Fast non-parametric action recognition

• Autores: Ubalde, S.; Goussies, N.A.
• Año de publicación: 2012
• Idioma: inglés
• Tipo de recurso: artículo
• Fuente: Lect. Notes Comput. Sci. 2012;7441 LNCS:268-275
• Materia: action recognition; image-to-class distance; nearest neighbor; Action recognition; Average running time; Classification performance; Data sets; image-to-class distance; Nearest neighbors; Non-parametric; Real-world problem; Training data; Image analysis; Computer vision
• Institución: Universidad Nacional de Buenos Aires. Facultad de Ciencias Exactas y Naturales

In this work we propose a method for action recognition which needs no intensive learning stage, and achieves state-of-the-art classification performance. Our work is based on a method presented in the context of image classification. Unlike that method, our approach is well-suited for working with large real-world problems, thanks to an efficient organization of the training data. We show results on the KTH and IXMAS datasets. On the challenging IXMAS dataset, the average running time is reduced by 50% when using our method. © 2012 Springer-Verlag.