# Image Classification on ImageNet-9

Source: [Image Classification on ImageNet-9](<https://www.sota2.com/research/sota/image-classification-on-imagenet-9>)

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

Data updated: 2026-02-26T05:54:20.717Z

Evaluation of Image Classification performance on the ImageNet-9 dataset, utilizing custom metrics related to mixed input conditions.

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) | ImageNet-9 |
| Metric | Original Accuracy |
| 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

- [Original Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet-9.md?metric=Original%20Accuracy>): Higher is better; unit: PERCENT.

- [Mixed-same Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet-9.md?metric=Mixed-same%20Accuracy>): Higher is better; unit: PERCENT.

- [Mixed-rand Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet-9.md?metric=Mixed-rand%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Original Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| CvT\*-LS-21 | 97.2 | Not recorded | 68fa6b5e1e209131cea2dbb8 |
| DeiT-S | 97.1 | Not recorded | 68fa6b5e1e209131cea2dbb8 |
| CvT\*-LS-13 | 97 | Not recorded | 68fa6b5e1e209131cea2dbb8 |
| ResNet-50 | 94.9 | Not recorded | 68fa6b5e1e209131cea2dbb8 |

### CvT\*-LS-21

Original Accuracy: 97.2. Result ID: 698e80c0ffc3901de09e258c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Implementation](<https://github.com/NVIDIA/transformer-ls>)

Recorded evaluation configuration: {"Params (M)":"32.1"}. This record does not assert independent reproduction.

### DeiT-S

Original Accuracy: 97.1. Result ID: 698e80c0ffc3901de09e258a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69181e209131cea202b4.md>)

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

Recorded evaluation configuration: {"Params (M)":"22.1"}. This record does not assert independent reproduction.

### CvT\*-LS-13

Original Accuracy: 97. Result ID: 698e80c0ffc3901de09e258b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Implementation](<https://github.com/NVIDIA/transformer-ls>)

Recorded evaluation configuration: {"Params (M)":"20.3"}. This record does not assert independent reproduction.

### ResNet-50

Original Accuracy: 94.9. Result ID: 698e80c0ffc3901de09e2589.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b5e1e209131cea2dbb8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62321e209131ce9f8328.md>)

Recorded evaluation configuration: {"Params (M)":"25.6"}. This record does not assert independent reproduction.

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