Current computational approaches to learning visual object categories
require thousands of training images, are slow, cannot learn in an
incremental manner and cannot incorporate prior information into
the learning process. In addition, no algorithm presented in the
literature has been tested on more than a handful of object categories.
We present an method for learning object categories from just a few
training images. It is quick and it uses prior information in a principled
way. We test it on a dataset composed of images of objects belonging
to 101 widely varied categories. Our proposed method is based on
making use of prior information, assembled from (unrelated) object
categories which were previously learnt. A generative probabilistic
model is used, which represents the shape and appearance of a constellation
of features belonging to the object. The parameters of the model
are learnt incrementally in a Bayesian manner. Our incremental algorithm
is compared experimentally to an earlier batch Bayesian algorithm,
as well as to one based on maximum likelihood. The incremental and
batch versions have comparable classification performance on small
training sets, but incremental learning is significantly faster,
making real-time learning feasible. Both Bayesian methods outperform
maximum likelihood on small training sets.
%0 Journal Article
%1 Fei-Fei2007
%A Fei-Fei, Li
%A Fergus, Rob
%A Perona, Pietro
%B Special issue on Generative Model Based Vision
%D 2007
%J Computer Vision and Image Understanding
%K Bayesian Categorization, Generative Incremental Object learning, model model, recognition,
%N 1
%P 59--70
%T Learning generative visual models from few training examples: An
incremental Bayesian approach tested on 101 object categories
%U http://www.sciencedirect.com/science/article/B6WCX-4N74JC7-1/2/528a670ebe68673ca7ffa04cf3552b2d
%V 106
%X Current computational approaches to learning visual object categories
require thousands of training images, are slow, cannot learn in an
incremental manner and cannot incorporate prior information into
the learning process. In addition, no algorithm presented in the
literature has been tested on more than a handful of object categories.
We present an method for learning object categories from just a few
training images. It is quick and it uses prior information in a principled
way. We test it on a dataset composed of images of objects belonging
to 101 widely varied categories. Our proposed method is based on
making use of prior information, assembled from (unrelated) object
categories which were previously learnt. A generative probabilistic
model is used, which represents the shape and appearance of a constellation
of features belonging to the object. The parameters of the model
are learnt incrementally in a Bayesian manner. Our incremental algorithm
is compared experimentally to an earlier batch Bayesian algorithm,
as well as to one based on maximum likelihood. The incremental and
batch versions have comparable classification performance on small
training sets, but incremental learning is significantly faster,
making real-time learning feasible. Both Bayesian methods outperform
maximum likelihood on small training sets.
@article{Fei-Fei2007,
abstract = {Current computational approaches to learning visual object categories
require thousands of training images, are slow, cannot learn in an
incremental manner and cannot incorporate prior information into
the learning process. In addition, no algorithm presented in the
literature has been tested on more than a handful of object categories.
We present an method for learning object categories from just a few
training images. It is quick and it uses prior information in a principled
way. We test it on a dataset composed of images of objects belonging
to 101 widely varied categories. Our proposed method is based on
making use of prior information, assembled from (unrelated) object
categories which were previously learnt. A generative probabilistic
model is used, which represents the shape and appearance of a constellation
of features belonging to the object. The parameters of the model
are learnt incrementally in a Bayesian manner. Our incremental algorithm
is compared experimentally to an earlier batch Bayesian algorithm,
as well as to one based on maximum likelihood. The incremental and
batch versions have comparable classification performance on small
training sets, but incremental learning is significantly faster,
making real-time learning feasible. Both Bayesian methods outperform
maximum likelihood on small training sets.},
added-at = {2009-09-12T19:19:34.000+0200},
author = {Fei-Fei, Li and Fergus, Rob and Perona, Pietro},
biburl = {https://www.bibsonomy.org/bibtex/2e640ef96ebc7c74cee23fdfbedc39ac4/mozaher},
booktitle = {Special issue on Generative Model Based Vision},
file = {:Fei-Fei2007.pdf:PDF},
interhash = {f2cb343ee4dcfa8ad74f7df1e4adc800},
intrahash = {e640ef96ebc7c74cee23fdfbedc39ac4},
journal = {Computer Vision and Image Understanding},
keywords = {Bayesian Categorization, Generative Incremental Object learning, model model, recognition,},
month = {April},
number = 1,
owner = {Mozaher},
pages = {59--70},
timestamp = {2009-09-12T19:19:38.000+0200},
title = {Learning generative visual models from few training examples: An
incremental Bayesian approach tested on 101 object categories},
url = {http://www.sciencedirect.com/science/article/B6WCX-4N74JC7-1/2/528a670ebe68673ca7ffa04cf3552b2d},
volume = 106,
year = 2007
}