Region-Aware CAM: High-Resolution Weakly-Supervised Defect Segmentation via Salient Region Perception Article Swipe

PDF

Hangcheng Dong , Lu Zou , Bingguo Liu , Guodong Liu ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2506.22866

Surface defect detection plays a critical role in industrial quality inspection. Recent advances in artificial intelligence have significantly enhanced the automation level of detection processes. However, conventional semantic segmentation and object detection models heavily rely on large-scale annotated datasets, which conflicts with the practical requirements of defect detection tasks. This paper proposes a novel weakly supervised semantic segmentation framework comprising two key components: a region-aware class activation map (CAM) and pseudo-label training. To address the limitations of existing CAM methods, especially low-resolution thermal maps, and insufficient detail preservation, we introduce filtering-guided backpropagation (FGBP), which refines target regions by filtering gradient magnitudes to identify areas with higher relevance to defects. Building upon this, we further develop a region-aware weighted module to enhance spatial precision. Finally, pseudo-label segmentation is implemented to refine the model's performance iteratively. Comprehensive experiments on industrial defect datasets demonstrate the superiority of our method. The proposed framework effectively bridges the gap between weakly supervised learning and high-precision defect segmentation, offering a practical solution for resource-constrained industrial scenarios.

Related Topics

Truth And Reconciliation Commission Of Canada

Reich Ministry Of Public Enlightenment And Propaganda

Rick Hurst

Degenerate Art Exhibition

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2506.22866
PDF: https://arxiv.org/pdf/2506.22866
OA Status: green
OpenAlex ID: https://openalex.org/W4416508502

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4416508502

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2506.22866

Digital Object Identifier
Title: Region-Aware CAM: High-Resolution Weakly-Supervised Defect Segmentation via Salient Region Perception

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-06-28

Full publication date if available
Authors: Hangcheng Dong, Lu Zou, Bingguo Liu, Guodong Liu

List of authors in order
Landing page: https://arxiv.org/abs/2506.22866

Publisher landing page
PDF URL: https://arxiv.org/pdf/2506.22866

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/2506.22866

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4416508502
doi	https://doi.org/10.48550/arxiv.2506.22866
ids.doi	https://doi.org/10.48550/arxiv.2506.22866
ids.openalex	https://openalex.org/W4416508502
fwci
type	preprint
title	Region-Aware CAM: High-Resolution Weakly-Supervised Defect Segmentation via Salient Region Perception
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2506.22866
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/2506.22866
locations[0].version	submittedVersion
locations[0].raw_type	text
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/2506.22866
locations[1].id	doi:10.48550/arxiv.2506.22866
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.2506.22866
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5029378855
authorships[0].author.orcid	https://orcid.org/0000-0002-4880-6762
authorships[0].author.display_name	Hangcheng Dong
authorships[0].author_position	first
authorships[0].raw_author_name	Dong, Hang-Cheng
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5100938712
authorships[1].author.orcid	https://orcid.org/0009-0008-2354-6002
authorships[1].author.display_name	Lu Zou
authorships[1].author_position	middle
authorships[1].raw_author_name	Zou, Lu
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5064288905
authorships[2].author.orcid
authorships[2].author.display_name	Bingguo Liu
authorships[2].author_position	middle
authorships[2].raw_author_name	Liu, Bingguo
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5100452278
authorships[3].author.orcid	https://orcid.org/0000-0002-3173-6462
authorships[3].author.display_name	Guodong Liu
authorships[3].author_position	middle
authorships[3].raw_author_name	Liu, Guodong
authorships[3].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/2506.22866
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	Region-Aware CAM: High-Resolution Weakly-Supervised Defect Segmentation via Salient Region Perception
has_fulltext	False
is_retracted	False
updated_date	2025-11-28T16:05:42.392888
primary_topic
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2506.22866
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/2506.22866
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
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/2506.22866
primary_location.id	pmh:oai:arXiv.org:2506.22866
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/2506.22866
primary_location.version	submittedVersion
primary_location.raw_type	text
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/2506.22866
publication_date	2025-06-28
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	4, 52, 63, 115, 162
abstract_inverted_index.To	72
abstract_inverted_index.by	97
abstract_inverted_index.in	7, 13
abstract_inverted_index.is	126
abstract_inverted_index.of	22, 45, 76, 143
abstract_inverted_index.on	35, 136
abstract_inverted_index.to	101, 107, 119, 128
abstract_inverted_index.we	88, 112
abstract_inverted_index.CAM	78
abstract_inverted_index.The	146
abstract_inverted_index.and	29, 69, 84, 157
abstract_inverted_index.for	165
abstract_inverted_index.gap	152
abstract_inverted_index.key	61
abstract_inverted_index.map	67
abstract_inverted_index.our	144
abstract_inverted_index.the	19, 42, 74, 130, 141, 151
abstract_inverted_index.two	60
abstract_inverted_index.This	49
abstract_inverted_index.have	16
abstract_inverted_index.rely	34
abstract_inverted_index.role	6
abstract_inverted_index.upon	110
abstract_inverted_index.with	41, 104
abstract_inverted_index.(CAM)	68
abstract_inverted_index.areas	103
abstract_inverted_index.class	65
abstract_inverted_index.level	21
abstract_inverted_index.maps,	83
abstract_inverted_index.novel	53
abstract_inverted_index.paper	50
abstract_inverted_index.plays	3
abstract_inverted_index.this,	111
abstract_inverted_index.which	39, 93
abstract_inverted_index.Recent	11
abstract_inverted_index.defect	1, 46, 138, 159
abstract_inverted_index.detail	86
abstract_inverted_index.higher	105
abstract_inverted_index.models	32
abstract_inverted_index.module	118
abstract_inverted_index.object	30
abstract_inverted_index.refine	129
abstract_inverted_index.target	95
abstract_inverted_index.tasks.	48
abstract_inverted_index.weakly	54, 154
abstract_inverted_index.(FGBP),	92
abstract_inverted_index.Surface	0
abstract_inverted_index.address	73
abstract_inverted_index.between	153
abstract_inverted_index.bridges	150
abstract_inverted_index.develop	114
abstract_inverted_index.enhance	120
abstract_inverted_index.further	113
abstract_inverted_index.heavily	33
abstract_inverted_index.method.	145
abstract_inverted_index.model's	131
abstract_inverted_index.quality	9
abstract_inverted_index.refines	94
abstract_inverted_index.regions	96
abstract_inverted_index.spatial	121
abstract_inverted_index.thermal	82
abstract_inverted_index.Building	109
abstract_inverted_index.Finally,	123
abstract_inverted_index.However,	25
abstract_inverted_index.advances	12
abstract_inverted_index.critical	5
abstract_inverted_index.datasets	139
abstract_inverted_index.defects.	108
abstract_inverted_index.enhanced	18
abstract_inverted_index.existing	77
abstract_inverted_index.gradient	99
abstract_inverted_index.identify	102
abstract_inverted_index.learning	156
abstract_inverted_index.methods,	79
abstract_inverted_index.offering	161
abstract_inverted_index.proposed	147
abstract_inverted_index.proposes	51
abstract_inverted_index.semantic	27, 56
abstract_inverted_index.solution	164
abstract_inverted_index.weighted	117
abstract_inverted_index.annotated	37
abstract_inverted_index.conflicts	40
abstract_inverted_index.datasets,	38
abstract_inverted_index.detection	2, 23, 31, 47
abstract_inverted_index.filtering	98
abstract_inverted_index.framework	58, 148
abstract_inverted_index.introduce	89
abstract_inverted_index.practical	43, 163
abstract_inverted_index.relevance	106
abstract_inverted_index.training.	71
abstract_inverted_index.activation	66
abstract_inverted_index.artificial	14
abstract_inverted_index.automation	20
abstract_inverted_index.comprising	59
abstract_inverted_index.especially	80
abstract_inverted_index.industrial	8, 137, 167
abstract_inverted_index.magnitudes	100
abstract_inverted_index.precision.	122
abstract_inverted_index.processes.	24
abstract_inverted_index.scenarios.	168
abstract_inverted_index.supervised	55, 155
abstract_inverted_index.components:	62
abstract_inverted_index.demonstrate	140
abstract_inverted_index.effectively	149
abstract_inverted_index.experiments	135
abstract_inverted_index.implemented	127
abstract_inverted_index.inspection.	10
abstract_inverted_index.large-scale	36
abstract_inverted_index.limitations	75
abstract_inverted_index.performance	132
abstract_inverted_index.superiority	142
abstract_inverted_index.conventional	26
abstract_inverted_index.insufficient	85
abstract_inverted_index.intelligence	15
abstract_inverted_index.iteratively.	133
abstract_inverted_index.pseudo-label	70, 124
abstract_inverted_index.region-aware	64, 116
abstract_inverted_index.requirements	44
abstract_inverted_index.segmentation	28, 57, 125
abstract_inverted_index.Comprehensive	134
abstract_inverted_index.preservation,	87
abstract_inverted_index.segmentation,	160
abstract_inverted_index.significantly	17
abstract_inverted_index.high-precision	158
abstract_inverted_index.low-resolution	81
abstract_inverted_index.backpropagation	91
abstract_inverted_index.filtering-guided	90
abstract_inverted_index.resource-constrained	166
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	4
citation_normalized_percentile