MRF is a generative model. Hence we need to model
i) the likelihood of image given label
ii) prior of label
the inference can be modeled from the joint probability (using Bayes theorm) as a conditional probability of label given the image.
To make the inference tractable only local relationship between labels are encoded into in the form ii).
CRF can directly model the conditional probability of label given image, hence we don't need to explicitly model i) and ii).
Apr 6, 2017
Subscribe to:
Post Comments (Atom)
Carlo Cipolla's Laws of Stupidity
"By creating a graph of Cipolla's two factors, we obtain four groups of people. Helpless people contribute to society but are...
-
To install pngwrite we need to install to helper library before we install pngwriter. 1. libpng 2. freetype2 We can use fink or macport to i...
-
"By creating a graph of Cipolla's two factors, we obtain four groups of people. Helpless people contribute to society but are...
-
Use 'Disk Utility' application to format the USB drive in MS-DOS (FAT) format Download the windows ISO file from the Windows websit...
No comments:
Post a Comment