Thanks wvantoorn for the link.
It looks like BoofCV is active again ... Woot !

Since there were no updates for a long time I think kwatters and myself lost interest in this project and MRL's corresponding service - http://myrobotlab.org/service/BoofCv

I know kwatters fought with dependency management to get JavaCv/OpenCv & DL4J working probably at the cost of BoofCV needed dependencies....  So I suspect BoofCV service is not currently working..

We may not want to tackle getting BoofCV worky at the moment .. but I do like the idea of a common serializable data structure which could be published by (all) CV services.  Like CvData as apposed to OpenCvData (currently being used).

There was a lot of benefits in refactoring all the Speech Synthesis services (less code, easier to implement new speech services, easier to interface, fewer bugs, more features common accross all).  CV services could benefit similarly.  

This is not a call to change everything now .. more just trying to seed the the idea of a "common computer vision object"

kwatters

5 years 6 months ago

I'm happt ro hear BoofCV isn't a dead project!  We do have some major issues with classpaths because BoofCV is not as up to date with the latest version of OpenCV as we typically are.  

Perhaps we can look at it again.  Reality is, it's just a reference implementation of code around java bindings to OpenCV.  We already have that with JavaCV.  Maybe BoofCV is faster? 

If nothing else, Boof provides us with examples of what can be done with the OpenCV API.  We've used examples from BoofCV in the past, we should continue to look at them and use them to help us implement OpenCVFilters.  It looks like a more complete example of fisheye calibration, that's cool.  We should try to use that.  I think the tracker api is pretty much the same one I just added via JavaCV... 

Anyway, It's always good to see a fellow open source project making headway with new releases!  Steam on BoofCV!

GroG

5 years 6 months ago

In reply to by kwatters

Reality is, it's just a reference implementation of code around java bindings to OpenCV.  We already have that with JavaCV.  Maybe BoofCV is faster? 

No and No.

Unless they completely changed their original manifesto .. they are (or were) "pure" Java.  The one and only jni reference they had was with Sarxos webcam - which was just a means to get the image into Java via camera hardware.  It (in the past) was never a wrapper around OpenCV.

The examples I have seen in the past were complete re-writes of OpenCV functions in Java.

(again .. maybe this has changed currently, but that is what they stated in the past)

Interesting, I stand corrected.  Perhaps I thought that because so many of the boofcv examples look so similar to the javacv examples that I've been working with.  If that's the case, I'm not sure why we had so many version dependency conflicts.  I thought it was because BoofCV was actually pulling in JavaCV as a dependency.  I'll try and clone it and check it out...  Looks like they build with gradle now.  (gee, that's all the rage it seems...)