Academic Journal
Objective evaluation-based efficient learning framework for hyperspectral image classification
| العنوان: | Objective evaluation-based efficient learning framework for hyperspectral image classification |
|---|---|
| المؤلفون: | Xuming Zhang, Jian Yan, Jia Tian, Wei Li, Xingfa Gu, Qingjiu Tian |
| المصدر: | GIScience & Remote Sensing, Vol 60, Iss 1 (2023) |
| بيانات النشر: | Taylor & Francis Group, 2023. |
| سنة النشر: | 2023 |
| المجموعة: | LCC:Mathematical geography. Cartography LCC:Environmental sciences |
| مصطلحات موضوعية: | deep learning, fully convolutional network, features extraction, sampling strategy, Mathematical geography. Cartography, GA1-1776, Environmental sciences, GE1-350 |
| الوصف: | Deep learning techniques with remarkable performance have been successfully applied to hyperspectral image (HSI) classification. Due to the limited availability of training data, earlier studies primarily adopted the patch-based classification framework, which divides images into overlapping patches for training and testing. However, this framework results in redundant computations and possible information leakage. This study proposes an objective evaluation-based efficient learning framework for HSI classification. It consists of two main parts: (i) a leakage-free balanced sampling strategy and (ii) an efficient fully convolutional network (EfficientFCN) optimized for the accuracy-efficiency trade-off. The leakage-free balanced sampling strategy first generates balanced and non-overlapping training and test data by partitioning the HSI and its ground truth image into non-overlapping windows. Then, the generated training and test data are used to train and test the proposed EfficientFCN. EfficientFCN exhibits a pixel-to-pixel architecture with modifications for faster inference speed and improved parameter efficiency. Experimental results demonstrate that the proposed sampling strategy can provide objective performance evaluation. EfficientFCN outperforms many state-of-the-art approaches concerning the speed-accuracy trade-off. For instance, compared to the recent efficient models EfficientNetV2 and ConvNeXt, EfficientFCN achieves 0.92% and 3.42% superior accuracy and 0.19s and 0.16s faster inference time, respectively, on the Houston dataset. Code is available at https://github.com/xmzhang2018. |
| نوع الوثيقة: | article |
| وصف الملف: | electronic resource |
| اللغة: | English |
| تدمد: | 1548-1603 1943-7226 15481603 |
| Relation: | https://doaj.org/toc/1548-1603; https://doaj.org/toc/1943-7226 |
| DOI: | 10.1080/15481603.2023.2225273 |
| URL الوصول: | https://doaj.org/article/870e603544d640aeb8a29b59cd86e227 |
| رقم الانضمام: | edsdoj.870e603544d640aeb8a29b59cd86e227 |
| قاعدة البيانات: | Directory of Open Access Journals |
| ResultId |
1 |
|---|---|
| Header |
edsdoj Directory of Open Access Journals edsdoj.870e603544d640aeb8a29b59cd86e227 994 3 Academic Journal academicJournal 994.193542480469 |
| PLink |
https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&scope=site&db=edsdoj&AN=edsdoj.870e603544d640aeb8a29b59cd86e227&custid=s6537998&authtype=sso |
| FullText |
Array
(
[Availability] => 0
)
Array ( [0] => Array ( [Url] => https://doaj.org/article/870e603544d640aeb8a29b59cd86e227 [Name] => EDS - DOAJ [Category] => fullText [Text] => View record in DOAJ [MouseOverText] => View record in DOAJ ) ) |
| Items |
Array
(
[Name] => Title
[Label] => Title
[Group] => Ti
[Data] => Objective evaluation-based efficient learning framework for hyperspectral image classification
)
Array ( [Name] => Author [Label] => Authors [Group] => Au [Data] => <searchLink fieldCode="AR" term="%22Xuming+Zhang%22">Xuming Zhang</searchLink><br /><searchLink fieldCode="AR" term="%22Jian+Yan%22">Jian Yan</searchLink><br /><searchLink fieldCode="AR" term="%22Jia+Tian%22">Jia Tian</searchLink><br /><searchLink fieldCode="AR" term="%22Wei+Li%22">Wei Li</searchLink><br /><searchLink fieldCode="AR" term="%22Xingfa+Gu%22">Xingfa Gu</searchLink><br /><searchLink fieldCode="AR" term="%22Qingjiu+Tian%22">Qingjiu Tian</searchLink> ) Array ( [Name] => TitleSource [Label] => Source [Group] => Src [Data] => GIScience & Remote Sensing, Vol 60, Iss 1 (2023) ) Array ( [Name] => Publisher [Label] => Publisher Information [Group] => PubInfo [Data] => Taylor & Francis Group, 2023. ) Array ( [Name] => DatePubCY [Label] => Publication Year [Group] => Date [Data] => 2023 ) Array ( [Name] => Subset [Label] => Collection [Group] => HoldingsInfo [Data] => LCC:Mathematical geography. Cartography<br />LCC:Environmental sciences ) Array ( [Name] => Subject [Label] => Subject Terms [Group] => Su [Data] => <searchLink fieldCode="DE" term="%22deep+learning%22">deep learning</searchLink><br /><searchLink fieldCode="DE" term="%22fully+convolutional+network%22">fully convolutional network</searchLink><br /><searchLink fieldCode="DE" term="%22features+extraction%22">features extraction</searchLink><br /><searchLink fieldCode="DE" term="%22sampling+strategy%22">sampling strategy</searchLink><br /><searchLink fieldCode="DE" term="%22Mathematical+geography%2E+Cartography%22">Mathematical geography. Cartography</searchLink><br /><searchLink fieldCode="DE" term="%22GA1-1776%22">GA1-1776</searchLink><br /><searchLink fieldCode="DE" term="%22Environmental+sciences%22">Environmental sciences</searchLink><br /><searchLink fieldCode="DE" term="%22GE1-350%22">GE1-350</searchLink> ) Array ( [Name] => Abstract [Label] => Description [Group] => Ab [Data] => Deep learning techniques with remarkable performance have been successfully applied to hyperspectral image (HSI) classification. Due to the limited availability of training data, earlier studies primarily adopted the patch-based classification framework, which divides images into overlapping patches for training and testing. However, this framework results in redundant computations and possible information leakage. This study proposes an objective evaluation-based efficient learning framework for HSI classification. It consists of two main parts: (i) a leakage-free balanced sampling strategy and (ii) an efficient fully convolutional network (EfficientFCN) optimized for the accuracy-efficiency trade-off. The leakage-free balanced sampling strategy first generates balanced and non-overlapping training and test data by partitioning the HSI and its ground truth image into non-overlapping windows. Then, the generated training and test data are used to train and test the proposed EfficientFCN. EfficientFCN exhibits a pixel-to-pixel architecture with modifications for faster inference speed and improved parameter efficiency. Experimental results demonstrate that the proposed sampling strategy can provide objective performance evaluation. EfficientFCN outperforms many state-of-the-art approaches concerning the speed-accuracy trade-off. For instance, compared to the recent efficient models EfficientNetV2 and ConvNeXt, EfficientFCN achieves 0.92% and 3.42% superior accuracy and 0.19s and 0.16s faster inference time, respectively, on the Houston dataset. Code is available at https://github.com/xmzhang2018. ) Array ( [Name] => TypeDocument [Label] => Document Type [Group] => TypDoc [Data] => article ) Array ( [Name] => Format [Label] => File Description [Group] => SrcInfo [Data] => electronic resource ) Array ( [Name] => Language [Label] => Language [Group] => Lang [Data] => English ) Array ( [Name] => ISSN [Label] => ISSN [Group] => ISSN [Data] => 1548-1603<br />1943-7226<br />15481603 ) Array ( [Name] => NoteTitleSource [Label] => Relation [Group] => SrcInfo [Data] => https://doaj.org/toc/1548-1603; https://doaj.org/toc/1943-7226 ) Array ( [Name] => DOI [Label] => DOI [Group] => ID [Data] => 10.1080/15481603.2023.2225273 ) Array ( [Name] => URL [Label] => Access URL [Group] => URL [Data] => <link linkTarget="URL" linkTerm="https://doaj.org/article/870e603544d640aeb8a29b59cd86e227" linkWindow="_blank">https://doaj.org/article/870e603544d640aeb8a29b59cd86e227</link> ) Array ( [Name] => AN [Label] => Accession Number [Group] => ID [Data] => edsdoj.870e603544d640aeb8a29b59cd86e227 ) |
| RecordInfo |
Array
(
[BibEntity] => Array
(
[Identifiers] => Array
(
[0] => Array
(
[Type] => doi
[Value] => 10.1080/15481603.2023.2225273
)
)
[Languages] => Array
(
[0] => Array
(
[Text] => English
)
)
[Subjects] => Array
(
[0] => Array
(
[SubjectFull] => deep learning
[Type] => general
)
[1] => Array
(
[SubjectFull] => fully convolutional network
[Type] => general
)
[2] => Array
(
[SubjectFull] => features extraction
[Type] => general
)
[3] => Array
(
[SubjectFull] => sampling strategy
[Type] => general
)
[4] => Array
(
[SubjectFull] => Mathematical geography. Cartography
[Type] => general
)
[5] => Array
(
[SubjectFull] => GA1-1776
[Type] => general
)
[6] => Array
(
[SubjectFull] => Environmental sciences
[Type] => general
)
[7] => Array
(
[SubjectFull] => GE1-350
[Type] => general
)
)
[Titles] => Array
(
[0] => Array
(
[TitleFull] => Objective evaluation-based efficient learning framework for hyperspectral image classification
[Type] => main
)
)
)
[BibRelationships] => Array
(
[HasContributorRelationships] => Array
(
[0] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Xuming Zhang
)
)
)
[1] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Jian Yan
)
)
)
[2] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Jia Tian
)
)
)
[3] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Wei Li
)
)
)
[4] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Xingfa Gu
)
)
)
[5] => Array
(
[PersonEntity] => Array
(
[Name] => Array
(
[NameFull] => Qingjiu Tian
)
)
)
)
[IsPartOfRelationships] => Array
(
[0] => Array
(
[BibEntity] => Array
(
[Dates] => Array
(
[0] => Array
(
[D] => 01
[M] => 12
[Type] => published
[Y] => 2023
)
)
[Identifiers] => Array
(
[0] => Array
(
[Type] => issn-print
[Value] => 15481603
)
[1] => Array
(
[Type] => issn-print
[Value] => 19437226
)
)
[Numbering] => Array
(
[0] => Array
(
[Type] => volume
[Value] => 60
)
[1] => Array
(
[Type] => issue
[Value] => 1
)
)
[Titles] => Array
(
[0] => Array
(
[TitleFull] => GIScience & Remote Sensing
[Type] => main
)
)
)
)
)
)
)
|
| IllustrationInfo |