Run in Google Collab?

[DanielRamosAcosta]

There are non AI-users (like me) out there that would like to test these kind of project without having all the dependencies locally installed. I would recomend creating a Google Collab project demonstrating this library for more for greater diffusion!

[loveexword]

!git clone https://github.com/adamian98/pulse
%cd pulse/
%mkdir input

#upload png file into input

!python align_face.py -input_dir input/ -output_dir input1/

!python run.py -input_dir input1/ -output_dir out/

#delete all files
%rm -rf input/
%mkdir input

[loveexword]

There are non AI-users (like me) out there that would like to test these kind of project without having all the dependencies locally installed. I would recomend creating a Google Collab project demonstrating this library for more for greater diffusion!

i finish it……

[DanielRamosAcosta]

I mean, it would be nice to have a "Open in Collab" badge, like deezer/spleeter does

[cchanyu]

!git clone https://github.com/adamian98/pulse
%cd pulse/
%mkdir input

#upload png file into input

!python align_face.py -input_dir input/ -output_dir input1/

!python run.py -input_dir input1/ -output_dir out/

#delete all files
%rm -rf input/
%mkdir input

MAD RESPECT!! Anyone having issues? USE THIS!!!

Let me just tell you, I've spent 8 hours debugging all the issues that I've encountered during the installation and downloaded over 15GB of VS,Python stuff, only to realize that I'm stuck with insufficient memory to run the final program, which outputs the virtual face.

I spent another 5 hours trying to reallocate/reduce my Nvidia graphics card memory usage, which I failed at. Idk what I was doing anyways. So at the last hope, I tried this.. AND IT WORKED.

---

Newbie Instruction on how to use code 'Loveexword' wrote If you have never used Google Collab.

• Google Collab is free- but requires a Google account.

• Create a "new Collab project"

• Click "Edit" -> "Notebook Settings" -> Select GPU on Hardware accelerator

• Create Code blocks (4 of them)
  Paste each one into the block
  !git clone https://github.com/adamian98/pulse %cd pulse/ %mkdir input
  1st code block
  !python align_face.py -input_dir input/ -output_dir input1/
  2nd code block
  !python run.py -input_dir input1/ -output_dir out/
  3rd code block
  #delete all files %rm -rf input/ %mkdir input
  4th code block

• Run the 1st code

• Click the "folder icon" in the left SIDEBAR. Don't confuse with "File" at the top.

• Press "Upload to session storage" and drop your face/drawing image

• Open pulse -> input folder and drag those images in.

• Run the 2nd code (Detected face images will be in Input 1 folder, if not found, that means ur picture is bad)

• Run the 3rd code (Your virtual face will be available to download at "out" folder)

• If you want to re-run, you could try repeating from 2nd code block or press "Run time" -> "Reset to Factory mode", and repeat the process of running the code and image again.