In this chapter, we will create a production system, from training to serving a model. Our system will have the ability to distinguish between 37 different species of dogs and cat. A user can upload an image to our system to receive the results. The system can also receive feedback from the user and automatically train itself every day to improve results.

This chapter will focus on several areas:

• How to apply transfer learning to a new dataset
• How to serve a production model with TensorFlow Serving
• Creating a system with crowd-sourced labeling of the dataset and automatic fine-tuning the model on user data