A fully convolutional network for weed mapping of unmanned aerial vehicle (UAV) imagery

Huasheng Huang; Jizhong Deng; Yubin Lan; Aqing Yang; Xiaoling Deng; Lei Zhang

doi:10.1371/journal.pone.0196302

A fully convolutional network for weed mapping of unmanned aerial vehicle (UAV) imagery

PLoS One. 2018 Apr 26;13(4):e0196302. doi: 10.1371/journal.pone.0196302. eCollection 2018.

Authors

Huasheng Huang^{1

2}, Jizhong Deng^{1

2}, Yubin Lan^{1

2}, Aqing Yang³, Xiaoling Deng^{2

3}, Lei Zhang^{2

4}

Affiliations

¹ College of Engineering, South China Agricultural University, Guangzhou, China.
² National Center for International Collaboration Research on Precision Agricultural Aviation Pesticide Spraying Technology, Guangzhou, China.
³ College of Electronical Engineering, South China Agricultural University, Guangzhou, China.
⁴ College of Agriculture, South China Agricultural University, Guangzhou, China.

Abstract

Appropriate Site Specific Weed Management (SSWM) is crucial to ensure the crop yields. Within SSWM of large-scale area, remote sensing is a key technology to provide accurate weed distribution information. Compared with satellite and piloted aircraft remote sensing, unmanned aerial vehicle (UAV) is capable of capturing high spatial resolution imagery, which will provide more detailed information for weed mapping. The objective of this paper is to generate an accurate weed cover map based on UAV imagery. The UAV RGB imagery was collected in 2017 October over the rice field located in South China. The Fully Convolutional Network (FCN) method was proposed for weed mapping of the collected imagery. Transfer learning was used to improve generalization capability, and skip architecture was applied to increase the prediction accuracy. After that, the performance of FCN architecture was compared with Patch_based CNN algorithm and Pixel_based CNN method. Experimental results showed that our FCN method outperformed others, both in terms of accuracy and efficiency. The overall accuracy of the FCN approach was up to 0.935 and the accuracy for weed recognition was 0.883, which means that this algorithm is capable of generating accurate weed cover maps for the evaluated UAV imagery.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Agriculture / methods*
Aircraft
Algorithms
China
Imagery, Psychotherapy
Models, Statistical
Oryza / physiology*
Plant Weeds*
Plants*
Remote Sensing Technology / methods*
Software

Associated data

figshare/5739669

Grants and funding

This work was supported by: 1. the National Key Research and Development Plan: High Efficient Ground and Aerial Spraying Technology and Intelligent Equipment, China; grant number: 2016YFD0200700; http://service.most.gov.cn/2015tztg_all/20160622/1108.html; funding institution: Ministry of Agriculture of the People's Republic of China; the author receiving the funding: Yubin Lan; role of funder in this study: study design, decision to publish. 2: Science and Technology Planning Project of Guangdong Province, China; grant number: 2017A020208046; http://www.gdstc.gov.cn/; funding institution: Department of Science and Technology of Guangdong Province, China; the author receiving the funding: Jizhong Deng; role of funder in this study: study design, decision to publish. 3: The National Natural Science Fund, China; grant number: 61675003; http://www.nsfc.gov.cn/publish/portal0/; funding institution: National Natural Science Foundation of China; the author receiving the funding: Xiaoling Deng; role of funder in this study: study design, preparation of the manuscript.