Research on Chess Text Recognition Based on ResNet
The main difficulties in recognizing chess characters are font differences and the rotation angle of the characters,among which the characters with rotation angles are the difficulties in recognition.This article constructs a deep learning network based on the ResNet principle.In order to improve the accuracy of text recognition,images are grayscale processed and the mean and variance of the training dataset are calculated to normalize image data.The research results show that compared to the form of color images,the form of grayscale images and standardized image data shows better performance in the recognition of chess text,improving recognition accuracy by 6.7 percentage points,verifying the effectiveness of the custom ResNet deep learning network in chess text recognition.
万方数据
维普
