A Method for Traditional Shadow Play Figure’s Head Regeneration Based on Generative Adversarial Network

The field of artistic style transfer and creation has introduced more and more intelligent methods. We apply the technology of generative adversarial network to sculpt creation for traditional shadow play figure’s head, and suggest a recreation method based on generative adversarial network for traditional shadow play figure’s head. Firstly, we collect and make data sets of side face of real people and tradition shadow play figure’s head; Secondly, we train a CycleGAN model using former data sets and obtain transformed results from side face to shadow play figure’s head. Lastly, we analyze experiment results, expose existing problems and present future work. Our application method provides thoughts for developing the art of shadow play with scientific approaches.