by J.R.R. Uijlings, A.W.M. Smeulders, R.J.H. Scha
Abstract:
We start from the state-of-the-art Bag of Words pipeline that in the 2008 benchmarks of TRECvid and PASCAL yielded the best performance scores. We have contributed to that pipeline, which now forms the basis to compare various fast alternatives for all of its components: (i) For descriptor extraction we propose a fast algorithm to densely sample SIFT and SURF, and we compare several variants of these descriptors. (ii) For descriptor projection we compare a k-means visual vocabulary with a Random Forest. As a preprojection step we experiment with PCA on the descriptors to decrease projection time. (iii) For classification we use Support Vector Machines and compare the Chi-square kernel with the RBF kernel. Our results lead to a 10-fold speed increase without any loss of accuracy and to a 30-fold speed increase with 17% loss of accuracy, where the latter system does real-time classification at 26 images per second.
Reference:
J.R.R. Uijlings, A.W.M. Smeulders, R.J.H. Scha, "Real-time Bag-of-Words, Approximately", In CIVR, 2009. (best paper award)
Bibtex Entry:
@INPROCEEDINGS{Uijlings09b,
author = {J.R.R. Uijlings and A.W.M. Smeulders and R.J.H. Scha},
title = {Real-time Bag-of-Words, Approximately},
booktitle = {CIVR},
year = {2009},
abstract = {We start from the state-of-the-art Bag of Words pipeline that in the
2008 benchmarks of TRECvid and PASCAL yielded the best performance
scores. We have contributed to that pipeline, which now forms the
basis to compare various fast alternatives for all of its components:
(i) For descriptor extraction we propose a fast algorithm to densely
sample SIFT and SURF, and we compare several variants of these descriptors.
(ii) For descriptor projection we compare a k-means visual vocabulary
with a Random Forest. As a preprojection step we experiment with
PCA on the descriptors to decrease projection time. (iii) For classification
we use Support Vector Machines and compare the Chi-square kernel
with the RBF kernel. Our results lead to a 10-fold speed increase
without any loss of accuracy and to a 30-fold speed increase with
17% loss of accuracy, where the latter system does real-time classification
at 26 images per second.},
comment = {<b>best paper award</b>},
doi = {10.1145/1646396.1646405},
owner = {jrruijli},
timestamp = {2009.05.20},
url = {http://www.huppelen.nl/publications/realTimeBoW.pdf}
}