Learning to count with CNN boosting

Elad Walach; Lior Wolf

doi:https://doi.org/10.1007/978-3-319-46475-6_41

Learning to count with CNN boosting

Elad Walach, Lior Wolf

School of Computer Science

Tel Aviv University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In this paper, we address the task of object counting in images. We follow modern learning approaches in which a density map is estimated directly from the input image. We employ CNNs and incorporate two significant improvements to the state of the art methods: layered boosting and selective sampling. As a result, we manage both to increase the counting accuracy and to reduce processing time. Moreover, we show that the proposed method is effective, even in the presence of labeling errors. Extensive experiments on five different datasets demonstrate the efficacy and robustness of our approach. Mean Absolute error was reduced by 20% to 35%. At the same time, the training time of each CNN has been reduced by 50 %.

Original language	English
Title of host publication	Computer Vision - 14th European Conference, ECCV 2016, Proceedings
Editors	Bastian Leibe, Nicu Sebe, Max Welling, Jiri Matas
Pages	660-676
Number of pages	17
DOIs	https://doi.org/10.1007/978-3-319-46475-6_41
State	Published - 2016
Event	14th European Conference on Computer Vision, ECCV 2016 - Amsterdam, Netherlands Duration: 8 Oct 2016 → 16 Oct 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9906 LNCS

Conference

Conference	14th European Conference on Computer Vision, ECCV 2016
Country/Territory	Netherlands
City	Amsterdam
Period	8/10/16 → 16/10/16

Keywords

Convolutional neural networks
Counting
Gradient boosting
Sample selection

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

https://doi.org/10.1007/978-3-319-46475-6_41

Cite this

Walach, E., & Wolf, L. (2016). Learning to count with CNN boosting. In B. Leibe, N. Sebe, M. Welling, & J. Matas (Eds.), Computer Vision - 14th European Conference, ECCV 2016, Proceedings (pp. 660-676). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9906 LNCS). https://doi.org/10.1007/978-3-319-46475-6_41

@inproceedings{8b7a20a538d8481bb361e766bddcd539,

title = "Learning to count with CNN boosting",

abstract = "In this paper, we address the task of object counting in images. We follow modern learning approaches in which a density map is estimated directly from the input image. We employ CNNs and incorporate two significant improvements to the state of the art methods: layered boosting and selective sampling. As a result, we manage both to increase the counting accuracy and to reduce processing time. Moreover, we show that the proposed method is effective, even in the presence of labeling errors. Extensive experiments on five different datasets demonstrate the efficacy and robustness of our approach. Mean Absolute error was reduced by 20% to 35%. At the same time, the training time of each CNN has been reduced by 50 %.",

keywords = "Convolutional neural networks, Counting, Gradient boosting, Sample selection",

author = "Elad Walach and Lior Wolf",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 14th European Conference on Computer Vision, ECCV 2016 ; Conference date: 08-10-2016 Through 16-10-2016",

year = "2016",

doi = "https://doi.org/10.1007/978-3-319-46475-6_41",

language = "الإنجليزيّة",

isbn = "9783319464749",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "660--676",

editor = "Bastian Leibe and Nicu Sebe and Max Welling and Jiri Matas",

booktitle = "Computer Vision - 14th European Conference, ECCV 2016, Proceedings",

}

Walach, E & Wolf, L 2016, Learning to count with CNN boosting. in B Leibe, N Sebe, M Welling & J Matas (eds), Computer Vision - 14th European Conference, ECCV 2016, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9906 LNCS, pp. 660-676, 14th European Conference on Computer Vision, ECCV 2016, Amsterdam, Netherlands, 8/10/16. https://doi.org/10.1007/978-3-319-46475-6_41

Learning to count with CNN boosting. / Walach, Elad; Wolf, Lior.
Computer Vision - 14th European Conference, ECCV 2016, Proceedings. ed. / Bastian Leibe; Nicu Sebe; Max Welling; Jiri Matas. 2016. p. 660-676 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9906 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Learning to count with CNN boosting

AU - Walach, Elad

AU - Wolf, Lior

N1 - Publisher Copyright: © Springer International Publishing AG 2016.

PY - 2016

Y1 - 2016

N2 - In this paper, we address the task of object counting in images. We follow modern learning approaches in which a density map is estimated directly from the input image. We employ CNNs and incorporate two significant improvements to the state of the art methods: layered boosting and selective sampling. As a result, we manage both to increase the counting accuracy and to reduce processing time. Moreover, we show that the proposed method is effective, even in the presence of labeling errors. Extensive experiments on five different datasets demonstrate the efficacy and robustness of our approach. Mean Absolute error was reduced by 20% to 35%. At the same time, the training time of each CNN has been reduced by 50 %.

AB - In this paper, we address the task of object counting in images. We follow modern learning approaches in which a density map is estimated directly from the input image. We employ CNNs and incorporate two significant improvements to the state of the art methods: layered boosting and selective sampling. As a result, we manage both to increase the counting accuracy and to reduce processing time. Moreover, we show that the proposed method is effective, even in the presence of labeling errors. Extensive experiments on five different datasets demonstrate the efficacy and robustness of our approach. Mean Absolute error was reduced by 20% to 35%. At the same time, the training time of each CNN has been reduced by 50 %.

KW - Convolutional neural networks

KW - Counting

KW - Gradient boosting

KW - Sample selection

UR - http://www.scopus.com/inward/record.url?scp=84990864917&partnerID=8YFLogxK

U2 - https://doi.org/10.1007/978-3-319-46475-6_41

DO - https://doi.org/10.1007/978-3-319-46475-6_41

M3 - منشور من مؤتمر

SN - 9783319464749

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 660

EP - 676

BT - Computer Vision - 14th European Conference, ECCV 2016, Proceedings

A2 - Leibe, Bastian

A2 - Sebe, Nicu

A2 - Welling, Max

A2 - Matas, Jiri

T2 - 14th European Conference on Computer Vision, ECCV 2016

Y2 - 8 October 2016 through 16 October 2016

ER -

Walach E, Wolf L. Learning to count with CNN boosting. In Leibe B, Sebe N, Welling M, Matas J, editors, Computer Vision - 14th European Conference, ECCV 2016, Proceedings. 2016. p. 660-676. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: https://doi.org/10.1007/978-3-319-46475-6_41

Learning to count with CNN boosting

Abstract

Publication series

Conference

Keywords

All Science Journal Classification (ASJC) codes

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this