AN EVALUATION OF LOCAL IMAGE FEATURES FOR OBJECT CLASS RECOGNITION
Saiful Islam, Andrzej Sluzek
2010
Abstract
The use of local image features (LIF) for object class recognition is becoming increasingly popular. To better understand the suitability and power of existing LIFs for object class recognition, a simple but useful method is proposed in evaluation of such features. We have compared the performance of eight frequently used LIFs by the proposed method on two popular databases. We have used F-measure criterion for this evaluation. It is found that the individual performance of SURF and SIFT features are better than that of the global features on ETH-80* database with considerably lower number of training objects. However, it may not be good enough for more challenging object class recognition problem (e.g. Caltech-101+). The evaluation of LIFs suggests the requirement for further investigation of more complementary LIFs.
DownloadPaper Citation
in Harvard Style
Islam S. and Sluzek A. (2010). AN EVALUATION OF LOCAL IMAGE FEATURES FOR OBJECT CLASS RECOGNITION . In Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 2: VISAPP, (VISIGRAPP 2010) ISBN 978-989-674-029-0, pages 519-523. DOI: 10.5220/0002844805190523
in Bibtex Style
@conference{visapp10,
author={Saiful Islam and Andrzej Sluzek},
title={AN EVALUATION OF LOCAL IMAGE FEATURES FOR OBJECT CLASS RECOGNITION},
booktitle={Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 2: VISAPP, (VISIGRAPP 2010)},
year={2010},
pages={519-523},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002844805190523},
isbn={978-989-674-029-0},
}
in EndNote Style
TY - CONF
JO - Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 2: VISAPP, (VISIGRAPP 2010)
TI - AN EVALUATION OF LOCAL IMAGE FEATURES FOR OBJECT CLASS RECOGNITION
SN - 978-989-674-029-0
AU - Islam S.
AU - Sluzek A.
PY - 2010
SP - 519
EP - 523
DO - 10.5220/0002844805190523