Heterogeneous multi-column ConvNets with a fusion framework for object recognition

Research output: Chapter in Book/Conference paperConference paper

2 Citations (Scopus)

Abstract

© 2015 IEEE. The purpose of this paper is to investigate heterogeneous multi-column ConvNets (MCCNN) and fusion methods for them. We first construct heterogeneous MCCNN by combining ConvNets with different structures. We then use different fusion methods to check their performances to find out the effect of fusion methods for MCCNN. We also propose a novel sliding window based fusion framework which defines a specific subset of columns to be picked up from MCCNN for fusion. Two different strategies (exhaustive sliding window and sliding window from training) are investigated to determine the best performance of the fusion process. We tested the heterogeneous MCCNN and sliding window fusion on the MNIST dataset for optical character recognition. Experiments show that MCCNN improved the accuracy of recognition compared with a single column of ConvNets. Moreover, sliding window fusion is a more generalized fusion method and consistently achieves better results compared with the traditional fusion methods. We also tested the MCCNN and sliding window fusion on CIFAR-10 and Caltech-256 datasets. We achieved superior results compared to existing state-of-the-art techniques.
Original languageEnglish
Title of host publication2015 IEEE Winter Conference on Applications of Computer Vision
Place of PublicationUnited States
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages773-780
VolumeN/A
ISBN (Print)9781479966820
DOIs
Publication statusPublished - 2015
Event2015 IEEE Winter Conference on Applications of Computer Vision - Waikoloa, United States
Duration: 5 Jan 20159 Mar 2015

Conference

Conference2015 IEEE Winter Conference on Applications of Computer Vision
CountryUnited States
CityWaikoloa
Period5/01/159/03/15

Fingerprint

Object recognition
Fusion reactions
Optical character recognition

Cite this

Li, Y., Sohel, F., Bennamoun, M., & Lei, H. (2015). Heterogeneous multi-column ConvNets with a fusion framework for object recognition. In 2015 IEEE Winter Conference on Applications of Computer Vision (Vol. N/A, pp. 773-780). United States: IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/WACV.2015.108
Li, Y. ; Sohel, Ferdous ; Bennamoun, Mohammed ; Lei, H. / Heterogeneous multi-column ConvNets with a fusion framework for object recognition. 2015 IEEE Winter Conference on Applications of Computer Vision. Vol. N/A United States : IEEE, Institute of Electrical and Electronics Engineers, 2015. pp. 773-780
@inproceedings{675fb493eed043d9a9c59f3a55ecf76d,
title = "Heterogeneous multi-column ConvNets with a fusion framework for object recognition",
abstract = "{\circledC} 2015 IEEE. The purpose of this paper is to investigate heterogeneous multi-column ConvNets (MCCNN) and fusion methods for them. We first construct heterogeneous MCCNN by combining ConvNets with different structures. We then use different fusion methods to check their performances to find out the effect of fusion methods for MCCNN. We also propose a novel sliding window based fusion framework which defines a specific subset of columns to be picked up from MCCNN for fusion. Two different strategies (exhaustive sliding window and sliding window from training) are investigated to determine the best performance of the fusion process. We tested the heterogeneous MCCNN and sliding window fusion on the MNIST dataset for optical character recognition. Experiments show that MCCNN improved the accuracy of recognition compared with a single column of ConvNets. Moreover, sliding window fusion is a more generalized fusion method and consistently achieves better results compared with the traditional fusion methods. We also tested the MCCNN and sliding window fusion on CIFAR-10 and Caltech-256 datasets. We achieved superior results compared to existing state-of-the-art techniques.",
author = "Y. Li and Ferdous Sohel and Mohammed Bennamoun and H. Lei",
year = "2015",
doi = "10.1109/WACV.2015.108",
language = "English",
isbn = "9781479966820",
volume = "N/A",
pages = "773--780",
booktitle = "2015 IEEE Winter Conference on Applications of Computer Vision",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
address = "United States",

}

Li, Y, Sohel, F, Bennamoun, M & Lei, H 2015, Heterogeneous multi-column ConvNets with a fusion framework for object recognition. in 2015 IEEE Winter Conference on Applications of Computer Vision. vol. N/A, IEEE, Institute of Electrical and Electronics Engineers, United States, pp. 773-780, 2015 IEEE Winter Conference on Applications of Computer Vision , Waikoloa, United States, 5/01/15. https://doi.org/10.1109/WACV.2015.108

Heterogeneous multi-column ConvNets with a fusion framework for object recognition. / Li, Y.; Sohel, Ferdous; Bennamoun, Mohammed; Lei, H.

2015 IEEE Winter Conference on Applications of Computer Vision. Vol. N/A United States : IEEE, Institute of Electrical and Electronics Engineers, 2015. p. 773-780.

Research output: Chapter in Book/Conference paperConference paper

TY - GEN

T1 - Heterogeneous multi-column ConvNets with a fusion framework for object recognition

AU - Li, Y.

AU - Sohel, Ferdous

AU - Bennamoun, Mohammed

AU - Lei, H.

PY - 2015

Y1 - 2015

N2 - © 2015 IEEE. The purpose of this paper is to investigate heterogeneous multi-column ConvNets (MCCNN) and fusion methods for them. We first construct heterogeneous MCCNN by combining ConvNets with different structures. We then use different fusion methods to check their performances to find out the effect of fusion methods for MCCNN. We also propose a novel sliding window based fusion framework which defines a specific subset of columns to be picked up from MCCNN for fusion. Two different strategies (exhaustive sliding window and sliding window from training) are investigated to determine the best performance of the fusion process. We tested the heterogeneous MCCNN and sliding window fusion on the MNIST dataset for optical character recognition. Experiments show that MCCNN improved the accuracy of recognition compared with a single column of ConvNets. Moreover, sliding window fusion is a more generalized fusion method and consistently achieves better results compared with the traditional fusion methods. We also tested the MCCNN and sliding window fusion on CIFAR-10 and Caltech-256 datasets. We achieved superior results compared to existing state-of-the-art techniques.

AB - © 2015 IEEE. The purpose of this paper is to investigate heterogeneous multi-column ConvNets (MCCNN) and fusion methods for them. We first construct heterogeneous MCCNN by combining ConvNets with different structures. We then use different fusion methods to check their performances to find out the effect of fusion methods for MCCNN. We also propose a novel sliding window based fusion framework which defines a specific subset of columns to be picked up from MCCNN for fusion. Two different strategies (exhaustive sliding window and sliding window from training) are investigated to determine the best performance of the fusion process. We tested the heterogeneous MCCNN and sliding window fusion on the MNIST dataset for optical character recognition. Experiments show that MCCNN improved the accuracy of recognition compared with a single column of ConvNets. Moreover, sliding window fusion is a more generalized fusion method and consistently achieves better results compared with the traditional fusion methods. We also tested the MCCNN and sliding window fusion on CIFAR-10 and Caltech-256 datasets. We achieved superior results compared to existing state-of-the-art techniques.

U2 - 10.1109/WACV.2015.108

DO - 10.1109/WACV.2015.108

M3 - Conference paper

SN - 9781479966820

VL - N/A

SP - 773

EP - 780

BT - 2015 IEEE Winter Conference on Applications of Computer Vision

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - United States

ER -

Li Y, Sohel F, Bennamoun M, Lei H. Heterogeneous multi-column ConvNets with a fusion framework for object recognition. In 2015 IEEE Winter Conference on Applications of Computer Vision. Vol. N/A. United States: IEEE, Institute of Electrical and Electronics Engineers. 2015. p. 773-780 https://doi.org/10.1109/WACV.2015.108