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

Input frame length for MPI-INF-3DHP dataset #34

Closed StevenSmith2000 closed 5 months ago

StevenSmith2000 commented 5 months ago

Many thanks for the great work!

I wonder why the input frame length is chosen as 81 instead of 243 for MPI-INF-3DHP dataset. Have you tried training on 243 frames and how is the performance? Thank you!

SoroushMehraban commented 5 months ago

@StevenSmith2000 No we haven't. We followed previous approaches and for a fair comparison used 81 instead. They mainly say in the paper that's because the original video lengths are not as long as Human3.6M.