Closed allo- closed 4 years ago
Thanks for bringing the installation issue to my attention- it is already fixed and will be part of the next release.
I will consider publishing the package on pypi once the next release is ready.
You seem not to have pushed this commit, yet.
It would be good when you can just push it, because I can include the repo into requirements.txt until you publish a package, but currently I need to document a separate installation step to first install tensorflowjs and then your package.
Thank you for the fix and your useful tool. I am using it in https://github.com/allo-/virtual_webcam_background to work with body-pix.
You seem not to have pushed this commit, yet.
Whoops - my bad, sorry 😀
@allo- Please reopen the issue in case the fix doesn't work for you for some reason. I tested it in an otherwise empty env and the pip install worked just just and pulled all requirements automatically.
@allo- PS: the next release will also include support for facemesh and other models that use activation functions not natively supported by TF.
The
setup.py
file importstfjs_graph_converter.version.py
and that indirectly importstfjs_graph_converter.__init__.py
.This makes it impossible to install your package using pip, when
tensorflowjs
is not installed, yet.The easiest way would probably be to move
version.py
to the top-level or to write the version directly intosetup.py
.In addition it would be great when you would publish your package to pypi.