princeton-vl / pytorch_stacked_hourglass

Pytorch implementation of the ECCV 2016 paper "Stacked Hourglass Networks for Human Pose Estimation"
BSD 3-Clause "New" or "Revised" License
465 stars 94 forks source link

How to get the h5py files from the raw annotations? #33

Closed BIGJUN777 closed 3 years ago

BIGJUN777 commented 3 years ago

Hello, thank you for your nice work.

I wanna know how to get the h5py data file (test.h5, train.h5, vaild.h5) from the raw annotation files in MPII dataset. Or can you explain what those data listed in ref.py (such as: center, scale, part and so on ) in the h5py file mean?

crockwell commented 3 years ago

Thanks. Yeah that comes essentially from MPII. I'd check out their work. Also https://github.com/akanazawa/hmr/blob/master/src/datasets/mpii_to_tfrecords.py has some really detailed code that might shed some light on this. It's basically for cropping around each individual person.

BIGJUN777 commented 3 years ago

Cool. Thank you for your nice reply.