R. Rosales, "Recognition of human action using moment-based features", Boston University Computer Science Technical Report BU 98-020, Nov. 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Mixture Densities for Video Objects Recognition - Hammoud, Mohr (2000) (Correct)
....its experimental behavior. Finally, sec. 6 discusses the performance of the proposed approach and section 7 gives some conclusions and perspectives. 2. Related work Mixture of Gaussian distribution is becoming more popular in the vision community. For the problem of motion recognition, Rosales [9] evaluates the performance of different classification approaches, K nearest neighbor, Gaussian, and Gaussian mixture, using a view based approach for motion representation. According to the results of his experiments on eight human actions, a mixture of Gaussians could be a good model for the ....
R. Rosales. Recognition of human action using momentbased features. Technical Report Report BU 98-020, Boston University Computer Science, Boston, MA 02215, November 1998.
Probabilistic Hierarchical Framework for Clustering Tracked.. - Hammoud, Mohr (2000) (Correct)
....by the input data (tracked objects; color distributions) and some related criterion related to mixture density estimation (number of Gaussian components is not fixed) Also, the classification technique we use is the AHC. Modeling with Gaussian mixture is now becoming very popular. Rosales [Ros98] McKenna et al. MGR98] and Hammoud at al. HM00a] use the Gaussian mixture model to recognize human actions, face colors and non rigid moving objects in videos, respectively. Then, they use the Gaussian mixture classifier to identify the appropriate class of new entities (action, face or ....
R. Rosales. Recognition of human action using moment-based features. Technical Report Report BU 98-020, Boston University Computer Science, Boston, MA 02215, November 1998.
3D Trajectory Recovery for Tracking Multiple Objects and.. - Rosales, Sclaroff (1999) Self-citation (Rosales) (Correct)
....best number of parameters to be used via MDL principle: MDL : arg max Theta i ;k (log P (OEj Theta i ) Gamma k 2 log n) 10) where k is the number of parameters in the model, and n is the number of samples in the training data. For further details about the model estimation module, see [25]. 5.1 Trajectory Guided Recognition In theory it is necessary to learn representations of every action from all possible directions. We denote P j to be the set of PDF s used to represent m actions under direction j. Each action i has its own PDF, P ( Theta (j) i jOE k ) 2 P j . Acquiring ....
R. Rosales. Recognition of human action using momentbased features. BU CS TR 98-020, 1998.
Trajectory Guided Tracking and Recognition of Actions - Rosales, Sclaroff (1999) (2 citations) Self-citation (Rosales) (Correct)
....from the sequence. The moving object is resampled in a canonical frame throughout the tracking sequence, despite changes in scale and position. The estimated bounding box is used to resize and resample the moving blob into a canonical view that can be used as input to motion recognition modules [38, 26, 42, 14, 45, 5], shape modeling [3, 12] or, if resolution allows, more detailed human tracking. The resulting translation scale stabilized images of the object are then fed to an action recognition module. Actions are represented in terms of motion energy images (MEI s) and motion history images (MHI s) 8, ....
.... vector: x f = x MEI ; xMHI ) The dimension of x f is reduced via principal components analysis (PCA) 18] The main motivation for this reduction is that this creates a more compact representation that improves distribution estimation accuracy, and statistically reduces empty space problems [45]. The reduction is achieved by solving the well known eigenvalue decomposition problem: Phi T Sigma Phi; 18) where Phi is the eigenvector matrix of the covariance of the data ( Sigma) and is the corresponding diagonal matrix of eigenvalues. Only M eigenvectors are kept corresponding to ....
[Article contains additional citation context not shown here]
R. Rosales. Recognition of human action using moment-based features. Technical Report BU 98-020, Boston University, 1998.
Video Analysis of Human Dynamics - a Survey - Wang, Singh (Correct)
No context found.
R. Rosales, "Recognition of human action using moment-based features", Boston University Computer Science Technical Report BU 98-020, Nov. 1998.
Video Analysis of Human Dynamics - a Survey - Wang, Singh (Correct)
No context found.
R. Rosales, "Recognition of human action using moment-based features", Boston University Computer Science Technical Report BU 98-020, Nov. 1998.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC