# Video-to-Text Retrieval on MSRVTT 1kA (Noise Robustness)

Source: [Video-to-Text Retrieval on MSRVTT 1kA (Noise Robustness)](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness>)

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

Data updated: 2026-04-24T07:00:47.835Z

Evaluation of Video-to-Text Retrieval performance on the MSRVTT 1kA subset, specifically measuring robustness using Recall@1 under increasing levels of video noise (20% to 80%).

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 | Video-to-Text Retrieval |
| Dataset (including any stated split) | MSRVTT 1kA |
| Metric | Recall@1 (20% Noise) |
| 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

- [Recall@1 (20% Noise)](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness.md?metric=Recall%401%20(20%25%20Noise)>): Higher is better; unit: PERCENT.

- [Recall@1 (40% Noise)](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness.md?metric=Recall%401%20(40%25%20Noise)>): Higher is better; unit: PERCENT.

- [Recall@1 (60% Noise)](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness.md?metric=Recall%401%20(60%25%20Noise)>): Higher is better; unit: PERCENT.

- [Recall@1 (80% Noise)](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness.md?metric=Recall%401%20(80%25%20Noise)>): Higher is better; unit: PERCENT.

- [Average Recall@1](<https://www.sota2.com/research/sota/video-to-text-retrieval-on-msrvtt-1ka-noise-robustness.md?metric=Average%20Recall%401>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Recall@1 (20% Noise) | Result date | Report paper |
| --- | --- | --- | --- |
| HAT-VTR | 41.6 | Not recorded | 69eaf90255a3c32af89f888d |
| EATA | 39.9 | Not recorded | 69eaf90255a3c32af89f888d |
| Tent | 39 | Not recorded | 69eaf90255a3c32af89f888d |
| SAR | 38.9 | Not recorded | 69eaf90255a3c32af89f888d |
| TCR | 38.6 | Not recorded | 69eaf90255a3c32af89f888d |
| READ | 38.5 | Not recorded | 69eaf90255a3c32af89f888d |
| Xpool | 38.4 | Not recorded | 69eaf90255a3c32af89f888d |
| HAT-VTR | 37.4 | Not recorded | 69eaf90255a3c32af89f888d |
| CLIP4Clip | 36.7 | Not recorded | 69eaf90255a3c32af89f888d |
| EATA | 36.6 | Not recorded | 69eaf90255a3c32af89f888d |
| SAR | 36.4 | Not recorded | 69eaf90255a3c32af89f888d |
| Tent | 36.3 | Not recorded | 69eaf90255a3c32af89f888d |
| READ | 36.1 | Not recorded | 69eaf90255a3c32af89f888d |
| TCR | 36.1 | Not recorded | 69eaf90255a3c32af89f888d |

### HAT-VTR

Recall@1 (20% Noise): 41.6. Result ID: 69eb151647f34ce924d2b9b4.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"Ours"}. This record does not assert independent reproduction.

### EATA

Recall@1 (20% Noise): 39.9. Result ID: 69eb151647f34ce924d2b9b2.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ca11e209131cea35123.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"EATA"}. This record does not assert independent reproduction.

### Tent

Recall@1 (20% Noise): 39. Result ID: 69eb151647f34ce924d2b9af.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69941e209131cea22d2b.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"Tent"}. This record does not assert independent reproduction.

### SAR

Recall@1 (20% Noise): 38.9. Result ID: 69eb151647f34ce924d2b9b1.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e961e209131cea40e77.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"SAR"}. This record does not assert independent reproduction.

### TCR

Recall@1 (20% Noise): 38.6. Result ID: 69eb151647f34ce924d2b9b3.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75561e209131cea6749e.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"TCR"}. This record does not assert independent reproduction.

### READ

Recall@1 (20% Noise): 38.5. Result ID: 69eb151647f34ce924d2b9b0.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"READ"}. This record does not assert independent reproduction.

### Xpool

Recall@1 (20% Noise): 38.4. Result ID: 69eb151647f34ce924d2b9ae.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c161e209131cea320b3.md>)

[Implementation](<https://www.sota2.com/layer6ai-labs.github.io/xpool/>)

Recorded evaluation configuration: {"Backbone":"Xpool","Adaptation Method":"None"}. This record does not assert independent reproduction.

### HAT-VTR

Recall@1 (20% Noise): 37.4. Result ID: 69eb151647f34ce924d2b9ad.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"Ours"}. This record does not assert independent reproduction.

### CLIP4Clip

Recall@1 (20% Noise): 36.7. Result ID: 69eb151647f34ce924d2b9a7.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"None"}. This record does not assert independent reproduction.

### EATA

Recall@1 (20% Noise): 36.6. Result ID: 69eb151647f34ce924d2b9ab.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ca11e209131cea35123.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"EATA"}. This record does not assert independent reproduction.

### SAR

Recall@1 (20% Noise): 36.4. Result ID: 69eb151647f34ce924d2b9aa.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e961e209131cea40e77.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"SAR"}. This record does not assert independent reproduction.

### Tent

Recall@1 (20% Noise): 36.3. Result ID: 69eb151647f34ce924d2b9a8.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69941e209131cea22d2b.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"Tent"}. This record does not assert independent reproduction.

### READ

Recall@1 (20% Noise): 36.1. Result ID: 69eb151647f34ce924d2b9a9.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"READ"}. This record does not assert independent reproduction.

### TCR

Recall@1 (20% Noise): 36.1. Result ID: 69eb151647f34ce924d2b9ac.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90255a3c32af89f888d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75561e209131cea6749e.md>)

Recorded evaluation configuration: {"Backbone":"CLIP4Clip","Adaptation Method":"TCR"}. This record does not assert independent reproduction.

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