Learning enhanced features and inferring twice for fine-grained image classification

Nie, X; Chai, B; Wang, L; Liao, Q; Xu, M

Learning enhanced features and inferring twice for fine-grained image classification

Nie, X Chai, B Wang, L Liao, Q Xu, M

Permalink

Publisher:: SPRINGER
Publication Type:: Journal Article
Citation:: Multimedia Tools and Applications, 2023, 82, (10), pp. 14799-14813
Issue Date:: 2023-04-01

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published versionAdobe PDF (1.06 MB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Nie, X
dc.contributor.author	Chai, B
dc.contributor.author	Wang, L
dc.contributor.author	Liao, Q
dc.contributor.author	Xu, M https://orcid.org/0000-0001-9581-8849
dc.date.accessioned	2023-04-12T03:42:50Z
dc.date.available	2023-04-12T03:42:50Z
dc.date.issued	2023-04-01
dc.identifier.citation	Multimedia Tools and Applications, 2023, 82, (10), pp. 14799-14813
dc.identifier.issn	1380-7501
dc.identifier.issn	1573-7721
dc.identifier.uri	http://hdl.handle.net/10453/169630
dc.description.abstract	Fine-Grained Visual Categorization (FGVC) aims to distinguish between extremely similar subordinate-level categories within the same basic-level category. Existing research has proven the great importance of the discriminative features in FGVC but ignored the contributions for correct classification from other features, and the extracted features always contain more information about the obvious regions but less about subtle regions. In this paper, firstly, a novel module named forcing module is proposed to force the network to extract more diverse features for FGVC, which generates a suppression mask based on the class activation maps to suppress the most distinguishable regions, so as to force the network to extract other secondary distinguishable features as the final features. The forcing module consists of the original branch and the forcing branch. The original branch focuses on the primary discriminative regions while the forcing branch focuses on secondary discriminative regions. Secondly, in order to solve the problem that information of small-scale distinguishable features is lost seriously after multi-layer down-sampling, according to the class activation maps of the first prediction, the object is cropped and scaled as the second input. To reduce the prediction error, the first and second prediction probabilities are fused as the final prediction result. Experimental results indicate that the proposed method not only outperforms the baseline model by a large margin (3.7%, 5.9%, 3.1% respectively) on CUB-200-2011, Stanford-Cars, and FGVC-Aircraft, but also achieves state-of-the-art performance on FGVC-Aircraft.
dc.language	English
dc.publisher	SPRINGER
dc.relation.ispartof	Multimedia Tools and Applications
dc.relation.isbasedon	10.1007/s11042-022-13619-z
dc.rights	info:eu-repo/semantics/openAccess
dc.subject	0801 Artificial Intelligence and Image Processing, 0803 Computer Software, 0805 Distributed Computing, 0806 Information Systems
dc.subject.classification	Artificial Intelligence & Image Processing
dc.subject.classification	Software Engineering
dc.title	Learning enhanced features and inferring twice for fine-grained image classification
dc.type	Journal Article
utslib.citation.volume	82
utslib.for	0801 Artificial Intelligence and Image Processing
utslib.for	0803 Computer Software
utslib.for	0805 Distributed Computing
utslib.for	0806 Information Systems
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Strength - INEXT - Innovation in IT Services and Applications
pubs.organisational-group	/University of Technology Sydney/Strength - GBDTC - Global Big Data Technologies
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/School of Electrical and Data Engineering
utslib.copyright.status	open_access	*
dc.date.updated	2023-04-12T03:42:49Z
pubs.issue	10
pubs.publication-status	Published
pubs.volume	82
utslib.citation.issue	10

Abstract:

Fine-Grained Visual Categorization (FGVC) aims to distinguish between extremely similar subordinate-level categories within the same basic-level category. Existing research has proven the great importance of the discriminative features in FGVC but ignored the contributions for correct classification from other features, and the extracted features always contain more information about the obvious regions but less about subtle regions. In this paper, firstly, a novel module named forcing module is proposed to force the network to extract more diverse features for FGVC, which generates a suppression mask based on the class activation maps to suppress the most distinguishable regions, so as to force the network to extract other secondary distinguishable features as the final features. The forcing module consists of the original branch and the forcing branch. The original branch focuses on the primary discriminative regions while the forcing branch focuses on secondary discriminative regions. Secondly, in order to solve the problem that information of small-scale distinguishable features is lost seriously after multi-layer down-sampling, according to the class activation maps of the first prediction, the object is cropped and scaled as the second input. To reduce the prediction error, the first and second prediction probabilities are fused as the final prediction result. Experimental results indicate that the proposed method not only outperforms the baseline model by a large margin (3.7%, 5.9%, 3.1% respectively) on CUB-200-2011, Stanford-Cars, and FGVC-Aircraft, but also achieves state-of-the-art performance on FGVC-Aircraft.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/169630