Group Cost-sensitive Boosting with Multi-scale Decorrelated Filters for Pedestrian Detection

Chengju Zhou, Meiqing Wu and SiewKei Lam

Abstract

We propose a novel two-stage pedestrian detection framework that combines multiscale decorrelated filters to extract more discriminative features and a novel group costsensitive boosting algorithm. The proposed boosting algorithm is based on mixture loss to alleviate the influence of annotation errors in training data and explores varying cost for different types of misclassification. Experiments on Caltech and INRIA datasets show that the proposed framework achieves the best detection performance among all state-of-the-art non-deep learning methods.

Session

Posters

Files

Paper (PDF)

DOI

10.5244/C.31.48
https://dx.doi.org/10.5244/C.31.48

Citation

Bibtex

            @inproceedings{BMVC2017_48,
                title={Group Cost-sensitive Boosting with Multi-scale Decorrelated Filters for Pedestrian Detection},
                author={Chengju Zhou, Meiqing Wu and SiewKei Lam},
                year={2017},
                month={September},
                pages={48.1-48.12},
                articleno={48},
                numpages={12},
                booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
                publisher={BMVA Press},
                editor={Tae-Kyun Kim, Stefanos Zafeiriou, Gabriel Brostow and Krystian Mikolajczyk},
                doi={10.5244/C.31.48},
                isbn={1-901725-60-X},
                url={https://dx.doi.org/10.5244/C.31.48}
            }