broken python wrappers for face module

[berak]

no methods are exposed:

>>> model = cv2.face.createLBPHFaceRecognizer()
>>> model
<face_LBPHFaceRecognizer 00ADB100>
>>> help(model)
Help on face_LBPHFaceRecognizer object:

class face_LBPHFaceRecognizer(__builtin__.object)
 |  Methods defined here:
 |
 |  __repr__(...)
 |      x.__repr__() <==> repr(x)
 |
 |  ----------------------------------------------------------------------
 |  Data and other attributes defined here:
 |
 |  __new__ = <built-in method __new__ of type object>
 |      T.__new__(S, ...) -> a new object with type S, a subtype of T

>>> model2 = cv2.face.createFisherFaceRecognizer()
>>> help(model2)
Help on face_BasicFaceRecognizer object:

class face_BasicFaceRecognizer(__builtin__.object)
 |  Methods defined here:
 |
 |  __repr__(...)
 |      x.__repr__() <==> repr(x)
 |
 |  ----------------------------------------------------------------------
 |  Data and other attributes defined here:
 |
 |  __new__ = <built-in method __new__ of type object>
 |      T.__new__(S, ...) -> a new object with type S, a subtype of T

code in pyopencv_generated_types.h has this:

static void pyopencv_face_BasicFaceRecognizer_specials(void)
{
    pyopencv_face_BasicFaceRecognizer_Type.tp_base = NULL; // should be: &pyopencv_face_FaceRecognizer_Type;

[berak]

due to the Algorithm removal, a new BasicFaceRecognizer class was introduced there, but it seems, the python wrappers can't handle more than one level of inheritance.

(btw, the java wrappers (currently not enabled) work fine !)

[mshabunin]

Looks like this problem only exists for modules inside nested namespaces. I've checked cv::StereoBM -> cv::StereoMatcher -> cv::Algorithm, and it works as expected, i.e. creates full inheritance chain and all parent methods.

class StereoBM(StereoMatcher)
 |  Method resolution order:
 |      StereoBM
 |      StereoMatcher
 |      Algorithm
 |      __builtin__.object

Nested modules members (I've checked ml and face) are shown like first level objects:

class ml_Boost(__builtin__.object)

I'll take a look into generator script to see how it can be fixed.

[mshabunin]

@berak, can you, please, check if the issue has been fixed for you with the latest master version (opencv + contrib)?

[berak]

yes, excellent, works now, thanks a lot !

>>> help(cv2.face)
Help on module cv2.face in cv2:

NAME
    cv2.face

FILE
    (built-in)

FUNCTIONS
    createEigenFaceRecognizer(...)
        createEigenFaceRecognizer([, num_components[, threshold]]) -> retval

    createFisherFaceRecognizer(...)
        createFisherFaceRecognizer([, num_components[, threshold]]) -> retval

    createLBPHFaceRecognizer(...)
        createLBPHFaceRecognizer([, radius[, neighbors[, grid_x[, grid_y[, thres
hold]]]]]) -> retval


>>> fr = createEigenFaceRecognizer()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'createEigenFaceRecognizer' is not defined
>>> fr = cv2.face.createEigenFaceRecognizer()
>>> help(fr)
Help on face_BasicFaceRecognizer object:

class face_BasicFaceRecognizer(face_FaceRecognizer)
 |  Method resolution order:
 |      face_BasicFaceRecognizer
 |      face_FaceRecognizer
 |      Algorithm
 |      __builtin__.object
 |
 |  Methods defined here:
 |
 |  __repr__(...)
 |      x.__repr__() <==> repr(x)
 |
 |  getEigenValues(...)
 |      getEigenValues() -> retval
 |
 |  getEigenVectors(...)
 |      getEigenVectors() -> retval
 |
 |  getLabels(...)
 |      getLabels() -> retval
 |
 |  getMean(...)
 |      getMean() -> retval
 |
 |  getNumComponents(...)
 |      getNumComponents() -> retval
 |
 |  getProjections(...)
 |      getProjections() -> retval
 |
 |  getThreshold(...)
 |      getThreshold() -> retval
 |
 |  setNumComponents(...)
 |      setNumComponents(val) -> None
 |
 |  setThreshold(...)
 |      setThreshold(val) -> None
 |
 |  ----------------------------------------------------------------------
 |  Data and other attributes defined here:
 |
 |  __new__ = <built-in method __new__ of type object>
 |      T.__new__(S, ...) -> a new object with type S, a subtype of T
 |
 |  ----------------------------------------------------------------------
 |  Methods inherited from face_FaceRecognizer:
 |
 |  getLabelInfo(...)
 |      getLabelInfo(label) -> retval
 |
 |  getLabelsByString(...)
 |      getLabelsByString(str) -> retval
 |
 |  load(...)
 |      load(filename) -> None
 |
 |  predict(...)
 |      predict(src) -> label, confidence
 |
 |  save(...)
 |      save(filename) -> None
 |
 |  setLabelInfo(...)
 |      setLabelInfo(label, strInfo) -> None
 |
 |  train(...)
 |      train(src, labels) -> None
 |
 |  update(...)
 |      update(src, labels) -> None

trying on vs2008 (!) i encountered another tiny issue:
it tried to wrap e:\code\opencv\modules\objdetect\include\opencv2\objdetect\detection_based_tracker.hpp , and the class is #idef'ed out (because no threading available).

the wrapper ofc. is not able to see the #ifdefs there, and it failed to compile/link.

adding the header to the blacklist in opencv\modules\python\common.cmake( ~line 31), like:

 ocv_list_filterout(opencv_hdrs "detection_based")

fixed this locally.