search

SamsungLabs / rome

Realistic mesh-based avatars. ECCV 2022
Other
428 stars 41 forks source link

Example doesn't work [might be a hardware problem] #11

Open DanielAronovich opened 2 years ago

DanielAronovich commented 2 years ago

Hi,

I have tried to use the example code: python3 infer.py --deca DECA --rome data

And got the following log: Start infer! creating the FLAME Decoder please check model path: DECA/data/deca_model.tar [W NNPACK.cpp:51] Could not initialize NNPACK! Reason: Unsupported hardware.

And when I try to visualize the output, I get:

image

Any feedback on what is happening? Thanks!

RobinROAR commented 1 year ago

Hi,

I have tried to use the example code: python3 infer.py --deca DECA --rome data

And got the following log: Start infer! creating the FLAME Decoder please check model path: DECA/data/deca_model.tar [W NNPACK.cpp:51] Could not initialize NNPACK! Reason: Unsupported hardware.

And when I try to visualize the output, I get: image

Any feedback on what is happening? Thanks!

same error occured!

khakhulin commented 1 year ago

Have you put all weights correctly?

RobinROAR commented 1 year ago

Have you put all weights correctly?

It seems that the all weights are loaded and all dependencies are installed. I can run "python3 infer.py --deca DECA --rome data" successfully without any errors occured. but still get the "green" rendered image and "empty" shape result image.

KabonMax commented 1 year ago

It seems that the all weights are loaded and all dependencies are installed. I can run "python3 infer.py --deca DECA --rome data" successfully without any errors occured. but still get the "green" rendered image and "empty" shape result image.

I encountered the same error and finally solved it. In my case, the reason is that the file ./DECA/data/deca_model.tar was not placed correctly.