I'm trying to write a new computer vision textbook. I'm going to post updated versions here as I do so. The plan is to first teach probability and machine learning and then present each chapter as a different model with associated learning and inference algorithms.
Just added the first list of errata to the printed version. Only one serious problem found in the few months that the book has been available: one of the update equations for factor analysis has a mistake in it. You can download this along with all the other material from http://www.computervisionmodels.com.