skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Tree, Shrub, and Grass Classification Using Only RGB Images

Abstract

In this work, a semantic segmentation-based deep learning method, DeepLabV3+, is applied to classify three vegetation land covers, which are tree, shrub, and grass using only three band color (RGB) images. DeepLabV3+’s detection performance has been studied on low and high resolution datasets that both contain tree, shrub, and grass and some other land cover types. The two datasets are heavily imbalanced where shrub pixels are much fewer than tree and grass pixels. A simple weighting strategy known as median frequency weighting was incorporated into DeepLabV3+ to mitigate the data imbalance issue, which originally used uniform weights. The tree, shrub, grass classification performances are compared when all land cover types are included in the classification and also when classification is limited to the three vegetation classes with both uniform and median frequency weights. Among the three vegetation types, shrub is found to be the most challenging one to classify correctly whereas correct classification accuracy was highest for tree. It is observed that even though the median frequency weighting did not improve the overall accuracy, it resulted in better classification accuracy for the underrepresented classes such as shrub in our case and it also significantly increased the average class accuracy. Themore » classification performance and computation time comparison of DeepLabV3+ with two other pixel-based classification methods on sampled pixels of the three vegetation classes showed that DeepLabV3+ achieves significantly higher accuracy than these methods with a trade-off for longer model training time.« less

Authors:
 [1]; ORCiD logo [1]
  1. Applied Research LLC, Rockville, MD (United States)
Publication Date:
Research Org.:
Applied Research LLC, Rockville, MD (United States)
Sponsoring Org.:
USDOE Office of Science (SC)
OSTI Identifier:
1668666
Grant/Contract Number:  
SC0019936
Resource Type:
Accepted Manuscript
Journal Name:
Remote Sensing
Additional Journal Information:
Journal Volume: 12; Journal Issue: 8; Journal ID: ISSN 2072-4292
Publisher:
MDPI
Country of Publication:
United States
Language:
English
Subject:
58 GEOSCIENCES; deep learning; vegetation classification; imbalanced data; median frequency weighting; DeepLabV3+

Citation Formats

Ayhan, Bulent, and Kwan, Chiman. Tree, Shrub, and Grass Classification Using Only RGB Images. United States: N. p., 2020. Web. https://doi.org/10.3390/rs12081333.
Ayhan, Bulent, & Kwan, Chiman. Tree, Shrub, and Grass Classification Using Only RGB Images. United States. https://doi.org/10.3390/rs12081333
Ayhan, Bulent, and Kwan, Chiman. Thu . "Tree, Shrub, and Grass Classification Using Only RGB Images". United States. https://doi.org/10.3390/rs12081333. https://www.osti.gov/servlets/purl/1668666.
@article{osti_1668666,
title = {Tree, Shrub, and Grass Classification Using Only RGB Images},
author = {Ayhan, Bulent and Kwan, Chiman},
abstractNote = {In this work, a semantic segmentation-based deep learning method, DeepLabV3+, is applied to classify three vegetation land covers, which are tree, shrub, and grass using only three band color (RGB) images. DeepLabV3+’s detection performance has been studied on low and high resolution datasets that both contain tree, shrub, and grass and some other land cover types. The two datasets are heavily imbalanced where shrub pixels are much fewer than tree and grass pixels. A simple weighting strategy known as median frequency weighting was incorporated into DeepLabV3+ to mitigate the data imbalance issue, which originally used uniform weights. The tree, shrub, grass classification performances are compared when all land cover types are included in the classification and also when classification is limited to the three vegetation classes with both uniform and median frequency weights. Among the three vegetation types, shrub is found to be the most challenging one to classify correctly whereas correct classification accuracy was highest for tree. It is observed that even though the median frequency weighting did not improve the overall accuracy, it resulted in better classification accuracy for the underrepresented classes such as shrub in our case and it also significantly increased the average class accuracy. The classification performance and computation time comparison of DeepLabV3+ with two other pixel-based classification methods on sampled pixels of the three vegetation classes showed that DeepLabV3+ achieves significantly higher accuracy than these methods with a trade-off for longer model training time.},
doi = {10.3390/rs12081333},
journal = {Remote Sensing},
number = 8,
volume = 12,
place = {United States},
year = {2020},
month = {4}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Save / Share:

Works referenced in this record:

Object-Based Change Detection Using Multiple Classifiers and Multi-Scale Uncertainty Analysis
journal, February 2019

  • Tan, Kun; Zhang, Yusha; Wang, Xue
  • Remote Sensing, Vol. 11, Issue 3
  • DOI: 10.3390/rs11030359

An Object-Based Approach for Mapping Shrub and Tree Cover on Grassland Habitats by Use of LiDAR and CIR Orthoimages
journal, January 2013

  • Hellesen, Thomas; Matikainen, Leena
  • Remote Sensing, Vol. 5, Issue 2
  • DOI: 10.3390/rs5020558

Semi-Automated Emergency Landing Site Selection Approach for UAVs
journal, August 2019

  • Ayhan, Bulent; Kwan, Chiman; Um, Yool-Bin
  • IEEE Transactions on Aerospace and Electronic Systems, Vol. 55, Issue 4
  • DOI: 10.1109/TAES.2018.2879529

Deep-learning Versus OBIA for Scattered Shrub Detection with Google Earth Imagery: Ziziphus lotus as Case Study
journal, November 2017

  • Guirado, Emilio; Tabik, Siham; Alcaraz-Segura, Domingo
  • Remote Sensing, Vol. 9, Issue 12
  • DOI: 10.3390/rs9121220

Estimation of shrub height for fuel-type mapping combining airborne LiDAR and simultaneous color infrared ortho imaging
journal, January 2007

  • Riaño, David; Chuvieco, Emilio; Ustin, Susan L.
  • International Journal of Wildland Fire, Vol. 16, Issue 3
  • DOI: 10.1071/WF06003

Progressive Cascaded Convolutional Neural Networks for Single Tree Detection with Google Earth Imagery
journal, July 2019

  • Dong, Tianyang; Shen, Yuqi; Zhang, Jian
  • Remote Sensing, Vol. 11, Issue 15
  • DOI: 10.3390/rs11151786

Urban land-use mapping using a deep convolutional neural network with high spatial resolution multispectral remote sensing imagery
journal, September 2018


Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning
journal, November 2018


Semantic Segmentation of Remote Sensing Imagery Using Object-Based Markov Random Field Model With Regional Penalties
journal, May 2015

  • Zheng, Chen; Wang, Leiguang
  • IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Vol. 8, Issue 5
  • DOI: 10.1109/JSTARS.2014.2361756

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
journal, April 2018

  • Chen, Liang-Chieh; Papandreou, George; Kokkinos, Iasonas
  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 40, Issue 4
  • DOI: 10.1109/TPAMI.2017.2699184

SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
journal, December 2017

  • Badrinarayanan, Vijay; Kendall, Alex; Cipolla, Roberto
  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 39, Issue 12
  • DOI: 10.1109/TPAMI.2016.2644615

Smallholder Crop Area Mapped with a Semantic Segmentation Deep Learning Method
journal, April 2019

  • Du, Zhenrong; Yang, Jianyu; Ou, Cong
  • Remote Sensing, Vol. 11, Issue 7
  • DOI: 10.3390/rs11070888

LIBSVM: A library for support vector machines
journal, April 2011

  • Chang, Chih-Chung; Lin, Chih-Jen
  • ACM Transactions on Intelligent Systems and Technology, Vol. 2, Issue 3
  • DOI: 10.1145/1961189.1961199