Photometric stereo using CNN-based feature-merging network

Euijeong Song; Minho Chang

doi:10.23919/ICCAS50221.2020.9268198

Photometric stereo using CNN-based feature-merging network

Euijeong Song, Minho Chang

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

2 Citations (Scopus)

Abstract

We propose a photometric stereo method using Convolutional Neural Network (CNN) based method, which is effective for deriving surface normal data from non-lambertian objects. Our method extracts feature maps from a set of images of object using shared feature extraction network, and merge the extracted feature maps using two pooling method: max-pooling and average-pooling. The merged feature maps are concatenated and passed to final CNN layers to derive the surface normal map. We tested our network on the most widely-used benchmark dataset and confirmed that our method performs better than existing deep learning based photometric stereo method.

Original language	English
Title of host publication	2020 20th International Conference on Control, Automation and Systems, ICCAS 2020
Publisher	IEEE Computer Society
Pages	865-868
Number of pages	4
ISBN (Electronic)	9788993215205
DOIs	https://doi.org/10.23919/ICCAS50221.2020.9268198
Publication status	Published - 2020 Oct 13
Event	20th International Conference on Control, Automation and Systems, ICCAS 2020 - Busan, Korea, Republic of Duration: 2020 Oct 13 → 2020 Oct 16

Publication series

Name	International Conference on Control, Automation and Systems
Volume	2020-October
ISSN (Print)	1598-7833

Conference

Conference	20th International Conference on Control, Automation and Systems, ICCAS 2020
Country/Territory	Korea, Republic of
City	Busan
Period	20/10/13 → 20/10/16

Bibliographical note

Funding Information:
This research was results of a study on the "HPC Support" Project, supported by the ‘Ministry of Science and ICT’ and NIPA.

Publisher Copyright:
© 2020 Institute of Control, Robotics, and Systems - ICROS.

Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.

Keywords

Computer Vision
Convolutional Neural Network
Feature Merge
Photometric Stereo

ASJC Scopus subject areas

Artificial Intelligence
Computer Science Applications
Control and Systems Engineering
Electrical and Electronic Engineering

Access to Document

10.23919/ICCAS50221.2020.9268198

Cite this

Photometric stereo using CNN-based feature-merging network. / Song, Euijeong; Chang, Minho.
2020 20th International Conference on Control, Automation and Systems, ICCAS 2020. IEEE Computer Society, 2020. p. 865-868 9268198 (International Conference on Control, Automation and Systems; Vol. 2020-October).

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

Song, E & Chang, M 2020, Photometric stereo using CNN-based feature-merging network. in 2020 20th International Conference on Control, Automation and Systems, ICCAS 2020., 9268198, International Conference on Control, Automation and Systems, vol. 2020-October, IEEE Computer Society, pp. 865-868, 20th International Conference on Control, Automation and Systems, ICCAS 2020, Busan, Korea, Republic of, 20/10/13. https://doi.org/10.23919/ICCAS50221.2020.9268198

@inproceedings{13b3e0d60824400bb89c1e709f2cea73,

title = "Photometric stereo using CNN-based feature-merging network",

abstract = "We propose a photometric stereo method using Convolutional Neural Network (CNN) based method, which is effective for deriving surface normal data from non-lambertian objects. Our method extracts feature maps from a set of images of object using shared feature extraction network, and merge the extracted feature maps using two pooling method: max-pooling and average-pooling. The merged feature maps are concatenated and passed to final CNN layers to derive the surface normal map. We tested our network on the most widely-used benchmark dataset and confirmed that our method performs better than existing deep learning based photometric stereo method.",

keywords = "Computer Vision, Convolutional Neural Network, Feature Merge, Photometric Stereo",

author = "Euijeong Song and Minho Chang",

note = "Funding Information: This research was results of a study on the {"}HPC Support{"} Project, supported by the {\textquoteleft}Ministry of Science and ICT{\textquoteright} and NIPA. Publisher Copyright: {\textcopyright} 2020 Institute of Control, Robotics, and Systems - ICROS. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.; 20th International Conference on Control, Automation and Systems, ICCAS 2020 ; Conference date: 13-10-2020 Through 16-10-2020",

year = "2020",

month = oct,

day = "13",

doi = "10.23919/ICCAS50221.2020.9268198",

language = "English",

series = "International Conference on Control, Automation and Systems",

publisher = "IEEE Computer Society",

pages = "865--868",

booktitle = "2020 20th International Conference on Control, Automation and Systems, ICCAS 2020",

}

TY - GEN

T1 - Photometric stereo using CNN-based feature-merging network

AU - Song, Euijeong

AU - Chang, Minho

N1 - Funding Information: This research was results of a study on the "HPC Support" Project, supported by the ‘Ministry of Science and ICT’ and NIPA. Publisher Copyright: © 2020 Institute of Control, Robotics, and Systems - ICROS. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.

PY - 2020/10/13

Y1 - 2020/10/13

N2 - We propose a photometric stereo method using Convolutional Neural Network (CNN) based method, which is effective for deriving surface normal data from non-lambertian objects. Our method extracts feature maps from a set of images of object using shared feature extraction network, and merge the extracted feature maps using two pooling method: max-pooling and average-pooling. The merged feature maps are concatenated and passed to final CNN layers to derive the surface normal map. We tested our network on the most widely-used benchmark dataset and confirmed that our method performs better than existing deep learning based photometric stereo method.

AB - We propose a photometric stereo method using Convolutional Neural Network (CNN) based method, which is effective for deriving surface normal data from non-lambertian objects. Our method extracts feature maps from a set of images of object using shared feature extraction network, and merge the extracted feature maps using two pooling method: max-pooling and average-pooling. The merged feature maps are concatenated and passed to final CNN layers to derive the surface normal map. We tested our network on the most widely-used benchmark dataset and confirmed that our method performs better than existing deep learning based photometric stereo method.

KW - Computer Vision

KW - Convolutional Neural Network

KW - Feature Merge

KW - Photometric Stereo

UR - http://www.scopus.com/inward/record.url?scp=85098048693&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85098048693&partnerID=8YFLogxK

U2 - 10.23919/ICCAS50221.2020.9268198

DO - 10.23919/ICCAS50221.2020.9268198

M3 - Conference contribution

AN - SCOPUS:85098048693

T3 - International Conference on Control, Automation and Systems

SP - 865

EP - 868

BT - 2020 20th International Conference on Control, Automation and Systems, ICCAS 2020

PB - IEEE Computer Society

T2 - 20th International Conference on Control, Automation and Systems, ICCAS 2020

Y2 - 13 October 2020 through 16 October 2020

ER -

Photometric stereo using CNN-based feature-merging network

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this