Data science guy
with background in Statistics and Fine Arts.


Twisting faces

Some usecases of clmtrackr


In the previous post, I explained how CLMtrackr was put together. Since then, my examples of face substitution and emotion detection has received a fair amount of attention, so in this post I’m going to explain a bit about how these are put together as well, plus comment on some fixes that I’ve done to CLMtrackr recently.


Fitting faces

An explanation of clmtrackr


A while ago I put out a semi-finished version of CLMtrackr, which is a javascript library for fitting a facial model to faces in images or video. Fitting a facial model is useful in cases where you need precise positions of facial features, such as for instance emotion detection, face masking and person identification. Though the tracking is pretty processor-intensive, we manage to reach real-time tracking in modern browsers, even though it’s implemented in javascript. If you’re interested in seeing some applications of CLMtrackr, check out the demos of face substitution, face deformation, or emotion detection.

In this post, I’ll explain a few details about how CLMtrackr is put together.


Building a budgeting service

A post-hoc analysis, part 2


As I wrote in my last blog post, around 3 years ago I decided to try to build a budgeting service like mint.com for the norwegian market. After around a year, having reached the prototype stage, I decided to take a short break from further building, to think about the business details. This quickly turned into an … extended break.

In this post I’ll write out some of the reasons I stopped working on it, and finally, some lessons learned.


Building a budgeting service

A post-hoc analysis, part 1


So, it’s been over a year since I “took a break” from working on my stealth startup project, and I guess it’s safe to say that I’m not going to pick it up again. Around 3 years ago, inspired by the success of the personal budgeting service mint.com in the US, and wanting something similar myself, I started investigating possibilities for making a personal budgeting service for the norwegian market. I ended up working on the project in my spare time for over a year.

In this post I’ll go through the challenges I encountered, some of the solutions, and in a later post I’ll go through the reasons I stopped working on it, and some lessons learned.


Fab failure

So, I was browsing exp.lore.com and came across these nifty little usb-sticks a couple of days ago. Huh, that’s a pretty decent just-in-time gift I thought - might be an idea to buy a couple of them for those occasions where you don’t really have time to buy a gift for someone. So I click the link, and end up on the fine site fab.com. For a brief couple of seconds, I see the beautiful website they seem to have prepared for me. I even manage to get a glimpse of the notification “Sale Ends in 3 hours”. Ooh, I must be quick! Unbeknownst to me, that will be the last I ever see of their sweet shop, at least in a functional state.


Head tracking with WebRTC

A lot of new exciting standards are coming to browsers these days, among them the WebRTC standard, which adds support for streaming video and audio from native devices such as a webcamera. One of the exciting things that this enables, is so called head tracking. We decided to do a little demonstration of this for the Opera 12 release, which is the first desktop browser to support video-streaming via the getUserMedia API.