# Image Classification on EuroSAT (mean per-class accuracy)

Source: [Image Classification on EuroSAT (mean per-class accuracy)](<https://www.sota2.com/research/sota/image-classification-on-eurosat-mean-per-class-accuracy>)

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

Data updated: 2026-02-26T05:52:13.198Z

Evaluation of Image Classification performance on the EuroSAT dataset, reporting mean per-class accuracy.

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 | Image Classification |
| Dataset (including any stated split) | EuroSAT |
| Metric | Mean Per-Class Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [Mean Per-Class Accuracy](<https://www.sota2.com/research/sota/image-classification-on-eurosat-mean-per-class-accuracy.md?metric=Mean%20Per-Class%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Mean Per-Class Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| REAL-Prompt | 57.5 | Not recorded | 68fa73411e209131cea5b4d9 |
| OpenAI templates | 51.5 | Not recorded | 68fa73411e209131cea5b4d9 |
| REAL-Linear | 51.5 | Not recorded | 68fa73411e209131cea5b4d9 |
| REACT | 51.1 | Not recorded | 68fa73411e209131cea5b4d9 |
| REACT | 45.4 | Not recorded | 68fa73411e209131cea5b4d9 |
| DCLIP | 36.8 | Not recorded | 68fa73411e209131cea5b4d9 |
| Prompting | 36.6 | Not recorded | 68fa73411e209131cea5b4d9 |
| Prompting | 28.6 | Not recorded | 68fa73411e209131cea5b4d9 |

### REAL-Prompt

Mean Per-Class Accuracy: 57.5. Result ID: 6992b2823a6c31484910d57b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-Shot Prompting"}. This record does not assert independent reproduction.

### OpenAI templates

Mean Per-Class Accuracy: 51.5. Result ID: 6992b2823a6c31484910d579.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

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

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

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-Shot Prompting"}. This record does not assert independent reproduction.

### REAL-Linear

Mean Per-Class Accuracy: 51.5. Result ID: 6992b2823a6c31484910d57e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Retrieval Augmented","Retrieval Size":500}. This record does not assert independent reproduction.

### REACT

Mean Per-Class Accuracy: 51.1. Result ID: 6992b2823a6c31484910d57c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e631e209131cea3fa16.md>)

[Implementation](<https://react-vl.github.io/>)

Recorded evaluation configuration: {"Evaluation Protocol":"Retrieval Augmented","Retrieval Size":"10K","Variant":"Locked-Text"}. This record does not assert independent reproduction.

### REACT

Mean Per-Class Accuracy: 45.4. Result ID: 6992b2823a6c31484910d57d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e631e209131cea3fa16.md>)

[Implementation](<https://react-vl.github.io/>)

Recorded evaluation configuration: {"Evaluation Protocol":"Retrieval Augmented","Retrieval Size":"10K","Variant":"Gated-Image"}. This record does not assert independent reproduction.

### DCLIP

Mean Per-Class Accuracy: 36.8. Result ID: 6992b2823a6c31484910d57a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e221e209131cea3e10c.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-Shot Prompting"}. This record does not assert independent reproduction.

### Prompting

Mean Per-Class Accuracy: 36.6. Result ID: 6992b2823a6c31484910d578.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-Shot Prompting","Prompt Template":"a photo of {concept}"}. This record does not assert independent reproduction.

### Prompting

Mean Per-Class Accuracy: 28.6. Result ID: 6992b2823a6c31484910d577.

[Reporting paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4d9.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-Shot Prompting","Prompt Template":"{concept}"}. This record does not assert independent reproduction.

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