Perspective-aware loss function for crowd density estimation

Bedir Yilmaz, Kok Ven Jyn , Mei Kuan Lim, Siti Norul Huda Sheikh Abdullah

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Estimation errors caused by perspective distortions are a long-standing problem in the domain of crowd counting. In this paper, we propose a novel loss function to allow filters in convolutional neural networks to learn features that are adaptive to the scale and perspective variation of individuals in crowd images. By exploring the crowd count error from regions close to the vanishing point of a perspective distorted image, we are able to penalize under-estimations. This is useful to train a network that is robust against perspective distortion for accurate density estimation. The proposed method is scene-independent and can be applied effectively to crowd scene with a variety of physical layout. Extensive comparative evaluations demonstrate that our proposed method achieves significant improvement over the state-of-the-art approaches on the challenging ShanghaiTech and UCF-QNRF datasets.

Original languageEnglish
Title of host publicationProceedings of the 16th International Conference on Machine Vision Applications, MVA 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9784901122184
DOIs
Publication statusPublished - 1 May 2019
Event16th International Conference on Machine Vision Applications, MVA 2019 - Tokyo, Japan
Duration: 27 May 201931 May 2019

Publication series

NameProceedings of the 16th International Conference on Machine Vision Applications, MVA 2019

Conference

Conference16th International Conference on Machine Vision Applications, MVA 2019
CountryJapan
CityTokyo
Period27/5/1931/5/19

Fingerprint

Error analysis
Neural networks

ASJC Scopus subject areas

  • Computer Science Applications
  • Signal Processing
  • Computer Vision and Pattern Recognition

Cite this

Yilmaz, B., Ven Jyn , K., Lim, M. K., & Sheikh Abdullah, S. N. H. (2019). Perspective-aware loss function for crowd density estimation. In Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019 [8758034] (Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/MVA.2019.8758034

Perspective-aware loss function for crowd density estimation. / Yilmaz, Bedir; Ven Jyn , Kok; Lim, Mei Kuan; Sheikh Abdullah, Siti Norul Huda.

Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019. Institute of Electrical and Electronics Engineers Inc., 2019. 8758034 (Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yilmaz, B, Ven Jyn , K, Lim, MK & Sheikh Abdullah, SNH 2019, Perspective-aware loss function for crowd density estimation. in Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019., 8758034, Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019, Institute of Electrical and Electronics Engineers Inc., 16th International Conference on Machine Vision Applications, MVA 2019, Tokyo, Japan, 27/5/19. https://doi.org/10.23919/MVA.2019.8758034
Yilmaz B, Ven Jyn  K, Lim MK, Sheikh Abdullah SNH. Perspective-aware loss function for crowd density estimation. In Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019. Institute of Electrical and Electronics Engineers Inc. 2019. 8758034. (Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019). https://doi.org/10.23919/MVA.2019.8758034
Yilmaz, Bedir ; Ven Jyn , Kok ; Lim, Mei Kuan ; Sheikh Abdullah, Siti Norul Huda. / Perspective-aware loss function for crowd density estimation. Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019. Institute of Electrical and Electronics Engineers Inc., 2019. (Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019).
@inproceedings{819ffa7a7c9145dc9e00bd2e4c393fa7,
title = "Perspective-aware loss function for crowd density estimation",
abstract = "Estimation errors caused by perspective distortions are a long-standing problem in the domain of crowd counting. In this paper, we propose a novel loss function to allow filters in convolutional neural networks to learn features that are adaptive to the scale and perspective variation of individuals in crowd images. By exploring the crowd count error from regions close to the vanishing point of a perspective distorted image, we are able to penalize under-estimations. This is useful to train a network that is robust against perspective distortion for accurate density estimation. The proposed method is scene-independent and can be applied effectively to crowd scene with a variety of physical layout. Extensive comparative evaluations demonstrate that our proposed method achieves significant improvement over the state-of-the-art approaches on the challenging ShanghaiTech and UCF-QNRF datasets.",
author = "Bedir Yilmaz and {Ven Jyn }, Kok and Lim, {Mei Kuan} and {Sheikh Abdullah}, {Siti Norul Huda}",
year = "2019",
month = "5",
day = "1",
doi = "10.23919/MVA.2019.8758034",
language = "English",
series = "Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019",

}

TY - GEN

T1 - Perspective-aware loss function for crowd density estimation

AU - Yilmaz, Bedir

AU - Ven Jyn , Kok

AU - Lim, Mei Kuan

AU - Sheikh Abdullah, Siti Norul Huda

PY - 2019/5/1

Y1 - 2019/5/1

N2 - Estimation errors caused by perspective distortions are a long-standing problem in the domain of crowd counting. In this paper, we propose a novel loss function to allow filters in convolutional neural networks to learn features that are adaptive to the scale and perspective variation of individuals in crowd images. By exploring the crowd count error from regions close to the vanishing point of a perspective distorted image, we are able to penalize under-estimations. This is useful to train a network that is robust against perspective distortion for accurate density estimation. The proposed method is scene-independent and can be applied effectively to crowd scene with a variety of physical layout. Extensive comparative evaluations demonstrate that our proposed method achieves significant improvement over the state-of-the-art approaches on the challenging ShanghaiTech and UCF-QNRF datasets.

AB - Estimation errors caused by perspective distortions are a long-standing problem in the domain of crowd counting. In this paper, we propose a novel loss function to allow filters in convolutional neural networks to learn features that are adaptive to the scale and perspective variation of individuals in crowd images. By exploring the crowd count error from regions close to the vanishing point of a perspective distorted image, we are able to penalize under-estimations. This is useful to train a network that is robust against perspective distortion for accurate density estimation. The proposed method is scene-independent and can be applied effectively to crowd scene with a variety of physical layout. Extensive comparative evaluations demonstrate that our proposed method achieves significant improvement over the state-of-the-art approaches on the challenging ShanghaiTech and UCF-QNRF datasets.

UR - http://www.scopus.com/inward/record.url?scp=85070466581&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85070466581&partnerID=8YFLogxK

U2 - 10.23919/MVA.2019.8758034

DO - 10.23919/MVA.2019.8758034

M3 - Conference contribution

AN - SCOPUS:85070466581

T3 - Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019

BT - Proceedings of the 16th International Conference on Machine Vision Applications, MVA 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -