# Semantic Segmentation on COCO-Stuff C=27 (Held-out Seen/Unseen)

Source: [Semantic Segmentation on COCO-Stuff C=27 (Held-out Seen/Unseen)](<https://www.sota2.com/research/sota/semantic-segmentation-on-coco-stuff-c-27-held-out-seen-unseen>)

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

Data updated: 2026-04-30T07:32:39.460Z

Evaluation of Semantic Segmentation performance on the COCO-Stuff coarse set annotation (C=27), specifically testing generalization using held-out seen and unseen categories.

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 | Semantic Segmentation |
| Dataset (including any stated split) | COCO-Stuff coarse set annotation (C=27) (held-out set (seen and 15 unseen categories)) |
| Metric | mIoU (Seen) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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 (Seen)](<https://www.sota2.com/research/sota/semantic-segmentation-on-coco-stuff-c-27-held-out-seen-unseen.md?metric=mIoU%20(Seen)>): Higher is better; unit: PERCENT.

- [mIoU (Unseen)](<https://www.sota2.com/research/sota/semantic-segmentation-on-coco-stuff-c-27-held-out-seen-unseen.md?metric=mIoU%20(Unseen)>): Higher is better; unit: PERCENT.

- [Harmonic mIoU](<https://www.sota2.com/research/sota/semantic-segmentation-on-coco-stuff-c-27-held-out-seen-unseen.md?metric=Harmonic%20mIoU>): Higher is better; unit: PERCENT.

## Recorded results

| Method | mIoU (Seen) | Result date | Report paper |
| --- | --- | --- | --- |
| LILA | 34.6 | Not recorded | 69f2e23255a3c32af89fc7d0 |
| LILA | 32.8 | Not recorded | 69f2e23255a3c32af89fc7d0 |
| DINO2-B14 | 32.6 | Not recorded | 69f2e23255a3c32af89fc7d0 |
| DINO2-S14 | 30.1 | Not recorded | 69f2e23255a3c32af89fc7d0 |

### LILA

mIoU (Seen): 34.6. Result ID: 69f304f347f34ce924d9190e.

[Reporting paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

Recorded evaluation configuration: {"Backbone":"ViT-B/14","Zero-shot setting":"true","Linear probing":"true","CLIP text embedding initialization":"true"}. This record does not assert independent reproduction.

### LILA

mIoU (Seen): 32.8. Result ID: 69f304f347f34ce924d9190c.

[Reporting paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

Recorded evaluation configuration: {"Backbone":"ViT-S/14","Zero-shot setting":"true","Linear probing":"true","CLIP text embedding initialization":"true"}. This record does not assert independent reproduction.

### DINO2-B14

mIoU (Seen): 32.6. Result ID: 69f304f347f34ce924d9190d.

[Reporting paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72181e209131cea546c3.md>)

[Implementation](<https://github.com/facebookresearch/dinov2>)

Recorded evaluation configuration: {"Backbone":"ViT-B/14","Zero-shot setting":"true","Linear probing":"true","CLIP text embedding initialization":"true"}. This record does not assert independent reproduction.

### DINO2-S14

mIoU (Seen): 30.1. Result ID: 69f304f347f34ce924d9190b.

[Reporting paper](<https://www.sota2.com/research/paper/69f2e23255a3c32af89fc7d0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72181e209131cea546c3.md>)

[Implementation](<https://github.com/facebookresearch/dinov2>)

Recorded evaluation configuration: {"Backbone":"ViT-S/14","Zero-shot setting":"true","Linear probing":"true","CLIP text embedding initialization":"true"}. This record does not assert independent reproduction.

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