* the first commit in the merged dnn: convert some public API from Blob's to Mat's

* temporarily or permantently removed OpenCL optimizations, which are not always stable nor usually very efficient; we'll likely use Halide instead

* got rid of Blob and BlobShape completely; use cv::Mat and std::vector<int> instead

* fixed a few compile errors

* got rid of separate .hpp files with layer declarations; instead, put everything into the respective .cpp files

* normalized all the layers' constructors; we concentrate on loading deep networks layers from files instead of constructing them from scratch, so we retained only SomeLayer::SomeLayer(const LayerParams& params); constructors

* fixed sample compilation

* suppress doxygen warnings

* trying to fix python bindings generation for DNN module

* temporarily disable python bindings while we refactor the module

* fix win32/win64 compile errors; remove trailing whitespaces

* fix win32/win64 compile errors; remove trailing whitespaces

- made some of dependencies explicit
- removed dependencies to highgui and some other modules where possible
- modified some samples to build without modules

* Fix several issues in TrackerMedianFlow implementation

Particularly, add possibility to tune optical flow parameters for a median
flow tracker.

* Improve code of TrackerMedianFlow

Replace a lot of calls of std::vector::erase by one call of
std::remove_if.

* Delete unused code, use norm from OpenCV

* medianFlow:turn getMedian method into function, small code cleanup

* TrackerMedianFlow:fixes in parameters I/O, add test for them

* TrackerMedianFlow:replace double with float in temp buffers

* Fix indentation

* TrackerMedianFlow:add absent parameter case handling in read()

* TrackerMedianFlow:use ROI instead of copy when getting a patch

* TrackerMedianFlow:don't calc image pyramids 2 times

* MedianFlowTracker: use cvIsNan()

* MedianFlow: refactor vector filtration code

* MedianFlow: change if statements layout in filterPointsInVectors

Using opencv/core/utility.hpp parallel primitives,
Intel TBB as backend is desirable.

* Fix kcfTracker when called from Python bindings. The variable boundingBox in updateImpl when called from Python is initialized with 0s and it needs to be updated with the tracker state.

* Fix to return the bbox correctly in the first frame in Python

  - remove double-check, make MSVC happy

Required for MSVC builds - shouldn't affect result otherwise.

For some applications it is useful to have an estimate of how uncertain
the specific variable is estimated. This could help to act accordingly
e.g. increase the measurement zone if the current estimate is very
uncertain. 

Signed-off-by: Yan Wang <yan.wang@linux.intel.com>

http://www.votchallenge.net/vot2015/dataset.html