@hunterjm Hi Jim – Thanks again for your great work on the stream component and tensorflow – I use them everyday in my home setup and always use them to show off the setup with the neural network detection of humans – it’s amazing.
I was wondering if you had any advice on the upcoming change in the base OS of the default Home Assistant docker image from Debian to Alpine. I know previously that balloob had approved making tensorflow work and adding the base libraries/wheels to the official docker image, and now it looks like in releases 0.98 and higher we won’t be able to use tensorflow any longer:
Are there any alternatives you could suggest? Support some how for an external docker tensorflow container?
Thanks for the kind words. Pascal and I are looking into building a version of TensorFlow that can be added to Home Assistant’s package repo so it can continue functioning as-is.
Thanks Jason! Running the integrated tensorflow natively has been so easy to setup and get going – it was my first introduction to tensorflow and now I’m using it for other projects – glad to hear it will continue to work with the alpine based docker image!
Sorry to bother you again, but this seems the most promising bit of information I found across the web.
I’m running home assistant core using the docker image and I’m trying to enable Tensorflow.
I’m trying to understand if there is something I’m missing, or if at the moment this is not supported due to the missing Tensorflow in the image.
No problem if you’ve not been able to include it yet, and thanks for your work!