Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features

Ankit Verma, Ramya Hebbalaguppe, Lovekesh Vig, Swagat Kumar, Ehtesham Hassan

Research output: Chapter in Book/Report/Conference proceedingChapter

11 Citations (Scopus)

Abstract

In this paper, we propose a two stage pedestrian detector. The first stage involves a cascade of Aggregated Channel Features (ACF) to extract potential pedestrian windows from an image. We further introduce a thresholding technique on the ACF confidence scores that segregates candidate windows lying at the extremes of the ACF score distribution. The windows with ACF scores in between the upper and lower bounds are passed on to a Mixture of Expert (MoE) CNNs for more refined classification in the second stage. Results show that the designed detector yields better than state-of-the-art performance on the INRIA benchmark dataset and yields a miss rate of 10.35% at FPPI=0.1.
Original languageEnglish
Title of host publicationProceedings of the IEEE International Conference on Computer Vision
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages555-563
Number of pages9
ISBN (Print)9781467383905
DOIs
Publication statusPublished - 11 Feb 2015

Publication series

NameProceedings of the IEEE International Conference on Computer Vision
Volume2015-February

Fingerprint

Detectors

Keywords

  • Convolutional codes
  • Deformable models
  • Detectors
  • Feature extraction
  • Neural networks
  • Support vector machines
  • Training

Cite this

Verma, A., Hebbalaguppe, R., Vig, L., Kumar, S., & Hassan, E. (2015). Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features. In Proceedings of the IEEE International Conference on Computer Vision (pp. 555-563). (Proceedings of the IEEE International Conference on Computer Vision; Vol. 2015-February). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCVW.2015.78
Verma, Ankit ; Hebbalaguppe, Ramya ; Vig, Lovekesh ; Kumar, Swagat ; Hassan, Ehtesham. / Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features. Proceedings of the IEEE International Conference on Computer Vision. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 555-563 (Proceedings of the IEEE International Conference on Computer Vision).
@inbook{9a37ee3cec49450d8e46127976d49e04,
title = "Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features",
abstract = "In this paper, we propose a two stage pedestrian detector. The first stage involves a cascade of Aggregated Channel Features (ACF) to extract potential pedestrian windows from an image. We further introduce a thresholding technique on the ACF confidence scores that segregates candidate windows lying at the extremes of the ACF score distribution. The windows with ACF scores in between the upper and lower bounds are passed on to a Mixture of Expert (MoE) CNNs for more refined classification in the second stage. Results show that the designed detector yields better than state-of-the-art performance on the INRIA benchmark dataset and yields a miss rate of 10.35{\%} at FPPI=0.1.",
keywords = "Convolutional codes, Deformable models, Detectors, Feature extraction, Neural networks, Support vector machines, Training",
author = "Ankit Verma and Ramya Hebbalaguppe and Lovekesh Vig and Swagat Kumar and Ehtesham Hassan",
year = "2015",
month = "2",
day = "11",
doi = "10.1109/ICCVW.2015.78",
language = "English",
isbn = "9781467383905",
series = "Proceedings of the IEEE International Conference on Computer Vision",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "555--563",
booktitle = "Proceedings of the IEEE International Conference on Computer Vision",
address = "United States",

}

Verma, A, Hebbalaguppe, R, Vig, L, Kumar, S & Hassan, E 2015, Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features. in Proceedings of the IEEE International Conference on Computer Vision. Proceedings of the IEEE International Conference on Computer Vision, vol. 2015-February, Institute of Electrical and Electronics Engineers Inc., pp. 555-563. https://doi.org/10.1109/ICCVW.2015.78

Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features. / Verma, Ankit; Hebbalaguppe, Ramya; Vig, Lovekesh; Kumar, Swagat; Hassan, Ehtesham.

Proceedings of the IEEE International Conference on Computer Vision. Institute of Electrical and Electronics Engineers Inc., 2015. p. 555-563 (Proceedings of the IEEE International Conference on Computer Vision; Vol. 2015-February).

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features

AU - Verma, Ankit

AU - Hebbalaguppe, Ramya

AU - Vig, Lovekesh

AU - Kumar, Swagat

AU - Hassan, Ehtesham

PY - 2015/2/11

Y1 - 2015/2/11

N2 - In this paper, we propose a two stage pedestrian detector. The first stage involves a cascade of Aggregated Channel Features (ACF) to extract potential pedestrian windows from an image. We further introduce a thresholding technique on the ACF confidence scores that segregates candidate windows lying at the extremes of the ACF score distribution. The windows with ACF scores in between the upper and lower bounds are passed on to a Mixture of Expert (MoE) CNNs for more refined classification in the second stage. Results show that the designed detector yields better than state-of-the-art performance on the INRIA benchmark dataset and yields a miss rate of 10.35% at FPPI=0.1.

AB - In this paper, we propose a two stage pedestrian detector. The first stage involves a cascade of Aggregated Channel Features (ACF) to extract potential pedestrian windows from an image. We further introduce a thresholding technique on the ACF confidence scores that segregates candidate windows lying at the extremes of the ACF score distribution. The windows with ACF scores in between the upper and lower bounds are passed on to a Mixture of Expert (MoE) CNNs for more refined classification in the second stage. Results show that the designed detector yields better than state-of-the-art performance on the INRIA benchmark dataset and yields a miss rate of 10.35% at FPPI=0.1.

KW - Convolutional codes

KW - Deformable models

KW - Detectors

KW - Feature extraction

KW - Neural networks

KW - Support vector machines

KW - Training

UR - http://www.mendeley.com/research/pedestrian-detection-via-mixture-cnn-experts-thresholded-aggregated-channel-features

U2 - 10.1109/ICCVW.2015.78

DO - 10.1109/ICCVW.2015.78

M3 - Chapter

SN - 9781467383905

T3 - Proceedings of the IEEE International Conference on Computer Vision

SP - 555

EP - 563

BT - Proceedings of the IEEE International Conference on Computer Vision

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Verma A, Hebbalaguppe R, Vig L, Kumar S, Hassan E. Pedestrian Detection via Mixture of CNN Experts and Thresholded Aggregated Channel Features. In Proceedings of the IEEE International Conference on Computer Vision. Institute of Electrical and Electronics Engineers Inc. 2015. p. 555-563. (Proceedings of the IEEE International Conference on Computer Vision). https://doi.org/10.1109/ICCVW.2015.78