Merge remote-tracking branch 'upstream/3.4' into merge-3.4

diff --cc CMakeLists.txt
index 0ca3375,169c385..4718c38
--- 1/CMakeLists.txt
--- 2/CMakeLists.txt
+++ b/CMakeLists.txt
@@@ -106,14 -113,8 +106,14 @@@ set(CMAKE_POSITION_INDEPENDENT_CODE ${E
  
  ocv_cmake_hook(PRE_CMAKE_BOOTSTRAP)
  
- # Bootstap CMake system: setup CMAKE_SYSTEM_NAME and other vars
+ # Bootstrap CMake system: setup CMAKE_SYSTEM_NAME and other vars
 +if(OPENCV_WORKAROUND_CMAKE_20989)
 +  set(CMAKE_SYSTEM_PROCESSOR_BACKUP ${CMAKE_SYSTEM_PROCESSOR})
 +endif()
  enable_language(CXX C)
 +if(OPENCV_WORKAROUND_CMAKE_20989)
 +  set(CMAKE_SYSTEM_PROCESSOR ${CMAKE_SYSTEM_PROCESSOR_BACKUP})
 +endif()
  
  ocv_cmake_hook(POST_CMAKE_BOOTSTRAP)
  

diff --cc modules/core/src/logger.cpp
index dbb25f9,94fa3fa..f395571
--- 1/modules/core/src/logger.cpp
--- 2/modules/core/src/logger.cpp
+++ b/modules/core/src/logger.cpp
@@@ -184,6 -81,15 +191,15 @@@ static int getShowTimestampMode(
  void writeLogMessage(LogLevel logLevel, const char* message)
  {
      const int threadID = cv::utils::getThreadID();
 -        case 1+2: message_id = cv::format("%d@%llu", threadID, getTimestampNS()); break;
+ 
+     std::string message_id;
+     switch (getShowTimestampMode())
+     {
+         case 1: message_id = cv::format("%d@%0.3f", threadID, getTimestampNS() * 1e-9); break;
++        case 1+2: message_id = cv::format("%d@%llu", threadID, (long long unsigned int)getTimestampNS()); break;
+         default: message_id = cv::format("%d", threadID); break;
+     }
+ 
      std::ostringstream ss;
      switch (logLevel)
      {

diff --cc modules/core/src/precomp.hpp
index 3057729,d30855e..d058260
--- 1/modules/core/src/precomp.hpp
--- 2/modules/core/src/precomp.hpp
+++ b/modules/core/src/precomp.hpp
@@@ -368,8 -374,19 +368,12 @@@ bool __termination;  // skip some clean
  
  cv::Mutex& getInitializationMutex();
  
 -// TODO Memory barriers?
+ /// @brief Returns timestamp in nanoseconds since program launch
+ int64 getTimestampNS();
+ 
+ 
  #define CV_SINGLETON_LAZY_INIT_(TYPE, INITIALIZER, RET_VALUE) \
 -    static TYPE* volatile instance = NULL; \
 -    if (instance == NULL) \
 -    { \
 -        cv::AutoLock lock(cv::getInitializationMutex()); \
 -        if (instance == NULL) \
 -            instance = INITIALIZER; \
 -    } \
 +    static TYPE* const instance = INITIALIZER; \
      return RET_VALUE;
  
  #define CV_SINGLETON_LAZY_INIT(TYPE, INITIALIZER) CV_SINGLETON_LAZY_INIT_(TYPE, INITIALIZER, instance)

diff --cc modules/core/src/system.cpp
Simple merge

diff --cc modules/dnn/test/test_backends.cpp
index e8c7e70,19e0729..cefd22c
--- 1/modules/dnn/test/test_backends.cpp
--- 2/modules/dnn/test/test_backends.cpp
+++ b/modules/dnn/test/test_backends.cpp
@@@ -217,23 -209,22 +217,31 @@@ TEST_P(DNNTestNetwork, MobileNet_SSD_Ca
  #if defined(INF_ENGINE_RELEASE)
      if ((backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 || backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH) &&
          target == DNN_TARGET_MYRIAD && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X)
-         applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X);
+         applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X, CV_TEST_TAG_DNN_SKIP_IE_NGRAPH, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
+ #endif
+ #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_EQ(2021040000)
+     // IE exception: Ngraph operation Transpose with name conv15_2_mbox_conf_perm has dynamic output shape on 0 port, but CPU plug-in supports only static shape
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))
+         applyTestTag(target == DNN_TARGET_OPENCL ? CV_TEST_TAG_DNN_SKIP_IE_OPENCL : CV_TEST_TAG_DNN_SKIP_IE_OPENCL_FP16,
+             CV_TEST_TAG_DNN_SKIP_IE_NGRAPH, CV_TEST_TAG_DNN_SKIP_IE_VERSION
+         );
  #endif
+ 
      Mat sample = imread(findDataFile("dnn/street.png"));
      Mat inp = blobFromImage(sample, 1.0f / 127.5, Size(300, 560), Scalar(127.5, 127.5, 127.5), false);
 -    float diffScores  = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.029 : 0.0;
 -    float diffSquares = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.09  : 0.0;
 +    float scoreDiff = 0.0, iouDiff = 0.0;
 +    if (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)
 +    {
 +        scoreDiff = 0.029;
 +        iouDiff = 0.09;
 +    }
 +    else if (target == DNN_TARGET_CUDA_FP16)
 +    {
 +        scoreDiff = 0.03;
 +        iouDiff = 0.08;
 +    }
      processNet("dnn/MobileNetSSD_deploy.caffemodel", "dnn/MobileNetSSD_deploy.prototxt",
 -                inp, "detection_out", "", diffScores, diffSquares);
 +                inp, "detection_out", "", scoreDiff, iouDiff);
      expectNoFallbacksFromIE(net);
  }
  

diff --cc modules/dnn/test/test_caffe_importer.cpp
index 24e4fa2,18b8d5a..e955443
--- 1/modules/dnn/test/test_caffe_importer.cpp
--- 2/modules/dnn/test/test_caffe_importer.cpp
+++ b/modules/dnn/test/test_caffe_importer.cpp
@@@ -501,26 -483,18 +501,28 @@@ TEST_P(Test_Caffe_nets, Colorization
      Mat out = net.forward();
  
      // Reference output values are in range [-29.1, 69.5]
 -    double l1 = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.25 : 4e-4;
 -    double lInf = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 5.3 : 3e-3;
 -    if (target == DNN_TARGET_MYRIAD && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X)
 +    double l1 = 4e-4, lInf = 3e-3;
 +    if (target == DNN_TARGET_OPENCL_FP16)
 +    {
 +        l1 = 0.25;
 +        lInf = 5.3;
 +    }
 +    else if (target == DNN_TARGET_MYRIAD)
 +    {
 +        l1 = (getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X) ? 0.5 : 0.25;
 +        lInf = (getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X) ? 11 : 5.3;
 +    }
 +    else if(target == DNN_TARGET_CUDA_FP16)
      {
 -        l1 = 0.5; lInf = 11;
 +        l1 = 0.21;
 +        lInf = 4.5;
      }
+ #if defined(INF_ENGINE_RELEASE)
      if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_OPENCL_FP16)
      {
-         l1 = 0.26; lInf = 6.5;
+         l1 = 0.3; lInf = 10;
      }
+ #endif
  
      normAssert(out, ref, "", l1, lInf);
      expectNoFallbacksFromIE(net);

diff --cc modules/dnn/test/test_common.hpp
Simple merge

diff --cc modules/dnn/test/test_darknet_importer.cpp
index d822efa,885bd6e..2113641
--- 1/modules/dnn/test/test_darknet_importer.cpp
--- 2/modules/dnn/test/test_darknet_importer.cpp
+++ b/modules/dnn/test/test_darknet_importer.cpp
@@@ -354,11 -355,13 +354,18 @@@ TEST_P(Test_Darknet_nets, YoloVoc
          scoreDiff = 1e-2;
          iouDiff = 0.018;
      }
 +    else if (target == DNN_TARGET_CUDA_FP16)
 +    {
 +        scoreDiff = 0.03;
 +        iouDiff = 0.018;
 +    }
+ #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_EQ(2021040000)
+     // accuracy
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_OPENCL_FP16)
+     {
+         iouDiff = std::numeric_limits<double>::quiet_NaN();
+     }
+ #endif
  
      std::string config_file = "yolo-voc.cfg";
      std::string weights_file = "yolo-voc.weights";

diff --cc modules/dnn/test/test_halide_layers.cpp
Simple merge

diff --cc modules/dnn/test/test_layers.cpp
index ee0629b,836b0aa..6b408f3
--- 1/modules/dnn/test/test_layers.cpp
--- 2/modules/dnn/test/test_layers.cpp
+++ b/modules/dnn/test/test_layers.cpp
@@@ -1469,12 -1415,15 +1469,20 @@@ TEST_P(Test_DLDT_two_inputs, as_backend
      // Output values are in range [0, 637.5].
      double l1 = (targetId == DNN_TARGET_OPENCL_FP16 || targetId == DNN_TARGET_MYRIAD) ? 0.06 : 1e-6;
      double lInf = (targetId == DNN_TARGET_OPENCL_FP16 || targetId == DNN_TARGET_MYRIAD) ? 0.3 : 1e-5;
 +    if (targetId == DNN_TARGET_CUDA_FP16)
 +    {
 +        l1 = 0.06;
 +        lInf = 0.3;
 +    }
      normAssert(out, ref, "", l1, lInf);
+     if (cvtest::debugLevel > 0 || HasFailure())
+     {
+         std::cout << "input1 scale=" << kScale << " input2 scale=" << kScaleInv << std::endl;
+         std::cout << "input1: " << firstInp.size << " " << firstInp.reshape(1, 1) << std::endl;
+         std::cout << "input2: " << secondInp.size << " " << secondInp.reshape(1, 1) << std::endl;
+         std::cout << "ref: " << ref.reshape(1, 1) << std::endl;
+         std::cout << "out: " << out.reshape(1, 1) << std::endl;
+     }
  }
  
  INSTANTIATE_TEST_CASE_P(/*nothing*/, Test_DLDT_two_inputs, Combine(

diff --cc modules/dnn/test/test_onnx_importer.cpp
index 56a6ab6,0fd5d08..f5696e5
--- 1/modules/dnn/test/test_onnx_importer.cpp
--- 2/modules/dnn/test/test_onnx_importer.cpp
+++ b/modules/dnn/test/test_onnx_importer.cpp
@@@ -1367,14 -1152,9 +1400,14 @@@ TEST_P(Test_ONNX_nets, TinyYolov2
      double l1 =  default_l1, lInf = default_lInf;
      if (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)
      {
-         l1 = 0.017;
-         lInf = 0.14;
+         l1 = 0.02;
+         lInf = 0.2;
      }
 +    else if (target == DNN_TARGET_CUDA_FP16)
 +    {
 +        l1 = 0.018;
 +        lInf = 0.16;
 +    }
  #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_EQ(2020040000)
      if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_OPENCL_FP16)
      {

diff --cc modules/dnn/test/test_tf_importer.cpp
index 8125233,d02b713..ed8e864
--- 1/modules/dnn/test/test_tf_importer.cpp
--- 2/modules/dnn/test/test_tf_importer.cpp
+++ b/modules/dnn/test/test_tf_importer.cpp
@@@ -715,13 -768,10 +838,14 @@@ TEST_P(Test_TensorFlow_nets, MobileNet_
      double scoreDiff = default_l1, iouDiff = default_lInf;
      if (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)
      {
-         scoreDiff = 0.0043;
-         iouDiff = 0.037;
+         scoreDiff = 0.01;
+         iouDiff = 0.1;
      }
 +    else if (target == DNN_TARGET_CUDA_FP16)
 +    {
 +        iouDiff = 0.04;
 +    }
+ 
      normAssertDetections(ref, out, "", 0.2, scoreDiff, iouDiff);
  #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE >= 2019010000
      expectNoFallbacksFromIE(net);
@@@ -839,27 -870,84 +961,96 @@@ TEST_P(Test_TensorFlow_nets, Faster_RCN
      // Assertion `prior_height > 0' failed.
      if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_OPENCL_FP16)
          applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_OPENCL_FP16, CV_TEST_TAG_DNN_SKIP_IE_NGRAPH);
+ #endif
  
-     if (backend == DNN_BACKEND_OPENCV && target == DNN_TARGET_OPENCL_FP16)
-         applyTestTag(CV_TEST_TAG_DNN_SKIP_OPENCL_FP16);
++    if (backend == DNN_BACKEND_CUDA && target == DNN_TARGET_CUDA_FP16)
++        applyTestTag(CV_TEST_TAG_DNN_SKIP_CUDA_FP16);
++
+     checkBackend();
+ 
+     double scoresDiff = 1e-5;
+     double iouDiff = 1e-4;
+ 
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 || backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH)
+     {
+         scoresDiff = 0.02;
+         iouDiff = 0.1;
+     }
+ 
+     std::string name = "faster_rcnn_inception_v2_coco_2018_01_28";
+     {
+         std::string proto = findDataFile("dnn/" + name + ".pbtxt");
+         std::string model = findDataFile("dnn/" + name + ".pb", false);
+ 
+         Net net = readNetFromTensorflow(model, proto);
+         net.setPreferableBackend(backend);
+         net.setPreferableTarget(target);
+         Mat img = imread(findDataFile("dnn/dog416.png"));
+         Mat blob = blobFromImage(img, 1.0f, Size(800, 600), Scalar(), true, false);
+ 
+         net.setInput(blob);
+         Mat out = net.forward();
+ 
+         Mat ref = blobFromNPY(findDataFile("dnn/tensorflow/" + name + ".detection_out.npy"));
+ 
+         // accuracy (both OpenCV & IE)
+         if (target == DNN_TARGET_OPENCL_FP16)
+             applyTestTag(CV_TEST_TAG_DNN_SKIP_OPENCL_FP16);
+ 
+         normAssertDetections(ref, out, name.c_str(), 0.3, scoresDiff, iouDiff);
+     }
+ }
+ 
+ TEST_P(Test_TensorFlow_nets, Faster_RCNN_resnet50_coco_2018_01_28)
+ {
+     applyTestTag(
+         (target == DNN_TARGET_CPU ? CV_TEST_TAG_MEMORY_1GB : CV_TEST_TAG_MEMORY_2GB),
+         CV_TEST_TAG_LONG,
+         CV_TEST_TAG_DEBUG_VERYLONG
+     );
+ 
+ #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_EQ(2021040000)
+     // IE exception: Ngraph operation Transpose with name FirstStageBoxPredictor/ClassPredictor/reshape_1/nhwc has dynamic output shape on 0 port, but CPU plug-in supports only static shape
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))
+         applyTestTag(target == DNN_TARGET_OPENCL ? CV_TEST_TAG_DNN_SKIP_IE_OPENCL : CV_TEST_TAG_DNN_SKIP_IE_OPENCL_FP16,
+             CV_TEST_TAG_DNN_SKIP_IE_NGRAPH, CV_TEST_TAG_DNN_SKIP_IE_VERSION
+         );
+ #endif
+ 
+ #ifdef INF_ENGINE_RELEASE
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 &&
+         (INF_ENGINE_VER_MAJOR_LT(2019020000) || target != DNN_TARGET_CPU))
+         applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
+ 
+     if (INF_ENGINE_VER_MAJOR_GT(2019030000) &&
+         backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_MYRIAD)
+         applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_NGRAPH);
+ #endif
+ #if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_LT(2021040000)
+     // segfault: inference-engine/thirdparty/clDNN/src/gpu/detection_output_cpu.cpp:111:
+     // Assertion `prior_height > 0' failed.
+     if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && target == DNN_TARGET_OPENCL_FP16)
+         applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_OPENCL_FP16, CV_TEST_TAG_DNN_SKIP_IE_NGRAPH);
+ #endif
  
 +    if (backend == DNN_BACKEND_CUDA && target == DNN_TARGET_CUDA_FP16)
 +        applyTestTag(CV_TEST_TAG_DNN_SKIP_CUDA_FP16);
 +
      checkBackend();
  
      double scoresDiff = backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 ? 2.9e-5 : 1e-5;
      double iouDiff = 1e-4;
-     for (int i = 0; i < 2; ++i)
 +    if (target == DNN_TARGET_CUDA)
 +    {
 +        // for faster_rcnn_resnet50_coco_2018_01_28
 +        scoresDiff = 0.06;
 +        iouDiff = 0.08;
 +    }
+ 
+     std::string name = "faster_rcnn_resnet50_coco_2018_01_28";
      {
-         std::string proto = findDataFile("dnn/" + names[i] + ".pbtxt");
-         std::string model = findDataFile("dnn/" + names[i] + ".pb", false);
+         std::string proto = findDataFile("dnn/" + name + ".pbtxt");
+         std::string model = findDataFile("dnn/" + name + ".pb", false);
  
          Net net = readNetFromTensorflow(model, proto);
          net.setPreferableBackend(backend);

diff --cc modules/dnn/test/test_torch_importer.cpp
Simple merge

diff --cc modules/highgui/CMakeLists.txt
Simple merge

diff --cc modules/js/generator/embindgen.py
Simple merge

diff --cc platforms/winpack_dldt/build_package.py
Simple merge

diff --cc samples/dnn/siamrpnpp.py
Simple merge