Merge pull request #9019 from alalek:dnn_trace

modules/dnn/src/caffe/caffe_importer.cpp

@@ -43,7 +43,7 @@
 using namespace cv;
 using namespace cv::dnn;
 
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 #include "caffe.pb.h"
 
 #include <iostream>
@@ -82,6 +82,8 @@ public:
 
     CaffeImporter(const char *pototxt, const char *caffeModel)
     {
+        CV_TRACE_FUNCTION();
+
         ReadNetParamsFromTextFileOrDie(pototxt, &net);
 
         if (caffeModel && caffeModel[0])
@@ -264,6 +266,8 @@ public:
 
     void populateNet(Net dstNet)
     {
+        CV_TRACE_FUNCTION();
+
         int layersSize = net.layer_size();
         layerCounter.clear();
         addedBlobs.clear();

modules/dnn/src/caffe/caffe_io.cpp

@@ -87,7 +87,7 @@
 //
 //M*/
 
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 #include <google/protobuf/io/coded_stream.h>
 #include <google/protobuf/io/zero_copy_stream_impl.h>
 #include <google/protobuf/text_format.h>

modules/dnn/src/caffe/caffe_io.hpp

@@ -89,7 +89,7 @@
 
 #ifndef __OPENCV_DNN_CAFFE_IO_HPP__
 #define __OPENCV_DNN_CAFFE_IO_HPP__
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 
 #include "caffe.pb.h"
 

modules/dnn/src/init.cpp

@@ -60,6 +60,8 @@ Mutex* __initialization_mutex_initializer = &getInitializationMutex();
 
 void initializeLayerFactory()
 {
+    CV_TRACE_FUNCTION();
+
     CV_DNN_REGISTER_LAYER_CLASS(Slice,          SliceLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Split,          SplitLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Concat,         ConcatLayer);

modules/dnn/src/layers/batch_norm_layer.cpp

@@ -104,6 +104,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(blobs.size() >= 2);
         CV_Assert(inputs.size() == 1);
 

modules/dnn/src/layers/blank_layer.cpp

@@ -64,6 +64,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (int i = 0, n = outputs.size(); i < n; ++i)
             if (outputs[i].data != inputs[i]->data)
                 inputs[i]->copyTo(outputs[i]);

modules/dnn/src/layers/concat_layer.cpp

@@ -96,6 +96,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         int cAxis = clamp(axis, inputs[0]->dims);
         Mat& outMat = outputs[0];
         std::vector<Range> ranges(outputs[0].dims, Range::all());

modules/dnn/src/layers/convolution_layer.cpp

@@ -627,6 +627,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         /*printf("conv %s: input (%d x %d x %d x %d), kernel (%d x %d), pad (%d x %d), stride (%d x %d), dilation (%d x %d)\n",
                name.c_str(), inputs[0]->size[0], inputs[0]->size[1], inputs[0]->size[2], inputs[0]->size[3],
                kernel.width, kernel.height, pad.width, pad.height,
@@ -1013,6 +1016,9 @@ public:
 
     void forward(std::vector<Mat *> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         int outCn = blobs[0].size[0];
         int inpCn = inputs[0]->size[1];
         bool is1x1flag = is1x1();

modules/dnn/src/layers/crop_layer.cpp

@@ -135,6 +135,9 @@ public:
 
     void forward(std::vector<Mat *> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         Mat &input = *inputs[0];
         Mat &output = outputs[0];
 

modules/dnn/src/layers/detection_output_layer.cpp

@@ -206,6 +206,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         const float* locationData = inputs[0]->ptr<float>();
         const float* confidenceData = inputs[1]->ptr<float>();
         const float* priorData = inputs[2]->ptr<float>();

modules/dnn/src/layers/elementwise_layers.cpp

@@ -156,6 +156,8 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+
         for (size_t i = 0; i < inputs.size(); i++)
         {
             const Mat &src = *inputs[i];

modules/dnn/src/layers/eltwise_layer.cpp

@@ -251,6 +251,9 @@ public:
 
     void forward(std::vector<Mat *> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(outputs.size() == 1);
         const int nstripes = getNumThreads();
         EltwiseInvoker::run((const Mat**)&inputs[0], (int)inputs.size(), outputs[0],

modules/dnn/src/layers/flatten_layer.cpp

@@ -106,6 +106,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (size_t i = 0; i < inputs.size(); i++)
         {
             MatShape outShape = shape(outputs[i]);

modules/dnn/src/layers/fully_connected_layer.cpp

@@ -233,6 +233,9 @@ public:
 
     void forward(std::vector<Mat*> &input, std::vector<Mat> &output, std::vector<Mat> &)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         int axisCan = clamp(axis, input[0]->dims);
         int outerSize = input[0]->total(0, axisCan);
 

modules/dnn/src/layers/lrn_layer.cpp

@@ -86,6 +86,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(inputs.size() == outputs.size());
         for (int i = 0; i < inputs.size(); i++)
         {

modules/dnn/src/layers/max_unpooling_layer.cpp

@@ -57,6 +57,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(inputs.size() == 2);
         Mat& input = *inputs[0];
         Mat& indices = *inputs[1];

modules/dnn/src/layers/mvn_layer.cpp

@@ -62,6 +62,9 @@ public:
 
     void forward(std::vector<Mat *> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (size_t inpIdx = 0; inpIdx < inputs.size(); inpIdx++)
         {
             Mat &inpBlob = *inputs[inpIdx];

modules/dnn/src/layers/normalize_bbox_layer.cpp

@@ -142,6 +142,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         checkInputs(inputs);
 
         Mat& buffer = internals[0], sumChannelMultiplier = internals[1],

modules/dnn/src/layers/padding_layer.cpp

@@ -61,6 +61,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for(int i = 0; i < inputs.size(); i++)
         {
             outputs[i] = paddingValue;

modules/dnn/src/layers/permute_layer.cpp

@@ -245,6 +245,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         size_t k, ninputs = inputs.size();
         if(!_needsPermute)
         {

modules/dnn/src/layers/pooling_layer.cpp

@@ -106,6 +106,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (size_t ii = 0; ii < inputs.size(); ii++)
         {
             switch (type)

modules/dnn/src/layers/prior_box_layer.cpp

@@ -228,6 +228,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         int _layerWidth = inputs[0]->size[3];
         int _layerHeight = inputs[0]->size[2];
 

modules/dnn/src/layers/recurrent_layers.cpp

@@ -221,6 +221,9 @@ public:
 
     void forward(std::vector<Mat*> &input, std::vector<Mat> &output, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         const Mat &Wh = blobs[0];
         const Mat &Wx = blobs[1];
         const Mat &bias = blobs[2];
@@ -406,6 +409,9 @@ public:
 
     void forward(std::vector<Mat*> &input, std::vector<Mat> &output, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         Mat xTs = input[0]->reshape(1, numSamplesTotal);
         Mat oTs = output[0].reshape(1, numSamplesTotal);
         Mat hTs = produceH ? output[1].reshape(1, numSamplesTotal) : Mat();

modules/dnn/src/layers/reshape_layer.cpp

@@ -196,6 +196,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (size_t i = 0; i < inputs.size(); i++)
         {
             Mat srcBlob = *inputs[i];

modules/dnn/src/layers/scale_layer.cpp

@@ -45,6 +45,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(blobs.size() == 1 + hasBias);
 
         for (size_t ii = 0; ii < outputs.size(); ii++)

modules/dnn/src/layers/shift_layer.cpp

@@ -38,6 +38,9 @@ public:
 
     virtual void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         CV_Assert(inputs.size() > 0);
         CV_Assert(blobs.size() > 0);
 

modules/dnn/src/layers/slice_layer.cpp

@@ -118,6 +118,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         const Mat& inpMat = *inputs[0];
         std::vector<Range> ranges(inpMat.dims, Range::all());
         int cAxis = clamp(axis, inpMat.dims);

modules/dnn/src/layers/softmax_layer.cpp

@@ -84,6 +84,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         const Mat &src = *inputs[0];
         Mat &dst = outputs[0];
 

modules/dnn/src/layers/split_layer.cpp

@@ -80,6 +80,9 @@ public:
 
     void forward(std::vector<Mat*> &inputs, std::vector<Mat> &outputs, std::vector<Mat> &internals)
     {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
         for (size_t i = 0; i < outputs.size(); i++)
         {
             CV_Assert(inputs[0]->total() == outputs[i].total());

modules/dnn/src/precomp.hpp

@@ -40,6 +40,7 @@
 //M*/
 
 #include <opencv2/core.hpp>
+#include <opencv2/core/utils/trace.hpp>
 #include "cvconfig.h"
 #include <opencv2/dnn.hpp>
 #include <opencv2/dnn/all_layers.hpp>

modules/dnn/src/tensorflow/tf_importer.cpp

@@ -13,7 +13,7 @@ Implementation of Tensorflow models parser
 using namespace cv;
 using namespace cv::dnn;
 
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 #include "graph.pb.h"
 
 #include <iostream>

modules/dnn/src/tensorflow/tf_io.cpp

@@ -9,7 +9,7 @@
 Implementation of various functions which are related to Tensorflow models reading.
 */
 
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 #include <google/protobuf/io/coded_stream.h>
 #include <google/protobuf/io/zero_copy_stream_impl.h>
 #include <google/protobuf/text_format.h>

modules/dnn/src/tensorflow/tf_io.hpp

@@ -11,7 +11,7 @@ Declaration of various functions which are related to Tensorflow models reading.
 
 #ifndef __OPENCV_DNN_TF_IO_HPP__
 #define __OPENCV_DNN_TF_IO_HPP__
-#if HAVE_PROTOBUF
+#ifdef HAVE_PROTOBUF
 
 #include "graph.pb.h"
 

modules/dnn/src/torch/torch_importer.cpp

@@ -115,6 +115,8 @@ struct TorchImporter : public ::cv::dnn::Importer
 
     TorchImporter(String filename, bool isBinary)
     {
+        CV_TRACE_FUNCTION();
+
         rootModule = curModule = NULL;
         moduleCounter = 0;
 
@@ -966,6 +968,8 @@ struct TorchImporter : public ::cv::dnn::Importer
 
     void populateNet(Net net_)
     {
+        CV_TRACE_FUNCTION();
+
         if (rootModule == NULL)
         {
             rootModule = new Module("Sequential");
@@ -1014,6 +1018,8 @@ Mat readTorchBlob(const String&, bool)
 
 Net readNetFromTorch(const String &model, bool isBinary)
 {
+    CV_TRACE_FUNCTION();
+
     Ptr<Importer> importer = createTorchImporter(model, isBinary);
     Net net;
     if (importer)

samples/dnn/caffe_googlenet.cpp

@@ -41,6 +41,7 @@
 #include <opencv2/dnn.hpp>
 #include <opencv2/imgproc.hpp>
 #include <opencv2/highgui.hpp>
+#include <opencv2/core/utils/trace.hpp>
 using namespace cv;
 using namespace cv::dnn;
 
@@ -84,6 +85,8 @@ static std::vector<String> readClassNames(const char *filename = "synset_words.t
 
 int main(int argc, char **argv)
 {
+    CV_TRACE_FUNCTION();
+
     String modelTxt = "bvlc_googlenet.prototxt";
     String modelBin = "bvlc_googlenet.caffemodel";
     String imageFile = (argc > 1) ? argv[1] : "space_shuttle.jpg";
@@ -117,13 +120,20 @@ int main(int argc, char **argv)
                                   Scalar(104, 117, 123));   //Convert Mat to batch of images
     //! [Prepare blob]
 
-    //! [Set input blob]
-    net.setInput(inputBlob, "data");        //set the network input
-    //! [Set input blob]
-
-    //! [Make forward pass]
-    Mat prob = net.forward("prob");                          //compute output
-    //! [Make forward pass]
+    Mat prob;
+    cv::TickMeter t;
+    for (int i = 0; i < 10; i++)
+    {
+        CV_TRACE_REGION("forward");
+        //! [Set input blob]
+        net.setInput(inputBlob, "data");        //set the network input
+        //! [Set input blob]
+        t.start();
+        //! [Make forward pass]
+        prob = net.forward("prob");                          //compute output
+        //! [Make forward pass]
+        t.stop();
+    }
 
     //! [Gather output]
     int classId;
@@ -136,6 +146,7 @@ int main(int argc, char **argv)
     std::cout << "Best class: #" << classId << " '" << classNames.at(classId) << "'" << std::endl;
     std::cout << "Probability: " << classProb * 100 << "%" << std::endl;
     //! [Print results]
+    std::cout << "Time: " << (double)t.getTimeMilli() / t.getCounter() << " ms (average from " << t.getCounter() << " iterations)" << std::endl;
 
     return 0;
 } //main