Cargando…

Dynamic Color Transform Networks for Wheat Head Detection

Wheat head detection can measure wheat traits such as head density and head characteristics. Standard wheat breeding largely relies on manual observation to detect wheat heads, yielding a tedious and inefficient procedure. The emergence of affordable camera platforms provides opportunities for deplo...

Descripción completa

Detalles Bibliográficos
Autores principales: Liu, Chengxin, Wang, Kewei, Lu, Hao, Cao, Zhiguo
Formato: Online Artículo Texto
Lenguaje:English
Publicado: AAAS 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8829536/
https://www.ncbi.nlm.nih.gov/pubmed/35198987
http://dx.doi.org/10.34133/2022/9818452
_version_ 1784648103749484544
author Liu, Chengxin
Wang, Kewei
Lu, Hao
Cao, Zhiguo
author_facet Liu, Chengxin
Wang, Kewei
Lu, Hao
Cao, Zhiguo
author_sort Liu, Chengxin
collection PubMed
description Wheat head detection can measure wheat traits such as head density and head characteristics. Standard wheat breeding largely relies on manual observation to detect wheat heads, yielding a tedious and inefficient procedure. The emergence of affordable camera platforms provides opportunities for deploying computer vision (CV) algorithms in wheat head detection, enabling automated measurements of wheat traits. Accurate wheat head detection, however, is challenging due to the variability of observation circumstances and the uncertainty of wheat head appearances. In this work, we propose a simple but effective idea—dynamic color transform (DCT)—for accurate wheat head detection. This idea is based on an observation that modifying the color channel of an input image can significantly alleviate false negatives and therefore improve detection results. DCT follows a linear color transform and can be easily implemented as a dynamic network. A key property of DCT is that the transform parameters are data-dependent such that illumination variations can be corrected adaptively. The DCT network can be incorporated into any existing object detectors. Experimental results on the Global Wheat Detection Dataset (GWHD) 2021 show that DCT can achieve notable improvements with negligible overhead parameters. In addition, DCT plays an important role in our solution participating in the Global Wheat Challenge (GWC) 2021, where our solution ranks the first on the initial public leaderboard, with an Average Domain Accuracy (ADA) of 0.821, and obtains the runner-up reward on the final private testing set, with an ADA of 0.695.
format Online
Article
Text
id pubmed-8829536
institution National Center for Biotechnology Information
language English
publishDate 2022
publisher AAAS
record_format MEDLINE/PubMed
spelling pubmed-88295362022-02-22 Dynamic Color Transform Networks for Wheat Head Detection Liu, Chengxin Wang, Kewei Lu, Hao Cao, Zhiguo Plant Phenomics Research Article Wheat head detection can measure wheat traits such as head density and head characteristics. Standard wheat breeding largely relies on manual observation to detect wheat heads, yielding a tedious and inefficient procedure. The emergence of affordable camera platforms provides opportunities for deploying computer vision (CV) algorithms in wheat head detection, enabling automated measurements of wheat traits. Accurate wheat head detection, however, is challenging due to the variability of observation circumstances and the uncertainty of wheat head appearances. In this work, we propose a simple but effective idea—dynamic color transform (DCT)—for accurate wheat head detection. This idea is based on an observation that modifying the color channel of an input image can significantly alleviate false negatives and therefore improve detection results. DCT follows a linear color transform and can be easily implemented as a dynamic network. A key property of DCT is that the transform parameters are data-dependent such that illumination variations can be corrected adaptively. The DCT network can be incorporated into any existing object detectors. Experimental results on the Global Wheat Detection Dataset (GWHD) 2021 show that DCT can achieve notable improvements with negligible overhead parameters. In addition, DCT plays an important role in our solution participating in the Global Wheat Challenge (GWC) 2021, where our solution ranks the first on the initial public leaderboard, with an Average Domain Accuracy (ADA) of 0.821, and obtains the runner-up reward on the final private testing set, with an ADA of 0.695. AAAS 2022-02-01 /pmc/articles/PMC8829536/ /pubmed/35198987 http://dx.doi.org/10.34133/2022/9818452 Text en Copyright © 2022 Chengxin Liu et al. https://creativecommons.org/licenses/by/4.0/Exclusive Licensee Nanjing Agricultural University. Distributed under a Creative Commons Attribution License (CC BY 4.0).
spellingShingle Research Article
Liu, Chengxin
Wang, Kewei
Lu, Hao
Cao, Zhiguo
Dynamic Color Transform Networks for Wheat Head Detection
title Dynamic Color Transform Networks for Wheat Head Detection
title_full Dynamic Color Transform Networks for Wheat Head Detection
title_fullStr Dynamic Color Transform Networks for Wheat Head Detection
title_full_unstemmed Dynamic Color Transform Networks for Wheat Head Detection
title_short Dynamic Color Transform Networks for Wheat Head Detection
title_sort dynamic color transform networks for wheat head detection
topic Research Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8829536/
https://www.ncbi.nlm.nih.gov/pubmed/35198987
http://dx.doi.org/10.34133/2022/9818452
work_keys_str_mv AT liuchengxin dynamiccolortransformnetworksforwheatheaddetection
AT wangkewei dynamiccolortransformnetworksforwheatheaddetection
AT luhao dynamiccolortransformnetworksforwheatheaddetection
AT caozhiguo dynamiccolortransformnetworksforwheatheaddetection