Assisted drawing 

In this section, we will briefly describe this chapter's project and what we aim to achieve. Recall from the previous chapter that we described an application capable of predicting what the user was trying to sketch, and fetched similar images based on the predicted categories, such as a sailboat. Based on this prediction, the application would search and download images of that category. After downloading, it would sort them based on their similarity with regards to the user's sketch. Then it would present the ordered alternatives to the user, which they could swap their sketch with.

The finished project is shown as follows: 

The model used for performing this classification was based on a Convolutional Neural Network (CNN), a type of neural network well suited for understanding images owing to its ability to find local patterns and build on top of these lower patterns to find more complex and interesting patterns. We took advantage of these higher order patterns by using...