Browse › Computer Vision › Semi-Supervised Semantic Segmentation › Cityscapes 25% labeled
Cityscapes 25% labeled Benchmark (Semi-Supervised Semantic Segmentation)
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).
The archive carries no text for this table; the description above is the archive's text for the task Semi-Supervised Semantic Segmentation. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Not inferred: Validation mIoU. Points are placed at the row's paper date; 30 of 30 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | |||||
|---|---|---|---|---|---|---|---|---|
| 1 | UniMatch V2 (DINOv2-B) | 84.5% | – | Paper | Code | 2024 | 2 of 5 ran · 3 unverified | report |
| 2 | SemiVL (ViT-B/16) | 80.3% | – | Paper | Code | 2023 | linked, not harvested | report |
| 3 | PrevMatch (ResNet-101) | 80.1% | – | Paper | Code | 2024 | 6 of 11 ran · 5 unverified | report |
| 4 | FARCLUSS | 80.0 | – | Paper | Code | 2025 | linked, not harvested | report |
| 5 | S4MC | 79.52% | – | Paper | Code | 2023 | linked, not harvested | report |
| 6 | Dual Teacher (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 79.46 | – | Paper | Code | 2023 | linked, not harvested | report |
| 7 | CorrMatch (Deeplabv3+ with ResNet-101) | 79.4% | – | Paper | Code | 2023 | linked, not harvested | report |
| 8 | UniMatch (DeepLab v3+ with ImageNet-pretrained ResNet-101, single scale inference) | 79.22% | – | Paper | Code | 2022 | 4 of 9 ran · 5 unverified | report |
| 9 | CPS (DeepLab v3+ with ImageNet-pretrained ResNet-101, single scale inference) | 79.21% | – | Paper | Code | 2021 | 22 of 39 ran · 17 unverified | report |
| 10 | AEL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 79.01% | – | Paper | Code | 2021 | 1 of 1 ran · 0 unverified | report |
| 11 | PrevMatch (ResNet-50) | 78.8% | – | Paper | Code | 2024 | 6 of 11 ran · 5 unverified | report |
| 12 | U2PL (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K, AEL) | 78.51% | – | Paper | Code | 2022 | 1 of 1 ran · 0 unverified | report |
| 13 | CW-BASS (DeepLab v3+ with ResNet-50) | 78.43% | – | Paper | Code | 2025 | linked, not harvested | report |
| 14 | n-CPS (ResNet-50) | 78.41% | – | Paper | – | 2021 | no code linked | report |
| 15 | PCR (DeepLab v3+ with ResNet-101 pretraind on ImageNet-1K) | 78.4% | – | Paper | Code | 2022 | linked, not harvested | report |
| 16 | PS-MT (DeepLab v3+ with ImageNet-pretrained ResNet-50, single scale inference) | 78.38% | – | Paper | Code | 2021 | linked, not harvested | report |
| 17 | LaserMix (DeepLab v3+, ImageNet pre- trained ResNet50, single scale inference) | 78.3% | – | Paper | Code | 2022 | linked, not harvested | report |
| 18 | SimpleBaseline(DeepLabv3+ with ImageNet pretrained Xception65, single scale inference) | 77.8% | – | Paper | Code | 2021 | linked, not harvested | report |
| 19 | CPCL (DeepLab v3+ with ResNet-50) | 76.98% | – | Paper | Code | 2022 | linked, not harvested | report |
| 20 | Error Localization Network (DeeplabV3 with ResNet-50) | 73.52% | – | Paper | Code | 2022 | 1 of 7 ran · 6 unverified | report |
| 21 | SegSDE (MTL decoder with ResNet101, ImageNet pretrained, unlabeled image sequences) | 69.38% | ✓ | Paper | Code | 2020 | linked, not harvested | report |
| 22 | ReCo (DeepLab v3+ with ResNet-101 backbone, ImageNet pretrained) | 68.50% | – | Paper | Code | 2021 | 3 of 6 ran · 3 unverified | report |
| 23 | ReCo (DeepLab v2 with ResNet-101 backbone, ImageNet pretrained) | 67.53% | – | Paper | Code | 2021 | 3 of 6 ran · 3 unverified | report |
| 24 | GuidedMix-Net(DeepLab v2 with ResNet101, ImageNet pretrained) | 67.5% | – | Paper | Code | 2021 | linked, not harvested | report |
| 25 | SemiSegContrast (DeepLab v2 with ResNet-101 backbone, MSCOCO pretrained) | 65.9% | – | Paper | Code | 2021 | 1 of 4 ran · 3 unverified | report |
| 26 | GIST and RIST (DeepLabv2 with ResNet101, MSCOCO pre-trained) | 65.14% | – | Paper | – | 2021 | no code linked | report |
| 27 | CutMix (DeepLab v2, ImageNet pre-trained) | 63.87% | – | Paper | Code | 2019 | 0 of 3 ran · 3 unverified | report |
| 28 | ClassMix (DeepLab v2 MSCOCO pretrained) | 63.63% | – | Paper | Code | 2020 | 5 of 19 ran · 14 unverified | report |
| 29 | s4GAN (DeepLab v2 ImageNet pre-trained) | 61.9% | – | Paper | Code | 2019 | 4 of 4 ran · 0 unverified | report |
| 30 | Adversarial (DeepLab v2 ImageNet pre-trained) | 60.5% | – | Paper | Code | 2018 | 1 of 1 ran · 0 unverified | report |
All 30 rows shown. 30 link to a paper page on this site; 1 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 15 rows have a graph line, from 13 distinct papers; 14 rows (12 papers) have at least one sample that ran. Counting each paper once: Syntology ran 51 of 110 samples; 59 unverified. Separately, 35 of those 110 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
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