weigq/3d_pose_baseline_pytorch

Stacked Hourglass Detections

Closed this issue · 10 comments

Hi @weigq,

Thanks for the great implementation. I'm trying to train on SHG detections but the .zip file doesn't contain the test_2d_ft.pth.tar file.

Could you provide detection files or processing script for that?

weigq commented

I used the fine-tuned 2d poses from the original implementation.

I will upload it asap

Thanks for quick answer. I look forward to it!

Hi @weigq, any update on this? Did you have a chance to upload it?

@weigq, any update?

weigq commented

@mkocabas
Sorry, I will update by the end of this month.

Thanks @weigq, but I don't need them anymore. I've adapted the original dataloaders to pytorch.

I can't reproduce results from stacked hourglass predictions for 3D pose baseline. Can anyone provide the results they obtained or the code for the stacked hourglass predictions?

@mkocabas @weigq

Wondering if there might be any update on this issue before I write the code to convert the predictions from https://drive.google.com/file/d/0BxWzojlLp259S2FuUXJ6aUNxZkE/view

Dear @vasusingla619, @benjiebob,

I modified the original tf preprocessing to pytorch instead of depending preprocessed *.tars provided by @weigq. I would share it if you need.

Dear @mkocabas,

Could you please share your modified version? Thanks!