This repository contains the dataset published with the ECCV 2022 paper "BRACE: The Breakdancing Competition Dataset for Dance Motion Synthesis".
- A new dataset for audio-conditioned dance motion synthesis
- Focuses on breakdancing sequences
- Contains high quality annotations for complex body poses and dance movements
Property | Value |
---|---|
Frames | 334,538 |
Manually annotated frames | 26,676 (8%) |
Duration | 3h 32m |
Dancers | 64 |
Videos | 81 |
Sequences | 465 |
Segments | 1,352 |
Avg. segments per sequence | 2.91 |
Avg. sequence duration | 27.48s |
Avg. segment duration | 9.45s |
https://github.com/dmoltisanti/brace/releases/download/latest/dataset.zip
@InProceedings{moltisanti22brace,
author = "Moltisanti, Davide and Wu, Jinyi and Bo, Dai and Loy, Chen Change",
title = "BRACE: The Breakdancing Competition Dataset for Dance Motion Synthesis",
booktitle = "European Conference on Computer Vision (ECCV)",
year = "2022"
}