Question about Human3.6M preprocessing scripts

karfly / learnable-triangulation-pytorch

This repository is an official PyTorch implementation of the paper "Learnable Triangulation of Human Pose" (ICCV 2019, oral). Proposed method archives state-of-the-art results in multi-view 3D human pose estimation!

MIT License

1.1k stars 181 forks source link

Question about Human3.6M preprocessing scripts #114

Closed Vegetebird closed 3 years ago

Vegetebird commented 4 years ago

When I run the python3 generate-labels-npy-multiview.py, why I can't get the warning saying "camera 54138969 isn't present in S11/Directions-2". And in your paper said, "some scenes of the ’S9’ validation actor (parts of ’Greeting’, ’SittingDown’ and ’Waiting’", so in "Human3.6M preprocessing scripts", these data are been ignored?

shrubb commented 4 years ago

Sorry for the late reply.

why I can't get the warning saying "camera 54138969 isn't present in S11/Directions-2"

What we see is:

$ ls datasets/Human3.6M/processed/S11/Directions-2/imageSequence/
55011271/ 58860488/ 60457274/

and there's no 54138969 folder. The dataset just doesn't have it (other actions have this folder). Check yours. If the warning didn't fire, it means the folder is there. Even if so, I'm sure it's empty. It's still fine.

so in "Human3.6M preprocessing scripts", these data are been ignored?

No, they are preprocessed too. You choose to ignore them later, here.

Vegetebird commented 4 years ago

Thank you for your reply! It seems I have the 54138969 folder and it's not empty.....

shrubb commented 4 years ago

Strange. Then I suggest that you manually check some images against the respective frames from other cameras, and you should be OK.

anas-zafar commented 1 year ago

@Vegetebird can you please share the Human3.6m dataset. I am unable to access it