# Grad-CAM Evaluation on ImageNet

Source: [Grad-CAM Evaluation on ImageNet](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet>)

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

Data updated: 2026-02-26T05:21:21.588Z

Evaluation of the fidelity and quality of Grad-CAM explanations generated for models trained on the ImageNet dataset, measured by metrics such as ADCC, Coherence, and Complexity.

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 | Grad-CAM Evaluation |
| Dataset (including any stated split) | ImageNet |
| Metric | A.D |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [A.D](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=A.D>): Higher is better; unit: SCALAR.

- [A.I](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=A.I>): Higher is better; unit: SCALAR.

- [A.G](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=A.G>): Higher is better; unit: SCALAR.

- [Coherence](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=Coherence>): Higher is better; unit: PERCENT.

- [Complexity](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=Complexity>): Higher is better; unit: SCALAR.

- [ADCC](<https://www.sota2.com/research/sota/grad-cam-evaluation-on-imagenet.md?metric=ADCC>): Higher is better; unit: PERCENT.

## Recorded results

| Method | A.D | Result date | Report paper |
| --- | --- | --- | --- |
| LIFT | 66.9 | Not recorded | 68fa79721e209131cea7daa3 |
| CARAFE | 49.9 | Not recorded | 68fa79721e209131cea7daa3 |
| Large Image (x8) | 48.8 | Not recorded | 68fa79721e209131cea7daa3 |
| Strided | 19.8 | Not recorded | 68fa79721e209131cea7daa3 |
| Bilinear | 19 | Not recorded | 68fa79721e209131cea7daa3 |
| Dysample | 17.8 | Not recorded | 68fa79721e209131cea7daa3 |
| JAFAR | 17.4 | Not recorded | 68fa79721e209131cea7daa3 |
| FeatUp | 15.3 | Not recorded | 68fa79721e209131cea7daa3 |
| ReSFU | 14.5 | Not recorded | 68fa79721e209131cea7daa3 |
| SAPA | 8.5 | Not recorded | 68fa79721e209131cea7daa3 |

### LIFT

A.D: 66.9. Result ID: 698b4f178d67a41d0a6cf85a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756a1e209131cea67b87.md>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Agnostic"}. This record does not assert independent reproduction.

### CARAFE

A.D: 49.9. Result ID: 698b4f178d67a41d0a6cf855.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66911e209131cea10f20.md>)

[Implementation](<https://github.com/open-mmlab/mmdetection>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Dependent"}. This record does not assert independent reproduction.

### Large Image (x8)

A.D: 48.8. Result ID: 698b4f178d67a41d0a6cf853.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

Recorded evaluation configuration: {"Upsampling Category":"Training-free"}. This record does not assert independent reproduction.

### Strided

A.D: 19.8. Result ID: 698b4f178d67a41d0a6cf854.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

Recorded evaluation configuration: {"Upsampling Category":"Training-free"}. This record does not assert independent reproduction.

### Bilinear

A.D: 19. Result ID: 698b4f178d67a41d0a6cf852.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

Recorded evaluation configuration: {"Upsampling Category":"Training-free"}. This record does not assert independent reproduction.

### Dysample

A.D: 17.8. Result ID: 698b4f178d67a41d0a6cf857.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70781e209131cea4b50b.md>)

[Implementation](<https://github.com/tiny-smart/dysample>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Dependent"}. This record does not assert independent reproduction.

### JAFAR

A.D: 17.4. Result ID: 698b4f178d67a41d0a6cf85b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Implementation](<https://jafar-upsampler.github.io/>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Agnostic"}. This record does not assert independent reproduction.

### FeatUp

A.D: 15.3. Result ID: 698b4f178d67a41d0a6cf859.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72c11e209131cea5869f.md>)

[Implementation](<https://tinyurl.com/28h3yppa>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Agnostic"}. This record does not assert independent reproduction.

### ReSFU

A.D: 14.5. Result ID: 698b4f178d67a41d0a6cf858.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76ea1e209131cea6fee4.md>)

[Implementation](<https://github.com/zmhhmz/ReSFU>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Dependent"}. This record does not assert independent reproduction.

### SAPA

A.D: 8.5. Result ID: 698b4f178d67a41d0a6cf856.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79721e209131cea7daa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e481e209131cea3ef7d.md>)

[Implementation](<https://github.com/poppinace/sapa>)

Recorded evaluation configuration: {"Upsampling Category":"Task-Dependent"}. This record does not assert independent reproduction.

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