This repository contains a solution of client/server to run an Deep Learning Classification model and gets the answer at the client side, it was implemented using Google Pub/Sub, OpenCV, Numpy and other technologies.
first you will have to create a project in the Gcloud platform an then go PubSub and create 2 topics, 1 will deal with request communication with the server and the other one with the answer from the server, then adjust the parameters depending on your project_id, subscription_ids and topic_ids. After adjusting the Gcloud parameters follow the instructions of application and model-serverfolders.
To use it with Kafka you will need to run a ZooKeper and also an Kafka server. Then you will need to create two topics client_request and client_response. After that you will need to get an bootstrap_server that the default is something like localhost:9092.
python3.7 -m pip install -r requirements.txt