Hands-On Deep Learning for Images with TensorFlow

[ 66 ]
Summary
Alright! We've learned about convolutions, which are a loosely connected way of moving
over an image to extract features; we've learned about pooling, which summarizes the most
important features; we've built a convolutional neural network using these techniques; and
then finally, we combined many layers of convolution and pooling in order to generate a
deep neural network.
In the next chapter, we're going to switch over to a bit more application development.
We're going to be building an image classification REST server that can take different
neural network models and serve those as APIs.

---

5
An Image Classification Server
In this chapter, we're going to be taking the machine learning models we've learned about
and turn them into REST servers for image classification.
In this chapter, we'll be covering the following topics:
Making a REST API definition with OpenAPI or Swagger
Creating a Docker container to create a repeatable build environment
Making predictions with our API and posting images over HTTP
REST API definition
Let's begin by defining the REST API. This is comprised of four activities: getting the
project source code from GitHub with git; installing the necessary packages and reviewing
the packages that will be needed in order to run our server; editing and creating the
OpenAPI or Swagger definition file in YAML; and then finally handling a POST-ed image
in that code that the REST API takes to turn an actual image file into a tensor.
First, we need to clone the repository that we've provided in order to have a REST service.
I'm getting this over HTTPS and cloning it with the command line:
$ git clone https://github.com/wballard/kerasvideo-server/tree/2018.git

---

An Image Classification Server
Chapter 5
[ 68 ]
You can put it in any directory you like. Afterwards, we'll be able to use this source code
for the rest of this section, and in the remaining chapters of this book:
Files in Kerasvideo-server-2018
Using my locally installed Python, I'm actually using 
pip
here to install the requirements
for this service: