# Corruption Detection on ImageNet100 label noise

Source: [Corruption Detection on ImageNet100 label noise](<https://www.sota2.com/research/sota/corruption-detection-on-imagenet100-label-noise>)

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

Data updated: 2026-07-02T09:45:31.510Z

Evaluation of model capability to detect label corruption within the ImageNet-100 dataset, measured using F1, Recall-AUC, and delta metrics.

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 | Corruption Detection |
| Dataset (including any stated split) | ImageNet100 label noise |
| Metric | F1 Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [F1 Score](<https://www.sota2.com/research/sota/corruption-detection-on-imagenet100-label-noise.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

- [Recall AUC](<https://www.sota2.com/research/sota/corruption-detection-on-imagenet100-label-noise.md?metric=Recall%20AUC>): Higher is better; unit: PERCENT.

- [Delta Metric](<https://www.sota2.com/research/sota/corruption-detection-on-imagenet100-label-noise.md?metric=Delta%20Metric>): Higher is better; unit: SCALAR.

## Recorded results

| Method | F1 Score | Result date | Report paper |
| --- | --- | --- | --- |
| GhostSuite | 100 | Not recorded | 68fa7c3c1e209131cea8d150 |
| NDDV | 100 | Not recorded | 68fa7c3c1e209131cea8d150 |
| NDDV | 98.5 | Not recorded | 68fa7c3c1e209131cea8d150 |
| NDDV | 98 | Not recorded | 68fa7c3c1e209131cea8d150 |
| NDDV | 98 | Not recorded | 68fa7c3c1e209131cea8d150 |
| Data-OOB | 97.9 | Not recorded | 68fa7c3c1e209131cea8d150 |
| Data-OOB | 93.7 | Not recorded | 68fa7c3c1e209131cea8d150 |
| NDDV | 93.3 | Not recorded | 68fa7c3c1e209131cea8d150 |
| LossVal | 86 | Not recorded | 68fa7c3c1e209131cea8d150 |
| GhostSuite | 45.2 | Not recorded | 68fa7c3c1e209131cea8d150 |

### GhostSuite

F1 Score: 100. Result ID: 6a46333860b88bb3398653b2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa792f1e209131cea7c7c1.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 100"}. This record does not assert independent reproduction.

### NDDV

F1 Score: 100. Result ID: 6a46333860b88bb3398653b3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 100"}. This record does not assert independent reproduction.

### NDDV

F1 Score: 98.5. Result ID: 6a46333860b88bb3398653b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 1,000"}. This record does not assert independent reproduction.

### NDDV

F1 Score: 98. Result ID: 6a46333860b88bb3398653af.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

Recorded evaluation configuration: {"Evaluation":"Head-to-head","Setting":"10% label noise, n = 7,500"}. This record does not assert independent reproduction.

### NDDV

F1 Score: 98. Result ID: 6a46333860b88bb3398653bb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 10,000"}. This record does not assert independent reproduction.

### Data-OOB

F1 Score: 97.9. Result ID: 6a46333860b88bb3398653ae.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f8c1e209131cea4622f.md>)

Recorded evaluation configuration: {"Evaluation":"Head-to-head","Setting":"10% label noise, n = 7,500"}. This record does not assert independent reproduction.

### Data-OOB

F1 Score: 93.7. Result ID: 6a46333860b88bb3398653b0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f8c1e209131cea4622f.md>)

Recorded evaluation configuration: {"Evaluation":"Label-noise mean","Setting":"Noise rates 5%–45%"}. This record does not assert independent reproduction.

### NDDV

F1 Score: 93.3. Result ID: 6a46333860b88bb3398653b1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

Recorded evaluation configuration: {"Evaluation":"Label-noise mean","Setting":"Noise rates 5%–45%"}. This record does not assert independent reproduction.

### LossVal

F1 Score: 86. Result ID: 6a46333860b88bb3398653ba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76511e209131cea6c627.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 10,000"}. This record does not assert independent reproduction.

### GhostSuite

F1 Score: 45.2. Result ID: 6a46333860b88bb3398653b6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c3c1e209131cea8d150.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa792f1e209131cea7c7c1.md>)

Recorded evaluation configuration: {"Evaluation":"Scale","Setting":"n = 1,000"}. This record does not assert independent reproduction.

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