Hetconv: Heterogeneous kernel-based convolutions for deep cnns

Pravendra Singh; Vinay Kumar Verma; Piyush Rai; Vinay P. Namboodiri

doi:10.1109/CVPR.2019.00497

Hetconv: Heterogeneous kernel-based convolutions for deep cnns

Pravendra Singh, Vinay Kumar Verma, Piyush Rai, Vinay P. Namboodiri

Department of Computer Science

Indian Institute of Technology Kanpur

Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding

109 Citations (SciVal)

117 Downloads (Pure)

Abstract

We present a novel deep learning architecture in which the convolution operation leverages heterogeneous kernels. The proposed HetConv (Heterogeneous Kernel-Based Convolution) reduces the computation (FLOPs) and the number of parameters as compared to standard convolution operation while still maintaining representational efficiency. To show the effectiveness of our proposed convolution, we present extensive experimental results on the standard convolutional neural network (CNN) architectures such as VGG and ResNet. We find that after replacing the standard convolutional filters in these architectures with our proposed HetConv filters, we achieve 3X to 8X FLOPs based improvement in speed while still maintaining (and sometimes improving) the accuracy. We also compare our proposed convolutions with group/depth wise convolutions and show that it achieves more FLOPs reduction with significantly higher accuracy.

Original language	English
Title of host publication	Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019
Publisher	IEEE
Pages	4830-4839
Number of pages	10
ISBN (Electronic)	9781728132938
DOIs	https://doi.org/10.1109/CVPR.2019.00497
Publication status	Published - 9 Jan 2020
Event	32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019 - Long Beach, USA United States Duration: 16 Jun 2019 → 20 Jun 2019

Publication series

Name	Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Volume	2019-June
ISSN (Print)	1063-6919

Conference

Conference	32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019
Country/Territory	USA United States
City	Long Beach
Period	16/06/19 → 20/06/19

Funding

Pravendra Singh acknowledges his travel support from Google. Vinay Verma acknowledges support from Visves-varaya fellowship.

Keywords

Categorization
Computer Vision Theory
Deep Learning
Recognition: Detection
Retrieval

ASJC Scopus subject areas

Software
Computer Vision and Pattern Recognition

Access to Document

10.1109/CVPR.2019.00497

HetConv Heterogeneous Kernel-Based Convolutions for Deep CNNs
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
Accepted author manuscript, 333 KB

Cite this

Singh, P., Verma, V. K., Rai, P., & Namboodiri, V. P. (2020). Hetconv: Heterogeneous kernel-based convolutions for deep cnns. In Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019 (pp. 4830-4839). Article 8953857 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition; Vol. 2019-June). IEEE. https://doi.org/10.1109/CVPR.2019.00497

Hetconv: Heterogeneous kernel-based convolutions for deep cnns. / Singh, Pravendra; Verma, Vinay Kumar; Rai, Piyush et al.
Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019. IEEE, 2020. p. 4830-4839 8953857 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition; Vol. 2019-June).

Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding

Singh, P, Verma, VK, Rai, P & Namboodiri, VP 2020, Hetconv: Heterogeneous kernel-based convolutions for deep cnns. in Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019., 8953857, Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2019-June, IEEE, pp. 4830-4839, 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, USA United States, 16/06/19. https://doi.org/10.1109/CVPR.2019.00497

@inproceedings{51e85d912fb44743a061b4c58e8bf182,

title = "Hetconv: Heterogeneous kernel-based convolutions for deep cnns",

abstract = "We present a novel deep learning architecture in which the convolution operation leverages heterogeneous kernels. The proposed HetConv (Heterogeneous Kernel-Based Convolution) reduces the computation (FLOPs) and the number of parameters as compared to standard convolution operation while still maintaining representational efficiency. To show the effectiveness of our proposed convolution, we present extensive experimental results on the standard convolutional neural network (CNN) architectures such as VGG and ResNet. We find that after replacing the standard convolutional filters in these architectures with our proposed HetConv filters, we achieve 3X to 8X FLOPs based improvement in speed while still maintaining (and sometimes improving) the accuracy. We also compare our proposed convolutions with group/depth wise convolutions and show that it achieves more FLOPs reduction with significantly higher accuracy.",

keywords = "Categorization, Computer Vision Theory, Deep Learning, Recognition: Detection, Retrieval",

author = "Pravendra Singh and Verma, {Vinay Kumar} and Piyush Rai and Namboodiri, {Vinay P.}",

year = "2020",

month = jan,

day = "9",

doi = "10.1109/CVPR.2019.00497",

language = "English",

series = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",

publisher = "IEEE",

pages = "4830--4839",

booktitle = "Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019",

address = "USA United States",

note = "32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019 ; Conference date: 16-06-2019 Through 20-06-2019",

}

TY - GEN

T1 - Hetconv: Heterogeneous kernel-based convolutions for deep cnns

AU - Singh, Pravendra

AU - Verma, Vinay Kumar

AU - Rai, Piyush

AU - Namboodiri, Vinay P.

PY - 2020/1/9

Y1 - 2020/1/9

N2 - We present a novel deep learning architecture in which the convolution operation leverages heterogeneous kernels. The proposed HetConv (Heterogeneous Kernel-Based Convolution) reduces the computation (FLOPs) and the number of parameters as compared to standard convolution operation while still maintaining representational efficiency. To show the effectiveness of our proposed convolution, we present extensive experimental results on the standard convolutional neural network (CNN) architectures such as VGG and ResNet. We find that after replacing the standard convolutional filters in these architectures with our proposed HetConv filters, we achieve 3X to 8X FLOPs based improvement in speed while still maintaining (and sometimes improving) the accuracy. We also compare our proposed convolutions with group/depth wise convolutions and show that it achieves more FLOPs reduction with significantly higher accuracy.

AB - We present a novel deep learning architecture in which the convolution operation leverages heterogeneous kernels. The proposed HetConv (Heterogeneous Kernel-Based Convolution) reduces the computation (FLOPs) and the number of parameters as compared to standard convolution operation while still maintaining representational efficiency. To show the effectiveness of our proposed convolution, we present extensive experimental results on the standard convolutional neural network (CNN) architectures such as VGG and ResNet. We find that after replacing the standard convolutional filters in these architectures with our proposed HetConv filters, we achieve 3X to 8X FLOPs based improvement in speed while still maintaining (and sometimes improving) the accuracy. We also compare our proposed convolutions with group/depth wise convolutions and show that it achieves more FLOPs reduction with significantly higher accuracy.

KW - Categorization

KW - Computer Vision Theory

KW - Deep Learning

KW - Recognition: Detection

KW - Retrieval

UR - http://www.scopus.com/inward/record.url?scp=85076968781&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2019.00497

DO - 10.1109/CVPR.2019.00497

M3 - Chapter in a published conference proceeding

AN - SCOPUS:85076968781

T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

SP - 4830

EP - 4839

BT - Proceedings - 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019

PB - IEEE

T2 - 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2019

Y2 - 16 June 2019 through 20 June 2019

ER -

Hetconv: Heterogeneous kernel-based convolutions for deep cnns

Abstract

Publication series

Conference

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this