Eliminating Gradient Conflict in Reference-based Line-Art Colorization Article Swipe

PDF

Zekun Li , Zhengyang Geng , Kang Zhao , Wenyu Chen , Yibo Yang ·

YOU? · · 2022 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2207.06095

Reference-based line-art colorization is a challenging task in computer vision. The color, texture, and shading are rendered based on an abstract sketch, which heavily relies on the precise long-range dependency modeling between the sketch and reference. Popular techniques to bridge the cross-modal information and model the long-range dependency employ the attention mechanism. However, in the context of reference-based line-art colorization, several techniques would intensify the existing training difficulty of attention, for instance, self-supervised training protocol and GAN-based losses. To understand the instability in training, we detect the gradient flow of attention and observe gradient conflict among attention branches. This phenomenon motivates us to alleviate the gradient issue by preserving the dominant gradient branch while removing the conflict ones. We propose a novel attention mechanism using this training strategy, Stop-Gradient Attention (SGA), outperforming the attention baseline by a large margin with better training stability. Compared with state-of-the-art modules in line-art colorization, our approach demonstrates significant improvements in Fréchet Inception Distance (FID, up to 27.21%) and structural similarity index measure (SSIM, up to 25.67%) on several benchmarks. The code of SGA is available at https://github.com/kunkun0w0/SGA .

Related Topics

Computer Science

Artificial Intelligence

Concepts

Computer science Sketch Artificial intelligence Context (archaeology) Similarity (geometry) Dependency (UML) Line (geometry) Range (aeronautics) Task (project management) Margin (machine learning) Computer vision Pattern recognition (psychology) Image (mathematics) Machine learning Algorithm Mathematics Management Geometry Economics Biology Materials science Paleontology Composite material

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2207.06095
PDF: https://arxiv.org/pdf/2207.06095
OA Status: green
Cited By: 2
Related Works: 10
OpenAlex ID: https://openalex.org/W4285483461

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4285483461

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2207.06095

Digital Object Identifier
Title: Eliminating Gradient Conflict in Reference-based Line-Art Colorization

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2022

Year of publication
Publication date: 2022-07-13

Full publication date if available
Authors: Zekun Li, Zhengyang Geng, Kang Zhao, Wenyu Chen, Yibo Yang

List of authors in order
Landing page: https://arxiv.org/abs/2207.06095

Publisher landing page
PDF URL: https://arxiv.org/pdf/2207.06095

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/2207.06095

Direct OA link when available
Concepts: Computer science, Sketch, Artificial intelligence, Context (archaeology), Similarity (geometry), Dependency (UML), Line (geometry), Range (aeronautics), Task (project management), Margin (machine learning), Computer vision, Pattern recognition (psychology), Image (mathematics), Machine learning, Algorithm, Mathematics, Management, Geometry, Economics, Biology, Materials science, Paleontology, Composite material

Top concepts (fields/topics) attached by OpenAlex
Cited by: 2

Total citation count in OpenAlex
Citations by year (recent): 2024: 2

Per-year citation counts (last 5 years)
Related works (count): 10

Other works algorithmically related by OpenAlex

Full payload

id	https://openalex.org/W4285483461
doi	https://doi.org/10.48550/arxiv.2207.06095
ids.doi	https://doi.org/10.48550/arxiv.2207.06095
ids.openalex	https://openalex.org/W4285483461
fwci
type	preprint
title	Eliminating Gradient Conflict in Reference-based Line-Art Colorization
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10775
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.982699990272522
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1707
topics[0].subfield.display_name	Computer Vision and Pattern Recognition
topics[0].display_name	Generative Adversarial Networks and Image Synthesis
topics[1].id	https://openalex.org/T11019
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.9243000149726868
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1707
topics[1].subfield.display_name	Computer Vision and Pattern Recognition
topics[1].display_name	Image Enhancement Techniques
is_xpac	False
apc_list
apc_paid
concepts[0].id	https://openalex.org/C41008148
concepts[0].level	0
concepts[0].score	0.7931402325630188
concepts[0].wikidata	https://www.wikidata.org/wiki/Q21198
concepts[0].display_name	Computer science
concepts[1].id	https://openalex.org/C2779231336
concepts[1].level	2
concepts[1].score	0.6895971894264221
concepts[1].wikidata	https://www.wikidata.org/wiki/Q7534724
concepts[1].display_name	Sketch
concepts[2].id	https://openalex.org/C154945302
concepts[2].level	1
concepts[2].score	0.5932052135467529
concepts[2].wikidata	https://www.wikidata.org/wiki/Q11660
concepts[2].display_name	Artificial intelligence
concepts[3].id	https://openalex.org/C2779343474
concepts[3].level	2
concepts[3].score	0.5181716680526733
concepts[3].wikidata	https://www.wikidata.org/wiki/Q3109175
concepts[3].display_name	Context (archaeology)
concepts[4].id	https://openalex.org/C103278499
concepts[4].level	3
concepts[4].score	0.49750497937202454
concepts[4].wikidata	https://www.wikidata.org/wiki/Q254465
concepts[4].display_name	Similarity (geometry)
concepts[5].id	https://openalex.org/C19768560
concepts[5].level	2
concepts[5].score	0.4952181875705719
concepts[5].wikidata	https://www.wikidata.org/wiki/Q320727
concepts[5].display_name	Dependency (UML)
concepts[6].id	https://openalex.org/C198352243
concepts[6].level	2
concepts[6].score	0.4566281735897064
concepts[6].wikidata	https://www.wikidata.org/wiki/Q37105
concepts[6].display_name	Line (geometry)
concepts[7].id	https://openalex.org/C204323151
concepts[7].level	2
concepts[7].score	0.44373977184295654
concepts[7].wikidata	https://www.wikidata.org/wiki/Q905424
concepts[7].display_name	Range (aeronautics)
concepts[8].id	https://openalex.org/C2780451532
concepts[8].level	2
concepts[8].score	0.4425400495529175
concepts[8].wikidata	https://www.wikidata.org/wiki/Q759676
concepts[8].display_name	Task (project management)
concepts[9].id	https://openalex.org/C774472
concepts[9].level	2
concepts[9].score	0.4212483763694763
concepts[9].wikidata	https://www.wikidata.org/wiki/Q6760393
concepts[9].display_name	Margin (machine learning)
concepts[10].id	https://openalex.org/C31972630
concepts[10].level	1
concepts[10].score	0.347167432308197
concepts[10].wikidata	https://www.wikidata.org/wiki/Q844240
concepts[10].display_name	Computer vision
concepts[11].id	https://openalex.org/C153180895
concepts[11].level	2
concepts[11].score	0.3311108946800232
concepts[11].wikidata	https://www.wikidata.org/wiki/Q7148389
concepts[11].display_name	Pattern recognition (psychology)
concepts[12].id	https://openalex.org/C115961682
concepts[12].level	2
concepts[12].score	0.3007064461708069
concepts[12].wikidata	https://www.wikidata.org/wiki/Q860623
concepts[12].display_name	Image (mathematics)
concepts[13].id	https://openalex.org/C119857082
concepts[13].level	1
concepts[13].score	0.2605000436306
concepts[13].wikidata	https://www.wikidata.org/wiki/Q2539
concepts[13].display_name	Machine learning
concepts[14].id	https://openalex.org/C11413529
concepts[14].level	1
concepts[14].score	0.17844626307487488
concepts[14].wikidata	https://www.wikidata.org/wiki/Q8366
concepts[14].display_name	Algorithm
concepts[15].id	https://openalex.org/C33923547
concepts[15].level	0
concepts[15].score	0.1108962893486023
concepts[15].wikidata	https://www.wikidata.org/wiki/Q395
concepts[15].display_name	Mathematics
concepts[16].id	https://openalex.org/C187736073
concepts[16].level	1
concepts[16].score	0.0
concepts[16].wikidata	https://www.wikidata.org/wiki/Q2920921
concepts[16].display_name	Management
concepts[17].id	https://openalex.org/C2524010
concepts[17].level	1
concepts[17].score	0.0
concepts[17].wikidata	https://www.wikidata.org/wiki/Q8087
concepts[17].display_name	Geometry
concepts[18].id	https://openalex.org/C162324750
concepts[18].level	0
concepts[18].score	0.0
concepts[18].wikidata	https://www.wikidata.org/wiki/Q8134
concepts[18].display_name	Economics
concepts[19].id	https://openalex.org/C86803240
concepts[19].level	0
concepts[19].score	0.0
concepts[19].wikidata	https://www.wikidata.org/wiki/Q420
concepts[19].display_name	Biology
concepts[20].id	https://openalex.org/C192562407
concepts[20].level	0
concepts[20].score	0.0
concepts[20].wikidata	https://www.wikidata.org/wiki/Q228736
concepts[20].display_name	Materials science
concepts[21].id	https://openalex.org/C151730666
concepts[21].level	1
concepts[21].score	0.0
concepts[21].wikidata	https://www.wikidata.org/wiki/Q7205
concepts[21].display_name	Paleontology
concepts[22].id	https://openalex.org/C159985019
concepts[22].level	1
concepts[22].score	0.0
concepts[22].wikidata	https://www.wikidata.org/wiki/Q181790
concepts[22].display_name	Composite material
keywords[0].id	https://openalex.org/keywords/computer-science
keywords[0].score	0.7931402325630188
keywords[0].display_name	Computer science
keywords[1].id	https://openalex.org/keywords/sketch
keywords[1].score	0.6895971894264221
keywords[1].display_name	Sketch
keywords[2].id	https://openalex.org/keywords/artificial-intelligence
keywords[2].score	0.5932052135467529
keywords[2].display_name	Artificial intelligence
keywords[3].id	https://openalex.org/keywords/context
keywords[3].score	0.5181716680526733
keywords[3].display_name	Context (archaeology)
keywords[4].id	https://openalex.org/keywords/similarity
keywords[4].score	0.49750497937202454
keywords[4].display_name	Similarity (geometry)
keywords[5].id	https://openalex.org/keywords/dependency
keywords[5].score	0.4952181875705719
keywords[5].display_name	Dependency (UML)
keywords[6].id	https://openalex.org/keywords/line
keywords[6].score	0.4566281735897064
keywords[6].display_name	Line (geometry)
keywords[7].id	https://openalex.org/keywords/range
keywords[7].score	0.44373977184295654
keywords[7].display_name	Range (aeronautics)
keywords[8].id	https://openalex.org/keywords/task
keywords[8].score	0.4425400495529175
keywords[8].display_name	Task (project management)
keywords[9].id	https://openalex.org/keywords/margin
keywords[9].score	0.4212483763694763
keywords[9].display_name	Margin (machine learning)
keywords[10].id	https://openalex.org/keywords/computer-vision
keywords[10].score	0.347167432308197
keywords[10].display_name	Computer vision
keywords[11].id	https://openalex.org/keywords/pattern-recognition
keywords[11].score	0.3311108946800232
keywords[11].display_name	Pattern recognition (psychology)
keywords[12].id	https://openalex.org/keywords/image
keywords[12].score	0.3007064461708069
keywords[12].display_name	Image (mathematics)
keywords[13].id	https://openalex.org/keywords/machine-learning
keywords[13].score	0.2605000436306
keywords[13].display_name	Machine learning
keywords[14].id	https://openalex.org/keywords/algorithm
keywords[14].score	0.17844626307487488
keywords[14].display_name	Algorithm
keywords[15].id	https://openalex.org/keywords/mathematics
keywords[15].score	0.1108962893486023
keywords[15].display_name	Mathematics
language	en
locations[0].id	pmh:oai:arXiv.org:2207.06095
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/2207.06095
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/2207.06095
locations[1].id	doi:10.48550/arxiv.2207.06095
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.2207.06095
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5100607162
authorships[0].author.orcid	https://orcid.org/0000-0002-0590-8873
authorships[0].author.display_name	Zekun Li
authorships[0].author_position	first
authorships[0].raw_author_name	Li, Zekun
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5102893697
authorships[1].author.orcid	https://orcid.org/0009-0006-9903-2716
authorships[1].author.display_name	Zhengyang Geng
authorships[1].author_position	middle
authorships[1].raw_author_name	Geng, Zhengyang
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5100731707
authorships[2].author.orcid	https://orcid.org/0000-0002-5373-6308
authorships[2].author.display_name	Kang Zhao
authorships[2].author_position	middle
authorships[2].raw_author_name	Kang, Zhao
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5100687323
authorships[3].author.orcid	https://orcid.org/0000-0002-9933-8014
authorships[3].author.display_name	Wenyu Chen
authorships[3].author_position	middle
authorships[3].raw_author_name	Chen, Wenyu
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5103072699
authorships[4].author.orcid	https://orcid.org/0000-0003-0530-7231
authorships[4].author.display_name	Yibo Yang
authorships[4].author_position	last
authorships[4].raw_author_name	Yang, Yibo
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/2207.06095
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	Eliminating Gradient Conflict in Reference-based Line-Art Colorization
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10775
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.982699990272522
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1707
primary_topic.subfield.display_name	Computer Vision and Pattern Recognition
primary_topic.display_name	Generative Adversarial Networks and Image Synthesis
related_works	https://openalex.org/W2378994405, https://openalex.org/W2385974820, https://openalex.org/W2373478030, https://openalex.org/W2378679551, https://openalex.org/W3149739944, https://openalex.org/W2392363776, https://openalex.org/W2063051341, https://openalex.org/W2591066345, https://openalex.org/W1494563618, https://openalex.org/W2357022711
cited_by_count	2
counts_by_year[0].year	2024
counts_by_year[0].cited_by_count	2
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2207.06095
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/2207.06095
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/2207.06095
primary_location.id	pmh:oai:arXiv.org:2207.06095
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/2207.06095
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/2207.06095
publication_date	2022-07-13
publication_year	2022
referenced_works_count	0
abstract_inverted_index..	183
abstract_inverted_index.a	4, 120, 136
abstract_inverted_index.To	78
abstract_inverted_index.We	118
abstract_inverted_index.an	19
abstract_inverted_index.at	181
abstract_inverted_index.by	107, 135
abstract_inverted_index.in	7, 53, 82, 147, 155
abstract_inverted_index.is	3, 179
abstract_inverted_index.of	56, 68, 89, 177
abstract_inverted_index.on	18, 25, 172
abstract_inverted_index.to	38, 102, 161, 170
abstract_inverted_index.up	160, 169
abstract_inverted_index.us	101
abstract_inverted_index.we	84
abstract_inverted_index.SGA	178
abstract_inverted_index.The	10, 175
abstract_inverted_index.and	13, 34, 43, 75, 91, 163
abstract_inverted_index.are	15
abstract_inverted_index.for	70
abstract_inverted_index.our	150
abstract_inverted_index.the	26, 32, 40, 45, 49, 54, 64, 80, 86, 104, 109, 115, 132
abstract_inverted_index.This	98
abstract_inverted_index.code	176
abstract_inverted_index.flow	88
abstract_inverted_index.task	6
abstract_inverted_index.this	125
abstract_inverted_index.with	139, 144
abstract_inverted_index.(FID,	159
abstract_inverted_index.among	95
abstract_inverted_index.based	17
abstract_inverted_index.index	166
abstract_inverted_index.issue	106
abstract_inverted_index.large	137
abstract_inverted_index.model	44
abstract_inverted_index.novel	121
abstract_inverted_index.ones.	117
abstract_inverted_index.using	124
abstract_inverted_index.which	22
abstract_inverted_index.while	113
abstract_inverted_index.would	62
abstract_inverted_index.(SGA),	130
abstract_inverted_index.(SSIM,	168
abstract_inverted_index.better	140
abstract_inverted_index.branch	112
abstract_inverted_index.bridge	39
abstract_inverted_index.color,	11
abstract_inverted_index.detect	85
abstract_inverted_index.employ	48
abstract_inverted_index.margin	138
abstract_inverted_index.relies	24
abstract_inverted_index.sketch	33
abstract_inverted_index.25.67%)	171
abstract_inverted_index.27.21%)	162
abstract_inverted_index.Popular	36
abstract_inverted_index.between	31
abstract_inverted_index.context	55
abstract_inverted_index.heavily	23
abstract_inverted_index.losses.	77
abstract_inverted_index.measure	167
abstract_inverted_index.modules	146
abstract_inverted_index.observe	92
abstract_inverted_index.precise	27
abstract_inverted_index.propose	119
abstract_inverted_index.several	60, 173
abstract_inverted_index.shading	14
abstract_inverted_index.sketch,	21
abstract_inverted_index.vision.	9
abstract_inverted_index.Compared	143
abstract_inverted_index.Distance	158
abstract_inverted_index.Fréchet	156
abstract_inverted_index.However,	52
abstract_inverted_index.abstract	20
abstract_inverted_index.approach	151
abstract_inverted_index.baseline	134
abstract_inverted_index.computer	8
abstract_inverted_index.conflict	94, 116
abstract_inverted_index.dominant	110
abstract_inverted_index.existing	65
abstract_inverted_index.gradient	87, 93, 105, 111
abstract_inverted_index.line-art	1, 58, 148
abstract_inverted_index.modeling	30
abstract_inverted_index.protocol	74
abstract_inverted_index.removing	114
abstract_inverted_index.rendered	16
abstract_inverted_index.texture,	12
abstract_inverted_index.training	66, 73, 126, 141
abstract_inverted_index.Attention	129
abstract_inverted_index.GAN-based	76
abstract_inverted_index.Inception	157
abstract_inverted_index.alleviate	103
abstract_inverted_index.attention	50, 90, 96, 122, 133
abstract_inverted_index.available	180
abstract_inverted_index.branches.	97
abstract_inverted_index.instance,	71
abstract_inverted_index.intensify	63
abstract_inverted_index.mechanism	123
abstract_inverted_index.motivates	100
abstract_inverted_index.strategy,	127
abstract_inverted_index.training,	83
abstract_inverted_index.attention,	69
abstract_inverted_index.dependency	29, 47
abstract_inverted_index.difficulty	67
abstract_inverted_index.long-range	28, 46
abstract_inverted_index.mechanism.	51
abstract_inverted_index.phenomenon	99
abstract_inverted_index.preserving	108
abstract_inverted_index.reference.	35
abstract_inverted_index.similarity	165
abstract_inverted_index.stability.	142
abstract_inverted_index.structural	164
abstract_inverted_index.techniques	37, 61
abstract_inverted_index.understand	79
abstract_inverted_index.benchmarks.	174
abstract_inverted_index.challenging	5
abstract_inverted_index.cross-modal	41
abstract_inverted_index.information	42
abstract_inverted_index.instability	81
abstract_inverted_index.significant	153
abstract_inverted_index.colorization	2
abstract_inverted_index.demonstrates	152
abstract_inverted_index.improvements	154
abstract_inverted_index.Stop-Gradient	128
abstract_inverted_index.colorization,	59, 149
abstract_inverted_index.outperforming	131
abstract_inverted_index.Reference-based	0
abstract_inverted_index.reference-based	57
abstract_inverted_index.self-supervised	72
abstract_inverted_index.state-of-the-art	145
abstract_inverted_index.https://github.com/kunkun0w0/SGA	182
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	5
citation_normalized_percentile