Sorting by visual similarity

So far, we have achieved what we set out to do, that is, inferring what the user is trying to draw and providing them with suggestions that they can swap their sketch with. But our solution currently falls short of understanding the user. Sure, it may predict correctly and provide the correct category of what the user is drawing, but it dismisses any style or details of the user's drawing. For example, if the user is drawing, and only wanting, a cats head, our model may predict correctly that the user is drawing a cat but ignore the fact that their drawing lacks a body. It is likely to suggest images of full-bodied cats. 

In this section, we will look at a technique to be more sensitive with respect to the user's ...

Get Machine Learning with Core ML now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.