pose-keypoints-to-spatiotemporal-tensor

transform

PoseKeypoints -> JointFeatureTensor

Map sequences of human skeleton joint coordinates and confidence scores into a structured tensor formatted for graph convolutional layers.

Problem it solves

Raw list-based pose estimator outputs cannot be processed directly by spatio-temporal graph convolutional networks (GCNs).

Consumes

PoseKeypoints

Emits

JointFeatureTensor

Distilled from 1 source

The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.