FineDance
收藏资源简介:
FineDance是由清华大学深圳国际研究生院创建的大型专业3D动作捕捉舞蹈数据集,包含14.6小时的音乐与舞蹈配对数据,涵盖22种精细分类的舞蹈风格,具有精确的姿势和精细的手部动作。该数据集是目前最大的音乐与舞蹈配对数据集,旨在推动AI编舞、运动先验和全身重建方法的发展。FineDance数据集通过专业的舞蹈家和动作捕捉系统创建,确保了数据的真实性和准确性。该数据集的应用领域包括从音乐生成舞蹈、舞蹈风格匹配以及长时舞蹈序列的生成,旨在解决现有数据集在手部动作和舞蹈风格多样性方面的不足。
FineDance is a large-scale professional 3D motion-captured dance dataset created by the Graduate School at Shenzhen, Tsinghua University. It contains 14.6 hours of paired music and dance data, covers 22 finely categorized dance styles, and features precise poses and delicate hand movements. As the largest paired music and dance dataset to date, it aims to advance the development of AI choreography, motion prior and full-body reconstruction methods. The FineDance dataset is constructed using professional dancers and motion capture systems, ensuring the authenticity and accuracy of the data. Its application scenarios include music-to-dance generation, dance style matching and long-duration dance sequence generation, which is designed to address the shortcomings of existing datasets in terms of hand movements and diversity of dance styles.

- 1FineDance: A Fine-grained Choreography Dataset for 3D Full Body Dance Generation清华大学深圳国际研究生院 · 2023年



