<div id="dwpdlx_short_desc">Selects best keypoints from over-sampled data.</div>
**Function:** Reduces an over-sampled (high FPS) keypoint dataset by selecting the highest-confidence individual keypoints from each frame chunk.
### Inputs
- **pose_keypoints:** The keypoint dataset (e.g., from a 2x VFI source).
### Parameters
- **factor:** The downsampling factor (e.g., 2 for 2x source to 1x output).
### Outputs
- **pose_keypoints:** The downsampled keypoint data with cherry-picked confident points.
### Notes
- This node is designed to combat jitter and noise from detection models by leveraging over-sampling.
- It is ideal for use after generating keypoints from a VFI (Video Frame Interpolation) output.
- Logs a summary of confidence gain per body point upon completion.
By hobinrude·Created 9 months ago·Updated 8 months ago· 10