# Image Classification on Oxford Pets (Accuracy and Calibration Metrics)

Source: [Image Classification on Oxford Pets (Accuracy and Calibration Metrics)](<https://www.sota2.com/research/sota/image-classification-on-oxford-pets-accuracy-and-calibration-metrics>)

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

Data updated: 2026-06-24T08:48:26.083Z

Evaluation of standard Image Classification performance on the Oxford Pets dataset, including accuracy, Expected Calibration Error (ECE), Brier score, and Negative Log-Likelihood (NLL).

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) | Oxford Pets |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 28 of 28 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/image-classification-on-oxford-pets-accuracy-and-calibration-metrics.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Expected Calibration Error (ECE)](<https://www.sota2.com/research/sota/image-classification-on-oxford-pets-accuracy-and-calibration-metrics.md?metric=Expected%20Calibration%20Error%20(ECE)>): Lower is better; unit: PERCENT.

- [Negative Log-Likelihood (NLL)](<https://www.sota2.com/research/sota/image-classification-on-oxford-pets-accuracy-and-calibration-metrics.md?metric=Negative%20Log-Likelihood%20(NLL)>): Lower is better; unit: SCALAR.

- [Brier Score](<https://www.sota2.com/research/sota/image-classification-on-oxford-pets-accuracy-and-calibration-metrics.md?metric=Brier%20Score>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| BioMedVR | 93.9 | Not recorded | 6a3b64e3206542251d7c9a11 |
| AttrVR | 93.1 | Not recorded | 6a3b64e3206542251d7c9a11 |
| AR | 92.6 | Not recorded | 6a3b64e3206542251d7c9a11 |
| VP | 91.3 | Not recorded | 6a3b64e3206542251d7c9a11 |
| SV-Ensemble | 90.1 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| Deep Ensemble | 89.2 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| BioMedVR | 88.4 | Not recorded | 6a3b64e3206542251d7c9a11 |
| MG-CLIP | 88.2 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| SeGP-CL | 87.7 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| zero-shot CLIP | 87.6 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| CLIP | 87.35 | Not recorded | 69e85bd47d0172140ec81123 |
| Single w/ SVF | 87.2 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| Ours | 86.32 | Not recorded | 69e85bd47d0172140ec81123 |
| LoRA-Ensemble | 86.1 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| SPU | 85.9 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| AoS | 85.8 | Not recorded | 69e85bd47d0172140ec81123 |
| Ours | 85.63 | Not recorded | 69e85bd47d0172140ec81123 |
| SeGP-CL | 85.2 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| FARE | 84.64 | Not recorded | 69e85bd47d0172140ec81123 |
| MoE-Adapter | 84.1 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| PMG-FT | 82.91 | Not recorded | 69e85bd47d0172140ec81123 |
| Single | 81.4 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| TeCoA | 79.42 | Not recorded | 69e85bd47d0172140ec81123 |
| CLIP | 77.6 | Not recorded | 6a3b64e3206542251d7c9a11 |
| Batch Ensemble | 70.5 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| RAPF | 28.6 | Not recorded | 69b39a2d0636e5bf4afbcc2d |
| MC Dropout | 25.9 | Not recorded | 697c3b79a8b01e76cc8da7d8 |
| PROOF | 22.8 | Not recorded | 69b39a2d0636e5bf4afbcc2d |

### BioMedVR

Accuracy: 93.9. Result ID: 6a3b99cefbc373592c44ec7e.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"few-shot","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### AttrVR

Accuracy: 93.1. Result ID: 6a3b99cefbc373592c44ec7d.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771c1e209131cea710ec.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"few-shot","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### AR

Accuracy: 92.6. Result ID: 6a3b99cefbc373592c44ec7b.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ee81e209131cea42cd9.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"few-shot","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### VP

Accuracy: 91.3. Result ID: 6a3b99cefbc373592c44ec7c.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cb81e209131cea3595b.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"few-shot","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### SV-Ensemble

Accuracy: 90.1. Result ID: 6998f5995698665ec45a9c71.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations","Number of Ensemble Members":4}. This record does not assert independent reproduction.

### Deep Ensemble

Accuracy: 89.2. Result ID: 6998f5995698665ec45a9c6e.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72111e209131cea54493.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations","Number of Ensemble Members":4}. This record does not assert independent reproduction.

### BioMedVR

Accuracy: 88.4. Result ID: 6a3b99cefbc373592c44ec80.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"zero-shot","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### MG-CLIP

Accuracy: 88.2. Result ID: 69b3d5fd47f34ce9249e36b1.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79f61e209131cea8046b.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### SeGP-CL

Accuracy: 87.7. Result ID: 69b3d5fd47f34ce9249e36b3.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

Recorded evaluation configuration: {"TSGR":"true"}. This record does not assert independent reproduction.

### zero-shot CLIP

Accuracy: 87.6. Result ID: 69b3d5fd47f34ce9249e36ac.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

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

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

Recorded evaluation configuration: {"mode":"zero-shot"}. This record does not assert independent reproduction.

### CLIP

Accuracy: 87.35. Result ID: 69e9cb8a47f34ce924d09db7.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

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

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

Recorded evaluation configuration: {"Year":"2021"}. This record does not assert independent reproduction.

### Single w/ SVF

Accuracy: 87.2. Result ID: 6998f5995698665ec45a9c6c.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d9f1e209131cea3aee5.md>)

[Implementation](<https://github.com/syp2ysy/SVF>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations"}. This record does not assert independent reproduction.

### Ours

Accuracy: 86.32. Result ID: 69e9cb8a47f34ce924d09dbd.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

Recorded evaluation configuration: {"Model Configuration":"5 trees"}. This record does not assert independent reproduction.

### LoRA-Ensemble

Accuracy: 86.1. Result ID: 6998f5995698665ec45a9c70.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70cf1e209131cea4d563.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations","Number of Ensemble Members":4}. This record does not assert independent reproduction.

### SPU

Accuracy: 85.9. Result ID: 69b3d5fd47f34ce9249e36b0.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### AoS

Accuracy: 85.8. Result ID: 69e9cb8a47f34ce924d09dbb.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

Recorded evaluation configuration: {"Year":"2025"}. This record does not assert independent reproduction.

### Ours

Accuracy: 85.63. Result ID: 69e9cb8a47f34ce924d09dbc.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

Recorded evaluation configuration: {"Model Configuration":"1 tree"}. This record does not assert independent reproduction.

### SeGP-CL

Accuracy: 85.2. Result ID: 69b3d5fd47f34ce9249e36b2.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

Recorded evaluation configuration: {"TSGR":"false"}. This record does not assert independent reproduction.

### FARE

Accuracy: 84.64. Result ID: 69e9cb8a47f34ce924d09dba.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738c1e209131cea5d24a.md>)

Recorded evaluation configuration: {"Year":"2024"}. This record does not assert independent reproduction.

### MoE-Adapter

Accuracy: 84.1. Result ID: 69b3d5fd47f34ce9249e36af.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73661e209131cea5c39b.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### PMG-FT

Accuracy: 82.91. Result ID: 69e9cb8a47f34ce924d09db9.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72df1e209131cea5912e.md>)

Recorded evaluation configuration: {"Year":"2024"}. This record does not assert independent reproduction.

### Single

Accuracy: 81.4. Result ID: 6998f5995698665ec45a9c6b.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations"}. This record does not assert independent reproduction.

### TeCoA

Accuracy: 79.42. Result ID: 69e9cb8a47f34ce924d09db8.

[Reporting paper](<https://www.sota2.com/research/paper/69e85bd47d0172140ec81123.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f271e209131cea44177.md>)

Recorded evaluation configuration: {"Year":"2023"}. This record does not assert independent reproduction.

### CLIP

Accuracy: 77.6. Result ID: 6a3b99cefbc373592c44ec7f.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64e3206542251d7c9a11.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","Backbone":"ViT-B/16"}. This record does not assert independent reproduction.

### Batch Ensemble

Accuracy: 70.5. Result ID: 6998f5995698665ec45a9c6f.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa671c1e209131cea1455e.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations","Number of Ensemble Members":4}. This record does not assert independent reproduction.

### RAPF

Accuracy: 28.6. Result ID: 69b3d5fd47f34ce9249e36ae.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74331e209131cea60657.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### MC Dropout

Accuracy: 25.9. Result ID: 6998f5995698665ec45a9c6d.

[Reporting paper](<https://www.sota2.com/research/paper/697c3b79a8b01e76cc8da7d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72111e209131cea54493.md>)

Recorded evaluation configuration: {"Backbone":"DINOv2 ViT-S/14","Evaluation Protocol":"Fine-tuning","Training Duration":"~200 iterations"}. This record does not assert independent reproduction.

### PROOF

Accuracy: 22.8. Result ID: 69b3d5fd47f34ce9249e36ad.

[Reporting paper](<https://www.sota2.com/research/paper/69b39a2d0636e5bf4afbcc2d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76f01e209131cea701a9.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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