Home >

news Help

Publication Information


Title
Japanese:暗黙的なニューラル表現を用いた可変長な人間動作の生成 
English:Implicit neural representations for variable length human motion generation 
Author
Japanese: CERVANTES BAQUE Pablo Alberto, Yusuke Sekikawa, 佐藤 育郎, 篠田 浩一.  
English: Pablo Cervantes, Yusuke Sekikawa, Ikuro Sato, Koichi Shinoda.  
Language English 
Journal/Book name
Japanese: 
English: 
Volume, Number, Page        
Published date Sept. 2023 
Publisher
Japanese: 
English: 
Conference name
Japanese:第22回情報科学技術フォーラム（FIT2023） 
English: 
Conference site
Japanese:大阪府堺市 
English: 
Official URL https://www.ipsj.or.jp/event/fit/fit2023/
https://onsite.gakkai-web.net/fit2023/abstract/data/html/event/event_TCS7-3.html
 
Abstract We propose an action-conditional human motion generation method using variational implicit neural representations (INR). The variational formalism enables action-conditional distributions of INRs, from which one can easily sample representations to generate novel human motion sequences. Our method offers variable-length sequence generation by construction because a part of INR is optimized for a whole sequence of arbitrary length with temporal embeddings. In contrast, previous works reported difficulties with modeling variable-length sequences. We confirm that our method with a Transformer decoder outperforms all relevant methods on HumanAct12, NTU-RGBD, and UESTC datasets in terms of realism and diversity of generated motions. Surprisingly, even our method with an MLP decoder consistently outperforms the state-of-the-art Transformer-based auto-encoder. In particular, we show that variable-length motions generated by our method are better than fixed-length motions generated by the state-of-the-art method in terms of realism and diversity.

©2007 Institute of Science Tokyo All rights reserved.