1

Back before deep learning, there were a lot of different attempts at computer vision. Some involved Conditional Random Fields and Markov Random Fields, which were both computationally difficult and hard to understand/implement.

Are these areas still being developed in the computer vision domain? What was the end result of this line of study? I haven't seen any papers on this topic be cited in top-performing benchmarks, so I assume nobody cares about them anymore, but I wanted to ask.

nbro
  • 39,006
  • 12
  • 98
  • 176
FourierFlux
  • 783
  • 1
  • 4
  • 14

1 Answers1

1

In the Image-to-Image Translation with Conditional Adversarial Networks paper (popularly known as pix2pix), they used a Markovian Discriminator to effectively model the image as a Markovian Random Field.

There were some papers in the last 5 years concerning Markov Fields. Here are some of them:-

  1. Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis

  2. Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks

Saurav Maheshkar
  • 756
  • 1
  • 7
  • 20