Header logo is ps

The Informed Sampler: A Discriminative Approach to Bayesian Inference in Generative Computer Vision Models

2015

Conference Paper

ps


Computer vision is hard because of a large variability in lighting, shape, and texture; in addition the image signal is non-additive due to occlusion. Generative models promised to account for this variability by accurately modelling the image formation process as a function of latent variables with prior beliefs. Bayesian posterior inference could then, in principle, explain the observation. While intuitively appealing, generative models for computer vision have largely failed to deliver on that promise due to the difficulty of posterior inference. As a result the community has favored efficient discriminative approaches. We still believe in the usefulness of generative models in computer vision, but argue that we need to leverage existing discriminative or even heuristic computer vision methods. We implement this idea in a principled way in our informed sampler and in careful experiments demonstrate it on challenging models which contain renderer programs as their components. The informed sampler, using simple discriminative proposals based on existing computer vision technology achieves dramatic improvements in inference. Our approach enables a new richness in generative models that was out of reach with existing inference technology.

Author(s): Varun Jampani and Sebastian Nowozin and Matthew Loper and Peter V. Gehler
Book Title: Special Issue on Generative Models in Computer Vision and Medical Imaging
Journal: Computer Vision and Image Understanding
Volume: 136
Pages: 32-44
Year: 2015
Month: July
Publisher: Elsevier

Department(s): Perceiving Systems
Research Project(s): Inverse Graphics
Bibtex Type: Conference Paper (inproceedings)
Paper Type: Journal

DOI: 10.1016/j.cviu.2015.03.002

Links: arXiv-preprint
Attachments: pdf

BibTex

@inproceedings{jampani15_cviu,
  title = {The Informed Sampler: A Discriminative Approach to Bayesian Inference in Generative Computer Vision Models},
  author = {Jampani, Varun and Nowozin, Sebastian and Loper, Matthew and Gehler, Peter V.},
  journal = {Computer Vision and Image Understanding},
  booktitle = {Special Issue on Generative Models in Computer Vision and Medical Imaging},
  volume = {136},
  pages = {32-44},
  publisher = {Elsevier},
  month = jul,
  year = {2015},
  month_numeric = {7}
}