Re: A few questions about OpenBR...

[Scott Klum]

Hi Hosein,

This is a good question and worth posting to the OpenBR mailing list (although there may be a duplicate question of this already answered).  In any case, I wouldn't compare OpenCV and OpenBR directly.  

OpenCV provides all sorts of low level matrix manipulations in addition to higher level image processing/computer vision algorithms that are encapsulated via their function calls.  Further, OpenCV's matrix library itself is very powerful and help one avoid reinventing the wheel for this sort of thing.

By comparison, OpenBR's strength is to be able to string a bunch of algorithmic steps together via the grammar (http://openbiometrics.org/docs/api_docs/cl_api/) making it very easy and simple to train statistical models for a lot of biometric tasks and evaluate their performance.  For example, it's possible to train a face detector on your own data without writing any code (see the SlidingWindowTransform and the associated ClassifierTransforms: CascadeClassifier and HaarRepresentation).  To do the same with OpenCV is challenging, especially with respect to manipulating parameters of the Cascade.  Further, all the code for evaluating algorithmic performance is already there for you.  You're absolutely right that OpenCV's documentation is significantly better than ours - we simply do not have the manpower to write the in depth tutorials like OpenCV.

So, I would classify OpenCV and OpenBR as serving different purposes.  Indeed, a lot of OpenBR transforms simply wrap calls to OpenCV functions.

I hope this helps.

Regards,

Scott

On Tue, Apr 11, 2017 at 6:39 AM, hosein saeidi <gh.s...@gmail.com> wrote:

Hi Scott,

    I’m Hosein. I’m working on a face recognition project as part-time and some time ago, I learned OpenBR because of Shervin Emami's suggestion.Shervin said to me: "OpenBR is a framework for doing many different types of face recognition algorithms. So it has lots of different face recognition techniques."
    I researched about OpenBR, I found just one face recognition algorithm is implemented in OpenBR. I mean 4SF. But there are some disadvantages in OpenBR. One of them, this is that the OpenBR is not updated (But OpenCV is updated) . And the another is there is a few document about OpenBR (opposite OpenCV).
    Do I understand correctly? I'm going to make a decision on how to proceed the project and this depends on your opinion. in your opinion for face recognition OpenCV is better or openBR?

cheers,

hosein

--

Scott Klum

[h.saeidi]

Hi Scott,

Thanks a lot for your time...

You didn't talk about updating OpenBR. Do you plan to update OpenBR?

[Scott Klum]

Hosein,

OpenBR is being updated regularly, although admittedly less regularly lately.  For example, Josh just pushed a new gallery yesterday related to parsing json data.  You may be alluding to updating such that we can support OpenCV3.  This is on the agenda (and we've had at least one user begin a patch for this), but I can't give you a firm date for when this will happen.

Regards,

Scott

On Wed, Apr 12, 2017 at 3:11 AM, hosein saeidi <gh.s...@gmail.com> wrote:

Hi Scott,

Thanks a lot for your time...

You didn't talk about updating OpenBR. Do you plan to update OpenBR?

cheers,

hosein

--

Scott Klum