UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation Article Swipe

PDF

Lihe Yang , Zhen Zhao , Hengshuang Zhao ·

YOU? · · 2024 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2410.10777

Semi-supervised semantic segmentation (SSS) aims at learning rich visual knowledge from cheap unlabeled images to enhance semantic segmentation capability. Among recent works, UniMatch improves its precedents tremendously by amplifying the practice of weak-to-strong consistency regularization. Subsequent works typically follow similar pipelines and propose various delicate designs. Despite the achieved progress, strangely, even in this flourishing era of numerous powerful vision models, almost all SSS works are still sticking to 1) using outdated ResNet encoders with small-scale ImageNet-1K pre-training, and 2) evaluation on simple Pascal and Cityscapes datasets. In this work, we argue that, it is necessary to switch the baseline of SSS from ResNet-based encoders to more capable ViT-based encoders (e.g., DINOv2) that are pre-trained on massive data. A simple update on the encoder (even using 2x fewer parameters) can bring more significant improvement than careful method designs. Built on this competitive baseline, we present our upgraded and simplified UniMatch V2, inheriting the core spirit of weak-to-strong consistency from V1, but requiring less training cost and providing consistently better results. Additionally, witnessing the gradually saturated performance on Pascal and Cityscapes, we appeal that we should focus on more challenging benchmarks with complex taxonomy, such as ADE20K and COCO datasets. Code, models, and logs of all reported values, are available at https://github.com/LiheYoung/UniMatch-V2.

Related Topics

Segmentation Fault

Computer Science

Artificial Intelligence

Mathematics

Mathematical Analysis

Concepts

Limit (mathematics) Segmentation Computer science Artificial intelligence Natural language processing Mathematics Mathematical analysis

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2410.10777
PDF: https://arxiv.org/pdf/2410.10777
OA Status: green
Cited By: 1
Related Works: 10
OpenAlex ID: https://openalex.org/W4403571352

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4403571352

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2410.10777

Digital Object Identifier
Title: UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2024

Year of publication
Publication date: 2024-10-14

Full publication date if available
Authors: Lihe Yang, Zhen Zhao, Hengshuang Zhao

List of authors in order
Landing page: https://arxiv.org/abs/2410.10777

Publisher landing page
PDF URL: https://arxiv.org/pdf/2410.10777

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/2410.10777

Direct OA link when available
Concepts: Limit (mathematics), Segmentation, Computer science, Artificial intelligence, Natural language processing, Mathematics, Mathematical analysis

Top concepts (fields/topics) attached by OpenAlex
Cited by: 1

Total citation count in OpenAlex
Citations by year (recent): 2025: 1

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

Other works algorithmically related by OpenAlex

Full payload

id	https://openalex.org/W4403571352
doi	https://doi.org/10.48550/arxiv.2410.10777
ids.doi	https://doi.org/10.48550/arxiv.2410.10777
ids.openalex	https://openalex.org/W4403571352
fwci
type	preprint
title	UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10181
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.4683000147342682
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1702
topics[0].subfield.display_name	Artificial Intelligence
topics[0].display_name	Natural Language Processing Techniques
is_xpac	False
apc_list
apc_paid
concepts[0].id	https://openalex.org/C151201525
concepts[0].level	2
concepts[0].score	0.6421110033988953
concepts[0].wikidata	https://www.wikidata.org/wiki/Q177239
concepts[0].display_name	Limit (mathematics)
concepts[1].id	https://openalex.org/C89600930
concepts[1].level	2
concepts[1].score	0.6377905607223511
concepts[1].wikidata	https://www.wikidata.org/wiki/Q1423946
concepts[1].display_name	Segmentation
concepts[2].id	https://openalex.org/C41008148
concepts[2].level	0
concepts[2].score	0.552157461643219
concepts[2].wikidata	https://www.wikidata.org/wiki/Q21198
concepts[2].display_name	Computer science
concepts[3].id	https://openalex.org/C154945302
concepts[3].level	1
concepts[3].score	0.5223894715309143
concepts[3].wikidata	https://www.wikidata.org/wiki/Q11660
concepts[3].display_name	Artificial intelligence
concepts[4].id	https://openalex.org/C204321447
concepts[4].level	1
concepts[4].score	0.4428941607475281
concepts[4].wikidata	https://www.wikidata.org/wiki/Q30642
concepts[4].display_name	Natural language processing
concepts[5].id	https://openalex.org/C33923547
concepts[5].level	0
concepts[5].score	0.23310008645057678
concepts[5].wikidata	https://www.wikidata.org/wiki/Q395
concepts[5].display_name	Mathematics
concepts[6].id	https://openalex.org/C134306372
concepts[6].level	1
concepts[6].score	0.06783083081245422
concepts[6].wikidata	https://www.wikidata.org/wiki/Q7754
concepts[6].display_name	Mathematical analysis
keywords[0].id	https://openalex.org/keywords/limit
keywords[0].score	0.6421110033988953
keywords[0].display_name	Limit (mathematics)
keywords[1].id	https://openalex.org/keywords/segmentation
keywords[1].score	0.6377905607223511
keywords[1].display_name	Segmentation
keywords[2].id	https://openalex.org/keywords/computer-science
keywords[2].score	0.552157461643219
keywords[2].display_name	Computer science
keywords[3].id	https://openalex.org/keywords/artificial-intelligence
keywords[3].score	0.5223894715309143
keywords[3].display_name	Artificial intelligence
keywords[4].id	https://openalex.org/keywords/natural-language-processing
keywords[4].score	0.4428941607475281
keywords[4].display_name	Natural language processing
keywords[5].id	https://openalex.org/keywords/mathematics
keywords[5].score	0.23310008645057678
keywords[5].display_name	Mathematics
keywords[6].id	https://openalex.org/keywords/mathematical-analysis
keywords[6].score	0.06783083081245422
keywords[6].display_name	Mathematical analysis
language	en
locations[0].id	pmh:oai:arXiv.org:2410.10777
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/2410.10777
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/2410.10777
locations[1].id	doi:10.48550/arxiv.2410.10777
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.2410.10777
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5038009961
authorships[0].author.orcid	https://orcid.org/0009-0007-5803-072X
authorships[0].author.display_name	Lihe Yang
authorships[0].author_position	first
authorships[0].raw_author_name	Yang, Lihe
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5072694516
authorships[1].author.orcid	https://orcid.org/0000-0002-0796-4078
authorships[1].author.display_name	Zhen Zhao
authorships[1].author_position	middle
authorships[1].raw_author_name	Zhao, Zhen
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5078109015
authorships[2].author.orcid	https://orcid.org/0000-0001-8277-2706
authorships[2].author.display_name	Hengshuang Zhao
authorships[2].author_position	last
authorships[2].raw_author_name	Zhao, Hengshuang
authorships[2].is_corresponding	False
has_content.pdf	True
has_content.grobid_xml	True
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2410.10777
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation
has_fulltext	True
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10181
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.4683000147342682
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1702
primary_topic.subfield.display_name	Artificial Intelligence
primary_topic.display_name	Natural Language Processing Techniques
related_works	https://openalex.org/W4391375266, https://openalex.org/W2899084033, https://openalex.org/W2748952813, https://openalex.org/W2390279801, https://openalex.org/W4391913857, https://openalex.org/W2358668433, https://openalex.org/W4396701345, https://openalex.org/W2376932109, https://openalex.org/W2001405890, https://openalex.org/W3204019825
cited_by_count	1
counts_by_year[0].year	2025
counts_by_year[0].cited_by_count	1
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2410.10777
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/2410.10777
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/2410.10777
primary_location.id	pmh:oai:arXiv.org:2410.10777
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/2410.10777
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/2410.10777
publication_date	2024-10-14
publication_year	2024
referenced_works_count	0
abstract_inverted_index.A	118
abstract_inverted_index.1)	69
abstract_inverted_index.2)	79
abstract_inverted_index.2x	126
abstract_inverted_index.In	87
abstract_inverted_index.as	194
abstract_inverted_index.at	5, 209
abstract_inverted_index.by	27
abstract_inverted_index.in	52
abstract_inverted_index.is	94
abstract_inverted_index.it	93
abstract_inverted_index.of	31, 56, 100, 155, 203
abstract_inverted_index.on	81, 115, 121, 139, 176, 186
abstract_inverted_index.to	14, 68, 96, 105
abstract_inverted_index.we	90, 143, 180, 183
abstract_inverted_index.SSS	63, 101
abstract_inverted_index.V1,	159
abstract_inverted_index.V2,	150
abstract_inverted_index.all	62, 204
abstract_inverted_index.and	41, 78, 84, 147, 165, 178, 196, 201
abstract_inverted_index.are	65, 113, 207
abstract_inverted_index.but	160
abstract_inverted_index.can	129
abstract_inverted_index.era	55
abstract_inverted_index.its	24
abstract_inverted_index.our	145
abstract_inverted_index.the	29, 47, 98, 122, 152, 172
abstract_inverted_index.COCO	197
abstract_inverted_index.aims	4
abstract_inverted_index.core	153
abstract_inverted_index.cost	164
abstract_inverted_index.even	51
abstract_inverted_index.from	10, 102, 158
abstract_inverted_index.less	162
abstract_inverted_index.logs	202
abstract_inverted_index.more	106, 131, 187
abstract_inverted_index.rich	7
abstract_inverted_index.such	193
abstract_inverted_index.than	134
abstract_inverted_index.that	112, 182
abstract_inverted_index.this	53, 88, 140
abstract_inverted_index.with	74, 190
abstract_inverted_index.(SSS)	3
abstract_inverted_index.(even	124
abstract_inverted_index.Among	19
abstract_inverted_index.Built	138
abstract_inverted_index.Code,	199
abstract_inverted_index.argue	91
abstract_inverted_index.bring	130
abstract_inverted_index.cheap	11
abstract_inverted_index.data.	117
abstract_inverted_index.fewer	127
abstract_inverted_index.focus	185
abstract_inverted_index.still	66
abstract_inverted_index.that,	92
abstract_inverted_index.using	70, 125
abstract_inverted_index.work,	89
abstract_inverted_index.works	36, 64
abstract_inverted_index.(e.g.,	110
abstract_inverted_index.ADE20K	195
abstract_inverted_index.Pascal	83, 177
abstract_inverted_index.ResNet	72
abstract_inverted_index.almost	61
abstract_inverted_index.appeal	181
abstract_inverted_index.better	168
abstract_inverted_index.follow	38
abstract_inverted_index.images	13
abstract_inverted_index.method	136
abstract_inverted_index.recent	20
abstract_inverted_index.should	184
abstract_inverted_index.simple	82, 119
abstract_inverted_index.spirit	154
abstract_inverted_index.switch	97
abstract_inverted_index.update	120
abstract_inverted_index.vision	59
abstract_inverted_index.visual	8
abstract_inverted_index.works,	21
abstract_inverted_index.DINOv2)	111
abstract_inverted_index.Despite	46
abstract_inverted_index.capable	107
abstract_inverted_index.careful	135
abstract_inverted_index.complex	191
abstract_inverted_index.encoder	123
abstract_inverted_index.enhance	15
abstract_inverted_index.massive	116
abstract_inverted_index.models,	60, 200
abstract_inverted_index.present	144
abstract_inverted_index.propose	42
abstract_inverted_index.similar	39
abstract_inverted_index.values,	206
abstract_inverted_index.various	43
abstract_inverted_index.UniMatch	22, 149
abstract_inverted_index.achieved	48
abstract_inverted_index.baseline	99
abstract_inverted_index.delicate	44
abstract_inverted_index.designs.	45, 137
abstract_inverted_index.encoders	73, 104, 109
abstract_inverted_index.improves	23
abstract_inverted_index.learning	6
abstract_inverted_index.numerous	57
abstract_inverted_index.outdated	71
abstract_inverted_index.powerful	58
abstract_inverted_index.practice	30
abstract_inverted_index.reported	205
abstract_inverted_index.results.	169
abstract_inverted_index.semantic	1, 16
abstract_inverted_index.sticking	67
abstract_inverted_index.training	163
abstract_inverted_index.upgraded	146
abstract_inverted_index.ViT-based	108
abstract_inverted_index.available	208
abstract_inverted_index.baseline,	142
abstract_inverted_index.datasets.	86, 198
abstract_inverted_index.gradually	173
abstract_inverted_index.knowledge	9
abstract_inverted_index.necessary	95
abstract_inverted_index.pipelines	40
abstract_inverted_index.progress,	49
abstract_inverted_index.providing	166
abstract_inverted_index.requiring	161
abstract_inverted_index.saturated	174
abstract_inverted_index.taxonomy,	192
abstract_inverted_index.typically	37
abstract_inverted_index.unlabeled	12
abstract_inverted_index.Cityscapes	85
abstract_inverted_index.Subsequent	35
abstract_inverted_index.amplifying	28
abstract_inverted_index.benchmarks	189
abstract_inverted_index.evaluation	80
abstract_inverted_index.inheriting	151
abstract_inverted_index.precedents	25
abstract_inverted_index.simplified	148
abstract_inverted_index.strangely,	50
abstract_inverted_index.witnessing	171
abstract_inverted_index.Cityscapes,	179
abstract_inverted_index.ImageNet-1K	76
abstract_inverted_index.capability.	18
abstract_inverted_index.challenging	188
abstract_inverted_index.competitive	141
abstract_inverted_index.consistency	33, 157
abstract_inverted_index.flourishing	54
abstract_inverted_index.improvement	133
abstract_inverted_index.parameters)	128
abstract_inverted_index.performance	175
abstract_inverted_index.pre-trained	114
abstract_inverted_index.significant	132
abstract_inverted_index.small-scale	75
abstract_inverted_index.ResNet-based	103
abstract_inverted_index.consistently	167
abstract_inverted_index.segmentation	2, 17
abstract_inverted_index.tremendously	26
abstract_inverted_index.Additionally,	170
abstract_inverted_index.pre-training,	77
abstract_inverted_index.weak-to-strong	32, 156
abstract_inverted_index.Semi-supervised	0
abstract_inverted_index.regularization.	34
abstract_inverted_index.https://github.com/LiheYoung/UniMatch-V2.	210
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	3
citation_normalized_percentile