Closed beraldofilippo closed 8 years ago
HI @beraldofilippo ,
Thank you. It is smart and nice to add a new method to add landmark in jni, could I move the code to dlib-android instead of building cpp source in dlib-android-app? I will like to separate the cpp and java code. Besides, I will like to keep mLabel in VisionDetRet because mLabel field will be used for object detection.
So I would like to merge the commits, d5c12ca and c71f8ed, first, and change it a little bit. Does it sound to you?
Hey @tzutalin sure, my intention was to use the .cpp source in dlib-android and then take the .so files here, just as you did until now! I've put it here just because I didn't have the correct config to build it...
As for the other stuff, yes it should be OK to put back the label field! I din't think about it because I was concentrating too much on the facial features :P
Hi @beraldofilippo , I sync your code manually, and change a little bit. I also sync the prebuilt shared library from dlib-android. https://github.com/tzutalin/dlib-android-app/commit/650afb63f611f775280117fb091714f91b4897ba
Great :D. Shall we close this PR now?
OK. Thanks for your pull-request
Please be careful, my configuration was completely different from the one in jnilib_ex, therefore I've added the working version of jni_people_det.cpp into JNIsources folder.
You should be ok in putting that into jnilib_ex (in dlib-android repo), compile and put the new .so files back into jniLibs folder of this repo.