Chapter 5. Annotation Automation


Too much work? Is annotation tedious? Quality poor?

In this chapter I show you how to solve these problems with automations.

The first topic is Pre-Labeling - the idea of running a model before annotation. I’ll cover the caveats you should be aware of and step through extensions of the idea such as ‘micro-model’ labeling.

Next, Interactive Automations are when a user adds information in order to help the algorithm. For example drawing a box to automatically get a tighter location marked by a polygon. Interactive improvements are often most relevant to spatial locations - however that’s just the start. The end goal of Interactive Automations is to make tedious UI work a more natural extension of human thought.

Quality Assurance (QA) is one of the common uses of training data tools. I cover exciting new methods like using the model to debug the ground truth. Other tools ...

Get Training Data for Machine Learning now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.