Open animeesh opened 2 years ago
Hi,
It will depend on how you want to overlay it, but it should be possible. You need to select few points (2, 3 or 4) in the origin (carton) image, and you just need to calculate the transform matrix between the coordinates of those points in the cartoon image and in the detected body landmarks. If you check the code, you will see the example for 2 (for the bones) and 3 points (for the torso and head). In the case of 4 points, instead of an affine transform, you should estimate the perspective transform.
is there any way we can keep anything instead of a skeleton ? like a cartoon hand.