3D human pose search using oriented cylinders
Author
Pehlivan, Selen
Duygulu, Pınar
Date
2009-09-10Source Title
IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops 2009
Publisher
IEEE
Pages
16 - 22
Language
English
Type
Conference PaperItem Usage Stats
185
views
views
122
downloads
downloads
Abstract
In this study, we present a representation based on a new 3D search technique for volumetric human poses which is then used to recognize actions in three dimensional video sequences. We generate a set of cylinder like 3D kernels in various sizes and orientations. These kernels are searched over 3D volumes to find high response regions. The distribution of these responses are then used to represent a 3D pose. We use the proposed representation for (i) pose retrieval using Nearest Neighbor (NN) based classification and Support Vector Machine (SVM) based classification methods, and for (ii) action recognition on a set of actions using Dynamic Time Warping (DTW) and Hidden Markov Model (HMM) based classification methods. Evaluations on IXMAS dataset supports the effectiveness of such a robust pose representation. ©2009 IEEE.
Keywords
Action recognitionClassification methods
Data sets
Dynamic time warping
High response
Human pose
Nearest neighbors
Search technique
Video sequences
Computer vision
Cylinders (shapes)
Hidden Markov models
Video recording
Three dimensional