A new SMPL-compatible signed distance field body model, VolumetricSMPL, uses neural blend weights to cut inference time and memory about 10x and 6x versus COAP while matching or improving accuracy.
imGHUM: Implicit generative models of 3d human shape and articulated pose
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
VolumetricSMPL: A Neural Volumetric Body Model for Efficient Interactions, Contacts, and Collisions
A new SMPL-compatible signed distance field body model, VolumetricSMPL, uses neural blend weights to cut inference time and memory about 10x and 6x versus COAP while matching or improving accuracy.