Research on Motion Capture and Digital Protection of Shadow Puppetry Based on Deep Learning and Machine Vision

This paper proposes a shadow puppetry motion capture and digital protection system based on deep learning and machine vision technology. The system uses a high-precision camera to collect the motion data of shadow puppet performances, combines deep learning algorithms for feature extraction and motion recognition, and accurately captures the complex shadow puppet movement trajectory. According to the characteristics of shadow puppetry movements, a CNN model is designed to efficiently extract motion features, and LSTM is used for timing prediction to improve the accuracy and fluency of motion capture. This paper takes a typical performance clip of shadow puppetry in a certain area as the research object, and verifies the effectiveness of the system through experiments. Simulation results show that the accuracy of motion capture of the system reaches 96.8%, and the timing prediction error is controlled at 2.3%, which is better than the existing methods. At the same time, this paper performs three-dimensional modeling and digital storage of the captured data, realizes the dynamic display and virtual reproduction of shadow puppetry performances, and provides a new technical approach for the protection of cultural heritage. Research shows that the combination of deep learning and machine vision technology can not only improve the accuracy of shadow puppetry motion capture, but also open up new ideas for the digital inheritance and promotion of traditional culture.