search

SangbumChoi / MobileHumanPose

This repo is official PyTorch implementation of MobileHumanPose: Toward real-time 3D human pose estimation in mobile devices(CVPRW 2021).
MIT License
235 stars 29 forks source link

Model modification #46

Open glostickk opened 7 months ago

glostickk commented 7 months ago

Hi,

Thanks for your work! I am planning on using/modifying your model for RGB-D pose estimation and have a couple of questions (please excuse my beginner level):

  1. I would like to confirm that I only need to change the embedding size (512 or 2048) and width multiplier (0.75 or 1.0) in the config file to change the model type (small or large)?
  2. Where can I change the number of input channels? I don't see it being referenced anywhere. (I need to it to be 4 channels since I'm feeding RGB-D).

Thanks so much in advance.