Ok, so we’ve still got a lot of work to do from here on (like making a distributable library package, for example), but this evening Stéphane and I at least got a pretty decent little edge detection example working with OpenCV humming along quite nicely inside of Processing. So we’re ready for the workshop that starts tomorrow. Of course, there are other libraries that already do this inside of Processing, but this will open a lot of future doors, so we’re pretty happy. Here’s a quick snapshot before I go to bed:
OpenCV, in relation to Myron, is huge. For example, in the little example we built, it gives us not only the shape of the blog but its perimeter, area, centroid (le barycentre), as well as the general information on its bounding rectangle, position, etc. You can then go on (we haven’t) to ask it to calculate the orientation of the shape. For example, you could know how a face is leaning, after you’ve recognized it. You can add filters for tracking an object over time. Etc.
Again, most of these things are things we have not added yet. The advantage for OpenFrameworks, is that artists can dig deeper into OpenCV without the OF people having to write anything extra. Same with Quicktime. Don’t like what OpenFrameworks builds for you? Well, just read the Quicktime docs and dig deeper yourself (with the advantage of not having to write a whole new application, though). It is the same process in Processing, in relation to Java. You can go dive into Java whenever you want something more complex. But in our case, we have no such luxury. We have to re-write, line for line, everything you’d want to do in OpenCV. This is a total pain, and worse is limited by my (limited) knowledge of OpenCV’s underlying theory.
i must say i’m having a hard time waiting to get my hands on it .. :) looks really promissing, thanks guys!
Cher Douglas, Stéphane et toi allez faire des heureux avec cette librairie. Des besoins en détection «intelligente» surviennent toujours avec les projets de nos étudiants. Surtout, vous continuez dans votre magnifique travail de contribution à Processing. Nous avons expérimenté avec votre librairie UDP cet automne et ça fonctionne vraiment bien.
Also, i am wandering if you could explain a bit about what will be the main difference between OpenCV technology and the options that was available with JMyron library. Edge detection is something you could already handle with JMyron, no?
Take care et bonne session d’hiver 2008
wow, looking forward to play with it. any release date in mind?
As for a release date, not too long I suppose. We’ll have the first bare-bones blob detection out pretty soon for Mac OS X. I don’t know about the other platforms, even if we wrote it cross-platform. Also, on that front PPC works, but isn’t accelerated, as opposed to on Intel chips.
We just finished the workshop which was used as an excuse to clean everything up and simplify the commands. Next week we have to get the Hypermedia pages moved to our new server, lots of junk like that. Don’t worry, we want to get it out there, even in the early stages, which is where we’re at right now.
Basically, since we were able to get a working piece out of it in 1 week, I consider it releasable. We just have to make sure it’s clean & simple to install with somewhat workable documentation (we’re using Javadoc, so it’s already 90% of the battle).
Original Comments:
Douglas Edric Stanley
OpenCV, in relation to Myron, is huge. For example, in the little example we built, it gives us not only the shape of the blog but its perimeter, area, centroid (le barycentre), as well as the general information on its bounding rectangle, position, etc. You can then go on (we haven’t) to ask it to calculate the orientation of the shape. For example, you could know how a face is leaning, after you’ve recognized it. You can add filters for tracking an object over time. Etc.
Again, most of these things are things we have not added yet. The advantage for OpenFrameworks, is that artists can dig deeper into OpenCV without the OF people having to write anything extra. Same with Quicktime. Don’t like what OpenFrameworks builds for you? Well, just read the Quicktime docs and dig deeper yourself (with the advantage of not having to write a whole new application, though). It is the same process in Processing, in relation to Java. You can go dive into Java whenever you want something more complex. But in our case, we have no such luxury. We have to re-write, line for line, everything you’d want to do in OpenCV. This is a total pain, and worse is limited by my (limited) knowledge of OpenCV’s underlying theory.
Florian Jenett
i must say i’m having a hard time waiting to get my hands on it .. :) looks really promissing, thanks guys!
Morpholux
Cher Douglas, Stéphane et toi allez faire des heureux avec cette librairie. Des besoins en détection «intelligente» surviennent toujours avec les projets de nos étudiants. Surtout, vous continuez dans votre magnifique travail de contribution à Processing. Nous avons expérimenté avec votre librairie UDP cet automne et ça fonctionne vraiment bien.
Also, i am wandering if you could explain a bit about what will be the main difference between OpenCV technology and the options that was available with JMyron library. Edge detection is something you could already handle with JMyron, no?
Take care et bonne session d’hiver 2008
andreas schlegel
wow, looking forward to play with it. any release date in mind?
Douglas Edric Stanley
As for a release date, not too long I suppose. We’ll have the first bare-bones blob detection out pretty soon for Mac OS X. I don’t know about the other platforms, even if we wrote it cross-platform. Also, on that front PPC works, but isn’t accelerated, as opposed to on Intel chips.
We just finished the workshop which was used as an excuse to clean everything up and simplify the commands. Next week we have to get the Hypermedia pages moved to our new server, lots of junk like that. Don’t worry, we want to get it out there, even in the early stages, which is where we’re at right now.
Basically, since we were able to get a working piece out of it in 1 week, I consider it releasable. We just have to make sure it’s clean & simple to install with somewhat workable documentation (we’re using Javadoc, so it’s already 90% of the battle).