colab example?

[molo32]

someone is so kind to share the notebook colab of this

[penincillin]

@molo32
That's really a good suggestion !
But we are just too busy to update this for now. Colab might be added after CVPR submission.

[stspanho]

@molo32 I was able to make it work with the following commands in a Google Colab instance:

!git clone https://github.com/facebookresearch/frankmocap.git
cd /content/frankmocap/
pip install -r docs/requirements.txt
pip install detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/torch1.6/index.html
pip install pytorch3d
!sh scripts/install_frankmocap.sh

Download SMPL Model (Neutral model: basicModel_neutral_lbs_10_207_0_v1.0.0.pkl): https://github.com/facebookresearch/frankmocap/blob/master/docs/INSTALL.md

Download SMPLX Model (Neutral model: SMPLX_NEUTRAL.pkl):
https://github.com/facebookresearch/frankmocap/blob/master/docs/INSTALL.md

!apt-get install -y xvfb x11-utils
!xvfb-run -a python -m demo.demo_frankmocap --input_path ./sample_data/han_short.mp4 --out_dir ./mocap_output

I got a segmentation fault at the end (ffmpeg), but the video seems rendered in the output folder.

[justinjohn0306]

@molo32 I was able to make it work with the following commands in a Google Colab instance:

!git clone https://github.com/facebookresearch/frankmocap.git
cd /content/frankmocap/
pip install -r docs/requirements.txt
pip install detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/torch1.6/index.html
pip install pytorch3d
!sh scripts/install_frankmocap.sh

Download SMPL Model (Neutral model: basicModel_neutral_lbs_10_207_0_v1.0.0.pkl): https://github.com/facebookresearch/frankmocap/blob/master/docs/INSTALL.md

Download SMPLX Model (Neutral model: SMPLX_NEUTRAL.pkl):
https://github.com/facebookresearch/frankmocap/blob/master/docs/INSTALL.md

!apt-get install -y xvfb x11-utils
!xvfb-run -a python -m demo.demo_frankmocap --input_path ./sample_data/han_short.mp4 --out_dir ./mocap_output

I got a segmentation fault at the end (ffmpeg), but the video seems rendered in the output folder.

can you share the notebook?

[justinjohn0306]

I made a notebook aswell

[LucaswasTaken]

Hey guys, it worked like a charm here, too, with four considerations, as already said above.

• In my case, I removed the opendr from the requirements.txt and installed it via the GitHub repository.
• Set the GPU;
• Sign up and download SMPL and SMPLX models (put it in the drive or someplace easy to access in the colab;
• The segmentation fault of FFmpeg will occur in the and, just ignore it.

The notebook is still raw and poorly documented, but if you guys want to take a look: https://colab.research.google.com/drive/1RaS5NKP2RsrbKuAy2ugCQDj-U_Qn1tXR?usp=sharing
Should work near out-of-the-box, respecting the four observations. OBS: once again, congratulations to the Frank-mocap team for the great work.

[Tikquuss]

I have executed the readme indications step by step with success (notebook)
I hope this can help.

[penincillin]

@Tikquuss
Thanks ! That's really helpful !

[neonascent]

Hey guys, it worked like a charm here, too, with four considerations, as already said above.

• In my case, I removed the opendr from the requirements.txt and installed it via the GitHub repository.
• Set the GPU;
• Sign up and download SMPL and SMPLX models (put it in the drive or someplace easy to access in the colab;
• The segmentation fault of FFmpeg will occur in the and, just ignore it.

The notebook is still raw and poorly documented, but if you guys want to take a look: https://colab.research.google.com/drive/1RaS5NKP2RsrbKuAy2ugCQDj-U_Qn1tXR?usp=sharing
Should work near out-of-the-box, respecting the four observations. OBS: once again, congratulations to the Frank-mocap team for the great work.

@LucaswasTaken - I've executed all the code cells -- thanks for sharing them -- but I'm getting "No module named demo.demo_bodymocap" from Python when I try to run tests. Is this from missing environment info that the install doc said to create via "conda create -n venv_frankmocap python=3.7; conda activate venv_frankmocap" right at the start?

[AhmedShalabe]

@LucaswasTaken - I've executed all the code cells -- thanks for sharing them -- but I'm getting "No module named demo.demo_bodymocap" from Python when I try to run tests. Is this from missing environment info that the install doc said to create via "conda create -n venv_frankmocap python=3.7; conda activate venv_frankmocap" right at the start?

Have you found a solution to the problem?

[AhmedShalabe]

Hey guys, it worked like a charm here, too, with four considerations, as already said above.

• In my case, I removed the opendr from the requirements.txt and installed it via the GitHub repository.
• Set the GPU;
• Sign up and download SMPL and SMPLX models (put it in the drive or someplace easy to access in the colab;
• The segmentation fault of FFmpeg will occur in the and, just ignore it.

The notebook is still raw and poorly documented, but if you guys want to take a look: https://colab.research.google.com/drive/1RaS5NKP2RsrbKuAy2ugCQDj-U_Qn1tXR?usp=sharing
Should work near out-of-the-box, respecting the four observations. OBS: once again, congratulations to the Frank-mocap team for the great work.

@LucaswasTaken - I've executed all the code cells -- thanks for sharing them -- but I'm getting "No module named demo.demo_bodymocap" from Python when I try to run tests. Is this from missing environment info that the install doc said to create via "conda create -n venv_frankmocap python=3.7; conda activate venv_frankmocap" right at the start?

I have found a solution
you have to turn on the GPU from the beginning of the code,
So turn on the GPU
and execute all the commands again