Another update

Since last time - did some more work on the tracking chapter although it isn't finished and no doubt full of errors. I've sorted out a bunch of mistakes in the first 8 chapters. I've started working on the geometry chapter - trying to do geometry without homogeneous coordinates (which are beautiful, but I think unnecessary). We'll see how it works out - don't see the point in writing a poor-man's version of Hartley and Zisserman anyway. I've also put in (invisible) hyperlinks so you can nip around the document faster...

Tracking

There's been a lot of progress on the tracking and temporal models chapter. There's now a reasonable description of the Kalman filter, the extended Kalman filter and the unscented Kalman filter. I'm working on the section on particle filters tomorrow.

Update

Working on the first draft of the text for the chapter on tracking. There should be something worth reading in a couple of weeks. In the mean time, I've corrected a lot of mistakes in the first eight chapters, fixed some of the early figures so they are up to the standard of the later ones and added some questions to the end of the early chapters.

Classification Chapter

The chapter on visual classification is now roughly comprehensible. As with all the other chapters, there are no references yet and the applications section is rather sketchy, but the main content should be okay. Might put in some more stuff about boosting methods, I'm not sure.

Since last time, I've also worked my way through the chapters on generative models and preprocessing trying to eliminate as many mistakes as possible. I'm going to start writing text on tracking (Kalman filter/Condensation etc.) this week.

New update today

I've finished the first draft of the section on preprocessing (its in the appendix at the moment although it may move soon). I've also been going over the chapter on regression and correcting the numerous errors therein. I think that things are getting better.

Another update...

Since last time - worked a lot of the kinks out of Chapter 6 - a frightening number of mistakes in it and it is still much too long but I think it reads better now. I've also been working on the figures for the image processing section which has now been relegated to an appendix. Yes - you heard me right - I've relegated what is the core part of most machine vision books to an appendix. Ouch.

Update

Since Thursdays version quite a lot of stuff has changed. I've finished my (horrible) first draft of the text on pre-processing images. I've worked on my (horrible) first draft of the text on subspace models. I've made some figures for the subspace models chapter. I've corrected some small errors in Chapters 3,5,6,7. I've added figures explaining the relationship between the models in chapters 7 and 8. I've done some serious editing to chapter 8.

Since yesterday

Worked on preprocessing chapter some more - wrote first draft of section on textons and local binary patterns. Corrected a bunch of mistakes in Chapters 5 and 8. Improved a figure in chapter 5. Added a couple of questions to chapter 3 and one for chapter 4. Added another algorithm to the recipes section. Phew.

Progress

Quite a lot of progress since last week. I've started working on the text for the preprocessing and subspace chapters. Both are in horrible shape, so don't look at them yet. I've also added the problems back in and added in a few more in the early chapters. I've ironed out a few remaining bugs in the first couple of chapters, rearranged the sections in Chapter 6 and worked quite a bit on the regression chapter (chapter 7). I added a new figure to Chapter 3 to try to explain the predictive distribution with Bayesian estimation better. Enjoy.

Another update

Pretty much worked my way to the end of the classification chapter now. Just figures out what applications I want to use. I've also fixed some small mistakes in chapters 1 and 2, added a figure to chapter 3 and put in some very preliminary problems.

Two for two

Two updates in two days. Who would have thought it? Since yesterday I have worked more on the classification section - added two more figures and improved the text a lot in the boosting section. I've also added in some problems to the early chapters and corrected a heinous error in the figure that shows the different covariances for the Normal distribution. This was pointed out to me by Oisin Mac Aodha. Ta.

Classification

Apologies if you've been having trouble downloading the book for the last couple of days. I think I've sorted this out now, but let me know if you have trouble with time-outs or whatever. In the meantime I've made all of the figures for the classification chapter and am now rewriting the text so that it is slightly more comprehensible. I've also written a draft preface and added an acknowledgements section, so if you mail me errata you will see your name added to this.

Back

Back from a not particularly relaxing vacation (a lot of 5am starts!) but in a good state of mind to make some serious progress. Since the last version I have rearranged the chapters a little. I've corrected a bunch of mistakes in the regression chapter (should now be just about readable) and I've also started work on the classification chapter. This should be ongoing for the next couple of weeks. Thanks to Oisin Mac Aodha who spotted a bunch of errata. These are fixed now as well.

Finally

I notice that I have impressively failed to make an update to the book for a month now. Actually there has been a lot of work going on - I have almost completed a first draft of the regression chapter. I have also corrected many mistakes from Chapter 6 that were helpfully identified by Yun Fu. Also Gabriel Brostow drew my attention to a few issues in Chapters 1 and 2 that have now been sorted out. I'm going on holiday for two weeks now, but I'm going to come back and attend to the classification chapter when I return on Sept 20th.

Various small changes

David Fleet found some errors in the chapter on directed models for images. This chapter needs a lot of work anyway. Various things need to be removed and added to make it more coherent.

Shelved working on the regression chapter to start writing about dimensionality reduction. Its surprisingly hard to explain PCA in an intuitive way!

Trying to make regression chapter comprehensible

I'm now going back over the regression chapter trying to make it a little more comprehensible. I've swapped the main application to Agarwal and Triggs body pose estimation. Every chapter can't be about faces.

I'm not super happy with the notation yet. There's a lot of maths and it could still be quite a lot clearer. Should get onto non-linear regression tomorrow and hopefully polish off the relevance vector machine on Saturday morning.

Regression

I'm still working on the regression chapter. I've been writing about relevance vector regression. It's taken me a while to figure it all out and even now, I think the notation is going to have to be substantially overhauled.

Regression

I've been sketching out the regression chapter. I'm going to cover linear regression,
Bayesian linear regression (i.e. GPs), kernel methods and the relevance vector machine. I'm still hunting around for some good applications to illustrate this all with.

Finally

I've been away from U of T for some time so I haven't updated the online copy until today. There have been a lot of changes in the mean time. I've finished the first full draft of the chapter on complex probability distributions (EM Algorithm, factor analysis, MoG and the like) and I'm now working on the chapters on regression and classification. I've also sorted out a whole bunch more mistakes from the first few chapters.

Generative models

Still working on the generative models chapter - made a lot of small updates to the sections on t-distributions and factor analysis. I've managed to cut out about a page of irrelevant information along the way. Less is more...

Editing...

Back from vacation... and back to working on the generative models chapter. Spent most of today trying to make the section on Mixtures of Gaussians somewhat more comprehensible. I think its gradually getting better. I'm slightly frustrated by the fact that I can't really explain the EM algorithm properly without ruining the flow of the chapter. The result is that the description of the fitting algorithm for mixtures of Gaussians is a bit unsatisfying. But I think the alternative is worse...

More applications

Today I wrote some text to go with all of the figures that I've been making for the applications section of the density modelling chapter. They try to show a variety of possible uses in classification and regression.

Applications of Density Models

I've been working for the last couple of days making figures showing applications for density modeling in Chapter 6. I'm going to give examples of skin detection, background subtraction and face recognition.

I've also corrected a bunch of mistakes in Chapter 5 that were spotted by my PhD student Yun Fu.

Hidden Variables

Worked on chapter 6 today which concerns modeling complex density functions using hidden variables and the EM algorithm. Sorted out numerous small mistakes. Made a figure illustrating a regression problem where we try to predict one pose from another. Made a second figure which concerns using hidden variables to register images of digits together.

Started this blog

I've started this blog about my book as progress is not going as fast as I wanted. I thought if I posted updates on the changes it might help motivate me. Hopefully some people will read this and find even the partial draft useful. I'm really hoping for feedback to eliminate typos, improve the clarity and make sure that everything is correct.