Fast Sliding Window Classification with Convolutional Neural Networks

Henry G. R. Gouk; Anthony M. Blake

doi:10.1145/2683405.2683429

Fast Sliding Window Classification with Convolutional Neural Networks

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

Abstract / Description of output

Convolutional Neural Networks (CNNs) have repeatedly been shown to be the state of the art method for natural signal classification -- image classification in particular. Unfortunately, due to the high model complexity CNNs often cannot be used for object detection tasks with real-time constraints, where many predictions have to be made on sub-windows of a large input image. We demonstrate how two recent advances in CNN efficiency can be combined, with modifications, to provide a substantial speedup for sliding window classification. An in depth analysis of the various factors that can impact performance is presented.

Original language	English
Title of host publication	Proceedings of the 29th International Conference on Image and Vision Computing New Zealand
Place of Publication	New York, NY, USA
Publisher	ACM
Pages	114-118
Number of pages	5
ISBN (Print)	978-1-4503-3184-5
DOIs	https://doi.org/10.1145/2683405.2683429
Publication status	Published - 19 Nov 2014
Event	29th International Conference on Image and Vision Computing New Zealand - Hamilton, New Zealand Duration: 19 Nov 2014 → 21 Nov 2014

Publication series

Name	IVCNZ '14
Publisher	ACM

Conference

Conference	29th International Conference on Image and Vision Computing New Zealand
Abbreviated title	IVCNZ
Country/Territory	New Zealand
City	Hamilton
Period	19/11/14 → 21/11/14

Keywords / Materials (for Non-textual outputs)

Convolutional Neural Networks
FFT
Object Detection
Sliding Window

Access to Document

10.1145/2683405.2683429

https://dl.acm.org/citation.cfm?id=2683429

Cite this

@inproceedings{b00895222788495db616d64d4f745c54,

title = "Fast Sliding Window Classification with Convolutional Neural Networks",

abstract = "Convolutional Neural Networks (CNNs) have repeatedly been shown to be the state of the art method for natural signal classification -- image classification in particular. Unfortunately, due to the high model complexity CNNs often cannot be used for object detection tasks with real-time constraints, where many predictions have to be made on sub-windows of a large input image. We demonstrate how two recent advances in CNN efficiency can be combined, with modifications, to provide a substantial speedup for sliding window classification. An in depth analysis of the various factors that can impact performance is presented.",

keywords = "Convolutional Neural Networks, FFT, Object Detection, Sliding Window",

author = "Gouk, {Henry G. R.} and Blake, {Anthony M.}",

year = "2014",

month = nov,

day = "19",

doi = "10.1145/2683405.2683429",

language = "English",

isbn = "978-1-4503-3184-5",

series = "IVCNZ '14",

publisher = "ACM",

pages = "114--118",

booktitle = "Proceedings of the 29th International Conference on Image and Vision Computing New Zealand",

note = "29th International Conference on Image and Vision Computing New Zealand, IVCNZ ; Conference date: 19-11-2014 Through 21-11-2014",

}

Gouk, HGR & Blake, AM 2014, Fast Sliding Window Classification with Convolutional Neural Networks. in Proceedings of the 29th International Conference on Image and Vision Computing New Zealand. IVCNZ '14, ACM, New York, NY, USA, pp. 114-118, 29th International Conference on Image and Vision Computing New Zealand, Hamilton, New Zealand, 19/11/14. https://doi.org/10.1145/2683405.2683429

TY - GEN

T1 - Fast Sliding Window Classification with Convolutional Neural Networks

AU - Gouk, Henry G. R.

AU - Blake, Anthony M.

PY - 2014/11/19

Y1 - 2014/11/19

N2 - Convolutional Neural Networks (CNNs) have repeatedly been shown to be the state of the art method for natural signal classification -- image classification in particular. Unfortunately, due to the high model complexity CNNs often cannot be used for object detection tasks with real-time constraints, where many predictions have to be made on sub-windows of a large input image. We demonstrate how two recent advances in CNN efficiency can be combined, with modifications, to provide a substantial speedup for sliding window classification. An in depth analysis of the various factors that can impact performance is presented.

AB - Convolutional Neural Networks (CNNs) have repeatedly been shown to be the state of the art method for natural signal classification -- image classification in particular. Unfortunately, due to the high model complexity CNNs often cannot be used for object detection tasks with real-time constraints, where many predictions have to be made on sub-windows of a large input image. We demonstrate how two recent advances in CNN efficiency can be combined, with modifications, to provide a substantial speedup for sliding window classification. An in depth analysis of the various factors that can impact performance is presented.

KW - Convolutional Neural Networks

KW - FFT

KW - Object Detection

KW - Sliding Window

U2 - 10.1145/2683405.2683429

DO - 10.1145/2683405.2683429

M3 - Conference contribution

SN - 978-1-4503-3184-5

T3 - IVCNZ '14

SP - 114

EP - 118

BT - Proceedings of the 29th International Conference on Image and Vision Computing New Zealand

PB - ACM

CY - New York, NY, USA

T2 - 29th International Conference on Image and Vision Computing New Zealand

Y2 - 19 November 2014 through 21 November 2014

ER -

Fast Sliding Window Classification with Convolutional Neural Networks

Abstract / Description of output

Publication series

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Cite this