On Learning Discriminative Features from Synthesized Data for Self-Supervised Fine-Grained Visual Recognition Article Swipe

PDF

Zihu Wang , Lingqiao Liu , Scott Ricardo Figueroa Weston , Sibo Tian , Peng Li ·

YOU? · · 2024 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2407.14676

Self-Supervised Learning (SSL) has become a prominent approach for acquiring visual representations across various tasks, yet its application in fine-grained visual recognition (FGVR) is challenged by the intricate task of distinguishing subtle differences between categories. To overcome this, we introduce an novel strategy that boosts SSL's ability to extract critical discriminative features vital for FGVR. This approach creates synthesized data pairs to guide the model to focus on discriminative features critical for FGVR during SSL. We start by identifying non-discriminative features using two main criteria: features with low variance that fail to effectively separate data and those deemed less important by Grad-CAM induced from the SSL loss. We then introduce perturbations to these non-discriminative features while preserving discriminative ones. A decoder is employed to reconstruct images from both perturbed and original feature vectors to create data pairs. An encoder is trained on such generated data pairs to become invariant to variations in non-discriminative dimensions while focusing on discriminative features, thereby improving the model's performance in FGVR tasks. We demonstrate the promising FGVR performance of the proposed approach through extensive evaluation on a wide variety of datasets.

Related Topics

Computer Science

Artificial Intelligence

Machine Learning

Concepts

Discriminative model Computer science Pattern recognition (psychology) Artificial intelligence Machine learning

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2407.14676
PDF: https://arxiv.org/pdf/2407.14676
OA Status: green
Related Works: 10
OpenAlex ID: https://openalex.org/W4406072111

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4406072111

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2407.14676

Digital Object Identifier
Title: On Learning Discriminative Features from Synthesized Data for Self-Supervised Fine-Grained Visual Recognition

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2024

Year of publication
Publication date: 2024-07-19

Full publication date if available
Authors: Zihu Wang, Lingqiao Liu, Scott Ricardo Figueroa Weston, Sibo Tian, Peng Li

List of authors in order
Landing page: https://arxiv.org/abs/2407.14676

Publisher landing page
PDF URL: https://arxiv.org/pdf/2407.14676

Direct link to full text PDF
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2407.14676

Direct OA link when available
Concepts: Discriminative model, Computer science, Pattern recognition (psychology), Artificial intelligence, Machine learning

Top concepts (fields/topics) attached by OpenAlex
Cited by: 0

Total citation count in OpenAlex
Related works (count): 10

Other works algorithmically related by OpenAlex

Full payload

id	https://openalex.org/W4406072111
doi	https://doi.org/10.48550/arxiv.2407.14676
ids.doi	https://doi.org/10.48550/arxiv.2407.14676
ids.openalex	https://openalex.org/W4406072111
fwci	0.0
type	preprint
title	On Learning Discriminative Features from Synthesized Data for Self-Supervised Fine-Grained Visual Recognition
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T12111
topics[0].field.id	https://openalex.org/fields/22
topics[0].field.display_name	Engineering
topics[0].score	0.6485999822616577
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/2209
topics[0].subfield.display_name	Industrial and Manufacturing Engineering
topics[0].display_name	Industrial Vision Systems and Defect Detection
is_xpac	False
apc_list
apc_paid
concepts[0].id	https://openalex.org/C97931131
concepts[0].level	2
concepts[0].score	0.8585612773895264
concepts[0].wikidata	https://www.wikidata.org/wiki/Q5282087
concepts[0].display_name	Discriminative model
concepts[1].id	https://openalex.org/C41008148
concepts[1].level	0
concepts[1].score	0.6667032241821289
concepts[1].wikidata	https://www.wikidata.org/wiki/Q21198
concepts[1].display_name	Computer science
concepts[2].id	https://openalex.org/C153180895
concepts[2].level	2
concepts[2].score	0.6298508644104004
concepts[2].wikidata	https://www.wikidata.org/wiki/Q7148389
concepts[2].display_name	Pattern recognition (psychology)
concepts[3].id	https://openalex.org/C154945302
concepts[3].level	1
concepts[3].score	0.6116368174552917
concepts[3].wikidata	https://www.wikidata.org/wiki/Q11660
concepts[3].display_name	Artificial intelligence
concepts[4].id	https://openalex.org/C119857082
concepts[4].level	1
concepts[4].score	0.33897897601127625
concepts[4].wikidata	https://www.wikidata.org/wiki/Q2539
concepts[4].display_name	Machine learning
keywords[0].id	https://openalex.org/keywords/discriminative-model
keywords[0].score	0.8585612773895264
keywords[0].display_name	Discriminative model
keywords[1].id	https://openalex.org/keywords/computer-science
keywords[1].score	0.6667032241821289
keywords[1].display_name	Computer science
keywords[2].id	https://openalex.org/keywords/pattern-recognition
keywords[2].score	0.6298508644104004
keywords[2].display_name	Pattern recognition (psychology)
keywords[3].id	https://openalex.org/keywords/artificial-intelligence
keywords[3].score	0.6116368174552917
keywords[3].display_name	Artificial intelligence
keywords[4].id	https://openalex.org/keywords/machine-learning
keywords[4].score	0.33897897601127625
keywords[4].display_name	Machine learning
language	en
locations[0].id	pmh:oai:arXiv.org:2407.14676
locations[0].is_oa	True
locations[0].source.id	https://openalex.org/S4306400194
locations[0].source.issn
locations[0].source.type	repository
locations[0].source.is_oa	True
locations[0].source.issn_l
locations[0].source.is_core	False
locations[0].source.is_in_doaj	False
locations[0].source.display_name	arXiv (Cornell University)
locations[0].source.host_organization	https://openalex.org/I205783295
locations[0].source.host_organization_name	Cornell University
locations[0].source.host_organization_lineage	https://openalex.org/I205783295
locations[0].license
locations[0].pdf_url	https://arxiv.org/pdf/2407.14676
locations[0].version	submittedVersion
locations[0].raw_type
locations[0].license_id
locations[0].is_accepted	False
locations[0].is_published	False
locations[0].raw_source_name
locations[0].landing_page_url	http://arxiv.org/abs/2407.14676
locations[1].id	doi:10.48550/arxiv.2407.14676
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published
locations[1].raw_source_name
locations[1].landing_page_url	https://doi.org/10.48550/arxiv.2407.14676
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5067730143
authorships[0].author.orcid	https://orcid.org/0009-0004-5132-2613
authorships[0].author.display_name	Zihu Wang
authorships[0].author_position	first
authorships[0].raw_author_name	Wang, Zihu
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5070976480
authorships[1].author.orcid	https://orcid.org/0000-0003-3584-795X
authorships[1].author.display_name	Lingqiao Liu
authorships[1].author_position	middle
authorships[1].raw_author_name	Liu, Lingqiao
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5114770071
authorships[2].author.orcid
authorships[2].author.display_name	Scott Ricardo Figueroa Weston
authorships[2].author_position	middle
authorships[2].raw_author_name	Weston, Scott Ricardo Figueroa
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5048786190
authorships[3].author.orcid	https://orcid.org/0000-0001-5018-4854
authorships[3].author.display_name	Sibo Tian
authorships[3].author_position	middle
authorships[3].raw_author_name	Tian, Samuel
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5100432782
authorships[4].author.orcid	https://orcid.org/0000-0003-3548-4589
authorships[4].author.display_name	Peng Li
authorships[4].author_position	last
authorships[4].raw_author_name	Li, Peng
authorships[4].is_corresponding	False
has_content.pdf	False
has_content.grobid_xml	False
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2407.14676
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	On Learning Discriminative Features from Synthesized Data for Self-Supervised Fine-Grained Visual Recognition
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T12111
primary_topic.field.id	https://openalex.org/fields/22
primary_topic.field.display_name	Engineering
primary_topic.score	0.6485999822616577
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/2209
primary_topic.subfield.display_name	Industrial and Manufacturing Engineering
primary_topic.display_name	Industrial Vision Systems and Defect Detection
related_works	https://openalex.org/W2961085424, https://openalex.org/W2404514746, https://openalex.org/W4306674287, https://openalex.org/W1652783584, https://openalex.org/W2082783427, https://openalex.org/W4387369504, https://openalex.org/W3046775127, https://openalex.org/W4394896187, https://openalex.org/W3170094116, https://openalex.org/W4386462264
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2407.14676
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2407.14676
best_oa_location.version	submittedVersion
best_oa_location.raw_type
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2407.14676
primary_location.id	pmh:oai:arXiv.org:2407.14676
primary_location.is_oa	True
primary_location.source.id	https://openalex.org/S4306400194
primary_location.source.issn
primary_location.source.type	repository
primary_location.source.is_oa	True
primary_location.source.issn_l
primary_location.source.is_core	False
primary_location.source.is_in_doaj	False
primary_location.source.display_name	arXiv (Cornell University)
primary_location.source.host_organization	https://openalex.org/I205783295
primary_location.source.host_organization_name	Cornell University
primary_location.source.host_organization_lineage	https://openalex.org/I205783295
primary_location.license
primary_location.pdf_url	https://arxiv.org/pdf/2407.14676
primary_location.version	submittedVersion
primary_location.raw_type
primary_location.license_id
primary_location.is_accepted	False
primary_location.is_published	False
primary_location.raw_source_name
primary_location.landing_page_url	http://arxiv.org/abs/2407.14676
publication_date	2024-07-19
publication_year	2024
referenced_works_count	0
abstract_inverted_index.A	119
abstract_inverted_index.a	5, 181
abstract_inverted_index.An	137
abstract_inverted_index.To	35
abstract_inverted_index.We	75, 107, 167
abstract_inverted_index.an	40
abstract_inverted_index.by	25, 77, 100
abstract_inverted_index.in	18, 151, 164
abstract_inverted_index.is	23, 121, 139
abstract_inverted_index.of	29, 173, 184
abstract_inverted_index.on	67, 141, 156, 180
abstract_inverted_index.to	47, 61, 65, 91, 111, 123, 133, 146, 149
abstract_inverted_index.we	38
abstract_inverted_index.SSL	105
abstract_inverted_index.and	95, 129
abstract_inverted_index.for	8, 53, 71
abstract_inverted_index.has	3
abstract_inverted_index.its	16
abstract_inverted_index.low	87
abstract_inverted_index.the	26, 63, 104, 161, 169, 174
abstract_inverted_index.two	82
abstract_inverted_index.yet	15
abstract_inverted_index.FGVR	72, 165, 171
abstract_inverted_index.SSL.	74
abstract_inverted_index.This	55
abstract_inverted_index.both	127
abstract_inverted_index.data	59, 94, 135, 144
abstract_inverted_index.fail	90
abstract_inverted_index.from	103, 126
abstract_inverted_index.less	98
abstract_inverted_index.main	83
abstract_inverted_index.such	142
abstract_inverted_index.task	28
abstract_inverted_index.that	43, 89
abstract_inverted_index.then	108
abstract_inverted_index.wide	182
abstract_inverted_index.with	86
abstract_inverted_index.(SSL)	2
abstract_inverted_index.FGVR.	54
abstract_inverted_index.SSL's	45
abstract_inverted_index.focus	66
abstract_inverted_index.guide	62
abstract_inverted_index.loss.	106
abstract_inverted_index.model	64
abstract_inverted_index.novel	41
abstract_inverted_index.ones.	118
abstract_inverted_index.pairs	60, 145
abstract_inverted_index.start	76
abstract_inverted_index.these	112
abstract_inverted_index.this,	37
abstract_inverted_index.those	96
abstract_inverted_index.using	81
abstract_inverted_index.vital	52
abstract_inverted_index.while	115, 154
abstract_inverted_index.(FGVR)	22
abstract_inverted_index.across	12
abstract_inverted_index.become	4, 147
abstract_inverted_index.boosts	44
abstract_inverted_index.create	134
abstract_inverted_index.deemed	97
abstract_inverted_index.during	73
abstract_inverted_index.images	125
abstract_inverted_index.pairs.	136
abstract_inverted_index.subtle	31
abstract_inverted_index.tasks,	14
abstract_inverted_index.tasks.	166
abstract_inverted_index.visual	10, 20
abstract_inverted_index.ability	46
abstract_inverted_index.between	33
abstract_inverted_index.creates	57
abstract_inverted_index.decoder	120
abstract_inverted_index.encoder	138
abstract_inverted_index.extract	48
abstract_inverted_index.feature	131
abstract_inverted_index.induced	102
abstract_inverted_index.model's	162
abstract_inverted_index.thereby	159
abstract_inverted_index.through	177
abstract_inverted_index.trained	140
abstract_inverted_index.variety	183
abstract_inverted_index.various	13
abstract_inverted_index.vectors	132
abstract_inverted_index.Grad-CAM	101
abstract_inverted_index.Learning	1
abstract_inverted_index.approach	7, 56, 176
abstract_inverted_index.critical	49, 70
abstract_inverted_index.employed	122
abstract_inverted_index.features	51, 69, 80, 85, 114
abstract_inverted_index.focusing	155
abstract_inverted_index.original	130
abstract_inverted_index.overcome	36
abstract_inverted_index.proposed	175
abstract_inverted_index.separate	93
abstract_inverted_index.strategy	42
abstract_inverted_index.variance	88
abstract_inverted_index.acquiring	9
abstract_inverted_index.criteria:	84
abstract_inverted_index.datasets.	185
abstract_inverted_index.extensive	178
abstract_inverted_index.features,	158
abstract_inverted_index.generated	143
abstract_inverted_index.important	99
abstract_inverted_index.improving	160
abstract_inverted_index.intricate	27
abstract_inverted_index.introduce	39, 109
abstract_inverted_index.invariant	148
abstract_inverted_index.perturbed	128
abstract_inverted_index.prominent	6
abstract_inverted_index.promising	170
abstract_inverted_index.challenged	24
abstract_inverted_index.dimensions	153
abstract_inverted_index.evaluation	179
abstract_inverted_index.preserving	116
abstract_inverted_index.variations	150
abstract_inverted_index.application	17
abstract_inverted_index.categories.	34
abstract_inverted_index.demonstrate	168
abstract_inverted_index.differences	32
abstract_inverted_index.effectively	92
abstract_inverted_index.identifying	78
abstract_inverted_index.performance	163, 172
abstract_inverted_index.recognition	21
abstract_inverted_index.reconstruct	124
abstract_inverted_index.synthesized	58
abstract_inverted_index.fine-grained	19
abstract_inverted_index.perturbations	110
abstract_inverted_index.discriminative	50, 68, 117, 157
abstract_inverted_index.distinguishing	30
abstract_inverted_index.Self-Supervised	0
abstract_inverted_index.representations	11
abstract_inverted_index.non-discriminative	79, 113, 152
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	5
citation_normalized_percentile