aimagelab / RefiNet

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Input_requirement

justintiger opened this issue · comments

Hi Thank for this nice model.
I have a question about the input parameters. Should i also have a file with keypoints of input? Or can i only have depth maps as input and return me 3D-skeleton?

Hi @justintiger, thank you for reaching out.
Our method need a starting set of keypoints which are then refined with the 3 steps.
You can use any of the already developed method to get a starting pose to work with, like:
Hrnet, OpenPose, or even using Shotton algorithm from kinect.