Fine-grained Choreography Dance Dataset
收藏资源简介:
Fine-grained Choreography Dance Dataset是一个高精度、细粒度的舞蹈编排数据集,包含超过 14.6 小时的数据,由 27 位专业舞者在动作捕捉系统下完成,共采集 346 组音乐—舞蹈配对片段,其中约 7.7 小时数据可通过百度网盘获取。数据集采用标准化的 52 关节骨架表示,完整记录身体与手指在三维空间中的精细运动信息,兼顾真实感与艺术表现力。在内容上,该数据集覆盖传统与现代风格在内的 22 种细粒度舞蹈类型,在专业舞蹈艺术家的指导下进行分类,不仅涵盖嘻哈、中国古典舞等常见风格,还首次引入民间舞动作,是目前舞蹈类型最为丰富的 3D 舞蹈数据集之一。所有动作均通过 Vicon 光学动作捕捉系统采集,并在 MotionBuilder 中统一重定向骨架,确保姿态数据的高精度与一致性;同时,舞蹈与音乐在节奏与风格上高度匹配,为舞蹈生成与音乐驱动建模提供了高质量训练数据。
Fine-grained Choreography Dance Dataset is a high-precision, fine-grained choreography dance dataset containing over 14.6 hours of motion data. It was collected by 27 professional dancers using a motion capture system, with a total of 346 pairs of music-dance clip segments. Approximately 7.7 hours of this data is available via Baidu Netdisk. The dataset adopts a standardized 52-joint skeleton representation, which fully records fine-grained motion information of the entire body and fingers in 3D space, balancing realism and artistic expressiveness. In terms of content, it covers 22 fine-grained dance styles including traditional and modern genres, which were classified under the guidance of professional dance artists. It not only includes common styles such as hip-hop and Chinese classical dance, but also introduces folk dance movements for the first time, making it one of the most diverse 3D dance datasets currently available in terms of dance categories. All motions were captured using the Vicon optical motion capture system, and the skeletons were uniformly retargeted in MotionBuilder to ensure high precision and consistency of pose data. Meanwhile, the dance and music are highly matched in terms of rhythm and style, providing high-quality training data for dance generation and music-driven modeling.




