# Image Classification on ImageNet (val) (Adversarial Accuracy)

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

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

Data updated: 2026-02-26T05:37:40.768Z

Evaluation of Image Classification performance on the validation split of the ImageNet dataset, focusing specifically on adversarial robustness metrics across various grid perturbations.

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 (val) |
| Metric | Baseline Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

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

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

- [Adversarial Accuracy (Half-Pixel Grid)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-val-adversarial-accuracy.md?metric=Adversarial%20Accuracy%20(Half-Pixel%20Grid)>): Higher is better; unit: PERCENT.

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

## Recorded results

| Method | Baseline Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| ConvNeXt-Baseline | 82.12 | Not recorded | 68fa6f921e209131cea46590 |
| ConvNeXt-APS | 82.11 | Not recorded | 68fa6f921e209131cea46590 |
| ConvNeXt-AFC | 81.04 | Not recorded | 68fa6f921e209131cea46590 |

### ConvNeXt-Baseline

Baseline Accuracy: 82.12. Result ID: 6990cc793a6c314849035f9d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f921e209131cea46590.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bf21e209131cea31132.md>)

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

Recorded evaluation configuration: {"adversarial translation parameter (k)":"12"}. This record does not assert independent reproduction.

### ConvNeXt-APS

Baseline Accuracy: 82.11. Result ID: 6990cc793a6c314849035f9e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f921e209131cea46590.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa699a1e209131cea23036.md>)

Recorded evaluation configuration: {"adversarial translation parameter (k)":"12"}. This record does not assert independent reproduction.

### ConvNeXt-AFC

Baseline Accuracy: 81.04. Result ID: 6990cc793a6c314849035f9f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f921e209131cea46590.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f921e209131cea46590.md>)

[Implementation](<https://github.com/hmichaeli/alias_free_convnets/>)

Recorded evaluation configuration: {"adversarial translation parameter (k)":"12"}. This record does not assert independent reproduction.

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