Fast unsupervised greedy learning of multiple objects and parts from video

Michalis K Titsias; Christopher K.I. Williams

doi:10.1109/CVPR.2004.72

Fast unsupervised greedy learning of multiple objects and parts from video

Michalis K Titsias
, Christopher K.I. Williams

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Williams and Titsias (2004) have shown how to carry out unsupervised greedy learning of multiple objects from images (GLOMO), building on the work of Jojic and Frey (2001). In this paper we show that the earlier work on GLOMO can be greatly speeded up for video sequence data by carrying out approximate tracking of the multiple objects in the scene. Our method is applied to raw image sequence data and extracts the objects one at a time. First, the moving background is learned, and moving objects are found at later stages. The algorithm recursively updates an appearance model of the tracked object so that possible occlusion of the object is taken into account which makes tracking stable. We apply this method to learn multiple objects in image sequences as well as articulated parts of the human body.

Original language	English
Title of host publication	Computer Vision and Pattern Recognition Workshop, 2004. CVPRW'04. Conference on
Publisher	Institute of Electrical and Electronics Engineers
Pages	179-179
Number of pages	1
DOIs	https://doi.org/10.1109/CVPR.2004.72
Publication status	Published - 2004

Access to Document

10.1109/CVPR.2004.72

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=1384979

Cite this

@inproceedings{77e54b5366994fd2959f5953e5245ff5,

title = "Fast unsupervised greedy learning of multiple objects and parts from video",

abstract = "Williams and Titsias (2004) have shown how to carry out unsupervised greedy learning of multiple objects from images (GLOMO), building on the work of Jojic and Frey (2001). In this paper we show that the earlier work on GLOMO can be greatly speeded up for video sequence data by carrying out approximate tracking of the multiple objects in the scene. Our method is applied to raw image sequence data and extracts the objects one at a time. First, the moving background is learned, and moving objects are found at later stages. The algorithm recursively updates an appearance model of the tracked object so that possible occlusion of the object is taken into account which makes tracking stable. We apply this method to learn multiple objects in image sequences as well as articulated parts of the human body.",

author = "Titsias, \{Michalis K\} and Williams, \{Christopher K.I.\}",

year = "2004",

doi = "10.1109/CVPR.2004.72",

language = "English",

pages = "179--179",

booktitle = "Computer Vision and Pattern Recognition Workshop, 2004. CVPRW'04. Conference on",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

}

TY - GEN

T1 - Fast unsupervised greedy learning of multiple objects and parts from video

AU - Titsias, Michalis K

AU - Williams, Christopher K.I.

PY - 2004

Y1 - 2004

N2 - Williams and Titsias (2004) have shown how to carry out unsupervised greedy learning of multiple objects from images (GLOMO), building on the work of Jojic and Frey (2001). In this paper we show that the earlier work on GLOMO can be greatly speeded up for video sequence data by carrying out approximate tracking of the multiple objects in the scene. Our method is applied to raw image sequence data and extracts the objects one at a time. First, the moving background is learned, and moving objects are found at later stages. The algorithm recursively updates an appearance model of the tracked object so that possible occlusion of the object is taken into account which makes tracking stable. We apply this method to learn multiple objects in image sequences as well as articulated parts of the human body.

AB - Williams and Titsias (2004) have shown how to carry out unsupervised greedy learning of multiple objects from images (GLOMO), building on the work of Jojic and Frey (2001). In this paper we show that the earlier work on GLOMO can be greatly speeded up for video sequence data by carrying out approximate tracking of the multiple objects in the scene. Our method is applied to raw image sequence data and extracts the objects one at a time. First, the moving background is learned, and moving objects are found at later stages. The algorithm recursively updates an appearance model of the tracked object so that possible occlusion of the object is taken into account which makes tracking stable. We apply this method to learn multiple objects in image sequences as well as articulated parts of the human body.

U2 - 10.1109/CVPR.2004.72

DO - 10.1109/CVPR.2004.72

M3 - Conference contribution

SP - 179

EP - 179

BT - Computer Vision and Pattern Recognition Workshop, 2004. CVPRW'04. Conference on

PB - Institute of Electrical and Electronics Engineers

ER -

Fast unsupervised greedy learning of multiple objects and parts from video

Abstract

Access to Document

Fingerprint

Cite this