Containerizing the application

We would rather install one package that has everything than install multiple individual packages (such as the individual modules and code required to run the application) and connect them later. Thus, it becomes important that we can wrap the entire code base and modules into a single package (something like a .exe file in Windows) so that the package can be deployed with as little as one command, still ensuring that it works the same on all hardware. To this end, we need to learn how to work with Docker, which is essentially a condensed operating system with code. The created Docker containers are lightweight and will perform only the tasks that we want them to perform. In our example, the Docker image we will create will run the API for the task of predicting the class of SDD images. But first, let’s understand some Docker jargon.

A Docker image is a standard unit of software that packages up code and all its dependencies. This way, the...