Title :
Efficient human action recognition by cascaded linear classifcation
Author :
Roth, Peter M. ; Mauthner, Thomas ; Khan, Inayatullah ; Bischof, Horst
Author_Institution :
Inst. for Comput. Graphics & Vision, Graz Univ. of Technol., Graz, Austria
fDate :
Sept. 27 2009-Oct. 4 2009
Abstract :
We present a human action recognition system suitable for very short sequences. In particular, we estimate Histograms of Oriented Gradients (HOGs) for the current frame as well as the corresponding dense flow field estimated from two frames. The thus obtained descriptors are then efficiently represented by the coefficients of a Nonnegative Matrix Factorization (NMF). To further speed up the overall process, we apply an efficient cascaded Linear Discriminant Analysis (CLDA) classifier. In the experimental results we show the benefits of the proposed approach on standard benchmark datasets as well as on more challenging and realistic videos. In addition, since other state-of-the-art methods apply weighting between different cues, we provide a detailed analysis of the importance of weighting for action recognition and show that weighting is not necessarily required for the given task.
Keywords :
image classification; matrix decomposition; video signal processing; cascaded linear classification; cascaded linear discriminant analysis classifier; efficient human action recognition; nonnegative matrix factorization; oriented gradients histogram; Biomedical optical imaging; Detectors; Histograms; Humans; Image motion analysis; Linear discriminant analysis; Optical filters; Optical sensors; Pipelines; Shape;
Conference_Titel :
Computer Vision Workshops (ICCV Workshops), 2009 IEEE 12th International Conference on
Conference_Location :
Kyoto
Print_ISBN :
978-1-4244-4442-7
Electronic_ISBN :
978-1-4244-4441-0
DOI :
10.1109/ICCVW.2009.5457655