yhw-yhw / mime Goto Github PK

i followed the guidance of doc/installation.md, but still can't run any of the scripts

First, thanks for your great works.
I'm interesting in scene generation with pose and I read both your MIME and Pose2Room.

And, I wonder how did you compare your model and Pose2Room,
because your human motion is based on SMPL, but Pose2Room used avatar from virtualhome, and it is different with smpl.
How did you map or change to each skeleton?

how to build this cpp extension?
Do I need to run the setup.py in the "MIME/thirdparty/Rotated_IoU/cuda_op"?

Hi,

what hardware specs are required to run this project?

Thanks

Hi,

I tried to run the preprocess pipeline, but for all three preprocessing steps (Add Static Humans, Put AMASS Walking Motions, Add Interactive Poses) I get the following error:

-> logging.getLogger("trimesh").setLevel(logging.ERROR)
(Pdb)

Any advice what I can do about this?

running it on ubuntu 18

Thanks

Nice work! Is it possible to put a simulated camera in the generated scene and record data like depth image, RGB image and camera pose? These data are very useful for 3D reconstruction works and, with the mesh you provide, the generated scenes will be valuable for 3D reconstruction benchmarking.