Accelerating Arrays of Linear Classifiers Using Approximate Range Queries

Abstract

Modern object detection methods apply binary linear classifiers on Euclidean feature vectors. This paper shows that projecting feature vectors onto a hypersphere allows an approximate range query to accelerate these detectors within acceptable levels of accuracy. The expense of constructing the k-d tree used by these range queries is justified when many detectors are used. We demonstrate our acceleration technique on several existing detection systems, including a state of the art logo detector, and show that approximate range queries can detect logos at least half as well at 11× the speed of the full fidelity method.

Cite

Text

Lu et al. "Accelerating Arrays of Linear Classifiers Using Approximate Range Queries." IEEE/CVF Winter Conference on Applications of Computer Vision, 2014. doi:10.1109/WACV.2014.6836092

Markdown

[Lu et al. "Accelerating Arrays of Linear Classifiers Using Approximate Range Queries." IEEE/CVF Winter Conference on Applications of Computer Vision, 2014.](https://mlanthology.org/wacv/2014/lu2014wacv-accelerating/) doi:10.1109/WACV.2014.6836092

BibTeX

@inproceedings{lu2014wacv-accelerating,
  title     = {{Accelerating Arrays of Linear Classifiers Using Approximate Range Queries}},
  author    = {Lu, Victor and Endres, Ian and Stroila, Matei and Hart, John C.},
  booktitle = {IEEE/CVF Winter Conference on Applications of Computer Vision},
  year      = {2014},
  pages     = {255-260},
  doi       = {10.1109/WACV.2014.6836092},
  url       = {https://mlanthology.org/wacv/2014/lu2014wacv-accelerating/}
}