Attention mechanism text recognition method based on deep learning

Abstract

The invention discloses an attention mechanism text recognition method based on deep learning, wherein the method comprises the following steps: S1, acquiring a model training data set; S2, preprocessing the model training data set; S3, constructing a model convolution layer; inputting the model training data set for feature extraction to obtain a feature map of an image, outputting to a subsequent recurrent neural network structure, wherein the visual features of the method are extracted through a multilayer convolutional neural network in which a convolutional layer and a maximum pool layerare staggered, the CNN accepts raw input, a feature grid V with the size of D * H * W is generated, wherein D represents the number of channels, and H and W represent the height and width of the result feature map. The method overcomes the problem of low text recognition accuracy of a neural OCR technology using a CTC-based model, can significantly reduce the amount of calculation of the network,and can ensure that the prediction precision of the model to the formula is not greatly reduced.

Description

technical field [0001] The present invention relates to the technical field of deep learning text recognition, in particular to a deep learning-based attention mechanism text recognition method. Background technique [0002] In the era of very large information, PDF and pictures account for a large part of the information, which has caused a large number of users' demand for picture and PDF text recognition, that is, optical character recognition (OCR, most commonly used to recognize natural language in images) , including various languages, handwriting, numbers, etc. Among them, there will be special signs in a large number of academic related texts, such as mathematical formulas, which are more complicated than text recognition. The recognition of mathematical formulas has become a special recognition field, and there are many difficulties. We use a real-world The formula recognition of the deep learning attention mechanism of the data set where the rendered mathematical ...

AI Technical Summary

Problems solved by technology

[0003] The traditional method is to use the neural OCR technology based on the CTC model. The RNN used in the encoding and decoding structure can handle certain short-term dependencies, but it cannot handle long-term dependencies, because when the sequence is long, the gradient at the back of the sequence is very large. It is difficult to backpropagate to the previous sequence. Similarly, RNN may also have a gradient explosion problem. Its model is relatively accurate for text recognition that is more complex (such as complex formula symbols)

Images