Research Project
Grant-in-Aid for Young Scientists (B)
Partly Locality Sensitive Hashing is proposed to efficiently extract frequent patterns from motion capture data or video so as to train recognizers automatically. The 3D shape and articulation of unknown objects found in frequent patterns are also recovered based on the variant of SFM methods. As a result, manipulation tasks can be trained and recognized from video in an unsupervised way.
All 2011 2010 2009
All Journal Article (4 results) (of which Peer Reviewed: 2 results) Presentation (20 results)
電子情報通信学会論文誌 Vol.J94-D, No.3
Pages: 593-603
110008460418
日本ロボット学会誌 Vol.29, No.1
Pages: 67-76
電子情報通信学会論文誌
Volume: Vol.J94-D, No.3 Pages: 593-603
日本ロボット学会誌
Volume: Vol.29, No. 1 Pages: 548-551
10027648570