learnable-triangulation-pytorch icon indicating copy to clipboard operation
learnable-triangulation-pytorch copied to clipboard

Published 20 hours ago • verified publisher icon karfly

Pretaining for 2D Backbone, with CMU Dataset

Open Samleo8 opened this issue 4 years ago • 1 comments

Hi, referencing #75. It seems that because the CMU dataset is looking for a different set of joints, if I would like to train using the CMU dataset, I will need to retrain the 2D backbone but using the CMU dataset instead. Is this correct?

Also, how did you guys train the 2D backbone? Are there any instructions on doing so? Of course, it would be even better if you could provide the pretrained backbone weights for the panoptic studio dataset!

Thank you!

Samleo8 avatar May 20 '20 17:05 Samleo8

@Samleo8 did you understand how do the authors train the 2D backbone? Thanks

anas-zafar avatar Apr 05 '23 20:04 anas-zafar