Calling Any Java Programmers - Computer Vision Project - Page 2

84RegalRider · 04-05-2006, 02:20 PM

Just some cheapy.
Fixed intensity.
Sorry its taking a bit long but just got my carputer reading my ecu/pcm and thus my gauges working. Just need to finish coding that before i want to dive into this.

sama · 04-07-2006, 01:30 AM

good that you have fixed intensity, let me know how it all goes. I've been really pressed for time recently so the carputer is on hold

sama · 04-12-2006, 01:56 AM

anyone had anytime to test this yet?

Bryan Pape · 04-12-2006, 03:57 AM

Crap... didn't know anyone was working on this stuff... I've been looking into AI / Computer Vision stuff for years. I got a grip on a lot of the theories / problems / etc, but the upper level math keeps me away from programming any real solutions. I can help work on integration stuff, but it would be in Delphi.

I'm D/L what ya got. Really interested to check it out!

Quote: Originally Posted by sama

anyone had anytime to test this yet?

sama · 04-12-2006, 06:08 AM

this is exactly my aim, to try to get a framework going so that people with little or no knowledge of vision algorithms can build applications, which in turn would drive the development of algorithm plugins to use.

my maths isn't the best either, but there are plenty of resources out there.

the current download offers a fairly simple motion detection algorithm. it learns the background by continously averaging pixel values over time. It detects foreground objects by subtracting the current frame from the learnt background.

the problems currently with it are that any major change in lighting will render the background model useless, since it no longer applies to the new lighting conditions. a way of detecting this would be useful, so a reset can be issued to the background learning loop. This infers that there would be a window where the foreground is not determined whilst the background is being learnt again.

Perhaps another motion detection algorithm can kick in at that stage, something like differential analysis (subtracting previous frame from current). hopefully, by using knowledge prior to the light change, the regions of last known moving objects can be applied to the differntial algorithm.

Im' thinking out loud here on how the impact of lighting changes can be minimised. you may need to study the code a little to to see what I'm gettin at.

ElKeeed · 04-12-2006, 06:42 AM

Wouldn't it be much more useful in .NET? Java is ok as a theoretical language and where stuff absolutely needs to be cross platform but is it really suited to this sort of thing?

alti · 04-12-2006, 07:50 AM

Quote: Originally Posted by ElKeeed

Wouldn't it be much more useful in .NET? Java is ok as a theoretical language and where stuff absolutely needs to be cross platform but is it really suited to this sort of thing?

actually java is perfectly suited for this sort of thing. C# is basicly nothing more then a ripped off copy of java. Java has had Generics and other powerful features in place long before dot net. And portibibly is key here, what if you want to the core classes of the SDK he is writing on systems other then windows systems (which for this application is very likely) Current OEM automakers might love this technology, but it will do them no good if it is in dot net.

BTW I still have to get a video capture device to try this out, I dont know much about them so i have been tring to learn, is there a capture device that you suggest sama?

sama · 04-12-2006, 11:39 AM

@ ElKEed

and that I have experience in Java/Vision/AI together + that I work with Java every day

.net is write in any language and run on windows (mainly), java is write in one language but run anywhere. It means that Linux/Mac users can also benefit from this project

Saying that, to make sure that everyone can take part, perhaps the framework can allow an algorithm to be defined outside of java, in a text file, and java would import it. in this way, mathmaticians with no java experience can write algorithms, whilst the algorithm modules can be used by application writers.

@alti
regarding a capture device, anything will do realy. as long as you can get a decent framerate. I'm currently using a webcam, but that has some major woes: it changes intensity when it wishes! I hope I can change that in software.

when you do get some time and you can capture, could you post/send me some footage from a night vision camera? I'd like to see what can be done with them.

alti · 04-12-2006, 01:00 PM

Quote: Originally Posted by sama

when you do get some time and you can capture, could you post/send me some footage from a night vision camera? I'd like to see what can be done with them.

sure thing, I ordered this today as a capture device, hopefully I can get my camera reassembled pretty quickly (right now it is in peices waiting to be installed in the car) in the interest of time, I might forgo the HUD i had planned temporarly and just use the computer to view the image.

SnyperBob · 04-16-2006, 10:10 AM

I'd be interested in helping with some of the algorithms. I have a strong math background, but a limited amount of programming experience. I have a little experience with some AI classes I've taken, just enough to get me interested.

I'll take a look at your attached file this week when I get some free time. This is definately promising

pronerd · 04-16-2006, 11:10 AM

Hi, Guys I do not know if I can help or not. It sounds like most of this is beyond my background, and I do not currently have working Car PC with a camera set up. I have done some work with image generation using JAI. If there is something I can add let me know.

pronerd · 04-16-2006, 11:16 AM

Quote: Originally Posted by ElKeeed

Wouldn't it be much more useful in .NET? Java is ok as a theoretical language and where stuff absolutely needs to be cross platform but is it really suited to this sort of thing?

It might help if you provided some supporting reasons as to why it would be better in .Net. What is your line of reasoning for this conclusion?

My first reaction is to disagree since one of goals of this idea is to have a development community using and supporting this code base. To switch to .Net everyone working on it would need access copy of Visual Studio. Where as Eclipse and Netbeans are freely available. Also from the reading I have done there are much richer graphics and image library options for Java than .Net.

As far as Java only being a “theoretical language” I can not follow this line of reasoning. Java is the technology running the back ends of many mission critical services. It is what runs Google, eBay, Amazon, etc.

sama · 04-16-2006, 11:17 AM

excellent!

I'm almost certain that to get motion detection whilst moving requires optic flow. Whilst I understand the high level concept, my maths aint great when it comes to reading greek! But I can understand it once it's in an algorithm that I can use.

If you don't mind, see if you can make sense of what this guy's talking about. I know that wavelets are a super elegant and fast way to do lots of different comupations. This guy proposes a method for doing optic flow using it. beyond me for now!

sama · 04-16-2006, 11:19 AM

added instructions in the first post to get the code up and running. So any new comers aren't scared away

hi pronerd. if you can give feedback on how it's working for your future setup, that would be more than enough contribution. Of course you're welcome to mess with the code

alti · 04-25-2006, 12:29 AM

sorry it took so long, but i finally took the first step and tested it out.

I cant seem to get it to work though, I have my cam up and running in centrafuse and in jmStudio (I can start capturing and it works) but in your program all i get is a black screen???

I will play more and post some screen shots, do you think this is because the camera is b/w?? I wish i had a color camera to test it... hmmm, maybe i will pick up a cheap web cam.