1 / 49

Segmentation Through Optimization

Segmentation Through Optimization. Pyry Matikainen. “He who fights with monsters should look to it that he himself does not become a monster.” - Friedrich Nietzsche, Beyond Good and Evil. Formulate Problem. Force problem into favorite algorithm. Retroactively justify decisions.

scano
Download Presentation

Segmentation Through Optimization

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Segmentation Through Optimization PyryMatikainen

  2. “He who fights with monsters should look to it that he himself does not become a monster.” -Friedrich Nietzsche, Beyond Good and Evil

  3. Formulate Problem Force problem into favorite algorithm Retroactively justify decisions Gradient ascent via parameter tweaking “Refine” Publish

  4. What is wrong with this? Formulate Problem • Difficult to use • Difficult to extend • Difficult to study Force problem into favorite algorithm Retroactively justify decisions Gradient ascent via parameter tweaking “Refine” Publish

  5. Z. Tu and S. C. Zhu (2002)to the rescue! and also Ren and Malik (2003)…

  6. Z. Tu and S. C. Zhu. Image Segmentation by Data-Driven Markov Chain Monte Carlo. PAMI, vol.24, no.5, pp. 657-673, May, 2002: The DDMCMC paradigmcombines and generalizes these [all other] segmentation methods in a principled way.

  7. Segmenter Evaluator Optimizer

  8. Everything is search.

  9. Evaluator Optimizer

  10. “What is a good segment?” Ren and Malik (2003)

  11. How do we model a segment? Contours Texture Raw pixel values

  12. x2 G(x) h(x) h(f(x)) G(b(x) - x)

  13. (gaussian) (histogram) (gabor) (Bezier)

  14. Region appearance model complexity Region area Region perimeter length (smoothness) Number of regions Notably absent: the data

  15. Brightness Superpixels (normalized cuts) Texture (textons) Oriented energy

  16. * G(W|I) Classifier

  17. Evaluator Optimizer

  18. MCMC is a technique for sampling from distributions.

  19. Number of regions Region Region? ? ? ?

  20. Ren and Malik The ‘data driven’ part revealed! Merge Split Boundary competition Model adaptation Switching image models

  21. Data driven = do some clustering to make the MCMC faster.

  22. Evaluator Optimizer

  23. Tu & Zhu

  24. Ren & Malik

  25. 1/2 1/2 0 1/2 0 0 1 1/3

  26. Optimizer Evaluator Evaluator Optimizer

  27. (gaussian) (mixture of gaussians) (3x Bezier spline)

  28. (g1) (gaussian) (g2) (histogram) (g3) (gabor filter) (g4) (Bezier spline)

  29. Number of regions Region appearance model parameters Region appearance model Pixels in region

  30. MCMC

  31. Xiaofeng Ren and Jitendra Malik. Learning a Classification Model for Segmentation. ICCV 2003.

  32. Boundary between i and j

  33. Classification certainty Ren and Malik 2003 Maximizing G(W|I) Discriminative models Superpixels Tu and Zhu 2002 Sampling P(W|I) Generative models Pixels

More Related