search

TaatiTeam / MotionAGFormer

Official implementation of the paper "MotionAGFormer: Enhancing 3D Pose Estimation with a Transformer-GCNFormer Network" (WACV 2024).
Apache License 2.0
128 stars 15 forks source link

3D pose result #9

Closed SunQiu88 closed 8 months ago

SunQiu88 commented 8 months ago

Hello,I want to ask a question? Tesing in the wild video,how to verify the result of 3D pose is right when are no GT? When different scene data is tested , the result wether is related to the camera parameters?

SoroushMehraban commented 8 months ago

@zh199110 It's not possible to quantitatively verify the result as we don't have any GT and we can only qualitatively observe and compare with other works. Note that as I explained in #8, since we're dealing with an ill-posed problem, reconstruction of output in millimeters from a monocular camera is not possible when only 2D pose sequence is received as input.