This paper describes a discriminatively trained, multi-scale, deformable
part model for object detection. Our sys- tem achieves a two-fold
improvement in average precision over the best performance in the
2006 PASCAL person detection challenge. It also outperforms the best
results in the 2007 challenge in ten out of twenty categories. The
system relies heavily on deformable parts. While deformable part
models have become quite popular, their value had not been demonstrated
on difficult benchmarks such as the PASCAL challenge. Our system
also relies heavily on new methods for discriminative training. We
combine a margin-sensitive approach for data mining hard negative
examples with a formalism we call latent SVM. A latent SVM, like
a hidden CRF, leads to a non-convex training problem. However, a
latent SVM is semi-convex and the training problem becomes convex
once latent information is specified for the positive examples. We
believe that our training methods will eventually make possible the
effective use of more latent information such as hierarchical (grammar)
models and models involving latent three dimensional pose.
%0 Journal Article
%1 citeulike:2873027
%A Felzenszwalb, P.
%A Mcallester, D.
%A Ramanan, D.
%D 2008
%J Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference
on
%K detection, object
%T A Discriminatively Trained, Multiscale, Deformable Part Model
%U http://www.ics.uci.edu/~dramanan/papers/latent.pdf
%X This paper describes a discriminatively trained, multi-scale, deformable
part model for object detection. Our sys- tem achieves a two-fold
improvement in average precision over the best performance in the
2006 PASCAL person detection challenge. It also outperforms the best
results in the 2007 challenge in ten out of twenty categories. The
system relies heavily on deformable parts. While deformable part
models have become quite popular, their value had not been demonstrated
on difficult benchmarks such as the PASCAL challenge. Our system
also relies heavily on new methods for discriminative training. We
combine a margin-sensitive approach for data mining hard negative
examples with a formalism we call latent SVM. A latent SVM, like
a hidden CRF, leads to a non-convex training problem. However, a
latent SVM is semi-convex and the training problem becomes convex
once latent information is specified for the positive examples. We
believe that our training methods will eventually make possible the
effective use of more latent information such as hierarchical (grammar)
models and models involving latent three dimensional pose.
@article{citeulike:2873027,
abstract = {This paper describes a discriminatively trained, multi-scale, deformable
part model for object detection. Our sys- tem achieves a two-fold
improvement in average precision over the best performance in the
2006 PASCAL person detection challenge. It also outperforms the best
results in the 2007 challenge in ten out of twenty categories. The
system relies heavily on deformable parts. While deformable part
models have become quite popular, their value had not been demonstrated
on difficult benchmarks such as the PASCAL challenge. Our system
also relies heavily on new methods for discriminative training. We
combine a margin-sensitive approach for data mining hard negative
examples with a formalism we call latent SVM. A latent SVM, like
a hidden CRF, leads to a non-convex training problem. However, a
latent SVM is semi-convex and the training problem becomes convex
once latent information is specified for the positive examples. We
believe that our training methods will eventually make possible the
effective use of more latent information such as hierarchical (grammar)
models and models involving latent three dimensional pose.},
added-at = {2009-08-24T22:23:57.000+0200},
author = {Felzenszwalb, P. and Mcallester, D. and Ramanan, D.},
biburl = {https://www.bibsonomy.org/bibtex/2395cacb1926c5b2fc29e3ec752fd74fa/sitrke},
citeulike-article-id = {2873027},
citeulike-linkout-0 = {http://www.ics.uci.edu/~dramanan/papers/latent.pdf},
description = {Human Detection},
file = {:paperpool\\deformpart.pdf:PDF},
interhash = {57e54252471b238614495f3ad644edaa},
intrahash = {395cacb1926c5b2fc29e3ec752fd74fa},
journal = {Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference
on},
keywords = {detection, object},
month = {June},
posted-at = {2008-06-08 00:42:17},
priority = {0},
timestamp = {2009-08-24T22:23:57.000+0200},
title = {A Discriminatively Trained, Multiscale, Deformable Part Model},
url = {http://www.ics.uci.edu/~dramanan/papers/latent.pdf},
year = 2008
}