Merge pull request #3032 from vpisarev:refactor_ml2

apps/traincascade/CMakeLists.txt

-set(OPENCV_TRAINCASCADE_DEPS opencv_core opencv_ml opencv_imgproc opencv_photo opencv_objdetect opencv_imgcodecs opencv_videoio opencv_highgui opencv_calib3d opencv_video opencv_features2d)
+set(OPENCV_TRAINCASCADE_DEPS opencv_core opencv_imgproc opencv_objdetect opencv_imgcodecs opencv_highgui opencv_calib3d opencv_features2d)
 ocv_check_dependencies(${OPENCV_TRAINCASCADE_DEPS})
 
 if(NOT OCV_DEPENDENCIES_FOUND)
@@ -10,13 +10,10 @@ project(traincascade)
 ocv_include_directories("${CMAKE_CURRENT_SOURCE_DIR}" "${OpenCV_SOURCE_DIR}/include/opencv")
 ocv_include_modules(${OPENCV_TRAINCASCADE_DEPS})
 
-set(traincascade_files traincascade.cpp
-  cascadeclassifier.cpp cascadeclassifier.h
-  boost.cpp boost.h features.cpp traincascade_features.h
-  haarfeatures.cpp haarfeatures.h
-  lbpfeatures.cpp lbpfeatures.h
-  HOGfeatures.cpp HOGfeatures.h
-  imagestorage.cpp imagestorage.h)
+file(GLOB SRCS *.cpp)
+file(GLOB HDRS *.h*)
+
+set(traincascade_files ${SRCS} ${HDRS})
 
 set(the_target opencv_traincascade)
 add_executable(${the_target} ${traincascade_files})

apps/traincascade/boost.h

@@ -2,7 +2,7 @@
 #define _OPENCV_BOOST_H_
 
 #include "traincascade_features.h"
-#include "ml.h"
+#include "old_ml.hpp"
 
 struct CvCascadeBoostParams : CvBoostParams
 {

apps/traincascade/cascadeclassifier.h

@@ -7,8 +7,6 @@
 #include "lbpfeatures.h"
 #include "HOGfeatures.h" //new
 #include "boost.h"
-#include "cv.h"
-#include "cxcore.h"
 
 #define CC_CASCADE_FILENAME "cascade.xml"
 #define CC_PARAMS_FILENAME "params.xml"

apps/traincascade/traincascade.cpp

 #include "opencv2/core.hpp"
-
-#include "cv.h"
 #include "cascadeclassifier.h"
 
 using namespace std;

apps/traincascade/traincascade_features.h

@@ -2,9 +2,6 @@
 #define _OPENCV_FEATURES_H_
 
 #include "imagestorage.h"
-#include "cxcore.h"
-#include "cv.h"
-#include "ml.h"
 #include <stdio.h>
 
 #define FEATURES "features"

modules/core/doc/basic_structures.rst

@@ -845,7 +845,6 @@ For convenience, the following types from the OpenCV C API already have such a s
 that calls the appropriate release function:
 
 * ``CvCapture``
-* :ocv:struct:`CvDTreeSplit`
 * :ocv:struct:`CvFileStorage`
 * ``CvHaarClassifierCascade``
 * :ocv:struct:`CvMat`

modules/core/src/lapack.cpp

@@ -1557,13 +1557,17 @@ static void _SVDcompute( InputArray _aarr, OutputArray _w,
     {
         if( !at )
         {
-            transpose(temp_u, _u);
-            temp_v.copyTo(_vt);
+            if( _u.needed() )
+                transpose(temp_u, _u);
+            if( _vt.needed() )
+                temp_v.copyTo(_vt);
         }
         else
         {
-            transpose(temp_v, _u);
-            temp_u.copyTo(_vt);
+            if( _u.needed() )
+                transpose(temp_v, _u);
+            if( _vt.needed() )
+                temp_u.copyTo(_vt);
         }
     }
 }

modules/imgproc/src/morph.cpp

@@ -1221,7 +1221,7 @@ static bool IPPMorphReplicate(int op, const Mat &src, Mat &dst, const Mat &kerne
         IPP_MORPH_CASE(CV_32FC3, 32f_C3R, 32f);
         IPP_MORPH_CASE(CV_32FC4, 32f_C4R, 32f);
         default:
-            return false;
+            ;
         }
 
         #undef IPP_MORPH_CASE
@@ -1253,14 +1253,11 @@ static bool IPPMorphReplicate(int op, const Mat &src, Mat &dst, const Mat &kerne
         IPP_MORPH_CASE(CV_32FC3, 32f_C3R, 32f);
         IPP_MORPH_CASE(CV_32FC4, 32f_C4R, 32f);
         default:
-            return false;
+            ;
         }
         #undef IPP_MORPH_CASE
-
-#if defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ >= 8
-        return false; /// It disables false positive warning in GCC 4.8 and further
-#endif
     }
+    return false;
 }
 
 static bool IPPMorphOp(int op, InputArray _src, OutputArray _dst,

modules/java/CMakeLists.txt

@@ -6,7 +6,7 @@ if(IOS OR NOT PYTHON_EXECUTABLE OR NOT ANT_EXECUTABLE OR NOT (JNI_FOUND OR (ANDR
 endif()
 
 set(the_description "The java bindings")
-ocv_add_module(java BINDINGS opencv_core opencv_imgproc OPTIONAL opencv_objdetect opencv_features2d opencv_video opencv_imgcodecs opencv_videoio opencv_ml opencv_calib3d opencv_photo opencv_nonfree opencv_contrib)
+ocv_add_module(java BINDINGS opencv_core opencv_imgproc OPTIONAL opencv_objdetect opencv_features2d opencv_video opencv_imgcodecs opencv_videoio opencv_calib3d opencv_photo opencv_nonfree opencv_contrib)
 ocv_module_include_directories("${CMAKE_CURRENT_SOURCE_DIR}/generator/src/cpp")
 
 if(NOT ANDROID)

modules/java/generator/rst_parser.py

@@ -2,7 +2,7 @@
 
 from __future__ import print_function
 import os, sys, re, string, fnmatch
-allmodules = ["core", "flann", "imgproc", "ml", "imgcodecs", "videoio", "highgui", "video", "features2d", "calib3d", "objdetect", "legacy", "contrib", "cuda", "androidcamera", "java", "python", "stitching", "ts", "photo", "nonfree", "videostab", "softcascade", "superres"]
+allmodules = ["core", "flann", "imgproc", "imgcodecs", "videoio", "highgui", "video", "features2d", "calib3d", "objdetect", "legacy", "contrib", "cuda", "androidcamera", "java", "python", "stitching", "ts", "photo", "nonfree", "videostab", "softcascade", "superres"]
 verbose = False
 show_warnings = True
 show_errors = True

modules/java/generator/src/cpp/jni_part.cpp

@@ -14,10 +14,6 @@
 #  include "opencv2/video.hpp"
 #endif
 
-#ifdef HAVE_OPENCV_ML
-#  include "opencv2/ml.hpp"
-#endif
-
 #ifdef HAVE_OPENCV_CONTRIB
 #  include "opencv2/contrib.hpp"
 #endif
@@ -41,10 +37,7 @@ JNI_OnLoad(JavaVM* vm, void* )
 #ifdef HAVE_OPENCV_VIDEO
     init &= cv::initModule_video();
 #endif
-#ifdef HAVE_OPENCV_ML
-    init &= cv::initModule_ml();
-#endif
-    #ifdef HAVE_OPENCV_CONTRIB
+#ifdef HAVE_OPENCV_CONTRIB
     init &= cv::initModule_contrib();
 #endif
 

modules/ml/doc/ertrees.rst

-Extremely randomized trees
-==========================
-
-Extremely randomized trees have been introduced by Pierre Geurts, Damien Ernst and Louis Wehenkel in the article "Extremely randomized trees", 2006 [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.65.7485&rep=rep1&type=pdf]. The algorithm of growing Extremely randomized trees is similar to :ref:`Random Trees` (Random Forest), but there are two differences:
-
-#. Extremely randomized trees don't apply the bagging procedure to construct a set of the training samples for each tree. The same input training set is used to train all trees.
-
-#. Extremely randomized trees pick a node split very extremely (both a variable index and variable splitting value are chosen randomly), whereas Random Forest finds the best split (optimal one by variable index and variable splitting value) among random subset of variables.
-
-
-CvERTrees
-----------
-.. ocv:class:: CvERTrees : public CvRTrees
-
-    The class implements the Extremely randomized trees algorithm. ``CvERTrees`` is inherited from :ocv:class:`CvRTrees` and has the same interface, so see description of :ocv:class:`CvRTrees` class to get details. To set the training parameters of Extremely randomized trees the same class :ocv:struct:`CvRTParams` is used.

modules/ml/doc/ml.rst

@@ -15,9 +15,7 @@ Most of the classification and regression algorithms are implemented as C++ clas
     support_vector_machines
     decision_trees
     boosting
-    gradient_boosted_trees
     random_trees
-    ertrees
     expectation_maximization
     neural_networks
     mldata

modules/ml/doc/normal_bayes_classifier.rst

@@ -9,55 +9,26 @@ This simple classification model assumes that feature vectors from each class ar
 
 .. [Fukunaga90] K. Fukunaga. *Introduction to Statistical Pattern Recognition*. second ed., New York: Academic Press, 1990.
 
-CvNormalBayesClassifier
+NormalBayesClassifier
 -----------------------
-.. ocv:class:: CvNormalBayesClassifier : public CvStatModel
+.. ocv:class:: NormalBayesClassifier : public StatModel
 
 Bayes classifier for normally distributed data.
 
-CvNormalBayesClassifier::CvNormalBayesClassifier
-------------------------------------------------
-Default and training constructors.
+NormalBayesClassifier::create
+-----------------------------
+Creates empty model
 
-.. ocv:function:: CvNormalBayesClassifier::CvNormalBayesClassifier()
+.. ocv:function:: Ptr<NormalBayesClassifier> NormalBayesClassifier::create(const NormalBayesClassifier::Params& params=Params())
 
-.. ocv:function:: CvNormalBayesClassifier::CvNormalBayesClassifier( const Mat& trainData, const Mat& responses, const Mat& varIdx=Mat(), const Mat& sampleIdx=Mat() )
+    :param params: The model parameters. There is none so far, the structure is used as a placeholder for possible extensions.
 
-.. ocv:function:: CvNormalBayesClassifier::CvNormalBayesClassifier( const CvMat* trainData, const CvMat* responses, const CvMat* varIdx=0, const CvMat* sampleIdx=0 )
+Use ``StatModel::train`` to train the model, ``StatModel::train<NormalBayesClassifier>(traindata, params)`` to create and train the model, ``StatModel::load<NormalBayesClassifier>(filename)`` to load the pre-trained model.
 
-.. ocv:pyfunction:: cv2.NormalBayesClassifier([trainData, responses[, varIdx[, sampleIdx]]]) -> <NormalBayesClassifier object>
-
-The constructors follow conventions of :ocv:func:`CvStatModel::CvStatModel`. See :ocv:func:`CvStatModel::train` for parameters descriptions.
-
-CvNormalBayesClassifier::train
-------------------------------
-Trains the model.
-
-.. ocv:function:: bool CvNormalBayesClassifier::train( const Mat& trainData, const Mat& responses, const Mat& varIdx = Mat(), const Mat& sampleIdx=Mat(), bool update=false )
-
-.. ocv:function:: bool CvNormalBayesClassifier::train( const CvMat* trainData, const CvMat* responses, const CvMat* varIdx = 0, const CvMat* sampleIdx=0, bool update=false )
-
-.. ocv:pyfunction:: cv2.NormalBayesClassifier.train(trainData, responses[, varIdx[, sampleIdx[, update]]]) -> retval
-
-    :param update: Identifies whether the model should be trained from scratch (``update=false``) or should be updated using the new training data (``update=true``).
-
-The method trains the Normal Bayes classifier. It follows the conventions of the generic :ocv:func:`CvStatModel::train` approach with the following limitations:
-
-* Only ``CV_ROW_SAMPLE`` data layout is supported.
-* Input variables are all ordered.
-* Output variable is categorical , which means that elements of ``responses`` must be integer numbers, though the vector may have the ``CV_32FC1`` type.
-* Missing measurements are not supported.
-
-CvNormalBayesClassifier::predict
---------------------------------
+NormalBayesClassifier::predictProb
+----------------------------------
 Predicts the response for sample(s).
 
-.. ocv:function:: float CvNormalBayesClassifier::predict(  const Mat& samples,  Mat* results=0, Mat* results_prob=0 ) const
-
-.. ocv:function:: float CvNormalBayesClassifier::predict( const CvMat* samples, CvMat* results=0, CvMat* results_prob=0 ) const
-
-.. ocv:pyfunction:: cv2.NormalBayesClassifier.predict(samples) -> retval, results
-
-The method estimates the most probable classes for input vectors. Input vectors (one or more) are stored as rows of the matrix ``samples``. In case of multiple input vectors, there should be one output vector ``results``. The predicted class for a single input vector is returned by the method. The vector ``results_prob`` contains the output probabilities coresponding to each element of ``result``.
+.. ocv:function:: float NormalBayesClassifier::predictProb( InputArray inputs, OutputArray outputs, OutputArray outputProbs, int flags=0 ) const
 
-The function is parallelized with the TBB library.
+The method estimates the most probable classes for input vectors. Input vectors (one or more) are stored as rows of the matrix ``inputs``. In case of multiple input vectors, there should be one output vector ``outputs``. The predicted class for a single input vector is returned by the method. The vector ``outputProbs`` contains the output probabilities corresponding to each element of ``result``.

modules/ml/src/gbt.cpp

@@ -2,6 +2,8 @@
 #include "precomp.hpp"
 #include <time.h>
 
+#if 0
+
 #define pCvSeq CvSeq*
 #define pCvDTreeNode CvDTreeNode*
 
@@ -1359,3 +1361,5 @@ float CvGBTrees::predict( const cv::Mat& sample, const cv::Mat& _missing,
     return predict(&_sample, _missing.empty() ? 0 : &miss, 0,
                    slice==cv::Range::all() ? CV_WHOLE_SEQ : cvSlice(slice.start, slice.end), k);
 }
+
+#endif

modules/ml/test/test_gbttest.cpp

 
 #include "test_precomp.hpp"
 
+#if 0
+
 #include <string>
 #include <fstream>
 #include <iostream>
@@ -284,3 +286,5 @@ void CV_GBTreesTest::run(int)
 /////////////////////////////////////////////////////////////////////////////
 
 TEST(ML_GBTrees, regression) { CV_GBTreesTest test; test.safe_run(); }
+
+#endif

modules/ml/test/test_mltests.cpp

@@ -65,7 +65,7 @@ int CV_AMLTest::run_test_case( int testCaseIdx )
         for (int k = 0; k < icount; k++)
         {
 #endif
-            data.mix_train_and_test_idx();
+            data->shuffleTrainTest();
             code = train( testCaseIdx );
 #ifdef GET_STAT
             float case_result = get_error();
@@ -101,9 +101,10 @@ int CV_AMLTest::validate_test_results( int testCaseIdx )
     {
         resultNode["mean"] >> mean;
         resultNode["sigma"] >> sigma;
-        float curErr = get_error( testCaseIdx, CV_TEST_ERROR );
+        model->save(format("/Users/vp/tmp/dtree/testcase_%02d.cur.yml", testCaseIdx));
+        float curErr = get_test_error( testCaseIdx );
         const int coeff = 4;
-        ts->printf( cvtest::TS::LOG, "Test case = %d; test error = %f; mean error = %f (diff=%f), %d*sigma = %f",
+        ts->printf( cvtest::TS::LOG, "Test case = %d; test error = %f; mean error = %f (diff=%f), %d*sigma = %f\n",
                                 testCaseIdx, curErr, mean, abs( curErr - mean), coeff, coeff*sigma );
         if ( abs( curErr - mean) > coeff*sigma )
         {
@@ -125,6 +126,6 @@ int CV_AMLTest::validate_test_results( int testCaseIdx )
 TEST(ML_DTree, regression) { CV_AMLTest test( CV_DTREE ); test.safe_run(); }
 TEST(ML_Boost, regression) { CV_AMLTest test( CV_BOOST ); test.safe_run(); }
 TEST(ML_RTrees, regression) { CV_AMLTest test( CV_RTREES ); test.safe_run(); }
-TEST(ML_ERTrees, regression) { CV_AMLTest test( CV_ERTREES ); test.safe_run(); }
+TEST(DISABLED_ML_ERTrees, regression) { CV_AMLTest test( CV_ERTREES ); test.safe_run(); }
 
 /* End of file. */