# Visual Hallucination Evaluation on POPE MS-COCO (Popular Sampling, val)

Source: [Visual Hallucination Evaluation on POPE MS-COCO (Popular Sampling, val)](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val>)

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

Data updated: 2026-02-26T05:29:31.976Z

Evaluation of model visual hallucination capabilities using the POPE methodology applied to the Popular sampling subset of the MS-COCO validation set.

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 Hallucination Evaluation |
| Dataset (including any stated split) | POPE MS-COCO Popular sampling (val) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Precision](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val.md?metric=Precision>): Higher is better; unit: PERCENT.

- [Recall](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val.md?metric=Recall>): Higher is better; unit: PERCENT.

- [F1 Score](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

- [Yes Rate](<https://www.sota2.com/research/sota/visual-hallucination-evaluation-on-pope-ms-coco-popular-sampling-val.md?metric=Yes%20Rate>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| InstructBLIP-14B | 82.77 | Not recorded | 68fa72941e209131cea5742f |
| DREAMLLM-7B | 80.07 | Not recorded | 68fa72941e209131cea5742f |
| MiniGPT-4-14B | 69.73 | Not recorded | 68fa72941e209131cea5742f |
| mPLUG-Owl-7B | 50.9 | Not recorded | 68fa72941e209131cea5742f |
| MMGPT-7B | 50 | Not recorded | 68fa72941e209131cea5742f |
| LLaVA-13B | 49.87 | Not recorded | 68fa72941e209131cea5742f |

### InstructBLIP-14B

Accuracy: 82.77. Result ID: 69925ff43a6c3148490e83ae.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fb71e209131cea47021.md>)

[Implementation](<https://github.com/salesforce/LAVIS/tree/main/projects/instructblip>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"14B"}. This record does not assert independent reproduction.

### DREAMLLM-7B

Accuracy: 80.07. Result ID: 69925ff43a6c3148490e83af.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Implementation](<https://dreamllm.github.io/>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"7B"}. This record does not assert independent reproduction.

### MiniGPT-4-14B

Accuracy: 69.73. Result ID: 69925ff43a6c3148490e83ad.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70bd1e209131cea4cf90.md>)

[Implementation](<https://minigpt-4.github.io/>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"14B"}. This record does not assert independent reproduction.

### mPLUG-Owl-7B

Accuracy: 50.9. Result ID: 69925ff43a6c3148490e83aa.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ae1e209131cea57d83.md>)

[Implementation](<https://github.com/X-PLUG/mPLUG-Owl>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"7B"}. This record does not assert independent reproduction.

### MMGPT-7B

Accuracy: 50. Result ID: 69925ff43a6c3148490e83ac.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fb01e209131cea46e99.md>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"7B"}. This record does not assert independent reproduction.

### LLaVA-13B

Accuracy: 49.87. Result ID: 69925ff43a6c3148490e83ab.

[Reporting paper](<https://www.sota2.com/research/paper/68fa72941e209131cea5742f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa717d1e209131cea51970.md>)

[Implementation](<https://llava-vl.github.io/>)

Recorded evaluation configuration: {"zero-shot":true,"backbone":"13B"}. This record does not assert independent reproduction.

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