Chapter 4. AI for media

 

This chapter covers

  • Using AI to understand the content of images
  • Using deep learning to solve challenges in computer vision
  • Detecting and labeling human faces
  • Creating original artwork with deep learning
  • Using AI with audio data
  • Case study: automating vegetable sorting with a $1,000 robot

Images and videos are probably the richest source of data that humans use to understand the world. Sadly, they were largely off-limits for computers until the 2010s. While cameras had been getting better and better for decades, computers still couldn’t understand the content of images. A pivotal moment was the development of a new field of machine learning called deep learning ( DL ). This family of algorithms has unsurpassed (for now) learning power that can reach superhuman accuracy in important tasks like detecting objects and faces in images. These abilities allow businesses to supercharge their existing products with abilities that would have been considered magical just 10 years ago, from automatically creating photo albums based on the content of images, to automating industrial processes with robots.

4.1 Improving products with computer vision

4.2 Using AI for image classification: deep learning?

4.3 Using transfer learning with small datasets

4.4 Face recognition: teaching computers to recognize people

4.5 Using content generation and style transfer

4.6 What to watch out for

4.7 AI for audio

4.8 Case study: optimizing agriculture with deep learning

4.8.1 Case questions

4.8.2 Case discussion

Summary