# Text-to-image person search on ICFG-PEDES 20% noise

Source: [Text-to-image person search on ICFG-PEDES 20% noise](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise>)

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

Data updated: 2026-03-04T07:33:56.805Z

Evaluation of Text-to-image person search performance on the ICFG-PEDES dataset configured with 20% noise, using rank-based metrics (R-1, R-5, R-10) and mAP/mINP.

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 | Text-to-image person search |
| Dataset (including any stated split) | ICFG-PEDES 20% noise |
| Metric | R-1 Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

- [R-1 Accuracy](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise.md?metric=R-1%20Accuracy>): Higher is better; unit: PERCENT.

- [R-5 Accuracy](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise.md?metric=R-5%20Accuracy>): Higher is better; unit: PERCENT.

- [R-10 Accuracy](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise.md?metric=R-10%20Accuracy>): Higher is better; unit: PERCENT.

- [mAP](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise.md?metric=mAP>): Higher is better; unit: PERCENT.

- [mINP](<https://www.sota2.com/research/sota/text-to-image-person-search-on-icfg-pedes-20-percent-noise.md?metric=mINP>): Higher is better; unit: PERCENT.

## Recorded results

| Method | R-1 Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| DURA | 66.62 | Not recorded | 68fa7ba11e209131cea8978c |
| RDE\* | 66.54 | Not recorded | 68fa7ba11e209131cea8978c |
| DURA | 66.53 | Not recorded | 68fa7ba11e209131cea8978c |
| RDE\* | 66.51 | Not recorded | 68fa7ba11e209131cea8978c |
| DECL | 61.95 | Not recorded | 68fa7ba11e209131cea8978c |
| DECL | 61.95 | Not recorded | 68fa7ba11e209131cea8978c |
| IRRA | 60.76 | Not recorded | 68fa7ba11e209131cea8978c |
| IRRA | 60.58 | Not recorded | 68fa7ba11e209131cea8978c |
| CLIP-C | 55.25 | Not recorded | 68fa7ba11e209131cea8978c |
| CLIP-C | 55.17 | Not recorded | 68fa7ba11e209131cea8978c |
| IVT | 50.21 | Not recorded | 68fa7ba11e209131cea8978c |
| IVT | 48.7 | Not recorded | 68fa7ba11e209131cea8978c |
| SSAN | 40.57 | Not recorded | 68fa7ba11e209131cea8978c |
| SSAN | 40.28 | Not recorded | 68fa7ba11e209131cea8978c |

### DURA

R-1 Accuracy: 66.62. Result ID: 69a7e01d6dbb7075b94c6584.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### RDE\*

R-1 Accuracy: 66.54. Result ID: 69a7e01d6dbb7075b94c6582.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ac1e209131cea57c03.md>)

[Implementation](<https://github.com/QinYang79/RDE>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### DURA

R-1 Accuracy: 66.53. Result ID: 69a7e01d6dbb7075b94c6585.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### RDE\*

R-1 Accuracy: 66.51. Result ID: 69a7e01d6dbb7075b94c6583.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ac1e209131cea57c03.md>)

[Implementation](<https://github.com/QinYang79/RDE>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### DECL

R-1 Accuracy: 61.95. Result ID: 69a7e01d6dbb7075b94c6580.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### DECL

R-1 Accuracy: 61.95. Result ID: 69a7e01d6dbb7075b94c6581.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### IRRA

R-1 Accuracy: 60.76. Result ID: 69a7e01d6dbb7075b94c657c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecd1e209131cea423e9.md>)

[Implementation](<https://github.com/anosorae/IRRA>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### IRRA

R-1 Accuracy: 60.58. Result ID: 69a7e01d6dbb7075b94c657d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ecd1e209131cea423e9.md>)

[Implementation](<https://github.com/anosorae/IRRA>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### CLIP-C

R-1 Accuracy: 55.25. Result ID: 69a7e01d6dbb7075b94c657e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### CLIP-C

R-1 Accuracy: 55.17. Result ID: 69a7e01d6dbb7075b94c657f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### IVT

R-1 Accuracy: 50.21. Result ID: 69a7e01d6dbb7075b94c657a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c97.md>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### IVT

R-1 Accuracy: 48.7. Result ID: 69a7e01d6dbb7075b94c657b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d491e209131cea38c97.md>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

### SSAN

R-1 Accuracy: 40.57. Result ID: 69a7e01d6dbb7075b94c6578.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a791e209131cea28668.md>)

[Implementation](<https://github.com/zifyloo/SSAN>)

Recorded evaluation configuration: {"Type":"Best"}. This record does not assert independent reproduction.

### SSAN

R-1 Accuracy: 40.28. Result ID: 69a7e01d6dbb7075b94c6579.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea8978c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a791e209131cea28668.md>)

[Implementation](<https://github.com/zifyloo/SSAN>)

Recorded evaluation configuration: {"Type":"Last"}. This record does not assert independent reproduction.

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