Awesome
Dataset
Introduction
This dataset is build to create relationship between dance motion and the music, which contains four types of dance, Cha-cha, Tango, Rumba and Waltz.
In each directory, audio.mp3
gives the audio files of the dance. skeletons.json
describes the skeleton points of the dance, the config.json
gives the start frames and the end frames (The dance sequence match part of the songs). The FPS of the dataset is 25 frames per second.
Paper
This dataset is build along with the ACM-Multimedia regular paper & demo paper:
@inproceedings{tang2018dance,
title={Dance with Melody: An LSTM-autoencoder Approach to Music-oriented Dance Synthesis},
author={Tang, Taoran and Jia, Jia and Mao, Hanyang},
booktitle={2018 ACM Multimedia Conference on Multimedia Conference},
pages={1598--1606},
year={2018},
organization={ACM}
}
@inproceedings{tang2018anidance,
title={AniDance: Real-Time Dance Motion Synthesize to the Song},
author={Tang, Taoran and Mao, Hanyang and Jia, Jia},
booktitle={2018 ACM Multimedia Conference on Multimedia Conference},
pages={1237--1239},
year={2018},
organization={ACM}
}
Notes
Let me know if something is wrong with the dataset.
Issues posted in this repository will not be noticed. If you have any question, please email me, or create an issue in https://github.com/mhy12345/Music-to-Dance-Motion-Synthesis.