[MNIST] Add output verification with tizen build

Add the training loss value verification with tizen build

**Self evaluation:**
1. Build test: [x]Passed [ ]Failed [ ]Skipped
2. Run test: [x]Passed [ ]Failed [ ]Skipped

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>

diff --git a/Applications/MNIST/jni/main.cpp b/Applications/MNIST/jni/main.cpp
index 802086d..d1380b9 100644 (file)
--- a/Applications/MNIST/jni/main.cpp
+++ b/Applications/MNIST/jni/main.cpp
@@ -29,6 +29,10 @@
 #include <stdlib.h>
 #include <time.h>
 
+#if defined(__TIZEN__)
+#include <gtest/gtest.h>
+#endif
+
 #include "databuffer.h"
 #include "databuffer_func.h"
 #include "neuralnet.h"
@@ -81,6 +85,8 @@ const float tolerance = 0.1;
 
 std::string data_path;
 
+float training_loss = 0.0;
+
 /**
  * @brief     step function
  * @param[in] x value to be distinguished
@@ -229,6 +235,12 @@ int getBatch_val(float **outVec, float **outLabel, bool *last,
   return ML_ERROR_NONE;
 }
 
+#if defined(__TIZEN__)
+TEST(MNIST_training, verify_accuracy) {
+  EXPECT_FLOAT_EQ(training_loss, 2.0374029);
+}
+#endif
+
 /**
  * @brief     create NN
  *            Get Feature from tflite & run foword & back propatation
@@ -236,6 +248,7 @@ int getBatch_val(float **outVec, float **outLabel, bool *last,
  * @param[in]  arg 2 : resource path
  */
 int main(int argc, char *argv[]) {
+  int status = 0;
   if (argc < 2) {
     std::cout << "./nntrainer_mnist mnist.ini\n";
     exit(0);
@@ -277,19 +290,42 @@ int main(int argc, char *argv[]) {
 
   NN.readModel();
   NN.setDataBuffer((DB));
+#if defined(__TIZEN__)
+  status = NN.setProperty({"epochs=5"});
+  if (status != ML_ERROR_NONE) {
+    std::cerr << "Error setting the number of epochs" << std::endl;
+    return 0;
+  }
+#endif
 
   /**
    * @brief     Neural Network Train & validation
    */
   try {
     NN.train();
+    training_loss = NN.getLoss();
   } catch (...) {
     std::cerr << "Error during train" << std::endl;
     return 0;
   }
 
+#if defined(__TIZEN__)
+  try {
+    testing::InitGoogleTest(&argc, argv);
+  } catch (...) {
+    std::cerr << "Error duing InitGoogleTest" << std::endl;
+    return 0;
+  }
+
+  try {
+    status = RUN_ALL_TESTS();
+  } catch (...) {
+    std::cerr << "Error duing RUN_ALL_TSETS()" << std::endl;
+  }
+#endif
+
   /**
    * @brief     Finalize NN
    */
-  return 0;
+  return status;
 }

diff --git a/Applications/MNIST/jni/meson.build b/Applications/MNIST/jni/meson.build
index 28f37c5..27f1bc4 100644 (file)
--- a/Applications/MNIST/jni/meson.build
+++ b/Applications/MNIST/jni/meson.build
@@ -1,10 +1,15 @@
 mnist_sources = [
   'main.cpp'
 ]
+if build_platform == 'tizen'
+  if not gtest_dep.found()
+    error('Gtest dependency not found for MNIST application')
+  endif
+endif
 
 executable('nntrainer_mnist',
   mnist_sources,
-  dependencies: [iniparser_dep, nntrainer_dep],
+  dependencies: [iniparser_dep, nntrainer_dep, gtest_dep],
   include_directories: include_directories('.'),
   install: get_option('install-app'),
   install_dir: application_install_dir

diff --git a/packaging/nntrainer.spec b/packaging/nntrainer.spec
index 4870b8c..d339047 100644 (file)
--- a/packaging/nntrainer.spec
+++ b/packaging/nntrainer.spec
@@ -213,8 +213,15 @@ bash %{test_script} ./test
 export NNSTREAMER_CONF=$(pwd)/test/nnstreamer_filter_nntrainer/nnstreamer-test.ini
 export NNSTREAMER_FILTERS=$(pwd)/build/nnstreamer/tensor_filter
 pushd build
+rm -rf model.bin
 TF_APP=Applications/TransferLearning/Draw_Classification
 ./${TF_APP}/jni/nntrainer_training ../${TF_APP}/res/Training.ini ../${TF_APP}/res
+
+rm -rf model.bin
+cp ../Applications/MNIST/jni/mnist_trainingSet.dat .
+MNIST_APP=Applications/MNIST
+./${MNIST_APP}/jni/nntrainer_mnist ../${MNIST_APP}/res/mnist.ini
+
 popd
 
 # unittest for nntrainer plugin for nnstreamer