Deploying tensorflow model to GCP without docker

Question

I want to deploy my existing Tensorflow model to Google Cloud Platform. I have an image detection model and I want to make a Python app that will have real time detection.

All tutorials that I found were using Docker. Is it possible to deploy Tensorflow model to server without Docker?

Answer 1

Yes, you should be able to deploy your Tensorflow model to either:

• Cloud Functions Python 3.7 runtime
• App Engine Python 3.7 runtime

neither of which require Docker.

Or, if you don't want to build an application around your model, you could also deploy it to Cloud ML Engine.