# Human Correlation for Image Captioning Evaluation on VICR

Source: [Human Correlation for Image Captioning Evaluation on VICR](<https://www.sota2.com/research/sota/human-correlation-for-image-captioning-evaluation-on-vicr>)

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

Data updated: 2026-05-22T06:58:32.848Z

Evaluation of Image Captioning correlation using Kendall's tau-c on the VICR dataset.

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 | Human Correlation for Image Captioning Evaluation |
| Dataset (including any stated split) | VICR |
| Metric | Kendall's tau-c |
| Unit | SCALAR |
| 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

- [Kendall's tau-c](<https://www.sota2.com/research/sota/human-correlation-for-image-captioning-evaluation-on-vicr.md?metric=Kendall%27s%20tau-c>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Kendall's tau-c | Result date | Report paper |
| --- | --- | --- | --- |
| BEiTScore-L/14 | 78.8 | Not recorded | 6a0fe2eff984739ed6739c5e |
| BEiTScore-L/14 w/o Stage 1 | 78.3 | Not recorded | 6a0fe2eff984739ed6739c5e |
| BEiTScore-B/32 | 78.2 | Not recorded | 6a0fe2eff984739ed6739c5e |
| BEiTScore-B/32 w/o Stage 1 | 77.7 | Not recorded | 6a0fe2eff984739ed6739c5e |
| EXPERT | 77.4 | Not recorded | 6a0fe2eff984739ed6739c5e |
| PAC-S++/ViT-L | 75.2 | Not recorded | 6a0fe2eff984739ed6739c5e |
| FLEUR | 74.7 | Not recorded | 6a0fe2eff984739ed6739c5e |
| PAC-S/ViT-L | 73.2 | Not recorded | 6a0fe2eff984739ed6739c5e |
| PAC-S++/ViT-B | 72.7 | Not recorded | 6a0fe2eff984739ed6739c5e |
| CLIP-S/ViT-L | 72.6 | Not recorded | 6a0fe2eff984739ed6739c5e |
| PAC-S/ViT-B | 71.4 | Not recorded | 6a0fe2eff984739ed6739c5e |
| CLIP-S/ViT-B | 70.8 | Not recorded | 6a0fe2eff984739ed6739c5e |
| LongCLIP-S/ViT-B | 70.2 | Not recorded | 6a0fe2eff984739ed6739c5e |
| VQAScore (CLIP-FlanT5-XXL) | 69.8 | Not recorded | 6a0fe2eff984739ed6739c5e |
| LongCLIP-S/ViT-L | 69 | Not recorded | 6a0fe2eff984739ed6739c5e |
| SPECS | 33.4 | Not recorded | 6a0fe2eff984739ed6739c5e |

### BEiTScore-L/14

Kendall's tau-c: 78.8. Result ID: 6a0ffe7afbc373592c0861cd.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

Recorded evaluation configuration: {"Size (M)":"674.69","GFLOPs":"296.73"}. This record does not assert independent reproduction.

### BEiTScore-L/14 w/o Stage 1

Kendall's tau-c: 78.3. Result ID: 6a0ffe7afbc373592c0861cb.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

Recorded evaluation configuration: {"Size (M)":"674.69","GFLOPs":"296.73"}. This record does not assert independent reproduction.

### BEiTScore-B/32

Kendall's tau-c: 78.2. Result ID: 6a0ffe7afbc373592c0861cc.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

Recorded evaluation configuration: {"Size (M)":"222.28","GFLOPs":"83.85"}. This record does not assert independent reproduction.

### BEiTScore-B/32 w/o Stage 1

Kendall's tau-c: 77.7. Result ID: 6a0ffe7afbc373592c0861ca.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

Recorded evaluation configuration: {"Size (M)":"222.28","GFLOPs":"83.85"}. This record does not assert independent reproduction.

### EXPERT

Kendall's tau-c: 77.4. Result ID: 6a0ffe7afbc373592c0861bf.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79c71e209131cea7f518.md>)

[Implementation](<https://github.com/hjkim811/EXPERT>)

Recorded evaluation configuration: {"Size (M)":"13350.84","GFLOPs":"9317.47"}. This record does not assert independent reproduction.

### PAC-S++/ViT-L

Kendall's tau-c: 75.2. Result ID: 6a0ffe7afbc373592c0861c8.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a441e209131cea820a0.md>)

Recorded evaluation configuration: {"Size (M)":"430.58","GFLOPs":"56.26"}. This record does not assert independent reproduction.

### FLEUR

Kendall's tau-c: 74.7. Result ID: 6a0ffe7afbc373592c0861be.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73881e209131cea5d06e.md>)

[Implementation](<https://github.com/Yebin46/FLEUR>)

Recorded evaluation configuration: {"Size (M)":"13350.84","GFLOPs":"9317.47"}. This record does not assert independent reproduction.

### PAC-S/ViT-L

Kendall's tau-c: 73.2. Result ID: 6a0ffe7afbc373592c0861c6.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70201e209131cea493e8.md>)

[Implementation](<https://github.com/aimagelab/pacscore>)

Recorded evaluation configuration: {"Size (M)":"427.62","GFLOPs":"56.26"}. This record does not assert independent reproduction.

### PAC-S++/ViT-B

Kendall's tau-c: 72.7. Result ID: 6a0ffe7afbc373592c0861c7.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a441e209131cea820a0.md>)

Recorded evaluation configuration: {"Size (M)":"155.42","GFLOPs":"4.88"}. This record does not assert independent reproduction.

### CLIP-S/ViT-L

Kendall's tau-c: 72.6. Result ID: 6a0ffe7afbc373592c0861c2.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c0f1e209131cea31cc9.md>)

Recorded evaluation configuration: {"Size (M)":"427.62","GFLOPs":"84.36"}. This record does not assert independent reproduction.

### PAC-S/ViT-B

Kendall's tau-c: 71.4. Result ID: 6a0ffe7afbc373592c0861c5.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70201e209131cea493e8.md>)

[Implementation](<https://github.com/aimagelab/pacscore>)

Recorded evaluation configuration: {"Size (M)":"151.28","GFLOPs":"4.88"}. This record does not assert independent reproduction.

### CLIP-S/ViT-B

Kendall's tau-c: 70.8. Result ID: 6a0ffe7afbc373592c0861c1.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c0f1e209131cea31cc9.md>)

Recorded evaluation configuration: {"Size (M)":"151.28","GFLOPs":"7.28"}. This record does not assert independent reproduction.

### LongCLIP-S/ViT-B

Kendall's tau-c: 70.2. Result ID: 6a0ffe7afbc373592c0861c3.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74361e209131cea60770.md>)

Recorded evaluation configuration: {"Size (M)":"149.84","GFLOPs":"26.24"}. This record does not assert independent reproduction.

### VQAScore (CLIP-FlanT5-XXL)

Kendall's tau-c: 69.8. Result ID: 6a0ffe7afbc373592c0861c0.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73a81e209131cea5dbb3.md>)

Recorded evaluation configuration: {"Size (M)":"11459.59","GFLOPs":"3742.1"}. This record does not assert independent reproduction.

### LongCLIP-S/ViT-L

Kendall's tau-c: 69. Result ID: 6a0ffe7afbc373592c0861c4.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74361e209131cea60770.md>)

Recorded evaluation configuration: {"Size (M)":"427.75","GFLOPs":"98.9"}. This record does not assert independent reproduction.

### SPECS

Kendall's tau-c: 33.4. Result ID: 6a0ffe7afbc373592c0861c9.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe2eff984739ed6739c5e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b401e209131cea874df.md>)

Recorded evaluation configuration: {"Size (M)":"151.49","GFLOPs":"9.2"}. This record does not assert independent reproduction.

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