I use OpenCV 2.0, there is a method in the cv::Mat_<_Tp>
class:
// conversion to vector.
operator vector<_Tp>() const;
When instantiated, the implementation does not compile on MSVS 2005:
template<typename _Tp> inline Mat_<_Tp>::operator vector<_Tp>() const
{
CV_Assert( rows == 1 || cols == 1 );
return isContinuous() ? vector<_Tp>((size_t)(rows + cols - 1), (_Tp*)data) :
(vector<_Tp>)((Mat_<_Tp>)this->t());
}
Is it a new fancy constructor of std::vector
that takes size_t
and _Tp*
from TR1 or something?
I thought they could just initialize the vector by an iterator instead:
vector<_Tp>((_Tp*)data, (_Tp*)data + rows + cols - 1)
Is it a bug, or don't I know something?
UPD. Compiler error text:
...\lib\opencv\include\opencv\cxmat.hpp(691) : error C2665: 'std::vector<_Ty>::vector' : none of the 6 overloads could convert all the argument types
with
[
_Ty=double
]
c:\program files\microsoft visual studio 8\vc\include\vector(473): could be 'std::vector<_Ty>::vector(__w64 unsigned int,const _Ty &)'
with
[
_Ty=double
]
while trying to match the argument list '(size_t, double *)'
...\lib\opencv\include\opencv\cxmat.hpp(689) : while compiling class template member function 'cv::Mat_<_Tp>::operator std::vector<_Ty>(void) const'
with
[
_Tp=double,
_Ty=double
]
z:\dev\mine\temp\temp\entry.cpp(37) : see reference to class template instantiation 'cv::Mat_<_Tp>' being compiled
with
[
_Tp=double
]