training method comprises the following steps: constructing an initial model; taking the output of the first text image data through the recurrent neural network as the character string input of the word embedding module, taking a first feature map obtained after the first text image data passes through a second convolutional neural network of the second part as the other input of the word embedding module, and training the initial model to obtain a convergent initial model; and obtaining a text recognition model based on the converged initial model; the initial model comprises a first part used for recognizing text content of the image, and provided with a first convolutional neural network and a recurrent neural network and a second part used for judging whether the given text is in the given image or not, and provided with a second convolutional neural network and a word embedding module; the device is used for executing the method. By means of the training method, the text recognition model which is high in recognition speed and higher in recognition precision can be obtained.

Description

technical field [0001] The present invention relates to text recognition technology, in particular to a text recognition model training method, text recognition method, device and equipment. Background technique [0002] Text detection and recognition have a wide range of applications and are the pre-steps of many computer vision tasks, such as image search, identity authentication, and visual navigation. The main purpose of text detection is to locate text lines or characters in the image, while text recognition is Transcribing an image with text lines into a string (recognizing its content), precise positioning and accurate recognition of text is both very important and challenging, because compared to general object detection and recognition, text has multiple directions, irregular shapes, Extreme aspect ratios, fonts, colors, backgrounds and other characteristics, therefore, algorithms that are often more successful in general object detection and recognition cannot be d...

AI Technical Summary

Problems solved by technology

[0002] Text detection and recognition have a wide range of applications, and are the pre-steps of many computer vision tasks, such as image search, identity authentication, and visual navigation. The main purpose of text detection is to locate the position of text lines or characters in the image, while text recognition is Transcribing an image with text lines into a string (recognizing its content), precise positioning and accurate recognition of text is both very important and challenging, because compared to general object detection and recognition, text has multiple directions, irregular shapes, Extreme aspect ratio, fonts, colors, backgrounds and other characteristics, therefore, algorithms that are often more successful in general object detection and recognition cannot be directly migrated to text detection

[0003] The recognition effect of existing text recognition models and methods is affected by many factors, and it is difficult to achieve both recognition speed and recognition accuracy, which cannot meet the needs of the rapid development of computer vision tasks.

Images