Topic models from the text understanding literature have shown promising
results in unsupervised image categorization and object localization.
Categories are treated as topics, and words are formed by vector
quantizing local descriptors of image patches. Limitations of topic
models include their weakness in localizing objects, and the requirement
of a fairly large proportion of words coming from the object. We
present a new approach that employs correspondences between images
to provide information about object configuration, which in turn
enhances the reliability of object localization and categorization.
This approach is efficient, as it requires only a small number of
correspondences. We show improved categorization and localization
performance on real and synthetic data. Moreover, we can push the
limits of topic models when the proportion of words coming from the
object is very low.
%0 Journal Article
%1 Liu2007
%A Liu, D.
%A Chen, Tsuhan
%D 2007
%J Computer Vision, 2007. ICCV 2007. IEEE 11th International Conference
on
%K categorization, descriptor, detection, image local localization, models, object patches, quantisationimage quantization recognition, topic unsupervised vector
%P 1-7
%R 10.1109/ICCV.2007.4408852
%T Unsupervised Image Categorization and Object Localization using Topic
Models and Correspondences between Images
%X Topic models from the text understanding literature have shown promising
results in unsupervised image categorization and object localization.
Categories are treated as topics, and words are formed by vector
quantizing local descriptors of image patches. Limitations of topic
models include their weakness in localizing objects, and the requirement
of a fairly large proportion of words coming from the object. We
present a new approach that employs correspondences between images
to provide information about object configuration, which in turn
enhances the reliability of object localization and categorization.
This approach is efficient, as it requires only a small number of
correspondences. We show improved categorization and localization
performance on real and synthetic data. Moreover, we can push the
limits of topic models when the proportion of words coming from the
object is very low.
@article{Liu2007,
abstract = {Topic models from the text understanding literature have shown promising
results in unsupervised image categorization and object localization.
Categories are treated as topics, and words are formed by vector
quantizing local descriptors of image patches. Limitations of topic
models include their weakness in localizing objects, and the requirement
of a fairly large proportion of words coming from the object. We
present a new approach that employs correspondences between images
to provide information about object configuration, which in turn
enhances the reliability of object localization and categorization.
This approach is efficient, as it requires only a small number of
correspondences. We show improved categorization and localization
performance on real and synthetic data. Moreover, we can push the
limits of topic models when the proportion of words coming from the
object is very low.},
added-at = {2009-09-12T19:19:34.000+0200},
author = {Liu, D. and Chen, Tsuhan},
biburl = {https://www.bibsonomy.org/bibtex/22129092e00b6ef37fd02f07810623139/mozaher},
doi = {10.1109/ICCV.2007.4408852},
file = {:Liu2007.pdf:PDF},
interhash = {5e94570acae31ab007ad3c7ea44986bc},
intrahash = {2129092e00b6ef37fd02f07810623139},
issn = {1550-5499},
journal = {Computer Vision, 2007. ICCV 2007. IEEE 11th International Conference
on},
keywords = {categorization, descriptor, detection, image local localization, models, object patches, quantisationimage quantization recognition, topic unsupervised vector},
month = {Oct.},
owner = {Mozaher},
pages = {1-7},
timestamp = {2009-09-12T19:19:41.000+0200},
title = {Unsupervised Image Categorization and Object Localization using Topic
Models and Correspondences between Images},
year = 2007
}