# Image-to-Sentence Retrieval on MS-COCO 5K full (test)

Source: [Image-to-Sentence Retrieval on MS-COCO 5K full (test)](<https://www.sota2.com/research/sota/image-to-sentence-retrieval-on-ms-coco-5k-full-test>)

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

Data updated: 2026-02-26T06:01:24.787Z

Evaluation of Image-to-Sentence Retrieval performance using the full 5K test subset of the MS-COCO dataset, measured using Recall@K metrics and rSum.

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 | Image-to-Sentence Retrieval |
| Dataset (including any stated split) | MS-COCO 5K full (test) |
| Metric | Recall@1 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 35 of 35 |
| 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

- [Recall@1](<https://www.sota2.com/research/sota/image-to-sentence-retrieval-on-ms-coco-5k-full-test.md?metric=Recall%401>): Higher is better; unit: PERCENT.

- [Recall@5](<https://www.sota2.com/research/sota/image-to-sentence-retrieval-on-ms-coco-5k-full-test.md?metric=Recall%405>): Higher is better; unit: PERCENT.

- [Recall@10](<https://www.sota2.com/research/sota/image-to-sentence-retrieval-on-ms-coco-5k-full-test.md?metric=Recall%4010>): Higher is better; unit: PERCENT.

- [rSum](<https://www.sota2.com/research/sota/image-to-sentence-retrieval-on-ms-coco-5k-full-test.md?metric=rSum>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Recall@1 | Result date | Report paper |
| --- | --- | --- | --- |
| ALIGN | 77 | Not recorded | 68fa730e1e209131cea5a297 |
| ALBEF | 73.1 | Not recorded | 68fa730e1e209131cea5a297 |
| OSCAR | 70 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 67.9 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 67.9 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 67.1 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 66.2 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 65.3 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 64.1 | Not recorded | 68fa730e1e209131cea5a297 |
| VSE∞ w/ Seg. | 63.9 | Not recorded | 68fa730e1e209131cea5a297 |
| 3SHNet | 63.8 | Not recorded | 68fa730e1e209131cea5a297 |
| Imp. | 63.5 | Not recorded | 68fa730e1e209131cea5a297 |
| UNITER | 63.3 | Not recorded | 68fa730e1e209131cea5a297 |
| VSE∞ | 62.5 | Not recorded | 68fa730e1e209131cea5a297 |
| Unicoder | 62.3 | Not recorded | 68fa730e1e209131cea5a297 |
| CMSEI | 61.5 | Not recorded | 68fa730e1e209131cea5a297 |
| CMCAN | 61.5 | Not recorded | 68fa730e1e209131cea5a297 |
| MKTLON | 61.4 | Not recorded | 68fa730e1e209131cea5a297 |
| DCIN | 60.8 | Not recorded | 68fa730e1e209131cea5a297 |
| KIDRR | 60.3 | Not recorded | 68fa730e1e209131cea5a297 |
| HGAN | 60 | Not recorded | 68fa730e1e209131cea5a297 |
| AME | 59.9 | Not recorded | 68fa730e1e209131cea5a297 |
| CHAN | 59.8 | Not recorded | 68fa730e1e209131cea5a297 |
| DIME | 59.3 | Not recorded | 68fa730e1e209131cea5a297 |
| VSE∞ | 59.1 | Not recorded | 68fa730e1e209131cea5a297 |
| NAAF | 58.9 | Not recorded | 68fa730e1e209131cea5a297 |
| ALIGN | 58.6 | Not recorded | 68fa730e1e209131cea5a297 |
| CLIP | 58.4 | Not recorded | 68fa730e1e209131cea5a297 |
| VSE∞ | 58.3 | Not recorded | 68fa730e1e209131cea5a297 |
| VILBERT | 57.5 | Not recorded | 68fa730e1e209131cea5a297 |
| RCTRN | 57.1 | Not recorded | 68fa730e1e209131cea5a297 |
| VSRN++ | 54.7 | Not recorded | 68fa730e1e209131cea5a297 |
| IMRAM | 53.7 | Not recorded | 68fa730e1e209131cea5a297 |
| CRGN | 51.2 | Not recorded | 68fa730e1e209131cea5a297 |
| SCO | 42.8 | Not recorded | 68fa730e1e209131cea5a297 |

### ALIGN

Recall@1: 77. Result ID: 6992a8e03a6c314849107ffc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a201e209131cea261e7.md>)

Recorded evaluation configuration: {"Pretrain":"1.8B","GPUs":"1024 TPUv3","Bs":"16384","Zero-shot":false}. This record does not assert independent reproduction.

### ALBEF

Recall@1: 73.1. Result ID: 6992a8e03a6c314849107ffd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6aef1e209131cea2b2da.md>)

[Implementation](<https://github.com/salesforce/ALBEF>)

Recorded evaluation configuration: {"Pretrain":"4.0M","GPUs":"8 A100","Bs":"512"}. This record does not assert independent reproduction.

### OSCAR

Recall@1: 70. Result ID: 6992a8e03a6c314849107ff9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa680e1e209131cea19edd.md>)

[Implementation](<https://github.com/microsoft/Oscar>)

Recorded evaluation configuration: {"Pretrain":"6.5M","GPUs":"16 V100","Bs":"1,024"}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 67.9. Result ID: 6992a8e03a6c314849107fdb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":true}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 67.9. Result ID: 6992a8e03a6c314849107ffe.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Pretrain":null,"GPUs":"1 TitanX","Bs":"256"}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 67.1. Result ID: 6992a8e03a6c314849107fda.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 66.2. Result ID: 6992a8e03a6c314849107fd3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Grid","Ensemble":true}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 65.3. Result ID: 6992a8e03a6c314849107fcf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 64.1. Result ID: 6992a8e03a6c314849107fd2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Grid","Ensemble":false}. This record does not assert independent reproduction.

### VSE∞ w/ Seg.

Recall@1: 63.9. Result ID: 6992a8e03a6c314849107fd9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a571e209131cea27881.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### 3SHNet

Recall@1: 63.8. Result ID: 6992a8e03a6c314849107fce.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Implementation](<https://github.com/XuriGe1995/3SHNet>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":false}. This record does not assert independent reproduction.

### Imp.

Recall@1: 63.5. Result ID: 6992a8e03a6c314849107fd7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":true}. This record does not assert independent reproduction.

### UNITER

Recall@1: 63.3. Result ID: 6992a8e03a6c314849107ff7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67f91e209131cea1989f.md>)

[Implementation](<https://github.com/ChenRocks/UNITER>)

Recorded evaluation configuration: {"Pretrain":"9.6M","GPUs":"16 V100","Bs":"192"}. This record does not assert independent reproduction.

### VSE∞

Recall@1: 62.5. Result ID: 6992a8e03a6c314849107fd5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a571e209131cea27881.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### Unicoder

Recall@1: 62.3. Result ID: 6992a8e03a6c314849107ff8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66cb1e209131cea1258f.md>)

Recorded evaluation configuration: {"Pretrain":"3.8M","GPUs":"4 V100"}. This record does not assert independent reproduction.

### CMSEI

Recall@1: 61.5. Result ID: 6992a8e03a6c314849107fc8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dad1e209131cea3b735.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### CMCAN

Recall@1: 61.5. Result ID: 6992a8e03a6c314849107fd6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### MKTLON

Recall@1: 61.4. Result ID: 6992a8e03a6c314849107fcd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### DCIN

Recall@1: 60.8. Result ID: 6992a8e03a6c314849107fcc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74911e209131cea6290d.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### KIDRR

Recall@1: 60.3. Result ID: 6992a8e03a6c314849107fcb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### HGAN

Recall@1: 60. Result ID: 6992a8e03a6c314849107fd8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e401e209131cea3eb77.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### AME

Recall@1: 59.9. Result ID: 6992a8e03a6c314849107fc7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### CHAN

Recall@1: 59.8. Result ID: 6992a8e03a6c314849107fc9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":false}. This record does not assert independent reproduction.

### DIME

Recall@1: 59.3. Result ID: 6992a8e03a6c314849107fc4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### VSE∞

Recall@1: 59.1. Result ID: 6992a8e03a6c314849107fd1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a571e209131cea27881.md>)

Recorded evaluation configuration: {"Type":"Grid","Ensemble":false}. This record does not assert independent reproduction.

### NAAF

Recall@1: 58.9. Result ID: 6992a8e03a6c314849107fc6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### ALIGN

Recall@1: 58.6. Result ID: 6992a8e03a6c314849107ffb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a201e209131cea261e7.md>)

Recorded evaluation configuration: {"Pretrain":"1.8B","GPUs":"1024 TPUv3","Bs":"16384","Zero-shot":true}. This record does not assert independent reproduction.

### CLIP

Recall@1: 58.4. Result ID: 6992a8e03a6c314849107ffa.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Pretrain":"400M","GPUs":"592 V100","Bs":"32,768","Zero-shot":true}. This record does not assert independent reproduction.

### VSE∞

Recall@1: 58.3. Result ID: 6992a8e03a6c314849107fc3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a571e209131cea27881.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":false}. This record does not assert independent reproduction.

### VILBERT

Recall@1: 57.5. Result ID: 6992a8e03a6c314849107ff6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa686c1e209131cea1c117.md>)

Recorded evaluation configuration: {"Pretrain":"3.3M","GPUs":"8 TitanX","Bs":"512"}. This record does not assert independent reproduction.

### RCTRN

Recall@1: 57.1. Result ID: 6992a8e03a6c314849107fca.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### VSRN++

Recall@1: 54.7. Result ID: 6992a8e03a6c314849107fc5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### IMRAM

Recall@1: 53.7. Result ID: 6992a8e03a6c314849107fc2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67341e209131cea14be0.md>)

[Implementation](<https://github.com/HuiChen24/IMRAM>)

Recorded evaluation configuration: {"Type":"Region","Ensemble":true}. This record does not assert independent reproduction.

### CRGN

Recall@1: 51.2. Result ID: 6992a8e03a6c314849107fd4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

Recorded evaluation configuration: {"Type":"Region+Grid","Ensemble":false}. This record does not assert independent reproduction.

### SCO

Recall@1: 42.8. Result ID: 6992a8e03a6c314849107fd0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa730e1e209131cea5a297.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63da1e209131cea01552.md>)

Recorded evaluation configuration: {"Type":"Grid","Ensemble":false}. This record does not assert independent reproduction.

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