News regarding the AIST Dance Video Database (AIST Dance DB), including planned maintenance and important updates. Please note that the schedule may be subject to change.
- Maintenance Period (Tue, Nov 26, 2024 12:00 to Fri, Dec 13, 2024 13:00): System downtime is scheduled due to essential power equipment maintenance. Access will be unavailable during this period.
- Network Line Construction (Sat, Dec 7, 2024 00:00 to 03:00): Expect brief communication interruptions (up to two 15-minute outages) due to network upgrades. This may affect connectivity between ABCI and external networks.
- ABCI 2.0/bb Termination (Tue, Dec 31, 2024): The ABCI 2.0/bb service will be discontinued. Users are advised to migrate any necessary data to alternative storage solutions before this date. Further details can be found here.
- ABCI 2.0 Cloud Storage Termination (Sun, Mar 30, 2025 17:00): The ABCI 2.0 Cloud Storage service will be terminated. Users requiring data stored here must migrate it to a different storage location. More information is available here.
Overview of the AIST Dance Video Database
The AIST Dance Video Database (AIST Dance DB) is a unique and valuable shared resource offering original street dance videos paired with copyright-cleared dance music. As the first large-scale database specifically curated for street dance videos, it is designed to significantly advance academic research in the burgeoning field of Dance Information Processing. This comprehensive dance video database is poised to stimulate innovation and facilitate a wide array of research tasks, including:
- Dance-motion genre classification
- Dancer identification within dance videos
- Dance-technique estimation from video footage
| Key Features of this Dance Video Database |
|---|
| Diverse Dance Genres: Encompassing ten major street dance genres, providing a broad spectrum of dance styles for analysis. |
| Multiple Dancers: Featuring solo and group performances by 40 professional dancers, enabling research across varied performance dynamics. |
| Multi-Camera Perspectives: Captured with up to nine video cameras surrounding each dancer, offering rich, multi-view dance video data for in-depth analysis. |
Getting Started with the Dance Video Database for Your Research
Before utilizing the database, please carefully review the Terms of Use to understand the conditions and disclaimers associated with its use. Detailed information regarding the database structure is available in the Database Structure documentation.
Researchers can access and download the database in several ways to suit specific research needs. The complete database (515 Gigabytes) can be downloaded via Getting the database. For users requiring a specific subset of data, the Filtered Search option allows for targeted data retrieval. Furthermore, the original dance music tracks used in the database are accessible separately through Get Only Music Data.
Alongside the database, we have introduced a novel benchmark task: dance-motion genre classification. Subset data specifically designed for this task is available under Tasks, providing a starting point for research in this area.
When utilizing the AIST Dance DB in your research, we kindly request that you cite the following publication to acknowledge its creators and contribute to its visibility within the academic community.
Citation:
Shuhei Tsuchida, Satoru Fukayama, Masahiro Hamasaki and Masataka Goto. AIST Dance Video Database: Multi-genre, Multi-dancer, and Multi-camera Database for Dance Information Processing. In Proceedings of the 20th International Society for Music Information Retrieval Conference (ISMIR 2019), 2019.
http://archives.ismir.net/ismir2019/paper/000060.pdf
@inproceedings{aist-dance-db, author = {Shuhei Tsuchida and Satoru Fukayama and Masahiro Hamasaki and Masataka Goto}, title = {AIST Dance Video Database: Multi-genre, Multi-dancer, and Multi-camera Database for Dance Information Processing}, booktitle = {Proceedings of the 20th International Society for Music Information Retrieval Conference, {ISMIR} 2019}, address = {Delft, Netherlands}, year = 2019, month = nov }Acknowledgments
The AIST Dance DB project received support from JST ACCEL Grant Number JPMJAC1602, Japan. ANOMALY INC. played a crucial role in music production and dance performance recording. Takahiro Inoue is responsible for managing the database server. Fabrizio Pedersoli and Hirohito Tanaka contributed to the refinement of the videos within the database. The database logo was thoughtfully designed by Misato Nakamura.
