Character recognition method and device, terminal and computer storage medium

Abstract

The invention discloses a character recognition method and device, a terminal and a computer storage medium thereof, and the method comprises the steps: carrying out the feature extraction of an input image through a convolutional neural network, inputting the features an attention mechanism module with a plurality of channels, obtaining the attention weight of each channel, scaling each channel of the depth feature map to obtain a plurality of attention feature maps, then inputting the attention feature maps into a full-connection layer for feature fusion to obtain a character category prediction result, designing a loss function according to character category labels of input pictures and the character category prediction result during model training, and optimizing attention weights, therefore, the character recognition accuracy is improved, and the recognition robustness of difficult samples is higher.

Description

technical field [0001] The embodiments of the present application relate to the technical field of computer vision, and more specifically, to a character recognition method, device, terminal and computer storage medium thereof. Background technique [0002] Handwritten Chinese Character Recognition (HCCR) has always been a very active and challenging research direction in the field of computer vision. It has been studied since the 1960s and has made great progress. Many real-life applications are closely related to it. , such as mail sorting, bank check reading, book and handwritten note transcription, and more. Although many studies have been done, the recognition of handwritten Chinese characters is still a very challenging task. On the one hand, due to the large number of Chinese People have huge differences in writing styles, so that even with the same type of characters, the visual differences are still obvious, which brings great difficulties to the recognition of han...

AI Technical Summary

Problems solved by technology

Although many studies have been done, the recognition of handwritten Chinese characters is still a very challenging task. On the one hand, due to the large number of Chinese People have huge differences in writing styles, so even with the same type of characters, the visual differences are still obvious, which brings great difficulties to the recognition of handwritten Chinese characters

[0003] Most of the existing deep learning-based methods utilize convolutional neural networks to classify handwritten Chinese characters by learning global semantic features from the whole image, but this is not enough for the recognition of visually similar characters because of the confusing There are often only minor differences between characters

Specifically, the global attention provided by these methods can well locate the entire character, but the attention regions between different classes of characters have a large overlap and lack of discrimination, which may lead to similar characters and large differences within the class. The recognition error rate is high

Images