Cyclecan-based answer sheet filling mode conversion research

The invention relates to the field of image style conversion, and discloses a research on answer sheet filling mode conversion based on cyclecan. In order to solve the problem, the invention provides a research on the conversion of the answer sheet filling mode based on the cyclecan, and a generator is composed of a residual network and is used for generating an image; the discriminator is composed of a convolutional neural network and is used for distinguishing true and false data. According to the method, a self-made data set is selected as a data source, the data set is divided according to the proportion of 8: 2, a self-made answer sheet picture set is selected as the data source, the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) are adopted as evaluation indexes, and the answer sheet filling mode can be effectively converted. 本发明涉及图像风格转换领域，公开了一种基于cyclegan的答题卡填涂方式变换的研究。本文针对此问题，提出了一种基于cyclegan的答题卡填涂方式变换的研究，生成器是由残差网络构成，其作用在于生成图像；判别器由卷积神经网络构成，作用在于区分真假数据。本文选取自制数据集作为数据源，并按照8：2比例来划分数据集，