Title: Continuous pattern detection and recognition in stream - a benchmark for online gesture recognition

Authors: Nehla Ghouaiel; Pierre-François Marteau; Marc Dupont

Addresses: IRISA, Université Bretagne Sud, Campus de Tohannic, Vannes, France ' IRISA, Université Bretagne Sud, Campus de Tohannic, Vannes, France ' IRISA, Université Bretagne Sud, Campus de Tohannic, Vannes, France; Thales Optronique, 2 Avenue Gay Lussac, Elancourt, France

Abstract: There exist few benchmarks for assessing pattern detection and recognition in streams in general and for gesture processing in particular. We propose a dedicated benchmark based on the construction of isolated gestures (patterns) and gesture sequences datasets. This benchmark is associated with a general assessment methodology for streaming processing which first consists in labelling the stream according to some heuristics (that can be optimised on training data) and then aligning the ground truth labelling with the predicted one. Six pattern recognition models (including DTW, KDTW, HMM, HCRF and SVM) have been accordingly evaluated using this benchmark. It turns out that the regularised kernelised version of DTW measure (KDTW) associated to a SVM is quite efficient, comparatively to the other models, for detecting and recognising continuous gestures in streams.

Keywords: stream processing; pattern detection; pattern recognition; gesture recognition; motion capture; MOCAP.

DOI: 10.1504/IJAPR.2017.085315

International Journal of Applied Pattern Recognition, 2017 Vol.4 No.2, pp.146 - 160

Received: 14 Oct 2016
Accepted: 20 Dec 2016
Published online: 21 Jul 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article