Browse State-of-the-Art › Semi-Supervised Semantic Segmentation

Semi-Supervised Semantic Segmentation

109 papers with code · 45 benchmarks · 13 datasets archive 2025-07-28

Computer CodeComputer VisionMedicalRobots

Models that are trained with a small number of labeled examples and a large number of unlabeled examples and whose aim is to learn to segment an image (i.e. assign a class to every pixel).

Description from the archive archive 2025-07-28.

Benchmarks archive 2025-07-28

45 leaderboard tables shown for this task, 45 with rows (a “benchmark” on this site is a table with at least one row, as on /sota), ordered by row count. “Best model” is the first row in the archive's own order at snapshot; nothing is re-ranked here and metric direction is not recorded in the archive. PwC's Trend sparklines are not in the archive, so that column is omitted. 10 shown of 45 until expanded.

DatasetBest model (first row in archive order)PaperCodeSyntologyCompare
Pascal VOC 2012 12.5% labeled (38 rows) SemiOVS (w/ UniMatch, ResNet-101) Leveraging Out-of-Distribution Unlabeled Images: Semi-Supervised... code — Compare
Cityscapes 12.5% labeled (33 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
Cityscapes 25% labeled (30 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 25% labeled (27 rows) SemiOVS (w/ PrevMatch, ResNet-101) Leveraging Out-of-Distribution Unlabeled Images: Semi-Supervised... code — Compare
Cityscapes 50% labeled (23 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
Pascal VOC 2012 6.25% labeled (19 rows) SemiOVS (w/ PrevMatch, ResNet-101) Leveraging Out-of-Distribution Unlabeled Images: Semi-Supervised... code — Compare
Cityscapes 6.25% labeled (18 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 1464 labels (17 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 92 labeled (17 rows) SemiOVS (w/ SemiVL, ViT-B/16) Leveraging Out-of-Distribution Unlabeled Images: Semi-Supervised... code — Compare
PASCAL VOC 2012 183 labeled (16 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 732 labeled (16 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 366 labeled (15 rows) UniMatch V2 (DINOv2-B) UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
PASCAL VOC 2012 50% (14 rows) S4MC Semi-Supervised Semantic Segmentation via Marginal Contextual Information code — Compare
Pascal VOC 2012 5% labeled (14 rows) ReCo (DeepLab v3+ with ResNet-101 backbone, ImageNet pretrained) Bootstrapping Semantic Segmentation with Regional Contrast code Syntology official (archive's flag): 2 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified Compare
Cityscapes 100 samples labeled (13 rows) SemiVL (ViT-B/16) SemiVL: Semi-Supervised Semantic Segmentation with Vision-Language Guidance code — Compare
SemanticKITTI (12 rows) PLE (Voxel) Learning from Spatio-temporal Correlation for Semi-Supervised... code — Compare
Pascal VOC 2012 2% labeled (12 rows) ReCo (DeepLab v3+ with ResNet-101 backbone, ImageNet pretrained) Bootstrapping Semantic Segmentation with Regional Contrast code Syntology official (archive's flag): 2 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified Compare
nuScenes (11 rows) PLE (Voxel) Learning from Spatio-temporal Correlation for Semi-Supervised... code — Compare
ScribbleKITTI (9 rows) LaserMix (Voxel) LaserMix for Semi-Supervised LiDAR Semantic Segmentation code — Compare
COCO 1/256 labeled (9 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
COCO 1/128 labeled (9 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
COCO 1/64 labeled (9 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
COCO 1/512 labeled (8 rows) SemiVL SemiVL: Semi-Supervised Semantic Segmentation with Vision-Language Guidance code — Compare
COCO 1/32 labeled (7 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
Pascal VOC 2012 1% labeled (6 rows) ReCo (DeepLab v3+ with ResNet-101 backbone, ImageNet pre-trained) Bootstrapping Semantic Segmentation with Regional Contrast code Syntology official (archive's flag): 2 ran · 3 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 0 honoured, 0 violated, 3 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified Compare
ADE20K 1/32 labeled (5 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
ADE20K 1/16 labeled (5 rows) UniMatch V2 UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation code Syntology official (archive's flag): 5 ran · 5 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 3 where Syntology's instrument failed) · 0 unverified Compare
Cityscapes 2% labeled (3 rows) GIST and RIST (DeepLabv2 with ResNet101, MSCOCO pre-trained) The GIST and RIST of Iterative Self-Training for Semi-Supervised... — — Compare
Cityscapes 5% labeled (3 rows) GIST and RIST (DeepLabv2 with ResNet101, MSCOCO pre-trained) The GIST and RIST of Iterative Self-Training for Semi-Supervised... — — Compare
Cityscapes 93 labeled (3 rows) AEL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Pascal VOC 2012 50% labeled (3 rows) AllSpark AllSpark: Reborn Labeled Features from Unlabeled in Transformer... code Syntology official (archive's flag): 1 ran · 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified; the one sample that ran constructed an object rather than computing a result Compare
PASCAL Context 12.5% labeled (2 rows) GuidedMix-Net(DeepLab v2 with ResNet101, ImageNet pretrained) GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled... code — Compare
PASCAL Context 25% labeled (2 rows) GuidedMix-Net(DeepLab v2 with ResNet101, ImageNet pretrained) GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled... code — Compare
Stanford 2D-3D (2 rows) M3L (Linear Fusion - Segformer B2) Missing Modality Robustness in Semi-Supervised Multi-Modal... code — Compare
Cityscapes with extra (no coarse labels) (2 rows) Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval) Dense FixMatch: a simple semi-supervised learning method for... code — Compare
WoodScape (2 rows) FishSegSSL FishSegSSL: A Semi-Supervised Semantic Segmentation Framework for... code — Compare
PASCAL VOC 2012 500 labels (1 row) GuidedMix-Net(DeepLab v2 with ResNet50, ImageNet pretrained) GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled... code — Compare
PASCAL VOC 2012 1000 labels (1 row) GuidedMix-Net(DeepLab v2 with ResNet50, ImageNet pretrained) GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled... code — Compare
2017 Robotic Instrument Segmentation Challenge (1 row) MMS (20% Labeled) Min-Max Similarity: A Contrastive Semi-Supervised Deep Learning... code — Compare
2D-3D-S (1 row) M3L (Linear Fusion B2) Missing Modality Robustness in Semi-Supervised Multi-Modal... code — Compare
Kvasir-Instrument (1 row) MMS(20% labeled) Min-Max Similarity: A Contrastive Semi-Supervised Deep Learning... code — Compare
KiTS19 (1 row) PatchCL Pseudo-Label Guided Contrastive Learning for Semi-Supervised... code — Compare
PASCAL VOC 2012 331 labeled (1 row) AEL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) Semi-Supervised Semantic Segmentation via Adaptive Equalization Learning code Syntology official (archive's flag): 1 ran · 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified Compare
Cityscapes 10% labeled (1 row) IM++ (416x208, 2.7m parameters, no pretraining) Inconsistency Masks: Removing the Uncertainty from Input-Pseudo-Label Pairs code — Compare
SUIM (1 row) AIM+ (256x256, 2.7m parameters, 10% labeled data, no pretraining) Inconsistency Masks: Removing the Uncertainty from Input-Pseudo-Label Pairs code — Compare

Syntology column: samples harvested from the paper's repositories and executed on synthesized fixtures; “ran” is not a correctness claim and does not order the table. Each line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the row's paper, the line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from. A dash means no Syntology record for that paper, not a recorded non-run. Each “ran” line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Libraries

Not in the archive: the export carries no per-task library table, so there is nothing to show at snapshot 2025-07-28.

Datasets archive 2025-07-28

13 datasets whose archive record lists this task, ordered by the archive's paper count.

Subtasks archive 2025-07-28

No subtask under this task in the archive's task tree.

Parent tasks archive 2025-07-28

Papers archive 2025-07-28

archive papers tagged: 190 · with a code link: 109 · where Syntology ran a sample: 35 (30 with a run with no instrument failure, 5 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (35 of 190 tagged: 30 with a run with no instrument failure, 5 where every run was a failure of Syntology's instrument)

Page 1 of 2: papers 1 to 100 of 190, in archive order: by repositories listed in the archive (most first), then newest first, not by stars (the archive holds no stars, so PwC's “Social” and “Latest” sorts cannot be reproduced). Papers that list no repository come after every paper that lists one.

Papers without a page here are shown as plain text. A Syntology line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the figure “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code); hover it for the repositories the samples that ran came from. The first sentence of the abstract is shown for the first 30 papers; every abstract is on its paper's page.

Syntology lines on 34 of the papers shown; no Syntology record for the others (a paper without an arXiv id cannot be joined to the graph, and absence from the graph layer is not a recorded non-run). “Ran” means the sample executed on a synthesized fixture, not that the paper's result was reproduced; each line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections