...Only Transformers and Human Pose Estimation Keypoint Data
Using this definition, a given x-axis coordinate value of a keypoint position in sequence frame 𝑠𝑖,𝑗si,j, denoted by 𝑠𝑖,𝑗,𝑝𝑥si,j,px, is scaled as 𝑠′𝑖,𝑗,𝑝𝑥=𝑠𝑖,𝑗,𝑝𝑥·𝑚𝑖,𝑗,𝑝𝑥|𝑠̲𝑖,𝑥2−𝑠̲𝑖,𝑥1|,...