In this work, a novel framework for the automatic evaluation of dance performances is presented. The proposed approach is based on acquiring the motion of a dance performer via Kinect-based human skeleton tracking. Then, the evaluation of his/her performance is achieved against a gold-standard performance of the teacher. Therefore, skeleton tracking is accompanied by a set of appropriate quaternionic vector-signal processing methodologies for temporally aligning (synchronizing) and comparing “dance motion signals”. The proposed quaternionic approach is approximately invariant to rigid transformations, as proved and demonstrated. The presented experimental results using the Huawei/3DLife/EMC2 dataset are promising and verify the effectiveness of the proposed method.
Digitally capturing unique skills involved in European Traditional Sports and Games
3D Living Interactions through Visual Environments
D. Alexiadis, P. Daras, “Quaternionic signal processing techniques for automatic evaluation of dance performances from MoCap data“, IEEE Transactions on Multimedia, Vol: 16, Issue: 5, Aug. 2014
A. Liutkus, A. Drémeau, D. Alexiadis, S. Essid, P. Daras, “Analysis of dance movements using Gaussian processes“, ACM Multimedia 2012, Oct. 29 – Nov. 2012, Nara, Japan
S. Essid, D. Alexiadis, R. Tournemenne, M. Gowing, P. Kelly, D. Monaghan, P. Daras, A. Drιmeau, N. O’connor, “An Advanced Virtual Dance Performance Evaluator“, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2012), Kyoto, Japan, 25-30 March 2012
D. Alexiadis, P. Kelly, P. Daras, N. O’ Connor, T. Boubekeur, M. Ben Moussa, “Evaluating a Dancer’s Performance using Kinect-based Skeleton Tracking“, ACM Multimedia 2011, Nov 28 – Dec 1, 2011, Scottsdale, Arizona, USA