Context Forest for Object Class Detection

Davide Modolo, Alexander Vezhnevets and Vittorio Ferrari

Abstract

We present Context Forest (ConF) - a technique for predicting properties of the objects in an image based on its global appearance. Compared to standard nearest-neighbour techniques, ConF is more accurate, fast and memory efficient. We demonstrate ConF by predicting three properties: aspects of appearance, location in the image, and class memebership. In extensive experiments we show that (i) ConF can automatically select which components of a multi-component detector to run on a given test image, obtaining a considerable speed-up for detectors trained from large sets (10x for EE-SVMs [36] and 2x for DPM [21]; (ii) ConF can improve object detection performance by removing false positive detections at unlikely locations (+2% mAP), and by (iii) removing false positives produced by classes unlikely to be present in the image (+5% mAP on a 200-class dataset [2]).

Session

Detection and Recognition

Files

PDF iconExtended Abstract (PDF, 2M)
PDF iconPaper (PDF, 4M)

DOI

10.5244/C.29.188
https://dx.doi.org/10.5244/C.29.188

Citation

Davide Modolo, Alexander Vezhnevets and Vittorio Ferrari. Context Forest for Object Class Detection. In Xianghua Xie, Mark W. Jones, and Gary K. L. Tam, editors, Proceedings of the British Machine Vision Conference (BMVC), pages 188.1-188.14. BMVA Press, September 2015.

Bibtex

@inproceedings{BMVC2015_188,
	title={Context Forest for Object Class Detection},
	author={Davide Modolo and Alexander Vezhnevets and Vittorio Ferrari},
	year={2015},
	month={September},
	pages={188.1-188.14},
	articleno={188},
	numpages={14},
	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.29.188},
	isbn={1-901725-53-7},
	url={https://dx.doi.org/10.5244/C.29.188}
}