Knee sweep
147 deg
Average left/right knee angle range in the processed clip.
Gallery
Championship 1500m sample
Josh Kerr
Kerr's sample is useful because the clip is crowded but still readable. The model has to keep choosing the right runner instead of the easiest body in frame.
Upright pack running, high rhythm proxy, and a strong knee-angle sweep from the usable frames.
Stride readout
These are clip-derived signals, not a coaching diagnosis. The point is to show what the pipeline can read and where the video still needs review.
Hip bounce
8.7%
Normalized hip vertical oscillation proxy from pose landmarks.
Readable frames
100%
Frames where the pipeline kept enough pose signal to use the stride.
Processed clips
5.83s146 of 146 usable frames
Championship pack segment
A 1500m race clip with multiple runners close together and a broadcast camera angle.
Stride rhythm proxy
9.42
Peak spacing in the lower-body motion signal.
Knee lift proxy
30%
Relative vertical knee travel in the normalized pose.
Pose visibility
72%
Mean landmark visibility across the usable clip.
- The rhythm proxy is the highest in the current gallery set.
- Knee sweep is large, especially on the right-side read.
- Pose visibility is lower than the cleaner clips, which makes this a useful identity and tracking test.
Runner isolation
The selected runner mask over the source segment.
Stride skeleton
Pose landmarks rendered frame by frame.