Gist for Move Mirror blog post: cosine distance

moveMirrorCosineDistance.js

// Great npm package for computing cosine similarity  
const similarity = require('compute-cosine-similarity');

// Cosine similarity as a distance function. The lower the number, the closer // the match
// poseVector1 and poseVector2 are a L2 normalized 34-float vectors (17 keypoints each  
// with an x and y. 17 * 2 = 34)
function cosineDistanceMatching(poseVector1, poseVector2) {
  let cosineSimilarity = similarity(poseVector1, poseVector2);
  let distance = 2 * (1 - cosineSimilarity);
  return Math.sqrt(distance);
}

The 32 at the end of the second comment should be 34.

thanks for that! fixed it

We used each person’s bounding box coordinates to crop and scale each image (and corresponding keypoint coordinates) to a consistent size.
how tom crop and scale? why height == weight?