# Keypoint Detection on MP-100 (All Splits (1-5))

Source: [Keypoint Detection on MP-100 (All Splits (1-5))](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5>)

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

Data updated: 2026-02-26T05:52:12.438Z

Evaluation of Keypoint Detection performance across all five splits (1 through 5) of the MP-100 dataset, typically reporting Percentage of Correct Keypoints (PCK) per split and the overall mean.

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 | Keypoint Detection |
| Dataset (including any stated split) | MP-100 (All Splits (1-5)) |
| Metric | Mean PCK |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [Mean PCK](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=Mean%20PCK>): Higher is better; unit: PERCENT.

- [PCK (Split 1)](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=PCK%20(Split%201)>): Higher is better; unit: PERCENT.

- [PCK (Split 2)](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=PCK%20(Split%202)>): Higher is better; unit: PERCENT.

- [PCK (Split 3)](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=PCK%20(Split%203)>): Higher is better; unit: PERCENT.

- [PCK (Split 4)](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=PCK%20(Split%204)>): Higher is better; unit: PERCENT.

- [PCK (Split 5)](<https://www.sota2.com/research/sota/keypoint-detection-on-mp-100-all-splits-1-5.md?metric=PCK%20(Split%205)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Mean PCK | Result date | Report paper |
| --- | --- | --- | --- |
| X-Pose | 86.14 | Not recorded | 68fa74211e209131cea5fd32 |
| CapeFormer | 85.31 | Not recorded | 68fa74211e209131cea5fd32 |
| CapeFormer | 81.31 | Not recorded | 68fa74211e209131cea5fd32 |
| POMNet | 79.7 | Not recorded | 68fa74211e209131cea5fd32 |
| Fine-tune | 63.58 | Not recorded | 68fa74211e209131cea5fd32 |
| MAML | 61.5 | Not recorded | 68fa74211e209131cea5fd32 |
| ProtoNet | 44.78 | Not recorded | 68fa74211e209131cea5fd32 |

### X-Pose

Mean PCK: 86.14. Result ID: 69931f893a6c31484913e028.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Implementation](<https://github.com/IDEA-Research/UniPose>)

Recorded evaluation configuration: {"Strategy":"End-to-End (E2E)","Backbone":"ResNet-50","Time":"59 ms"}. This record does not assert independent reproduction.

### CapeFormer

Mean PCK: 85.31. Result ID: 69931f893a6c31484913e026.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50","Require keypoint identifiers during test":true,"Time":"57x N ms"}. This record does not assert independent reproduction.

### CapeFormer

Mean PCK: 81.31. Result ID: 69931f893a6c31484913e027.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50","Require keypoint identifiers during test":false,"Time":"57x N ms"}. This record does not assert independent reproduction.

### POMNet

Mean PCK: 79.7. Result ID: 69931f893a6c31484913e025.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d101e209131cea37640.md>)

[Implementation](<https://github.com/luminxu/Pose-for-Everything>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50","Time":"151x N ms"}. This record does not assert independent reproduction.

### Fine-tune

Mean PCK: 63.58. Result ID: 69931f893a6c31484913e024.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66751e209131cea1039b.md>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### MAML

Mean PCK: 61.5. Result ID: 69931f893a6c31484913e023.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63861e209131ce9ff3e8.md>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50"}. This record does not assert independent reproduction.

### ProtoNet

Mean PCK: 44.78. Result ID: 69931f893a6c31484913e022.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74211e209131cea5fd32.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa636e1e209131ce9febd3.md>)

Recorded evaluation configuration: {"Strategy":"Top-Down (TD)","Backbone":"ResNet-50"}. This record does not assert independent reproduction.

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