计算机工程与应用 ›› 2023, Vol. 59 ›› Issue (5): 281-288.DOI: 10.3778/j.issn.1002-8331.2109-0398

• 工程与应用 • 上一篇    下一篇

基于骨架信息的民族舞蹈典型动作识别

秦晴,王卫星,刘清华,蒙德庆   

  1. 1.贵州大学 机械工程学院,贵阳 550025
    2.贵州大学 现代制造技术教育部重点实验室,贵阳 550025
  • 出版日期:2023-03-01 发布日期:2023-03-01

Classical Ethnic Dance Action Recognition Based on Skeleton Information

QIN Qing, WANG Weixing, LIU Qinghua, MENG Deqing   

  1. 1.School of Mechanical Engineering, Guizhou University, Guiyang 550025, China
    2.Key Laboratory of Advanced Manufacturing Technology, Ministry of Education, Guizhou University, Guiyang 550025, China
  • Online:2023-03-01 Published:2023-03-01

摘要: 民族舞蹈是中国重要的非物质文化遗产,在动作识别技术应用十分广泛的环境下,对于民族舞蹈文化保护和传承的研究领域还比较空白。为了更好地保护和传承少数民族舞蹈,筛选了5类少数民族舞蹈典型动作片段,结合深度摄像机开发了动作数据采集系统,通过该系统采集民族舞蹈动作骨架信息,构建了民族舞蹈典型动作数据集,使用改进的三维卷积神经网络(3D CNNs)模型对民族舞蹈动作骨架序列进行了识别与分类,并且与其他经典动作识别方法进行了对比,该方法在实验中获得了95%的识别率。研究结果表明,该民族舞蹈数据集构建方法合理,识别模型对民族舞蹈分类有良好的性能,对民族舞蹈动作进行了有效的记录和保存,为民族舞蹈数字化保护提供一种新的方式。

关键词: 民族舞蹈, 动作识别, 骨架信息, 三维卷积神经网络(3D CNNs)

Abstract: Chinese ethnic dance is an important intangible cultural heritage. Action recognition technology is widely used in the environment for research, and the protection of the cultural heritage of ethnic dance is still relatively empty. In order to better protect and pass national dance, five types of classical, ethnic dance action clips are screened. The movement data collection system is developed using the depth camera. The skeleton information of ethnic dance actions is obtained using this system, and a dataset of typical ethnic dance actions is constructed. The improved three-dimensional convolutional neural network(3D CNNs) model recognizes and classifies the skeleton sequence of ethnic dance movements and compares it with other classic action recognition methods. The proposed method has achieved an accuracy of 95% in the experiment. The results show that the method used to construct this ethnic dance dataset is reasonable. The recognition model performs well in recognizing ethnic dances, recording and conserving ethnic dance actions, and provides a novel method for ethnic dance digital preservation. The model provides a new way for the digital protection of ethnic dance.

Key words: ethnic dance, action recognition, skeleton information, three-dimensional convolutional neural network(3D CNNs)