Research on the Design of a Computer Interaction System for Zhangzhou Bag Puppet Theater Based on a Condition-Based Diffusion Model and OpenCV

This paper focuses on the digital innovation and development of Zhangzhou Bag Puppet Theater, a non-material cultural heritage, and conducts an in-depth study and application of cutting-edge computer technology to achieve breakthroughs in its interactive design. By adopting a conditional diffusion model and combining the VGG network to extract style features, we carefully designed and trained the model to achieve high-quality digital generation of Zhangzhou Bag Puppet Theater characters, preserving traditional styles while showcasing rich diversity. A real-time interactive system is constructed using OpenCV and MediaPipe. Leveraging OpenCV’s powerful computer vision capabilities and MediaPipe’s efficient machine learning deployment capabilities, a multi-module collaborative architecture is employed to achieve high-precision recognition of 21 key points on the human hand. Combined with smooth interpolation algorithms and component decomposition dynamic deformation techniques, this system accurately maps human movements to digital puppet performances. This research has established a new interactive design innovation method for Zhangzhou puppet theater, constructing a computer interaction system that integrates computer vision technology with intangible cultural heritage.