You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
How do we convert the 16-channel embedding into a meaningful number (e.g. nearest vertex?) I'm using the model trained on humans, thus I know the embedding in some way maps to points on the surface of the SMPL model, but is there a function in the API to get back e.g. SMPL vertex index from the embedding?
background:
I'm playing around a bit with fitting SMPL back into the segmentations I get from DensePose on multiple views. I have a script that will apply the DensePose model and output the IUV data in a format suitable for loading into my fitter. However, what with various flaws in the output (like left-right swaps on the feet) I want to try fitting to the CSE output instead - but thus need to know how that embedding maps to a coordinate on the SMPL model's surface (nearest vertex or some such thing).
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
How do we convert the 16-channel embedding into a meaningful number (e.g. nearest vertex?) I'm using the model trained on humans, thus I know the embedding in some way maps to points on the surface of the SMPL model, but is there a function in the API to get back e.g. SMPL vertex index from the embedding?
background:
I'm playing around a bit with fitting SMPL back into the segmentations I get from DensePose on multiple views. I have a script that will apply the DensePose model and output the IUV data in a format suitable for loading into my fitter. However, what with various flaws in the output (like left-right swaps on the feet) I want to try fitting to the CSE output instead - but thus need to know how that embedding maps to a coordinate on the SMPL model's surface (nearest vertex or some such thing).
Beta Was this translation helpful? Give feedback.
All reactions