# Visual Explanation Localization on ImageNet-S (val)

Source: [Visual Explanation Localization on ImageNet-S (val)](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val>)

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

Data updated: 2026-05-08T08:08:26.829Z

Evaluation of visual explanation localization methods on the validation split of the ImageNet-S dataset, reporting metrics such as PG, Pixel Accuracy, AP, and mask IoU.

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 | Visual Explanation Localization |
| Dataset (including any stated split) | ImageNet-S (val) |
| Metric | PG |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [PG](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val.md?metric=PG>): Higher is better; unit: PERCENT.

- [Energy-PG](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val.md?metric=Energy-PG>): Higher is better; unit: PERCENT.

- [Pixel Accuracy](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val.md?metric=Pixel%20Accuracy>): Higher is better; unit: PERCENT.

- [AP](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val.md?metric=AP>): Higher is better; unit: PERCENT.

- [maskIoU](<https://www.sota2.com/research/sota/visual-explanation-localization-on-imagenet-s-val.md?metric=maskIoU>): Higher is better; unit: PERCENT.

## Recorded results

| Method | PG | Result date | Report paper |
| --- | --- | --- | --- |
| Grad-ECLIP | 88.99 | Not recorded | 68fa772a1e209131cea71801 |
| Grad-ECLIP | 83.56 | Not recorded | 68fa772a1e209131cea71801 |
| CLIPSurgery | 57.59 | Not recorded | 68fa772a1e209131cea71801 |
| GAME | 47.06 | Not recorded | 68fa772a1e209131cea71801 |
| MaskCLIP | 40.41 | Not recorded | 68fa772a1e209131cea71801 |
| M2IB | 26.4 | Not recorded | 68fa772a1e209131cea71801 |
| Grad-CAM | 18.45 | Not recorded | 68fa772a1e209131cea71801 |
| Rollout | 13.75 | Not recorded | 68fa772a1e209131cea71801 |
| raw attention | 12.19 | Not recorded | 68fa772a1e209131cea71801 |

### Grad-ECLIP

PG: 88.99. Result ID: 69fd99f047f34ce924e077a5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

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

### Grad-ECLIP

PG: 83.56. Result ID: 69fd99f047f34ce924e077a6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

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

### CLIPSurgery

PG: 57.59. Result ID: 69fd99f047f34ce924e077a3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74dd1e209131cea64460.md>)

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

### GAME

PG: 47.06. Result ID: 69fd99f047f34ce924e077a1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69a11e209131cea233a8.md>)

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

### MaskCLIP

PG: 40.41. Result ID: 69fd99f047f34ce924e077a2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d1f1e209131cea37ece.md>)

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

### M2IB

PG: 26.4. Result ID: 69fd99f047f34ce924e077a4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b61e209131cea5e107.md>)

[Implementation](<https://github.com/YingWANGG/M2IB>)

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

### Grad-CAM

PG: 18.45. Result ID: 69fd99f047f34ce924e077a0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66cb1e209131cea12568.md>)

[Implementation](<https://github.com/ramprs/grad-cam/>)

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

### Rollout

PG: 13.75. Result ID: 69fd99f047f34ce924e0779f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67ae1e209131cea17be6.md>)

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

### raw attention

PG: 12.19. Result ID: 69fd99f047f34ce924e0779e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa772a1e209131cea71801.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.