ostadabbas / Infant-Pose-Estimation

Invariant representation learning for infant pose estimation with small data (FG2021)
44 stars 13 forks source link

FiDIP pretrained models #15

Closed diegosancarmen closed 2 months ago

diegosancarmen commented 3 months ago

Hello,

In your paper "Invariant Representation Learning for Infant Pose Estimation with Small Data" specifically in Table IV, you provide figures for generalization of the FiDIP method on three baseline models: SimpleBaseline, DarkPose, and MobileNet. However, in the models Google Drive folder, there are only two FiDIP models, HRNet (DarkPose) and MobileNet. Would you also be able to release the SimpleBaseline + FiDIP model?

Thank you!

FayeHuang0621 commented 3 months ago

Since the SimpleBaseline + FiDIP model is built on another pose estimation repository, I hadn't planned to release it. It's been three years, and I'm unsure if I can dig it out. However, I'll send it to you if I find it.

diegosancarmen commented 3 months ago

Hello @FayeHuang0621,

Thank you for your quick reply. Actually, would you be able to share the config and training files used for generating this model instead of the actual weights?

Thank you!

FayeHuang0621 commented 3 months ago

Please look at the old version of this repo