# Image Classification on ImageNet (test) (Standard and Robust Accuracy)

Source: [Image Classification on ImageNet (test) (Standard and Robust Accuracy)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-test-standard-and-robust-accuracy>)

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

Data updated: 2026-02-26T06:04:45.396Z

Evaluation of Image Classification performance on the ImageNet test set, reporting both standard accuracy and a general robustness accuracy metric.

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 (test) |
| Metric | Standard Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

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

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

## Recorded results

| Method | Standard Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| NADD | 76.23 | Not recorded | 695c96f953fdd5c361743643 |
| Nie et al. | 71.48 | Not recorded | 695c96f953fdd5c361743643 |
| Lee et al. | 70.74 | Not recorded | 695c96f953fdd5c361743643 |
| Bai et al. | 67.38 | Not recorded | 695c96f953fdd5c361743643 |
| Salman et al. | 63.86 | Not recorded | 695c96f953fdd5c361743643 |
| Engstrom et al. | 62.42 | Not recorded | 695c96f953fdd5c361743643 |
| Wong et al. | 53.83 | Not recorded | 695c96f953fdd5c361743643 |

### NADD

Standard Accuracy: 76.23. Result ID: 6996f6e75698665ec4439e42.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

Recorded evaluation configuration: {"Defense Type":"AP","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

### Nie et al.

Standard Accuracy: 71.48. Result ID: 6996f6e75698665ec4439e40.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c841e209131cea3472d.md>)

Recorded evaluation configuration: {"Defense Type":"AP","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

### Lee et al.

Standard Accuracy: 70.74. Result ID: 6996f6e75698665ec4439e41.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71551e209131cea5091d.md>)

Recorded evaluation configuration: {"Defense Type":"AP","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack","Number of diffusion cycles":"8"}. This record does not assert independent reproduction.

### Bai et al.

Standard Accuracy: 67.38. Result ID: 6996f6e75698665ec4439e3d.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b371e209131cea2cfa6.md>)

[Implementation](<https://github.com/ytongbai/ViTs-vs-CNNs>)

Recorded evaluation configuration: {"Defense Type":"AT","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

### Salman et al.

Standard Accuracy: 63.86. Result ID: 6996f6e75698665ec4439e3c.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68e21e209131cea1ef57.md>)

[Implementation](<https://github.com/Microsoft/robust-models-transfer>)

Recorded evaluation configuration: {"Defense Type":"AT","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

### Engstrom et al.

Standard Accuracy: 62.42. Result ID: 6996f6e75698665ec4439e3e.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

Recorded evaluation configuration: {"Defense Type":"AT","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

### Wong et al.

Standard Accuracy: 53.83. Result ID: 6996f6e75698665ec4439e3f.

[Reporting paper](<https://www.sota2.com/research/paper/695c96f953fdd5c361743643.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66e51e209131cea12efc.md>)

Recorded evaluation configuration: {"Defense Type":"AT","Attack setting":"L-infinity norm (epsilon = 4/255)","Attack":"AutoAttack"}. This record does not assert independent reproduction.

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