Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net

Miguel Ángel Chicchón Apaza; Héctor Manuel Bedón Monzón; Ramón Pablo Alcarria Garrido

doi:10.1007/978-3-030-42520-3_38

Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net

Miguel Ángel Chicchón Apaza, Héctor Manuel Bedón Monzón, Ramón Pablo Alcarria Garrido

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

Abstract

A first step in the process of automating weed removal in precision agriculture is the semantic segmentation of crops, weeds and soil. Deep learning techniques based on convolutional neural networks are successfully applied today and one of the most popular network architectures in semantic segmentation problems is U-Net. In this article, the variants in the U-Net architecture were evaluated based on the aggregation of residual and recurring blocks to improve their performance. For training and testing, a set of data available on the Internet was used, consisting of 60 multispectral images with unbalanced pixels, so techniques were applied to increase and balance the data. Experimental results show a slight increase in quality metrics compared to the classic U-Net architecture.

Original language	English
Title of host publication	Applied Technologies - 1st International Conference, ICAT 2019, Proceedings
Editors	Miguel Botto-Tobar, Marcelo Zambrano Vizuete, Pablo Torres-Carrión, Sergio Montes León, Guillermo Pizarro Vásquez, Benjamin Durakovic
Publisher	Springer
Pages	473-485
Number of pages	13
ISBN (Print)	9783030425197
DOIs	https://doi.org/10.1007/978-3-030-42520-3_38
State	Published - 1 Jan 2020
Event	1st International Conference on Applied Technologies, ICAT 2019 - Quito, Ecuador Duration: 3 Dec 2019 → 5 Dec 2019

Publication series

Name	Communications in Computer and Information Science
Volume	1194 CCIS
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	1st International Conference on Applied Technologies, ICAT 2019
Country	Ecuador
City	Quito
Period	3/12/19 → 5/12/19

Keywords

Convolutional neural network
Deep learning
Precision agriculture
Semantic segmentation
U-Net

Access to Document

10.1007/978-3-030-42520-3_38

Link to publication in Scopus

Fingerprint Dive into the research topics of 'Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net'. Together they form a unique fingerprint.

Projects

1 Finished

Quinua Smartapp: Quinoa Smartapp: intelligent rural platform prototype, in real time, to increase productivity in the organic quinoa value chain (chenopodium quinoa willd)

Bedón Monzón, H. M., Garcia Lopez, Y. J. & Carhuancho Lucen, C. A.

6/01/17 → 4/12/19

Project: Research

Cite this

Chicchón Apaza, M. Á., Monzón, H. M. B., & Garrido, R. P. A. (2020). Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net. In M. Botto-Tobar, M. Zambrano Vizuete, P. Torres-Carrión, S. Montes León, G. Pizarro Vásquez, & B. Durakovic (Eds.), Applied Technologies - 1st International Conference, ICAT 2019, Proceedings (pp. 473-485). (Communications in Computer and Information Science; Vol. 1194 CCIS). Springer. https://doi.org/10.1007/978-3-030-42520-3_38

Chicchón Apaza, Miguel Ángel ; Monzón, Héctor Manuel Bedón ; Garrido, Ramón Pablo Alcarria. / Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net. Applied Technologies - 1st International Conference, ICAT 2019, Proceedings. editor / Miguel Botto-Tobar ; Marcelo Zambrano Vizuete ; Pablo Torres-Carrión ; Sergio Montes León ; Guillermo Pizarro Vásquez ; Benjamin Durakovic. Springer, 2020. pp. 473-485 (Communications in Computer and Information Science).

@inproceedings{1e4978a22572452ab37a2d525dbf7fe8,

title = "Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net",

abstract = "A first step in the process of automating weed removal in precision agriculture is the semantic segmentation of crops, weeds and soil. Deep learning techniques based on convolutional neural networks are successfully applied today and one of the most popular network architectures in semantic segmentation problems is U-Net. In this article, the variants in the U-Net architecture were evaluated based on the aggregation of residual and recurring blocks to improve their performance. For training and testing, a set of data available on the Internet was used, consisting of 60 multispectral images with unbalanced pixels, so techniques were applied to increase and balance the data. Experimental results show a slight increase in quality metrics compared to the classic U-Net architecture.",

keywords = "Convolutional neural network, Deep learning, Precision agriculture, Semantic segmentation, U-Net",

author = "{Chicch{\'o}n Apaza}, {Miguel {\'A}ngel} and Monz{\'o}n, {H{\'e}ctor Manuel Bed{\'o}n} and Garrido, {Ram{\'o}n Pablo Alcarria}",

year = "2020",

month = jan,

day = "1",

doi = "10.1007/978-3-030-42520-3_38",

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

isbn = "9783030425197",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "473--485",

editor = "Miguel Botto-Tobar and {Zambrano Vizuete}, Marcelo and Pablo Torres-Carri{\'o}n and {Montes Le{\'o}n}, Sergio and {Pizarro V{\'a}squez}, Guillermo and Benjamin Durakovic",

booktitle = "Applied Technologies - 1st International Conference, ICAT 2019, Proceedings",

note = "null ; Conference date: 03-12-2019 Through 05-12-2019",

}

Chicchón Apaza, MÁ, Monzón, HMB & Garrido, RPA 2020, Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net. in M Botto-Tobar, M Zambrano Vizuete, P Torres-Carrión, S Montes León, G Pizarro Vásquez & B Durakovic (eds), Applied Technologies - 1st International Conference, ICAT 2019, Proceedings. Communications in Computer and Information Science, vol. 1194 CCIS, Springer, pp. 473-485, 1st International Conference on Applied Technologies, ICAT 2019, Quito, Ecuador, 3/12/19. https://doi.org/10.1007/978-3-030-42520-3_38

Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net. / Chicchón Apaza, Miguel Ángel; Monzón, Héctor Manuel Bedón; Garrido, Ramón Pablo Alcarria.

Applied Technologies - 1st International Conference, ICAT 2019, Proceedings. ed. / Miguel Botto-Tobar; Marcelo Zambrano Vizuete; Pablo Torres-Carrión; Sergio Montes León; Guillermo Pizarro Vásquez; Benjamin Durakovic. Springer, 2020. p. 473-485 (Communications in Computer and Information Science; Vol. 1194 CCIS).

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

TY - GEN

T1 - Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net

AU - Chicchón Apaza, Miguel Ángel

AU - Monzón, Héctor Manuel Bedón

AU - Garrido, Ramón Pablo Alcarria

PY - 2020/1/1

Y1 - 2020/1/1

N2 - A first step in the process of automating weed removal in precision agriculture is the semantic segmentation of crops, weeds and soil. Deep learning techniques based on convolutional neural networks are successfully applied today and one of the most popular network architectures in semantic segmentation problems is U-Net. In this article, the variants in the U-Net architecture were evaluated based on the aggregation of residual and recurring blocks to improve their performance. For training and testing, a set of data available on the Internet was used, consisting of 60 multispectral images with unbalanced pixels, so techniques were applied to increase and balance the data. Experimental results show a slight increase in quality metrics compared to the classic U-Net architecture.

AB - A first step in the process of automating weed removal in precision agriculture is the semantic segmentation of crops, weeds and soil. Deep learning techniques based on convolutional neural networks are successfully applied today and one of the most popular network architectures in semantic segmentation problems is U-Net. In this article, the variants in the U-Net architecture were evaluated based on the aggregation of residual and recurring blocks to improve their performance. For training and testing, a set of data available on the Internet was used, consisting of 60 multispectral images with unbalanced pixels, so techniques were applied to increase and balance the data. Experimental results show a slight increase in quality metrics compared to the classic U-Net architecture.

KW - Convolutional neural network

KW - Deep learning

KW - Precision agriculture

KW - Semantic segmentation

KW - U-Net

UR - http://www.scopus.com/inward/record.url?scp=85082396508&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-42520-3_38

DO - 10.1007/978-3-030-42520-3_38

M3 - Articulo (Contribución a conferencia)

AN - SCOPUS:85082396508

SN - 9783030425197

T3 - Communications in Computer and Information Science

SP - 473

EP - 485

BT - Applied Technologies - 1st International Conference, ICAT 2019, Proceedings

A2 - Botto-Tobar, Miguel

A2 - Zambrano Vizuete, Marcelo

A2 - Torres-Carrión, Pablo

A2 - Montes León, Sergio

A2 - Pizarro Vásquez, Guillermo

A2 - Durakovic, Benjamin

PB - Springer

Y2 - 3 December 2019 through 5 December 2019

ER -

Chicchón Apaza MÁ, Monzón HMB, Garrido RPA. Semantic Segmentation of Weeds and Crops in Multispectral Images by Using a Convolutional Neural Networks Based on U-Net. In Botto-Tobar M, Zambrano Vizuete M, Torres-Carrión P, Montes León S, Pizarro Vásquez G, Durakovic B, editors, Applied Technologies - 1st International Conference, ICAT 2019, Proceedings. Springer. 2020. p. 473-485. (Communications in Computer and Information Science). https://doi.org/10.1007/978-3-030-42520-3_38