What We Learned Labeling One Million Images

At Figure Eight, we’ve labeled over a million images, helping some of the world’s most innovative companies power and validate their computer vision models. In this guide, we share what we’ve learned along the way.

You’ll learn how to scope a computer vision project, what kind of source data you need to make it successful, what kind of tools fit your project best, how to label your dataset so your algorithms can learn, and a whole lot more.


eBook Insights

  • When it comes to training data, it’s not just about quality. It’s about quantity too.
  • Overfitting is a real problem with computer vision models. Find out how to solve it.
  • Simple image classification can be just as important as meticulous pixel labeling, if not more.
  • Every tool and strategy has its pros and cons. Learn how to decide which is right for your project.