Shape Detection with Nearest Neighbour Contour Fragments
Kasim Terzić, Hussein Adnan Mohammed and J.M.H. du Buf
Abstract
We present a novel method for shape detection in natural scenes based on incomplete contour fragments and nearest neighbour search. In contrast to popular methods based on sliding windows, chamfer matching and SVMs, we characterise each contour fragment using a local descriptor and perform a fast nearest-neighbour search to find similar fragments in the training set. Based on this idea, we show how to learn robust object models from training images, generate reliable object hypotheses, and verify them. Despite its simplicity and speed, our method produces good detection results on the challenging ETHZ dataset.
Session
Poster 1
Files
Extended Abstract (PDF, 288K)
Paper (PDF, 2M)
DOI
10.5244/C.29.59
https://dx.doi.org/10.5244/C.29.59
Citation
Kasim Terzić, Hussein Adnan Mohammed and J.M.H. du Buf. Shape Detection with Nearest Neighbour Contour Fragments. In Xianghua Xie, Mark W. Jones, and Gary K. L. Tam, editors, Proceedings of the British Machine Vision Conference (BMVC), pages 59.1-59.12. BMVA Press, September 2015.
Bibtex
@inproceedings{BMVC2015_59,
title={Shape Detection with Nearest Neighbour Contour Fragments},
author={Kasim Terzić and Hussein Adnan Mohammed and J.M.H. du Buf},
year={2015},
month={September},
pages={59.1-59.12},
articleno={59},
numpages={12},
booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
publisher={BMVA Press},
editor={Xianghua Xie, Mark W. Jones, and Gary K. L. Tam},
doi={10.5244/C.29.59},
isbn={1-901725-53-7},
url={https://dx.doi.org/10.5244/C.29.59}
}