A Study on the Visual Construction and Pathways for the Transmission of Cultural Memory in the Yao Long Drum Dance, Empowered by Digital Humanities

Abstract

In 2008, the Yao Long Drum Dance was included in the second batch of the National Intangible Cultural Heritage List, bringing to an end its long and tortuous history of marginalization and marking its transition from a regional dance to a national one. This study constructed a Yao Long Drum Dance dataset. To address the issue of self-occlusion of skeletal points during data acquisition, existing algorithms for restoring motion information in self-occlusion scenarios were improved, thereby recovering the position information of occluded skeletal points. Skeletal data of professional Yao Long Drum Dance performers were captured and stored using a Kinect, and a 3D CNN was designed for the recognition of typical movements in the Yao Long Drum Dance. Test results demonstrate that the dance motion capture method proposed in this paper provides significant support for dance motion acquisition and training. The 3D CNNs achieved a recognition accuracy of up to 96.45%, thereby facilitating the digital preservation and transmission of the Yao Long Drum Dance.