How to build CLM-framework in Mac OSX.? about clm-framework HOT 15 CLOSED

I don't have a machine with Mac OSX and don't have much experience developing on it. Are you using the gcc compiler and have you installed tbb separately?

from clm-framework.

Hi ,
I builded the CLM-framework IN Mac OSX and executed simpleCLM tracker. I builded the CLM-framework for Mobile platform(IOS). But Mobile platform(IOS) did not support TBB .So I want to build CLM-framework without TBB for Mobile platform.How to build CLM-framework without TBB.?

I tried to set TBB flag:
cmake .. -DWITH_TBB=OFF

But still LibCLM.a need TBB library.

Can you tell how to disable the TBB flag.?

from clm-framework.

Hi,

I don't plan on adding support for iOS anytime soon, as I'm not a mobile developer and don't really have a way to test it as well.

One way you could try fixing this is by removing the code that relies on TBB (you would need to convert it back to regular loops though). There are a couple of places where TBB is used in the code (the face detection in dlib), patch expert computation and multiple face tracking (you can just search where tbb is used in the source files).

Thanks,
Tadas

from clm-framework.

Thanks for your answer.

I removed TBB and dlib from this CLM-framework, and I used Haar cascade for face detection instead of Dlib. I got wrong face detection coordinates from CLM face detector code. Is wrong i used for Haar cascade face detection instead of dlib in CLM-framework.?

How to solve this wrong face coordinates from CLM face detection.?

from clm-framework.

The Haar face detector is much less reliable, especially in noisy or not well lit environments, but it should still work. You can test if it works by using:

clm_parameters.curr_face_detector = CLMTracker::CLMParameters::HAAR_DETECTOR;

This forces the tracker to use Haar instead of dlib face detector. Alternatively, if the Haar detector is not accurate enough for your purposes you can try using the default dlib code (http://dlib.net/) (as the one included with my tracker was modified to use TBB for speed), however, it will be somewhat slower.

from clm-framework.

Thanks for for your reply.

from clm-framework.

I tried HAAR_DETECTOR by removing dlib package. But it is run slow in IOS.also I tried another method, dlib package without TBB for IOS. I generated static library of default dlib package. I builded everything for IOS. But i did not get face. Everytime I got face size is Zero. Am i did anything wrong.? can you help me why dlib frontal face detection is not detect face in IOS.?

from clm-framework.

I haven't got experience with running things on IOS so can't really help you. But if HAAR_DETECTOR is too slow for you dlib is very unlikely to be able to run things faster.

Also I'm not the best suited person to ask about dlib face detection issues on IOS, you might have better luck contacting the author of dlib directly.

from clm-framework.

Thanks for your reply.

from clm-framework.

now dou you have CLM for ios? l really hurry,thanks. thanks..

from clm-framework.

For instructions of how to compile for Mac OSX, have a look at:
http://www.learnopencv.com/facial-landmark-detection/?utm_source=tuicool&utm_medium=referral

I do not have a version for use with iOS though and it is not being developed at the moment.

from clm-framework.

thanks for your reply.

from clm-framework.

Yes. I had builded CLM for IOS by modified DLIB face detection with HAAR detector. Because IOS did not support TBB support. So I removed TBB from DLIB .

from clm-framework.

so great, can you send me that lib, i am a student and l am doing a project, what you said is too difficult for me,and the deadline is coming,l would be grateful if you could send me a lib,my email is [email protected], thanks,thank you very much.-_-||

from clm-framework.

Hi all
I tried to make a port of this library for iOs, but without success ((
Could you please provide tips and instructions how I can do this

Thank you!

from clm-framework.