Improving Evolution-COnstructed Features Using Speciation for General Object Detection
Abstract
Object recognition is a well studied but extremely challenging field. Evolution COnstructed (ECO) features have been shown to be effective for general object recognition while at the same time self-tuning itself to the target object without the need of a human expert. ECO features use simulated evolution to build series of transforms that are used for object discrimination. We improve on the successful ECO features algorithm by employing speciation during evolution to create more diverse and effective ECO features. Speciation allows candidate solutions during evolution to compete within niches rather than against a large population. On the INRIA person dataset we show a 5% increase in accuracy at 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-4</sup> false positive rate.
Cite
Text
Lillywhite et al. "Improving Evolution-COnstructed Features Using Speciation for General Object Detection." IEEE/CVF Winter Conference on Applications of Computer Vision, 2012. doi:10.1109/WACV.2012.6163019Markdown
[Lillywhite et al. "Improving Evolution-COnstructed Features Using Speciation for General Object Detection." IEEE/CVF Winter Conference on Applications of Computer Vision, 2012.](https://mlanthology.org/wacv/2012/lillywhite2012wacv-improving/) doi:10.1109/WACV.2012.6163019BibTeX
@inproceedings{lillywhite2012wacv-improving,
title = {{Improving Evolution-COnstructed Features Using Speciation for General Object Detection}},
author = {Lillywhite, Kirt D. and Lee, Dah-Jye and Tippetts, Beau J.},
booktitle = {IEEE/CVF Winter Conference on Applications of Computer Vision},
year = {2012},
pages = {441-446},
doi = {10.1109/WACV.2012.6163019},
url = {https://mlanthology.org/wacv/2012/lillywhite2012wacv-improving/}
}