To run locally:

• Make env with Python<=3.8 (for TF1.15 to work)
• Install the follwing packages (from Dockerfile): Flask gunicorn tensorflow==1.15 opencv-python Pillow requests
• For opencv to work (on Linux), you need to install the following packages libglib2.0-0 libsm6 libxext6 libxrender-dev. This can be done with sudo apt-get install libglib2.0-0 libsm6 libxext6 libxrender-dev
• Run app.py and you can send post requests to port 8080.

To deploy to Google Cloud Run: Assuming gcloud is properly set up,

1. First, make container,

gcloud builds submit --tag gcr.io/PROJECT-ID/MODEL-NAME

--> Replace PROJECT-ID with the id of current project in cloud console. MODEL-NAME is the name of project to store on cloud 2. Deploy :)

gcloud builds submit --tag gcr.io/PROJECT-ID/MODEL-NAME

Follow simple steps for configuration, but should be fine after that (select location, security, etc.)