Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification

Geovanni Figueroa-Mata; Erick Mata-Montero; Juan Carlos Valverde-Otárola; Dagoberto Arias-Aguilar

doi:10.1109/CONCAPAN.2018.8596406

Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification

Costa Rica Institute of Technology

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

3 Scopus citations

Abstract

With the goal of quantifying the importance of each of the cutting planes of wood samples in the training process of a convolutional neural network that identifies forest species based on images of those cutting planes, we propose a convolutional model that is trained from scratch with images of transverse, radial, and tangential sections of Costa Rican forest species wood samples. The best Top1-accuracy achieved is 89.58% when the network is trained with transverse sections only. Because this is more than 20% better than the accuracy achieved when using any of the other two sections individually, we conclude that this is the most significant section of all three. This is consistent with current practice of experts, who prefer this cutting plane when conducting manual identifications based on anatomical features of wood samples.

Original language	English
Title of host publication	Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018
Editors	Manuel N. Cardona
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781538661222
DOIs	https://doi.org/10.1109/CONCAPAN.2018.8596406
State	Published - 28 Dec 2018
Event	2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018 - San Salvador, El Salvador Duration: 7 Nov 2018 → 9 Nov 2018

Publication series

Name	Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018

Conference

Conference	2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018
Country/Territory	El Salvador
City	San Salvador
Period	7/11/18 → 9/11/18

Keywords

Automated image-based species identification
Convolutional neural network
Costa rican forest species
Cutting planes

Access to Document

10.1109/CONCAPAN.2018.8596406

Cite this

Figueroa-Mata, G., Mata-Montero, E., Valverde-Otárola, J. C., & Arias-Aguilar, D. (2018). Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification. In M. N. Cardona (Ed.), Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018 Article 8596406 (Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CONCAPAN.2018.8596406

Figueroa-Mata, Geovanni ; Mata-Montero, Erick ; Valverde-Otárola, Juan Carlos et al. / Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification. Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018. editor / Manuel N. Cardona. Institute of Electrical and Electronics Engineers Inc., 2018. (Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018).

@inproceedings{8ee37eea497d4d9c9fd6a926efe8ed60,

title = "Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification",

abstract = "With the goal of quantifying the importance of each of the cutting planes of wood samples in the training process of a convolutional neural network that identifies forest species based on images of those cutting planes, we propose a convolutional model that is trained from scratch with images of transverse, radial, and tangential sections of Costa Rican forest species wood samples. The best Top1-accuracy achieved is 89.58\% when the network is trained with transverse sections only. Because this is more than 20\% better than the accuracy achieved when using any of the other two sections individually, we conclude that this is the most significant section of all three. This is consistent with current practice of experts, who prefer this cutting plane when conducting manual identifications based on anatomical features of wood samples.",

keywords = "Automated image-based species identification, Convolutional neural network, Costa rican forest species, Cutting planes",

author = "Geovanni Figueroa-Mata and Erick Mata-Montero and Valverde-Ot{\'a}rola, \{Juan Carlos\} and Dagoberto Arias-Aguilar",

note = "Publisher Copyright: {\textcopyright} 2018 IEEE.; 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018 ; Conference date: 07-11-2018 Through 09-11-2018",

year = "2018",

month = dec,

day = "28",

doi = "10.1109/CONCAPAN.2018.8596406",

language = "Ingl{\'e}s",

series = "Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

editor = "Cardona, \{Manuel N.\}",

booktitle = "Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018",

}

Figueroa-Mata, G, Mata-Montero, E, Valverde-Otárola, JC & Arias-Aguilar, D 2018, Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification. in MN Cardona (ed.), Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018., 8596406, Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018, Institute of Electrical and Electronics Engineers Inc., 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018, San Salvador, El Salvador, 7/11/18. https://doi.org/10.1109/CONCAPAN.2018.8596406

Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification. / Figueroa-Mata, Geovanni; Mata-Montero, Erick; Valverde-Otárola, Juan Carlos et al.
Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018. ed. / Manuel N. Cardona. Institute of Electrical and Electronics Engineers Inc., 2018. 8596406 (Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018).

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

TY - GEN

T1 - Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification

AU - Figueroa-Mata, Geovanni

AU - Mata-Montero, Erick

AU - Valverde-Otárola, Juan Carlos

AU - Arias-Aguilar, Dagoberto

PY - 2018/12/28

Y1 - 2018/12/28

N2 - With the goal of quantifying the importance of each of the cutting planes of wood samples in the training process of a convolutional neural network that identifies forest species based on images of those cutting planes, we propose a convolutional model that is trained from scratch with images of transverse, radial, and tangential sections of Costa Rican forest species wood samples. The best Top1-accuracy achieved is 89.58% when the network is trained with transverse sections only. Because this is more than 20% better than the accuracy achieved when using any of the other two sections individually, we conclude that this is the most significant section of all three. This is consistent with current practice of experts, who prefer this cutting plane when conducting manual identifications based on anatomical features of wood samples.

AB - With the goal of quantifying the importance of each of the cutting planes of wood samples in the training process of a convolutional neural network that identifies forest species based on images of those cutting planes, we propose a convolutional model that is trained from scratch with images of transverse, radial, and tangential sections of Costa Rican forest species wood samples. The best Top1-accuracy achieved is 89.58% when the network is trained with transverse sections only. Because this is more than 20% better than the accuracy achieved when using any of the other two sections individually, we conclude that this is the most significant section of all three. This is consistent with current practice of experts, who prefer this cutting plane when conducting manual identifications based on anatomical features of wood samples.

KW - Automated image-based species identification

KW - Convolutional neural network

KW - Costa rican forest species

KW - Cutting planes

UR - https://www.scopus.com/pages/publications/85061505168

U2 - 10.1109/CONCAPAN.2018.8596406

DO - 10.1109/CONCAPAN.2018.8596406

M3 - Contribución a la conferencia

AN - SCOPUS:85061505168

T3 - Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018

BT - Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018

A2 - Cardona, Manuel N.

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018

Y2 - 7 November 2018 through 9 November 2018

ER -

Figueroa-Mata G, Mata-Montero E, Valverde-Otárola JC , Arias-Aguilar D. Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification. In Cardona MN, editor, Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018. Institute of Electrical and Electronics Engineers Inc. 2018. 8596406. (Proceedings of the 2018 IEEE 38th Central America and Panama Convention, CONCAPAN 2018). doi: 10.1109/CONCAPAN.2018.8596406

Evaluating the significance of cutting planes of wood samples when training CNNs for forest species identification

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this