# Open-Vocabulary Segmentation on Potsdam (test)

Source: [Open-Vocabulary Segmentation on Potsdam (test)](<https://www.sota2.com/research/sota/open-vocabulary-segmentation-on-potsdam-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-06T07:14:10.510Z

Evaluation of Open-Vocabulary Segmentation performance on the test split of the Potsdam dataset, reporting mean Intersection over Union (mIoU).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Open-Vocabulary Segmentation |
| Dataset (including any stated split) | Potsdam (test) |
| Metric | mIoU |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 19 of 19 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [mIoU](<https://www.sota2.com/research/sota/open-vocabulary-segmentation-on-potsdam-test.md?metric=mIoU>): Higher is better; unit: PERCENT.

## Recorded results

| Method | mIoU | Result date | Report paper |
| --- | --- | --- | --- |
| CONCEPTBANK | 60.5 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| SegEarth-OV3 | 57.8 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| CAFe-DINO | 56.4 | Not recorded | 69facafbc0f56ed84308a303 |
| SkySense-O | 54.1 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| SAM3 | 54.1 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| CorrCLIP | 51.9 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| ProxyCLIP | 49 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| SegEarth-OV | 48.5 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| SegEarth-OV | 46.2 | Not recorded | 69facafbc0f56ed84308a303 |
| SCLIP | 39.6 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| GEM | 39.1 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| RSKT-Seg | 38.4 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| SAN | 37.3 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| Cat-Seg | 35.8 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| GSNet | 34.9 | Not recorded | 69facafbc0f56ed84308a303 |
| MaskCLIP | 33.9 | Not recorded | 698969d5c97a0b3e3b5dc30d |
| DINOv3.txt | 23.8 | Not recorded | 69facafbc0f56ed84308a303 |
| OVRS | 17.9 | Not recorded | 69facafbc0f56ed84308a303 |
| CLIP | 15.6 | Not recorded | 698969d5c97a0b3e3b5dc30d |

### CONCEPTBANK

mIoU: 60.5. Result ID: 699735e15698665ec48599d8.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

Recorded evaluation configuration: {"Backbone (Size)":"SAM3 (PE-L+/14)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":null}. This record does not assert independent reproduction.

### SegEarth-OV3

mIoU: 57.8. Result ID: 699735e15698665ec48599d7.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6938fe926cfd488761a53d4c.md>)

[Implementation](<https://github.com/earth-insights/SegEarth-OV-3>)

Recorded evaluation configuration: {"Backbone (Size)":"SAM3 (PE-L+/14)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"arXiv’25"}. This record does not assert independent reproduction.

### CAFe-DINO

mIoU: 56.4. Result ID: 69faea3d47f34ce924dd6459.

[Reporting paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

Recorded evaluation configuration: {"RS Training":"No","Background Inclusion":"Yes","Trained on target dataset":"false"}. This record does not assert independent reproduction.

### SkySense-O

mIoU: 54.1. Result ID: 699735e15698665ec48599cd.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-L/14@512px)","Training Paradigm":"Training-based","Evaluation Protocol":"with background","Pub. & Year":"CVPR’25"}. This record does not assert independent reproduction.

### SAM3

mIoU: 54.1. Result ID: 699735e15698665ec48599d6.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6923e69bd6b0dd334292f0ad.md>)

[Implementation](<https://github.com/facebookresearch/sam3>)

Recorded evaluation configuration: {"Backbone (Size)":"SAM3 (PE-L+/14)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ICLR’26"}. This record does not assert independent reproduction.

### CorrCLIP

mIoU: 51.9. Result ID: 699735e15698665ec48599d5.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a561e209131cea8293c.md>)

[Implementation](<https://github.com/zdk258/CorrCLIP>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP + DINO (ViT-B/8) + SAM2 (Hiera-L)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ICCV’25"}. This record does not assert independent reproduction.

### ProxyCLIP

mIoU: 49. Result ID: 699735e15698665ec48599d2.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa746f1e209131cea61bca.md>)

[Implementation](<https://github.com/mc-lan/ProxyCLIP>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP + DINOv2 (ViT-B/14)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ECCV’24"}. This record does not assert independent reproduction.

### SegEarth-OV

mIoU: 48.5. Result ID: 699735e15698665ec48599d4.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75911e209131cea68af9.md>)

[Implementation](<https://likyoo.github.io/SegEarth-OV>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-B/16)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"CVPR’25"}. This record does not assert independent reproduction.

### SegEarth-OV

mIoU: 46.2. Result ID: 69faea3d47f34ce924dd6458.

[Reporting paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75911e209131cea68af9.md>)

[Implementation](<https://likyoo.github.io/SegEarth-OV>)

Recorded evaluation configuration: {"RS Training":"Self-Supervised","Background Inclusion":"Yes","Trained on target dataset":"false"}. This record does not assert independent reproduction.

### SCLIP

mIoU: 39.6. Result ID: 699735e15698665ec48599d3.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756e1e209131cea67d7c.md>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-B/16)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ECCV’24"}. This record does not assert independent reproduction.

### GEM

mIoU: 39.1. Result ID: 699735e15698665ec48599d1.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71711e209131cea51429.md>)

[Implementation](<https://github.com/WalBouss/GEM>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-B/16)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"CVPR’24"}. This record does not assert independent reproduction.

### RSKT-Seg

mIoU: 38.4. Result ID: 699735e15698665ec48599ce.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b581e209131cea87ee7.md>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-L/14@336px) + RemoteCLIP (ViT-B/16) + DINO (ViT-B/32)","Training Paradigm":"Training-based","Evaluation Protocol":"with background","Pub. & Year":"AAAI’26"}. This record does not assert independent reproduction.

### SAN

mIoU: 37.3. Result ID: 699735e15698665ec48599cb.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecc1e209131cea42379.md>)

[Implementation](<https://mendelxu.github.io/SAN>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-L/14@336px)","Training Paradigm":"Training-based","Evaluation Protocol":"with background","Pub. & Year":"CVPR’23"}. This record does not assert independent reproduction.

### Cat-Seg

mIoU: 35.8. Result ID: 699735e15698665ec48599cc.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72b11e209131cea57ef4.md>)

[Implementation](<https://ku-cvlab.github.io/CAT-Seg/>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-L/14@336px)","Training Paradigm":"Training-based","Evaluation Protocol":"with background","Pub. & Year":"CVPR’24"}. This record does not assert independent reproduction.

### GSNet

mIoU: 34.9. Result ID: 69faea3d47f34ce924dd6457.

[Reporting paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76711e209131cea6d035.md>)

Recorded evaluation configuration: {"RS Training":"Yes","Background Inclusion":"Yes","Trained on target dataset":"false"}. This record does not assert independent reproduction.

### MaskCLIP

mIoU: 33.9. Result ID: 699735e15698665ec48599d0.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d1f1e209131cea37ece.md>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-B/16)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ECCV’22"}. This record does not assert independent reproduction.

### DINOv3.txt

mIoU: 23.8. Result ID: 69faea3d47f34ce924dd6455.

[Reporting paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a931e209131cea83fc5.md>)

Recorded evaluation configuration: {"RS Training":"No","Background Inclusion":"Yes","Trained on target dataset":"false"}. This record does not assert independent reproduction.

### OVRS

mIoU: 17.9. Result ID: 69faea3d47f34ce924dd6456.

[Reporting paper](<https://www.sota2.com/research/paper/69facafbc0f56ed84308a303.md>)

Recorded evaluation configuration: {"RS Training":"Yes","Background Inclusion":"Yes","Trained on target dataset":"false"}. This record does not assert independent reproduction.

### CLIP

mIoU: 15.6. Result ID: 699735e15698665ec48599cf.

[Reporting paper](<https://www.sota2.com/research/paper/698969d5c97a0b3e3b5dc30d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Backbone (Size)":"CLIP (ViT-B/16)","Training Paradigm":"Parameter-free","Evaluation Protocol":"with background","Pub. & Year":"ICML’21"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.