To solve the problems of time-consuming

high cost and low efficiency in Chinese paper-cutting design

a new convolutional neural network model paper cutting Unet（PC-Unet）is designed to realize the character paper-cutting design.PC-Unet algorithm uses the LeakyReLU activation function in the encoding stage to enhance the network’s ability to extract weaker characters.multiple hole convolutions are added between the encoding and decoding stages to perform multi-scale extraction and fusion of the features extracted from the encoding segment to obtain a more expressive feature description.To verify the actual design effect of the PC-Unet network model

the article conducts qualitative and quantitative test evaluation on the self-made paper-cut datasets.The algorithm test results can reach 89.8 % average pixel accuracy and 69.5 % cross-to-parallel ratio

compared with the original algorithm has improved by 3.5 % and 5.8 %.The results show that the paper-cut design drawings obtained by this method can achieve relatively ideal effects.