COMPMID-417: Extract common toolchain support file

The support file is needed as Android toolchains are missing some C++11
functions. It further includes functions that have been introduced in
C++14 but which we already need.

Eventually the support file might become unnecessary if the Android
compilers improve and we switch to C++14.

Change-Id: Ib5ec506f1a17c78149c315ab2096b901b852c180
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79247
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Pablo Tello <pablo.tello@arm.com>

101 files changed, 284 insertions, 279 deletions

diff --git a/src/runtime/CL/CLMultiHOG.cpp b/src/runtime/CL/CLMultiHOG.cpp
index b9e8739454..88d45acd12 100644
--- a/src/runtime/CL/CLMultiHOG.cpp
+++ b/src/runtime/CL/CLMultiHOG.cpp
@@ -25,12 +25,12 @@
 
 #include "arm_compute/core/CL/ICLHOG.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 CLMultiHOG::CLMultiHOG(size_t num_models)
-    : _num_models(num_models), _model(arm_compute::cpp14::make_unique<CLHOG[]>(_num_models))
+    : _num_models(num_models), _model(arm_compute::support::cpp14::make_unique<CLHOG[]>(_num_models))
 {
 }
 
diff --git a/src/runtime/CL/CLPyramid.cpp b/src/runtime/CL/CLPyramid.cpp
index 41d81ea0f8..865f389f7f 100644
--- a/src/runtime/CL/CLPyramid.cpp
+++ b/src/runtime/CL/CLPyramid.cpp
@@ -24,10 +24,10 @@
 #include "arm_compute/runtime/CL/CLPyramid.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PyramidInfo.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/TensorShape.h"
+#include "support/ToolchainSupport.h"
 
 #include <array>
 #include <cmath>
@@ -52,7 +52,7 @@ void CLPyramid::init_auto_padding(const PyramidInfo &info)
 void CLPyramid::internal_init(const PyramidInfo &info, bool auto_padding)
 {
     _info    = info;
-    _pyramid = arm_compute::cpp14::make_unique<CLTensor[]>(_info.num_levels());
+    _pyramid = arm_compute::support::cpp14::make_unique<CLTensor[]>(_info.num_levels());
 
     size_t      w            = _info.width();
     size_t      h            = _info.height();
diff --git a/src/runtime/CL/functions/CLAbsoluteDifference.cpp b/src/runtime/CL/functions/CLAbsoluteDifference.cpp
index 5097dd4710..5613e6c78d 100644
--- a/src/runtime/CL/functions/CLAbsoluteDifference.cpp
+++ b/src/runtime/CL/functions/CLAbsoluteDifference.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLAbsoluteDifference.h"
 
 #include "arm_compute/core/CL/kernels/CLAbsoluteDifferenceKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLAbsoluteDifference::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLAbsoluteDifferenceKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLAbsoluteDifferenceKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLAccumulate.cpp b/src/runtime/CL/functions/CLAccumulate.cpp
index 56c519984c..78f25fc015 100644
--- a/src/runtime/CL/functions/CLAccumulate.cpp
+++ b/src/runtime/CL/functions/CLAccumulate.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLAccumulate.h"
 
 #include "arm_compute/core/CL/kernels/CLAccumulateKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,21 +32,21 @@ using namespace arm_compute;
 
 void CLAccumulate::configure(const ICLTensor *input, ICLTensor *accum)
 {
-    auto k = arm_compute::cpp14::make_unique<CLAccumulateKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLAccumulateKernel>();
     k->configure(input, accum);
     _kernel = std::move(k);
 }
 
 void CLAccumulateWeighted::configure(const ICLTensor *input, float alpha, ICLTensor *accum)
 {
-    auto k = arm_compute::cpp14::make_unique<CLAccumulateWeightedKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLAccumulateWeightedKernel>();
     k->configure(input, alpha, accum);
     _kernel = std::move(k);
 }
 
 void CLAccumulateSquared::configure(const ICLTensor *input, uint32_t shift, ICLTensor *accum)
 {
-    auto k = arm_compute::cpp14::make_unique<CLAccumulateSquaredKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLAccumulateSquaredKernel>();
     k->configure(input, shift, accum);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLActivationLayer.cpp b/src/runtime/CL/functions/CLActivationLayer.cpp
index 0d0da0ca9a..b64739a008 100644
--- a/src/runtime/CL/functions/CLActivationLayer.cpp
+++ b/src/runtime/CL/functions/CLActivationLayer.cpp
@@ -24,13 +24,13 @@
 #include "arm_compute/runtime/CL/functions/CLActivationLayer.h"
 
 #include "arm_compute/core/CL/kernels/CLActivationLayerKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void CLActivationLayer::configure(ICLTensor *input, ICLTensor *output, ActivationLayerInfo act_info)
 {
-    auto k = arm_compute::cpp14::make_unique<CLActivationLayerKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLActivationLayerKernel>();
     k->configure(input, output, act_info);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLArithmeticAddition.cpp b/src/runtime/CL/functions/CLArithmeticAddition.cpp
index 36bff4285c..5ca384d6a4 100644
--- a/src/runtime/CL/functions/CLArithmeticAddition.cpp
+++ b/src/runtime/CL/functions/CLArithmeticAddition.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLArithmeticAddition.h"
 
 #include "arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLArithmeticAddition::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy)
 {
-    auto k = arm_compute::cpp14::make_unique<CLArithmeticAdditionKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLArithmeticAdditionKernel>();
     k->configure(input1, input2, output, policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLArithmeticSubtraction.cpp b/src/runtime/CL/functions/CLArithmeticSubtraction.cpp
index 97f0a1caf4..651f51a24a 100644
--- a/src/runtime/CL/functions/CLArithmeticSubtraction.cpp
+++ b/src/runtime/CL/functions/CLArithmeticSubtraction.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLArithmeticSubtraction.h"
 
 #include "arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLArithmeticSubtraction::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy)
 {
-    auto k = arm_compute::cpp14::make_unique<CLArithmeticSubtractionKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLArithmeticSubtractionKernel>();
     k->configure(input1, input2, output, policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLBitwiseAnd.cpp b/src/runtime/CL/functions/CLBitwiseAnd.cpp
index 7c85043206..f8a5a859bd 100644
--- a/src/runtime/CL/functions/CLBitwiseAnd.cpp
+++ b/src/runtime/CL/functions/CLBitwiseAnd.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLBitwiseAnd.h"
 
 #include "arm_compute/core/CL/kernels/CLBitwiseAndKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLBitwiseAnd::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLBitwiseAndKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLBitwiseAndKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLBitwiseNot.cpp b/src/runtime/CL/functions/CLBitwiseNot.cpp
index 17ae5dea3c..dc002e5cac 100644
--- a/src/runtime/CL/functions/CLBitwiseNot.cpp
+++ b/src/runtime/CL/functions/CLBitwiseNot.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLBitwiseNot.h"
 
 #include "arm_compute/core/CL/kernels/CLBitwiseNotKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLBitwiseNot::configure(const ICLTensor *input, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLBitwiseNotKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLBitwiseNotKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLBitwiseOr.cpp b/src/runtime/CL/functions/CLBitwiseOr.cpp
index c84a279bae..4a10bb2f36 100644
--- a/src/runtime/CL/functions/CLBitwiseOr.cpp
+++ b/src/runtime/CL/functions/CLBitwiseOr.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLBitwiseOr.h"
 
 #include "arm_compute/core/CL/kernels/CLBitwiseOrKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLBitwiseOr::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLBitwiseOrKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLBitwiseOrKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLBitwiseXor.cpp b/src/runtime/CL/functions/CLBitwiseXor.cpp
index fd49c7d818..d23622ab01 100644
--- a/src/runtime/CL/functions/CLBitwiseXor.cpp
+++ b/src/runtime/CL/functions/CLBitwiseXor.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLBitwiseXor.h"
 
 #include "arm_compute/core/CL/kernels/CLBitwiseXorKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLBitwiseXor::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLBitwiseXorKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLBitwiseXorKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLBox3x3.cpp b/src/runtime/CL/functions/CLBox3x3.cpp
index 20d5dae94e..f28be44d72 100644
--- a/src/runtime/CL/functions/CLBox3x3.cpp
+++ b/src/runtime/CL/functions/CLBox3x3.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLBox3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLBox3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLBox3x3::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLBox3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLBox3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, BorderSize(1), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLCannyEdge.cpp b/src/runtime/CL/functions/CLCannyEdge.cpp
index 1d018b8347..6c321ad69a 100644
--- a/src/runtime/CL/functions/CLCannyEdge.cpp
+++ b/src/runtime/CL/functions/CLCannyEdge.cpp
@@ -26,12 +26,12 @@
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/OpenCL.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 #include "arm_compute/runtime/CL/functions/CLSobel3x3.h"
 #include "arm_compute/runtime/CL/functions/CLSobel5x5.h"
 #include "arm_compute/runtime/CL/functions/CLSobel7x7.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -86,19 +86,19 @@ void CLCannyEdge::configure(ICLTensor *input, ICLTensor *output, int32_t upper_t
     // Configure/Init sobelNxN
     if(gradient_size == 3)
     {
-        auto k = arm_compute::cpp14::make_unique<CLSobel3x3>();
+        auto k = arm_compute::support::cpp14::make_unique<CLSobel3x3>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
     else if(gradient_size == 5)
     {
-        auto k = arm_compute::cpp14::make_unique<CLSobel5x5>();
+        auto k = arm_compute::support::cpp14::make_unique<CLSobel5x5>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
     else if(gradient_size == 7)
     {
-        auto k = arm_compute::cpp14::make_unique<CLSobel7x7>();
+        auto k = arm_compute::support::cpp14::make_unique<CLSobel7x7>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
diff --git a/src/runtime/CL/functions/CLChannelCombine.cpp b/src/runtime/CL/functions/CLChannelCombine.cpp
index 79a3676bd7..11605cf95e 100644
--- a/src/runtime/CL/functions/CLChannelCombine.cpp
+++ b/src/runtime/CL/functions/CLChannelCombine.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLChannelCombine.h"
 
 #include "arm_compute/core/CL/kernels/CLChannelCombineKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,14 +32,14 @@ using namespace arm_compute;
 
 void CLChannelCombine::configure(const ICLTensor *plane0, const ICLTensor *plane1, const ICLTensor *plane2, const ICLTensor *plane3, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLChannelCombineKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLChannelCombineKernel>();
     k->configure(plane0, plane1, plane2, plane3, output);
     _kernel = std::move(k);
 }
 
 void CLChannelCombine::configure(const ICLImage *plane0, const ICLImage *plane1, const ICLImage *plane2, ICLMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLChannelCombineKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLChannelCombineKernel>();
     k->configure(plane0, plane1, plane2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLChannelExtract.cpp b/src/runtime/CL/functions/CLChannelExtract.cpp
index 2c6174b9ee..50903826d3 100644
--- a/src/runtime/CL/functions/CLChannelExtract.cpp
+++ b/src/runtime/CL/functions/CLChannelExtract.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLChannelExtract.h"
 
 #include "arm_compute/core/CL/kernels/CLChannelExtractKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,14 +32,14 @@ using namespace arm_compute;
 
 void CLChannelExtract::configure(const ICLTensor *input, Channel channel, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLChannelExtractKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLChannelExtractKernel>();
     k->configure(input, channel, output);
     _kernel = std::move(k);
 }
 
 void CLChannelExtract::configure(const ICLMultiImage *input, Channel channel, ICLImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLChannelExtractKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLChannelExtractKernel>();
     k->configure(input, channel, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLColorConvert.cpp b/src/runtime/CL/functions/CLColorConvert.cpp
index 2fe465aeb8..65f8ac3740 100644
--- a/src/runtime/CL/functions/CLColorConvert.cpp
+++ b/src/runtime/CL/functions/CLColorConvert.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLColorConvert.h"
 
 #include "arm_compute/core/CL/kernels/CLColorConvertKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,28 +32,28 @@ using namespace arm_compute;
 
 void CLColorConvert::configure(const ICLTensor *input, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void CLColorConvert::configure(const ICLImage *input, ICLMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void CLColorConvert::configure(const ICLMultiImage *input, ICLImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void CLColorConvert::configure(const ICLMultiImage *input, ICLMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLConvolution.cpp b/src/runtime/CL/functions/CLConvolution.cpp
index 21b5d47679..641044451d 100644
--- a/src/runtime/CL/functions/CLConvolution.cpp
+++ b/src/runtime/CL/functions/CLConvolution.cpp
@@ -26,13 +26,13 @@
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLConvolutionKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Utils.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 #include "arm_compute/runtime/ITensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -40,7 +40,7 @@ using namespace arm_compute;
 
 void CLConvolution3x3::configure(ICLTensor *input, ICLTensor *output, const int16_t *conv, uint32_t scale, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLConvolution3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLConvolution3x3Kernel>();
     k->configure(input, output, conv, scale, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
@@ -107,7 +107,7 @@ template class arm_compute::CLConvolutionSquare<9>;
 
 void CLConvolutionRectangle::configure(ICLTensor *input, ICLTensor *output, const int16_t *conv, uint32_t rows, uint32_t cols, uint32_t scale, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLConvolutionRectangleKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLConvolutionRectangleKernel>();
     k->configure(input, output, conv, rows, cols, scale, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLDepthConcatenate.cpp b/src/runtime/CL/functions/CLDepthConcatenate.cpp
index d967d9865f..34778c6980 100644
--- a/src/runtime/CL/functions/CLDepthConcatenate.cpp
+++ b/src/runtime/CL/functions/CLDepthConcatenate.cpp
@@ -27,10 +27,10 @@
 #include "arm_compute/core/CL/kernels/CLDepthConcatenateKernel.h"
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -47,8 +47,8 @@ void CLDepthConcatenate::configure(std::vector<ICLTensor *> inputs_vector, ICLTe
 
     unsigned int depth_offset = 0;
 
-    _concat_kernels_vector  = arm_compute::cpp14::make_unique<CLDepthConcatenateKernel[]>(_num_inputs);
-    _border_handlers_vector = arm_compute::cpp14::make_unique<CLFillBorderKernel[]>(_num_inputs);
+    _concat_kernels_vector  = arm_compute::support::cpp14::make_unique<CLDepthConcatenateKernel[]>(_num_inputs);
+    _border_handlers_vector = arm_compute::support::cpp14::make_unique<CLFillBorderKernel[]>(_num_inputs);
 
     for(unsigned int i = 0; i < _num_inputs; i++)
     {
diff --git a/src/runtime/CL/functions/CLDepthConvert.cpp b/src/runtime/CL/functions/CLDepthConvert.cpp
index edcd4928ab..b64d05b8b1 100644
--- a/src/runtime/CL/functions/CLDepthConvert.cpp
+++ b/src/runtime/CL/functions/CLDepthConvert.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLDepthConvert.h"
 
 #include "arm_compute/core/CL/kernels/CLDepthConvertKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLDepthConvert::configure(const ICLTensor *input, ICLTensor *output, ConvertPolicy policy, uint32_t shift)
 {
-    auto k = arm_compute::cpp14::make_unique<CLDepthConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLDepthConvertKernel>();
     k->configure(input, output, policy, shift);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLDerivative.cpp b/src/runtime/CL/functions/CLDerivative.cpp
index 9357de52f1..ae49996f3e 100644
--- a/src/runtime/CL/functions/CLDerivative.cpp
+++ b/src/runtime/CL/functions/CLDerivative.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLDerivative.h"
 
 #include "arm_compute/core/CL/kernels/CLDerivativeKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLDerivative::configure(ICLTensor *input, ICLTensor *output_x, ICLTensor *output_y, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLDerivativeKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLDerivativeKernel>();
     k->configure(input, output_x, output_y, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, BorderSize(1), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLDilate.cpp b/src/runtime/CL/functions/CLDilate.cpp
index 0a4abe30c2..59c5ea5439 100644
--- a/src/runtime/CL/functions/CLDilate.cpp
+++ b/src/runtime/CL/functions/CLDilate.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLDilate.h"
 
 #include "arm_compute/core/CL/kernels/CLDilateKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLDilate::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLDilateKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLDilateKernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, BorderSize(1), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLErode.cpp b/src/runtime/CL/functions/CLErode.cpp
index c41371cf9e..eb1f6e4bd8 100644
--- a/src/runtime/CL/functions/CLErode.cpp
+++ b/src/runtime/CL/functions/CLErode.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLErode.h"
 
 #include "arm_compute/core/CL/kernels/CLErodeKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLErode::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLErodeKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLErodeKernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, BorderSize(1), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLFillBorder.cpp b/src/runtime/CL/functions/CLFillBorder.cpp
index 172f09da1a..54c096e25d 100644
--- a/src/runtime/CL/functions/CLFillBorder.cpp
+++ b/src/runtime/CL/functions/CLFillBorder.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLFillBorder.h"
 
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLFillBorder::configure(ICLTensor *tensor, unsigned int border_width, BorderMode border_mode, const PixelValue &constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLFillBorderKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLFillBorderKernel>();
     k->configure(tensor, BorderSize(border_width), border_mode, constant_border_value);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLGEMMInterleave4x4.cpp b/src/runtime/CL/functions/CLGEMMInterleave4x4.cpp
index 9dc77156ef..45547e4cb6 100644
--- a/src/runtime/CL/functions/CLGEMMInterleave4x4.cpp
+++ b/src/runtime/CL/functions/CLGEMMInterleave4x4.cpp
@@ -24,13 +24,13 @@
 #include "arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h"
 
 #include "arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void CLGEMMInterleave4x4::configure(const ICLTensor *input, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLGEMMInterleave4x4Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLGEMMInterleave4x4Kernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLGEMMTranspose1xW.cpp b/src/runtime/CL/functions/CLGEMMTranspose1xW.cpp
index 8f8a19e709..d054e01611 100644
--- a/src/runtime/CL/functions/CLGEMMTranspose1xW.cpp
+++ b/src/runtime/CL/functions/CLGEMMTranspose1xW.cpp
@@ -25,14 +25,14 @@
 
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLGEMMTranspose1xWKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/Types.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void CLGEMMTranspose1xW::configure(const ICLTensor *input, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLGEMMTranspose1xWKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLGEMMTranspose1xWKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
\ No newline at end of file
diff --git a/src/runtime/CL/functions/CLGaussian3x3.cpp b/src/runtime/CL/functions/CLGaussian3x3.cpp
index 362a3fe920..7ebabd756d 100644
--- a/src/runtime/CL/functions/CLGaussian3x3.cpp
+++ b/src/runtime/CL/functions/CLGaussian3x3.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLGaussian3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLGaussian3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLGaussian3x3::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLGaussian3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLGaussian3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLGaussianPyramid.cpp b/src/runtime/CL/functions/CLGaussianPyramid.cpp
index 8a4279e99b..8450d5ee69 100644
--- a/src/runtime/CL/functions/CLGaussianPyramid.cpp
+++ b/src/runtime/CL/functions/CLGaussianPyramid.cpp
@@ -27,11 +27,11 @@
 #include "arm_compute/core/CL/kernels/CLGaussianPyramidKernel.h"
 #include "arm_compute/core/CL/kernels/CLScaleKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/core/Window.h"
+#include "support/ToolchainSupport.h"
 
 #include "arm_compute/runtime/CL/CLPyramid.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
@@ -70,9 +70,9 @@ void CLGaussianPyramidHalf::configure(ICLTensor *input, CLPyramid *pyramid, Bord
 
     if(num_levels > 1)
     {
-        _border_handler       = arm_compute::cpp14::make_unique<CLFillBorderKernel[]>(num_levels - 1);
-        _horizontal_reduction = arm_compute::cpp14::make_unique<CLGaussianPyramidHorKernel[]>(num_levels - 1);
-        _vertical_reduction   = arm_compute::cpp14::make_unique<CLGaussianPyramidVertKernel[]>(num_levels - 1);
+        _border_handler       = arm_compute::support::cpp14::make_unique<CLFillBorderKernel[]>(num_levels - 1);
+        _horizontal_reduction = arm_compute::support::cpp14::make_unique<CLGaussianPyramidHorKernel[]>(num_levels - 1);
+        _vertical_reduction   = arm_compute::support::cpp14::make_unique<CLGaussianPyramidVertKernel[]>(num_levels - 1);
 
         // Apply half scale to the X dimension of the tensor shape
         TensorShape tensor_shape = pyramid->info()->tensor_shape();
@@ -141,8 +141,8 @@ void CLGaussianPyramidOrb::configure(ICLTensor *input, CLPyramid *pyramid, Borde
 
     if(num_levels > 1)
     {
-        _gauss5x5      = arm_compute::cpp14::make_unique<CLGaussian5x5[]>(num_levels - 1);
-        _scale_nearest = arm_compute::cpp14::make_unique<CLScaleKernel[]>(num_levels - 1);
+        _gauss5x5      = arm_compute::support::cpp14::make_unique<CLGaussian5x5[]>(num_levels - 1);
+        _scale_nearest = arm_compute::support::cpp14::make_unique<CLScaleKernel[]>(num_levels - 1);
 
         PyramidInfo pyramid_info(num_levels - 1, SCALE_PYRAMID_ORB, pyramid->info()->tensor_shape(), Format::U8);
 
diff --git a/src/runtime/CL/functions/CLHOGMultiDetection.cpp b/src/runtime/CL/functions/CLHOGMultiDetection.cpp
index b8f2224ac8..6e864464a4 100644
--- a/src/runtime/CL/functions/CLHOGMultiDetection.cpp
+++ b/src/runtime/CL/functions/CLHOGMultiDetection.cpp
@@ -25,11 +25,11 @@
 
 #include "arm_compute/core/CL/OpenCL.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/runtime/CL/CLArray.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -114,12 +114,12 @@ void CLHOGMultiDetection::configure(ICLTensor *input, const ICLMultiHOG *multi_h
     _num_block_norm_kernel  = input_block_norm.size(); // Number of CLHOGBlockNormalizationKernel kernels to compute
     _num_hog_detect_kernel  = input_hog_detect.size(); // Number of CLHOGDetector functions to compute
 
-    _orient_bin_kernel = arm_compute::cpp14::make_unique<CLHOGOrientationBinningKernel[]>(_num_orient_bin_kernel);
-    _block_norm_kernel = arm_compute::cpp14::make_unique<CLHOGBlockNormalizationKernel[]>(_num_block_norm_kernel);
-    _hog_detect_kernel = arm_compute::cpp14::make_unique<CLHOGDetector[]>(_num_hog_detect_kernel);
-    _non_maxima_kernel = arm_compute::cpp14::make_unique<CPPDetectionWindowNonMaximaSuppressionKernel>();
-    _hog_space         = arm_compute::cpp14::make_unique<CLTensor[]>(_num_orient_bin_kernel);
-    _hog_norm_space    = arm_compute::cpp14::make_unique<CLTensor[]>(_num_block_norm_kernel);
+    _orient_bin_kernel = arm_compute::support::cpp14::make_unique<CLHOGOrientationBinningKernel[]>(_num_orient_bin_kernel);
+    _block_norm_kernel = arm_compute::support::cpp14::make_unique<CLHOGBlockNormalizationKernel[]>(_num_block_norm_kernel);
+    _hog_detect_kernel = arm_compute::support::cpp14::make_unique<CLHOGDetector[]>(_num_hog_detect_kernel);
+    _non_maxima_kernel = arm_compute::support::cpp14::make_unique<CPPDetectionWindowNonMaximaSuppressionKernel>();
+    _hog_space         = arm_compute::support::cpp14::make_unique<CLTensor[]>(_num_orient_bin_kernel);
+    _hog_norm_space    = arm_compute::support::cpp14::make_unique<CLTensor[]>(_num_block_norm_kernel);
 
     // Allocate tensors for magnitude and phase
     TensorInfo info_mag(shape_img, Format::S16);
diff --git a/src/runtime/CL/functions/CLHarrisCorners.cpp b/src/runtime/CL/functions/CLHarrisCorners.cpp
index e1c44416ae..87d573a7ad 100644
--- a/src/runtime/CL/functions/CLHarrisCorners.cpp
+++ b/src/runtime/CL/functions/CLHarrisCorners.cpp
@@ -27,7 +27,6 @@
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
 #include "arm_compute/core/CL/kernels/CLHarrisCornersKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/CL/CLScheduler.h"
@@ -36,6 +35,7 @@
 #include "arm_compute/runtime/CL/functions/CLSobel7x7.h"
 #include "arm_compute/runtime/ITensorAllocator.h"
 #include "arm_compute/runtime/Scheduler.h"
+#include "support/ToolchainSupport.h"
 
 #include <cmath>
 #include <utility>
@@ -69,28 +69,28 @@ void CLHarrisCorners::configure(ICLImage *input, float threshold, float min_dist
     _score.allocator()->init(info_f32);
     _nonmax.allocator()->init(info_f32);
 
-    _corners_list = arm_compute::cpp14::make_unique<InternalKeypoint[]>(shape.x() * shape.y());
+    _corners_list = arm_compute::support::cpp14::make_unique<InternalKeypoint[]>(shape.x() * shape.y());
 
     /* Set/init Sobel kernel accordingly with gradient_size */
     switch(gradient_size)
     {
         case 3:
         {
-            auto k = arm_compute::cpp14::make_unique<CLSobel3x3>();
+            auto k = arm_compute::support::cpp14::make_unique<CLSobel3x3>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
         }
         case 5:
         {
-            auto k = arm_compute::cpp14::make_unique<CLSobel5x5>();
+            auto k = arm_compute::support::cpp14::make_unique<CLSobel5x5>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
         }
         case 7:
         {
-            auto k = arm_compute::cpp14::make_unique<CLSobel7x7>();
+            auto k = arm_compute::support::cpp14::make_unique<CLSobel7x7>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
diff --git a/src/runtime/CL/functions/CLLaplacianPyramid.cpp b/src/runtime/CL/functions/CLLaplacianPyramid.cpp
index d7ce20653d..3fc93426ee 100644
--- a/src/runtime/CL/functions/CLLaplacianPyramid.cpp
+++ b/src/runtime/CL/functions/CLLaplacianPyramid.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/CL/functions/CLLaplacianPyramid.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IPyramid.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
@@ -33,6 +32,7 @@
 #include "arm_compute/runtime/CL/functions/CLDepthConvert.h"
 #include "arm_compute/runtime/CL/functions/CLGaussian5x5.h"
 #include "arm_compute/runtime/CL/functions/CLGaussianPyramid.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -64,8 +64,8 @@ void CLLaplacianPyramid::configure(ICLTensor *input, CLPyramid *pyramid, ICLTens
     // Create Gaussian Pyramid function
     _gaussian_pyr_function.configure(input, &_gauss_pyr, border_mode, constant_border_value);
 
-    _convf = arm_compute::cpp14::make_unique<CLGaussian5x5[]>(_num_levels);
-    _subf  = arm_compute::cpp14::make_unique<CLArithmeticSubtraction[]>(_num_levels);
+    _convf = arm_compute::support::cpp14::make_unique<CLGaussian5x5[]>(_num_levels);
+    _subf  = arm_compute::support::cpp14::make_unique<CLArithmeticSubtraction[]>(_num_levels);
 
     for(unsigned int i = 0; i < _num_levels; ++i)
     {
diff --git a/src/runtime/CL/functions/CLLaplacianReconstruct.cpp b/src/runtime/CL/functions/CLLaplacianReconstruct.cpp
index 1dfab740d7..7fd2273230 100644
--- a/src/runtime/CL/functions/CLLaplacianReconstruct.cpp
+++ b/src/runtime/CL/functions/CLLaplacianReconstruct.cpp
@@ -24,11 +24,11 @@
 #include "arm_compute/runtime/CL/functions/CLLaplacianReconstruct.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IPyramid.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <cstddef>
 
@@ -60,8 +60,8 @@ void CLLaplacianReconstruct::configure(const CLPyramid *pyramid, const ICLTensor
     _tmp_pyr.init(pyramid_info);
 
     // Allocate add and scale functions. Level 0 does not need to be scaled.
-    _addf   = arm_compute::cpp14::make_unique<CLArithmeticAddition[]>(num_levels);
-    _scalef = arm_compute::cpp14::make_unique<CLScale[]>(num_levels - 1);
+    _addf   = arm_compute::support::cpp14::make_unique<CLArithmeticAddition[]>(num_levels);
+    _scalef = arm_compute::support::cpp14::make_unique<CLScale[]>(num_levels - 1);
 
     const size_t last_level = num_levels - 1;
 
diff --git a/src/runtime/CL/functions/CLMagnitude.cpp b/src/runtime/CL/functions/CLMagnitude.cpp
index 51088cb71f..68b8c3545a 100644
--- a/src/runtime/CL/functions/CLMagnitude.cpp
+++ b/src/runtime/CL/functions/CLMagnitude.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLMagnitude.h"
 
 #include "arm_compute/core/CL/kernels/CLMagnitudePhaseKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLMagnitude::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, MagnitudeType mag_type)
 {
-    auto k = arm_compute::cpp14::make_unique<CLMagnitudePhaseKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLMagnitudePhaseKernel>();
     k->configure(input1, input2, output, nullptr, mag_type);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLMedian3x3.cpp b/src/runtime/CL/functions/CLMedian3x3.cpp
index 0c10f9aa08..55f9eaa907 100644
--- a/src/runtime/CL/functions/CLMedian3x3.cpp
+++ b/src/runtime/CL/functions/CLMedian3x3.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLMedian3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLMedian3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLMedian3x3::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLMedian3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLMedian3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLNonLinearFilter.cpp b/src/runtime/CL/functions/CLNonLinearFilter.cpp
index b593a6cced..d37412f47f 100644
--- a/src/runtime/CL/functions/CLNonLinearFilter.cpp
+++ b/src/runtime/CL/functions/CLNonLinearFilter.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLNonLinearFilter.h"
 
 #include "arm_compute/core/CL/kernels/CLNonLinearFilterKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 void CLNonLinearFilter::configure(ICLTensor *input, ICLTensor *output, NonLinearFilterFunction function, unsigned int mask_size, MatrixPattern pattern, const uint8_t *mask,
                                   BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLNonLinearFilterKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLNonLinearFilterKernel>();
     k->configure(input, output, function, mask_size, pattern, mask, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLNonMaximaSuppression3x3.cpp b/src/runtime/CL/functions/CLNonMaximaSuppression3x3.cpp
index ca7d5aede7..c0a0cef885 100644
--- a/src/runtime/CL/functions/CLNonMaximaSuppression3x3.cpp
+++ b/src/runtime/CL/functions/CLNonMaximaSuppression3x3.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLNonMaximaSuppression3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLNonMaximaSuppression3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLNonMaximaSuppression3x3::configure(ICLTensor *input, ICLTensor *output, BorderMode border_mode)
 {
-    auto k = arm_compute::cpp14::make_unique<CLNonMaximaSuppression3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLNonMaximaSuppression3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
 
diff --git a/src/runtime/CL/functions/CLOpticalFlow.cpp b/src/runtime/CL/functions/CLOpticalFlow.cpp
index a6b0eb3bec..542f5961fa 100644
--- a/src/runtime/CL/functions/CLOpticalFlow.cpp
+++ b/src/runtime/CL/functions/CLOpticalFlow.cpp
@@ -26,7 +26,6 @@
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLLKTrackerKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Window.h"
 #include "arm_compute/runtime/CL/CLPyramid.h"
@@ -34,6 +33,7 @@
 #include "arm_compute/runtime/CL/CLTensor.h"
 #include "arm_compute/runtime/CL/CLTensorAllocator.h"
 #include "arm_compute/runtime/CL/functions/CLScharr3x3.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -70,21 +70,21 @@ void CLOpticalFlow::configure(const CLPyramid *old_pyramid, const CLPyramid *new
     const int   old_values_list_length = list_length * window_dimension * window_dimension;
 
     // Create kernels and tensors
-    _tracker_init_kernel   = arm_compute::cpp14::make_unique<CLLKTrackerInitKernel[]>(_num_levels);
-    _tracker_stage0_kernel = arm_compute::cpp14::make_unique<CLLKTrackerStage0Kernel[]>(_num_levels);
-    _tracker_stage1_kernel = arm_compute::cpp14::make_unique<CLLKTrackerStage1Kernel[]>(_num_levels);
-    _func_scharr           = arm_compute::cpp14::make_unique<CLScharr3x3[]>(_num_levels);
-    _scharr_gx             = arm_compute::cpp14::make_unique<CLTensor[]>(_num_levels);
-    _scharr_gy             = arm_compute::cpp14::make_unique<CLTensor[]>(_num_levels);
+    _tracker_init_kernel   = arm_compute::support::cpp14::make_unique<CLLKTrackerInitKernel[]>(_num_levels);
+    _tracker_stage0_kernel = arm_compute::support::cpp14::make_unique<CLLKTrackerStage0Kernel[]>(_num_levels);
+    _tracker_stage1_kernel = arm_compute::support::cpp14::make_unique<CLLKTrackerStage1Kernel[]>(_num_levels);
+    _func_scharr           = arm_compute::support::cpp14::make_unique<CLScharr3x3[]>(_num_levels);
+    _scharr_gx             = arm_compute::support::cpp14::make_unique<CLTensor[]>(_num_levels);
+    _scharr_gy             = arm_compute::support::cpp14::make_unique<CLTensor[]>(_num_levels);
 
     // Create internal keypoint arrays
-    _old_points_internal = arm_compute::cpp14::make_unique<CLLKInternalKeypointArray>(list_length);
+    _old_points_internal = arm_compute::support::cpp14::make_unique<CLLKInternalKeypointArray>(list_length);
     _old_points_internal->resize(list_length);
-    _new_points_internal = arm_compute::cpp14::make_unique<CLLKInternalKeypointArray>(list_length);
+    _new_points_internal = arm_compute::support::cpp14::make_unique<CLLKInternalKeypointArray>(list_length);
     _new_points_internal->resize(list_length);
-    _coefficient_table = arm_compute::cpp14::make_unique<CLCoefficientTableArray>(list_length);
+    _coefficient_table = arm_compute::support::cpp14::make_unique<CLCoefficientTableArray>(list_length);
     _coefficient_table->resize(list_length);
-    _old_values = arm_compute::cpp14::make_unique<CLOldValueArray>(old_values_list_length);
+    _old_values = arm_compute::support::cpp14::make_unique<CLOldValueArray>(old_values_list_length);
     _old_values->resize(old_values_list_length);
     _new_points->resize(list_length);
 
diff --git a/src/runtime/CL/functions/CLPhase.cpp b/src/runtime/CL/functions/CLPhase.cpp
index a8cb22b06e..cf3fa7e7fb 100644
--- a/src/runtime/CL/functions/CLPhase.cpp
+++ b/src/runtime/CL/functions/CLPhase.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLPhase.h"
 
 #include "arm_compute/core/CL/kernels/CLMagnitudePhaseKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLPhase::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, PhaseType phase_type)
 {
-    auto k = arm_compute::cpp14::make_unique<CLMagnitudePhaseKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLMagnitudePhaseKernel>();
     k->configure(input1, input2, nullptr, output, MagnitudeType::L1NORM, phase_type);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLPixelWiseMultiplication.cpp b/src/runtime/CL/functions/CLPixelWiseMultiplication.cpp
index 8a86c2e203..139d466b39 100644
--- a/src/runtime/CL/functions/CLPixelWiseMultiplication.cpp
+++ b/src/runtime/CL/functions/CLPixelWiseMultiplication.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h"
 
 #include "arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 void CLPixelWiseMultiplication::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, float scale,
                                           ConvertPolicy overflow_policy, RoundingPolicy rounding_policy)
 {
-    auto k = arm_compute::cpp14::make_unique<CLPixelWiseMultiplicationKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLPixelWiseMultiplicationKernel>();
     k->configure(input1, input2, output, scale, overflow_policy, rounding_policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLPoolingLayer.cpp b/src/runtime/CL/functions/CLPoolingLayer.cpp
index 1ef70f4a2b..2cb7d63aa8 100644
--- a/src/runtime/CL/functions/CLPoolingLayer.cpp
+++ b/src/runtime/CL/functions/CLPoolingLayer.cpp
@@ -24,14 +24,14 @@
 #include "arm_compute/runtime/CL/functions/CLPoolingLayer.h"
 
 #include "arm_compute/core/CL/kernels/CLPoolingLayerKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void CLPoolingLayer::configure(ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info)
 {
     // Configure pooling kernel
-    auto k = arm_compute::cpp14::make_unique<CLPoolingLayerKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLPoolingLayerKernel>();
     k->configure(input, output, pool_info);
     _kernel = std::move(k);
 
diff --git a/src/runtime/CL/functions/CLRemap.cpp b/src/runtime/CL/functions/CLRemap.cpp
index f6b1713c58..bc3fd4ed53 100644
--- a/src/runtime/CL/functions/CLRemap.cpp
+++ b/src/runtime/CL/functions/CLRemap.cpp
@@ -26,10 +26,10 @@
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLRemapKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -43,7 +43,7 @@ void CLRemap::configure(ICLTensor *input, const ICLTensor *map_x, const ICLTenso
     ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(map_y, 1, DataType::F32);
     ARM_COMPUTE_ERROR_ON_MSG(policy == InterpolationPolicy::AREA, "Area interpolation is not supported");
 
-    auto k = arm_compute::cpp14::make_unique<CLRemapKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLRemapKernel>();
     k->configure(input, map_x, map_y, output, policy, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLScale.cpp b/src/runtime/CL/functions/CLScale.cpp
index 043f873028..05522b4af5 100644
--- a/src/runtime/CL/functions/CLScale.cpp
+++ b/src/runtime/CL/functions/CLScale.cpp
@@ -26,8 +26,8 @@
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLScaleKernel.h"
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -38,7 +38,7 @@ void CLScale::configure(ICLTensor *input, ICLTensor *output, InterpolationPolicy
     ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::U8, DataType::S16);
     ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
 
-    auto k = arm_compute::cpp14::make_unique<CLScaleKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLScaleKernel>();
     k->configure(input, output, policy, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, constant_border_value);
diff --git a/src/runtime/CL/functions/CLScharr3x3.cpp b/src/runtime/CL/functions/CLScharr3x3.cpp
index c8bc465be6..73f8673e78 100644
--- a/src/runtime/CL/functions/CLScharr3x3.cpp
+++ b/src/runtime/CL/functions/CLScharr3x3.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLScharr3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLScharr3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLScharr3x3::configure(ICLTensor *input, ICLTensor *output_x, ICLTensor *output_y, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLScharr3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLScharr3x3Kernel>();
     k->configure(input, output_x, output_y, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLSobel3x3.cpp b/src/runtime/CL/functions/CLSobel3x3.cpp
index 6b74ebaedb..e227e58bb2 100644
--- a/src/runtime/CL/functions/CLSobel3x3.cpp
+++ b/src/runtime/CL/functions/CLSobel3x3.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLSobel3x3.h"
 
 #include "arm_compute/core/CL/kernels/CLSobel3x3Kernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLSobel3x3::configure(ICLTensor *input, ICLTensor *output_x, ICLTensor *output_y, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLSobel3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLSobel3x3Kernel>();
     k->configure(input, output_x, output_y, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLTableLookup.cpp b/src/runtime/CL/functions/CLTableLookup.cpp
index 743ed5e73e..d18765093f 100644
--- a/src/runtime/CL/functions/CLTableLookup.cpp
+++ b/src/runtime/CL/functions/CLTableLookup.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLTableLookup.h"
 
 #include "arm_compute/core/CL/kernels/CLTableLookupKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLTableLookup::configure(const ICLTensor *input, const ICLLut *lut, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLTableLookupKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLTableLookupKernel>();
     k->configure(input, lut, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLThreshold.cpp b/src/runtime/CL/functions/CLThreshold.cpp
index e70f932d66..1b30b77f1d 100644
--- a/src/runtime/CL/functions/CLThreshold.cpp
+++ b/src/runtime/CL/functions/CLThreshold.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLThreshold.h"
 
 #include "arm_compute/core/CL/kernels/CLThresholdKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLThreshold::configure(const ICLTensor *input, ICLTensor *output, uint8_t threshold, uint8_t false_value, uint8_t true_value, ThresholdType type, uint8_t upper)
 {
-    auto k = arm_compute::cpp14::make_unique<CLThresholdKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLThresholdKernel>();
     k->configure(input, output, threshold, false_value, true_value, type, upper);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/CL/functions/CLTranspose.cpp b/src/runtime/CL/functions/CLTranspose.cpp
index d802b4fe77..cd19e255bd 100644
--- a/src/runtime/CL/functions/CLTranspose.cpp
+++ b/src/runtime/CL/functions/CLTranspose.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/CL/functions/CLTranspose.h"
 
 #include "arm_compute/core/CL/kernels/CLTransposeKernel.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void CLTranspose::configure(const ICLTensor *input, ICLTensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<CLTransposeKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLTransposeKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
\ No newline at end of file
diff --git a/src/runtime/CL/functions/CLWarpAffine.cpp b/src/runtime/CL/functions/CLWarpAffine.cpp
index 537e0d9397..f785c75a72 100644
--- a/src/runtime/CL/functions/CLWarpAffine.cpp
+++ b/src/runtime/CL/functions/CLWarpAffine.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLWarpAffine.h"
 
 #include "arm_compute/core/CL/kernels/CLWarpAffineKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLWarpAffine::configure(ICLTensor *input, ICLTensor *output, const float *matrix, InterpolationPolicy policy, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLWarpAffineKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLWarpAffineKernel>();
     k->configure(input, output, matrix, policy);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/CL/functions/CLWarpPerspective.cpp b/src/runtime/CL/functions/CLWarpPerspective.cpp
index a552ab480d..b445b3b9c5 100644
--- a/src/runtime/CL/functions/CLWarpPerspective.cpp
+++ b/src/runtime/CL/functions/CLWarpPerspective.cpp
@@ -24,8 +24,8 @@
 #include "arm_compute/runtime/CL/functions/CLWarpPerspective.h"
 
 #include "arm_compute/core/CL/kernels/CLWarpPerspectiveKernel.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void CLWarpPerspective::configure(ICLTensor *input, ICLTensor *output, const float *matrix, InterpolationPolicy policy, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<CLWarpPerspectiveKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<CLWarpPerspectiveKernel>();
     k->configure(input, output, matrix, policy);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/Distribution1D.cpp b/src/runtime/Distribution1D.cpp
index b06767499b..3431834c48 100644
--- a/src/runtime/Distribution1D.cpp
+++ b/src/runtime/Distribution1D.cpp
@@ -24,14 +24,14 @@
 #include "arm_compute/runtime/Distribution1D.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 #include <cstdint>
 
 using namespace arm_compute;
 
 Distribution1D::Distribution1D(size_t num_bins, int32_t offset, uint32_t range)
-    : IDistribution1D(num_bins, offset, range), _data(arm_compute::cpp14::make_unique<uint32_t[]>(num_bins))
+    : IDistribution1D(num_bins, offset, range), _data(arm_compute::support::cpp14::make_unique<uint32_t[]>(num_bins))
 {
 }
 
diff --git a/src/runtime/HOG.cpp b/src/runtime/HOG.cpp
index 5d533dded4..01640bb0ac 100644
--- a/src/runtime/HOG.cpp
+++ b/src/runtime/HOG.cpp
@@ -24,7 +24,7 @@
 #include "arm_compute/runtime/HOG.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -37,7 +37,7 @@ void HOG::init(const HOGInfo &input)
 {
     ARM_COMPUTE_ERROR_ON(nullptr != _descriptor);
     _info       = input;
-    _descriptor = arm_compute::cpp14::make_unique<float[]>(_info.descriptor_size());
+    _descriptor = arm_compute::support::cpp14::make_unique<float[]>(_info.descriptor_size());
 }
 
 float *HOG::descriptor() const
diff --git a/src/runtime/LutAllocator.cpp b/src/runtime/LutAllocator.cpp
index 17baf21f45..eb9051cc0a 100644
--- a/src/runtime/LutAllocator.cpp
+++ b/src/runtime/LutAllocator.cpp
@@ -23,7 +23,7 @@
  */
 #include "arm_compute/runtime/LutAllocator.h"
 
-#include "arm_compute/core/Helpers.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -39,7 +39,7 @@ uint8_t *LutAllocator::data() const
 
 void LutAllocator::allocate()
 {
-    _buffer = arm_compute::cpp14::make_unique<uint8_t[]>(size());
+    _buffer = arm_compute::support::cpp14::make_unique<uint8_t[]>(size());
 }
 
 uint8_t *LutAllocator::lock()
diff --git a/src/runtime/MultiHOG.cpp b/src/runtime/MultiHOG.cpp
index 003dc93895..e0b60b1137 100644
--- a/src/runtime/MultiHOG.cpp
+++ b/src/runtime/MultiHOG.cpp
@@ -24,13 +24,13 @@
 #include "arm_compute/runtime/MultiHOG.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IMultiHOG.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 MultiHOG::MultiHOG(size_t num_models)
-    : _num_models(num_models), _model(arm_compute::cpp14::make_unique<HOG[]>(_num_models))
+    : _num_models(num_models), _model(arm_compute::support::cpp14::make_unique<HOG[]>(_num_models))
 {
 }
 
diff --git a/src/runtime/NEON/functions/NEAbsoluteDifference.cpp b/src/runtime/NEON/functions/NEAbsoluteDifference.cpp
index b39feb3a2b..b4620f16dd 100644
--- a/src/runtime/NEON/functions/NEAbsoluteDifference.cpp
+++ b/src/runtime/NEON/functions/NEAbsoluteDifference.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEAbsoluteDifference.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEAbsoluteDifferenceKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEAbsoluteDifference::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEAbsoluteDifferenceKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEAbsoluteDifferenceKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEAccumulate.cpp b/src/runtime/NEON/functions/NEAccumulate.cpp
index c39abfc540..49524d2202 100644
--- a/src/runtime/NEON/functions/NEAccumulate.cpp
+++ b/src/runtime/NEON/functions/NEAccumulate.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEAccumulate.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEAccumulateKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEAccumulate::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEAccumulateKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEAccumulateKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
@@ -41,13 +41,13 @@ void NEAccumulateWeighted::configure(const ITensor *input, float alpha, ITensor
 {
     if(use_fp16)
     {
-        auto k = arm_compute::cpp14::make_unique<NEAccumulateWeightedFP16Kernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEAccumulateWeightedFP16Kernel>();
         k->configure(input, alpha, output);
         _kernel = std::move(k);
     }
     else
     {
-        auto k = arm_compute::cpp14::make_unique<NEAccumulateWeightedKernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEAccumulateWeightedKernel>();
         k->configure(input, alpha, output);
         _kernel = std::move(k);
     }
@@ -55,7 +55,7 @@ void NEAccumulateWeighted::configure(const ITensor *input, float alpha, ITensor
 
 void NEAccumulateSquared::configure(const ITensor *input, uint32_t shift, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEAccumulateSquaredKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEAccumulateSquaredKernel>();
     k->configure(input, shift, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEActivationLayer.cpp b/src/runtime/NEON/functions/NEActivationLayer.cpp
index 447ae6411b..57a1738f85 100644
--- a/src/runtime/NEON/functions/NEActivationLayer.cpp
+++ b/src/runtime/NEON/functions/NEActivationLayer.cpp
@@ -23,14 +23,14 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEActivationLayerKernel.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void NEActivationLayer::configure(ITensor *input, ITensor *output, ActivationLayerInfo activation_info)
 {
-    auto k = arm_compute::cpp14::make_unique<NEActivationLayerKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEActivationLayerKernel>();
     k->configure(input, output, activation_info);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEArithmeticAddition.cpp b/src/runtime/NEON/functions/NEArithmeticAddition.cpp
index 50cc38b489..11f5aa74e4 100644
--- a/src/runtime/NEON/functions/NEArithmeticAddition.cpp
+++ b/src/runtime/NEON/functions/NEArithmeticAddition.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEArithmeticAdditionKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEArithmeticAddition::configure(const ITensor *input1, const ITensor *input2, ITensor *output, ConvertPolicy policy)
 {
-    auto k = arm_compute::cpp14::make_unique<NEArithmeticAdditionKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEArithmeticAdditionKernel>();
     k->configure(input1, input2, output, policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEArithmeticSubtraction.cpp b/src/runtime/NEON/functions/NEArithmeticSubtraction.cpp
index a3d27c0ed6..37586af751 100644
--- a/src/runtime/NEON/functions/NEArithmeticSubtraction.cpp
+++ b/src/runtime/NEON/functions/NEArithmeticSubtraction.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEArithmeticSubtraction::configure(const ITensor *input1, const ITensor *input2, ITensor *output, ConvertPolicy policy)
 {
-    auto k = arm_compute::cpp14::make_unique<NEArithmeticSubtractionKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEArithmeticSubtractionKernel>();
     k->configure(input1, input2, output, policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEBitwiseAnd.cpp b/src/runtime/NEON/functions/NEBitwiseAnd.cpp
index 5aafc51dc0..79820952bb 100644
--- a/src/runtime/NEON/functions/NEBitwiseAnd.cpp
+++ b/src/runtime/NEON/functions/NEBitwiseAnd.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEBitwiseAnd.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseAndKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEBitwiseAnd::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEBitwiseAndKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEBitwiseAndKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEBitwiseNot.cpp b/src/runtime/NEON/functions/NEBitwiseNot.cpp
index af3df6e46a..c55957e2b4 100644
--- a/src/runtime/NEON/functions/NEBitwiseNot.cpp
+++ b/src/runtime/NEON/functions/NEBitwiseNot.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEBitwiseNot.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseNotKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEBitwiseNot::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEBitwiseNotKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEBitwiseNotKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEBitwiseOr.cpp b/src/runtime/NEON/functions/NEBitwiseOr.cpp
index d12c5e5f6f..01036daaef 100644
--- a/src/runtime/NEON/functions/NEBitwiseOr.cpp
+++ b/src/runtime/NEON/functions/NEBitwiseOr.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEBitwiseOr.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseOrKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEBitwiseOr::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEBitwiseOrKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEBitwiseOrKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEBitwiseXor.cpp b/src/runtime/NEON/functions/NEBitwiseXor.cpp
index 65c943e64c..45916984b0 100644
--- a/src/runtime/NEON/functions/NEBitwiseXor.cpp
+++ b/src/runtime/NEON/functions/NEBitwiseXor.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEBitwiseXor.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseXorKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEBitwiseXor::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEBitwiseXorKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEBitwiseXorKernel>();
     k->configure(input1, input2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEBox3x3.cpp b/src/runtime/NEON/functions/NEBox3x3.cpp
index 7f0b45d34c..46cf259073 100644
--- a/src/runtime/NEON/functions/NEBox3x3.cpp
+++ b/src/runtime/NEON/functions/NEBox3x3.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEBox3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEBox3x3Kernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -35,13 +35,13 @@ void NEBox3x3::configure(ITensor *input, ITensor *output, BorderMode border_mode
 {
     if(use_fp16)
     {
-        auto k = arm_compute::cpp14::make_unique<NEBox3x3FP16Kernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEBox3x3FP16Kernel>();
         k->configure(input, output, border_mode == BorderMode::UNDEFINED);
         _kernel = std::move(k);
     }
     else
     {
-        auto k = arm_compute::cpp14::make_unique<NEBox3x3Kernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEBox3x3Kernel>();
         k->configure(input, output, border_mode == BorderMode::UNDEFINED);
         _kernel = std::move(k);
     }
diff --git a/src/runtime/NEON/functions/NECannyEdge.cpp b/src/runtime/NEON/functions/NECannyEdge.cpp
index 26f31f557b..029d30c5d6 100644
--- a/src/runtime/NEON/functions/NECannyEdge.cpp
+++ b/src/runtime/NEON/functions/NECannyEdge.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NECannyEdge.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NECannyEdgeKernel.h"
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
@@ -35,6 +34,7 @@
 #include "arm_compute/runtime/NEON/functions/NESobel5x5.h"
 #include "arm_compute/runtime/NEON/functions/NESobel7x7.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <cstring>
 #include <utility>
@@ -85,19 +85,19 @@ void NECannyEdge::configure(ITensor *input, ITensor *output, int32_t upper_thr,
     // Configure/Init sobelNxN
     if(gradient_size == 3)
     {
-        auto k = arm_compute::cpp14::make_unique<NESobel3x3>();
+        auto k = arm_compute::support::cpp14::make_unique<NESobel3x3>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
     else if(gradient_size == 5)
     {
-        auto k = arm_compute::cpp14::make_unique<NESobel5x5>();
+        auto k = arm_compute::support::cpp14::make_unique<NESobel5x5>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
     else if(gradient_size == 7)
     {
-        auto k = arm_compute::cpp14::make_unique<NESobel7x7>();
+        auto k = arm_compute::support::cpp14::make_unique<NESobel7x7>();
         k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
         _sobel = std::move(k);
     }
@@ -109,13 +109,13 @@ void NECannyEdge::configure(ITensor *input, ITensor *output, int32_t upper_thr,
     // Configure gradient
     if(use_fp16)
     {
-        auto k = arm_compute::cpp14::make_unique<NEGradientFP16Kernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEGradientFP16Kernel>();
         k->configure(&_gx, &_gy, &_magnitude, &_phase, norm_type);
         _gradient = std::move(k);
     }
     else
     {
-        auto k = arm_compute::cpp14::make_unique<NEGradientKernel>();
+        auto k = arm_compute::support::cpp14::make_unique<NEGradientKernel>();
         k->configure(&_gx, &_gy, &_magnitude, &_phase, norm_type);
         _gradient = std::move(k);
     }
diff --git a/src/runtime/NEON/functions/NEChannelCombine.cpp b/src/runtime/NEON/functions/NEChannelCombine.cpp
index 84d4fff4ff..9166aa97db 100644
--- a/src/runtime/NEON/functions/NEChannelCombine.cpp
+++ b/src/runtime/NEON/functions/NEChannelCombine.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEChannelCombine.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEChannelCombineKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,14 +32,14 @@ using namespace arm_compute;
 
 void NEChannelCombine::configure(const ITensor *plane0, const ITensor *plane1, const ITensor *plane2, const ITensor *plane3, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEChannelCombineKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEChannelCombineKernel>();
     k->configure(plane0, plane1, plane2, plane3, output);
     _kernel = std::move(k);
 }
 
 void NEChannelCombine::configure(const IImage *plane0, const IImage *plane1, const IImage *plane2, IMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEChannelCombineKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEChannelCombineKernel>();
     k->configure(plane0, plane1, plane2, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEChannelExtract.cpp b/src/runtime/NEON/functions/NEChannelExtract.cpp
index 634e918eac..7b8a99335f 100644
--- a/src/runtime/NEON/functions/NEChannelExtract.cpp
+++ b/src/runtime/NEON/functions/NEChannelExtract.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEChannelExtract.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEChannelExtractKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,14 +32,14 @@ using namespace arm_compute;
 
 void NEChannelExtract::configure(const ITensor *input, Channel channel, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEChannelExtractKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEChannelExtractKernel>();
     k->configure(input, channel, output);
     _kernel = std::move(k);
 }
 
 void NEChannelExtract::configure(const IMultiImage *input, Channel channel, IImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEChannelExtractKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEChannelExtractKernel>();
     k->configure(input, channel, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEColorConvert.cpp b/src/runtime/NEON/functions/NEColorConvert.cpp
index bbaa832284..b9fe1ff11d 100644
--- a/src/runtime/NEON/functions/NEColorConvert.cpp
+++ b/src/runtime/NEON/functions/NEColorConvert.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEColorConvert.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEColorConvertKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,28 +32,28 @@ using namespace arm_compute;
 
 void NEColorConvert::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void NEColorConvert::configure(const IMultiImage *input, IImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void NEColorConvert::configure(const IImage *input, IMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
 
 void NEColorConvert::configure(const IMultiImage *input, IMultiImage *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEColorConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEColorConvertKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEConvolution.cpp b/src/runtime/NEON/functions/NEConvolution.cpp
index 3f39ae2cbd..4ad6450c67 100644
--- a/src/runtime/NEON/functions/NEConvolution.cpp
+++ b/src/runtime/NEON/functions/NEConvolution.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NEConvolution.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NEConvolutionKernel.h"
 #include "arm_compute/core/PixelValue.h"
@@ -33,6 +32,7 @@
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/NEON/NEScheduler.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <array>
 #include <utility>
@@ -41,7 +41,7 @@ using namespace arm_compute;
 
 void NEConvolution3x3::configure(ITensor *input, ITensor *output, const int16_t *conv, uint32_t scale, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEConvolution3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEConvolution3x3Kernel>();
     k->configure(input, output, conv, scale, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
@@ -113,7 +113,7 @@ template class arm_compute::NEConvolutionSquare<9>;
 
 void NEConvolutionRectangle::configure(ITensor *input, ITensor *output, const int16_t *conv, uint32_t rows, uint32_t cols, uint32_t scale, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEConvolutionRectangleKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEConvolutionRectangleKernel>();
     k->configure(input, output, conv, rows, cols, scale, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NEDepthConcatenate.cpp b/src/runtime/NEON/functions/NEDepthConcatenate.cpp
index 7d2c5494a9..8661fe5fda 100644
--- a/src/runtime/NEON/functions/NEDepthConcatenate.cpp
+++ b/src/runtime/NEON/functions/NEDepthConcatenate.cpp
@@ -24,13 +24,13 @@
 #include "arm_compute/runtime/NEON/functions/NEDepthConcatenate.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NEDepthConcatenateKernel.h"
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/NEON/NEScheduler.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -44,8 +44,8 @@ void NEDepthConcatenate::configure(std::vector<ITensor *> inputs_vector, ITensor
     ARM_COMPUTE_ERROR_ON(inputs_vector.size() < 2);
 
     _num_inputs             = inputs_vector.size();
-    _concat_kernels_vector  = arm_compute::cpp14::make_unique<NEDepthConcatenateKernel[]>(_num_inputs);
-    _border_handlers_vector = arm_compute::cpp14::make_unique<NEFillBorderKernel[]>(_num_inputs);
+    _concat_kernels_vector  = arm_compute::support::cpp14::make_unique<NEDepthConcatenateKernel[]>(_num_inputs);
+    _border_handlers_vector = arm_compute::support::cpp14::make_unique<NEFillBorderKernel[]>(_num_inputs);
 
     unsigned int depth_offset = 0;
     for(unsigned int i = 0; i < _num_inputs; ++i)
diff --git a/src/runtime/NEON/functions/NEDepthConvert.cpp b/src/runtime/NEON/functions/NEDepthConvert.cpp
index a339cae316..011e366bf3 100644
--- a/src/runtime/NEON/functions/NEDepthConvert.cpp
+++ b/src/runtime/NEON/functions/NEDepthConvert.cpp
@@ -24,8 +24,9 @@
 #include "arm_compute/runtime/NEON/functions/NEDepthConvert.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEDepthConvertKernel.h"
+#include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -38,7 +39,7 @@ void NEDepthConvert::configure(const ITensor *input, ITensor *output, ConvertPol
     ARM_COMPUTE_ERROR_ON(input == output);
     ARM_COMPUTE_ERROR_ON(input->info()->data_type() == output->info()->data_type());
 
-    auto k = arm_compute::cpp14::make_unique<NEDepthConvertKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEDepthConvertKernel>();
     k->configure(input, output, policy, shift);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEDilate.cpp b/src/runtime/NEON/functions/NEDilate.cpp
index 0c016f14f9..5c733a8278 100644
--- a/src/runtime/NEON/functions/NEDilate.cpp
+++ b/src/runtime/NEON/functions/NEDilate.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEDilate.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEDilateKernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEDilate::configure(ITensor *input, ITensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEDilateKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEDilateKernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NEErode.cpp b/src/runtime/NEON/functions/NEErode.cpp
index 9b011db845..3609572de6 100644
--- a/src/runtime/NEON/functions/NEErode.cpp
+++ b/src/runtime/NEON/functions/NEErode.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEErode.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEErodeKernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEErode::configure(ITensor *input, ITensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEErodeKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEErodeKernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NEGEMMInterleave4x4.cpp b/src/runtime/NEON/functions/NEGEMMInterleave4x4.cpp
index 4c77c88656..63f330be6c 100644
--- a/src/runtime/NEON/functions/NEGEMMInterleave4x4.cpp
+++ b/src/runtime/NEON/functions/NEGEMMInterleave4x4.cpp
@@ -23,14 +23,14 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEGEMMInterleave4x4.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMInterleave4x4Kernel.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void NEGEMMInterleave4x4::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEGEMMInterleave4x4Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEGEMMInterleave4x4Kernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEGEMMTranspose1xW.cpp b/src/runtime/NEON/functions/NEGEMMTranspose1xW.cpp
index dc40ecec14..571bf2bc74 100644
--- a/src/runtime/NEON/functions/NEGEMMTranspose1xW.cpp
+++ b/src/runtime/NEON/functions/NEGEMMTranspose1xW.cpp
@@ -24,17 +24,17 @@
 #include "arm_compute/runtime/NEON/functions/NEGEMMTranspose1xW.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMTranspose1xWKernel.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void NEGEMMTranspose1xW::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEGEMMTranspose1xWKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEGEMMTranspose1xWKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEGaussian3x3.cpp b/src/runtime/NEON/functions/NEGaussian3x3.cpp
index 95ba5cbdf9..db8eb63c90 100644
--- a/src/runtime/NEON/functions/NEGaussian3x3.cpp
+++ b/src/runtime/NEON/functions/NEGaussian3x3.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEGaussian3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEGaussian3x3Kernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEGaussian3x3::configure(ITensor *input, ITensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEGaussian3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEGaussian3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NEGaussianPyramid.cpp b/src/runtime/NEON/functions/NEGaussianPyramid.cpp
index e1d64f11f6..75c439511e 100644
--- a/src/runtime/NEON/functions/NEGaussianPyramid.cpp
+++ b/src/runtime/NEON/functions/NEGaussianPyramid.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NEGaussianPyramid.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NEGaussianPyramidKernel.h"
 #include "arm_compute/core/NEON/kernels/NEScaleKernel.h"
@@ -36,6 +35,7 @@
 #include "arm_compute/runtime/Pyramid.h"
 #include "arm_compute/runtime/Tensor.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <cstddef>
 
@@ -68,9 +68,9 @@ void NEGaussianPyramidHalf::configure(const ITensor *input, IPyramid *pyramid, B
 
     if(num_levels > 1)
     {
-        _border_handler       = arm_compute::cpp14::make_unique<NEFillBorderKernel[]>(num_levels - 1);
-        _horizontal_reduction = arm_compute::cpp14::make_unique<NEGaussianPyramidHorKernel[]>(num_levels - 1);
-        _vertical_reduction   = arm_compute::cpp14::make_unique<NEGaussianPyramidVertKernel[]>(num_levels - 1);
+        _border_handler       = arm_compute::support::cpp14::make_unique<NEFillBorderKernel[]>(num_levels - 1);
+        _horizontal_reduction = arm_compute::support::cpp14::make_unique<NEGaussianPyramidHorKernel[]>(num_levels - 1);
+        _vertical_reduction   = arm_compute::support::cpp14::make_unique<NEGaussianPyramidVertKernel[]>(num_levels - 1);
 
         // Apply half scale to the X dimension of the tensor shape
         TensorShape tensor_shape = pyramid->info()->tensor_shape();
@@ -135,9 +135,9 @@ void NEGaussianPyramidOrb::configure(const ITensor *input, IPyramid *pyramid, Bo
 
     if(num_levels > 1)
     {
-        _gaus5x5       = arm_compute::cpp14::make_unique<NEGaussian5x5[]>(num_levels - 1);
-        _scale_nearest = arm_compute::cpp14::make_unique<NEScaleKernel[]>(num_levels - 1);
-        _offsets       = arm_compute::cpp14::make_unique<Image[]>(num_levels - 1);
+        _gaus5x5       = arm_compute::support::cpp14::make_unique<NEGaussian5x5[]>(num_levels - 1);
+        _scale_nearest = arm_compute::support::cpp14::make_unique<NEScaleKernel[]>(num_levels - 1);
+        _offsets       = arm_compute::support::cpp14::make_unique<Image[]>(num_levels - 1);
 
         PyramidInfo pyramid_info(num_levels - 1, SCALE_PYRAMID_ORB, pyramid->info()->tensor_shape(), Format::U8);
         _tmp.init(pyramid_info);
diff --git a/src/runtime/NEON/functions/NEHOGDetector.cpp b/src/runtime/NEON/functions/NEHOGDetector.cpp
index e8ed29d0b9..49d0778eaa 100644
--- a/src/runtime/NEON/functions/NEHOGDetector.cpp
+++ b/src/runtime/NEON/functions/NEHOGDetector.cpp
@@ -23,14 +23,14 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEHOGDetector.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEHOGDetectorKernel.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void NEHOGDetector::configure(const ITensor *input, const IHOG *hog, IDetectionWindowArray *detection_windows, const Size2D &detection_window_stride, float threshold, size_t idx_class)
 {
-    auto k = arm_compute::cpp14::make_unique<NEHOGDetectorKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEHOGDetectorKernel>();
     k->configure(input, hog, detection_windows, detection_window_stride, threshold, idx_class);
     _kernel = std::move(k);
 }
\ No newline at end of file
diff --git a/src/runtime/NEON/functions/NEHOGGradient.cpp b/src/runtime/NEON/functions/NEHOGGradient.cpp
index 2f4b8802e3..563d4c2640 100644
--- a/src/runtime/NEON/functions/NEHOGGradient.cpp
+++ b/src/runtime/NEON/functions/NEHOGGradient.cpp
@@ -23,10 +23,10 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEHOGGradient.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEMagnitudePhaseKernel.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/NEON/NEScheduler.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -54,13 +54,13 @@ void NEHOGGradient::configure(ITensor *input, ITensor *output_magnitude, ITensor
     // Initialise magnitude/phase kernel
     if(PhaseType::UNSIGNED == phase_type)
     {
-        auto k = arm_compute::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::UNSIGNED>>();
+        auto k = arm_compute::support::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::UNSIGNED>>();
         k->configure(&_gx, &_gy, output_magnitude, output_phase);
         _mag_phase = std::move(k);
     }
     else
     {
-        auto k = arm_compute::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
+        auto k = arm_compute::support::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
         k->configure(&_gx, &_gy, output_magnitude, output_phase);
         _mag_phase = std::move(k);
     }
diff --git a/src/runtime/NEON/functions/NEHOGMultiDetection.cpp b/src/runtime/NEON/functions/NEHOGMultiDetection.cpp
index 173b8f4c42..635b793d0c 100644
--- a/src/runtime/NEON/functions/NEHOGMultiDetection.cpp
+++ b/src/runtime/NEON/functions/NEHOGMultiDetection.cpp
@@ -24,10 +24,11 @@
 #include "arm_compute/runtime/NEON/functions/NEHOGMultiDetection.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/TensorInfo.h"
+#include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/NEON/NEScheduler.h"
 #include "arm_compute/runtime/Tensor.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -112,12 +113,12 @@ void NEHOGMultiDetection::configure(ITensor *input, const IMultiHOG *multi_hog,
     _num_block_norm_kernel  = input_block_norm.size(); // Number of NEHOGBlockNormalizationKernel kernels to compute
     _num_hog_detect_kernel  = input_hog_detect.size(); // Number of NEHOGDetector functions to compute
 
-    _orient_bin_kernel = arm_compute::cpp14::make_unique<NEHOGOrientationBinningKernel[]>(_num_orient_bin_kernel);
-    _block_norm_kernel = arm_compute::cpp14::make_unique<NEHOGBlockNormalizationKernel[]>(_num_block_norm_kernel);
-    _hog_detect_kernel = arm_compute::cpp14::make_unique<NEHOGDetector[]>(_num_hog_detect_kernel);
-    _non_maxima_kernel = arm_compute::cpp14::make_unique<CPPDetectionWindowNonMaximaSuppressionKernel>();
-    _hog_space         = arm_compute::cpp14::make_unique<Tensor[]>(_num_orient_bin_kernel);
-    _hog_norm_space    = arm_compute::cpp14::make_unique<Tensor[]>(_num_block_norm_kernel);
+    _orient_bin_kernel = arm_compute::support::cpp14::make_unique<NEHOGOrientationBinningKernel[]>(_num_orient_bin_kernel);
+    _block_norm_kernel = arm_compute::support::cpp14::make_unique<NEHOGBlockNormalizationKernel[]>(_num_block_norm_kernel);
+    _hog_detect_kernel = arm_compute::support::cpp14::make_unique<NEHOGDetector[]>(_num_hog_detect_kernel);
+    _non_maxima_kernel = arm_compute::support::cpp14::make_unique<CPPDetectionWindowNonMaximaSuppressionKernel>();
+    _hog_space         = arm_compute::support::cpp14::make_unique<Tensor[]>(_num_orient_bin_kernel);
+    _hog_norm_space    = arm_compute::support::cpp14::make_unique<Tensor[]>(_num_block_norm_kernel);
 
     // Allocate tensors for magnitude and phase
     TensorInfo info_mag(shape_img, Format::S16);
diff --git a/src/runtime/NEON/functions/NEHarrisCorners.cpp b/src/runtime/NEON/functions/NEHarrisCorners.cpp
index b54fb67ab7..53b1968db2 100644
--- a/src/runtime/NEON/functions/NEHarrisCorners.cpp
+++ b/src/runtime/NEON/functions/NEHarrisCorners.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NEHarrisCorners.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
 #include "arm_compute/core/NEON/kernels/NEHarrisCornersKernel.h"
 #include "arm_compute/core/TensorInfo.h"
@@ -35,6 +34,7 @@
 #include "arm_compute/runtime/NEON/functions/NESobel5x5.h"
 #include "arm_compute/runtime/NEON/functions/NESobel7x7.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <cmath>
 #include <utility>
@@ -73,28 +73,28 @@ void NEHarrisCorners::configure(IImage *input, float threshold, float min_dist,
     _score.allocator()->init(tensor_info_score);
     _nonmax.allocator()->init(tensor_info_score);
 
-    _corners_list = arm_compute::cpp14::make_unique<InternalKeypoint[]>(shape.x() * shape.y());
+    _corners_list = arm_compute::support::cpp14::make_unique<InternalKeypoint[]>(shape.x() * shape.y());
 
     // Set/init Sobel kernel accordingly with gradient_size
     switch(gradient_size)
     {
         case 3:
         {
-            auto k = arm_compute::cpp14::make_unique<NESobel3x3>();
+            auto k = arm_compute::support::cpp14::make_unique<NESobel3x3>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
         }
         case 5:
         {
-            auto k = arm_compute::cpp14::make_unique<NESobel5x5>();
+            auto k = arm_compute::support::cpp14::make_unique<NESobel5x5>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
         }
         case 7:
         {
-            auto k = arm_compute::cpp14::make_unique<NESobel7x7>();
+            auto k = arm_compute::support::cpp14::make_unique<NESobel7x7>();
             k->configure(input, &_gx, &_gy, border_mode, constant_border_value);
             _sobel = std::move(k);
             break;
@@ -112,21 +112,21 @@ void NEHarrisCorners::configure(IImage *input, float threshold, float min_dist,
         {
             case 3:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreFP16Kernel<3>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreFP16Kernel<3>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
             break;
             case 5:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreFP16Kernel<5>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreFP16Kernel<5>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
             break;
             case 7:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreFP16Kernel<7>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreFP16Kernel<7>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
@@ -141,21 +141,21 @@ void NEHarrisCorners::configure(IImage *input, float threshold, float min_dist,
         {
             case 3:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreKernel<3>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreKernel<3>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
             break;
             case 5:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreKernel<5>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreKernel<5>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
             break;
             case 7:
             {
-                auto k = arm_compute::cpp14::make_unique<NEHarrisScoreKernel<7>>();
+                auto k = arm_compute::support::cpp14::make_unique<NEHarrisScoreKernel<7>>();
                 k->configure(&_gx, &_gy, &_score, norm_factor, threshold, sensitivity, border_mode == BorderMode::UNDEFINED);
                 _harris_score = std::move(k);
             }
diff --git a/src/runtime/NEON/functions/NEHistogram.cpp b/src/runtime/NEON/functions/NEHistogram.cpp
index c42b2a56e0..f333ecb1f8 100644
--- a/src/runtime/NEON/functions/NEHistogram.cpp
+++ b/src/runtime/NEON/functions/NEHistogram.cpp
@@ -24,17 +24,17 @@
 #include "arm_compute/runtime/NEON/functions/NEHistogram.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IDistribution1D.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/NEON/NEScheduler.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 NEHistogram::NEHistogram()
-    : _histogram_kernel(), _local_hist(), _window_lut(arm_compute::cpp14::make_unique<uint32_t[]>(window_lut_default_size)), _local_hist_size(0)
+    : _histogram_kernel(), _local_hist(), _window_lut(arm_compute::support::cpp14::make_unique<uint32_t[]>(window_lut_default_size)), _local_hist_size(0)
 {
 }
 
@@ -45,7 +45,7 @@ void NEHistogram::configure(const IImage *input, IDistribution1D *output)
 
     // Allocate space for threads local histograms
     _local_hist_size = output->num_bins() * NEScheduler::get().num_threads();
-    _local_hist      = arm_compute::cpp14::make_unique<uint32_t[]>(_local_hist_size);
+    _local_hist      = arm_compute::support::cpp14::make_unique<uint32_t[]>(_local_hist_size);
 
     // Configure kernel
     _histogram_kernel.configure(input, output, _local_hist.get(), _window_lut.get());
diff --git a/src/runtime/NEON/functions/NEIntegralImage.cpp b/src/runtime/NEON/functions/NEIntegralImage.cpp
index af604e9295..2e94ed5f9a 100644
--- a/src/runtime/NEON/functions/NEIntegralImage.cpp
+++ b/src/runtime/NEON/functions/NEIntegralImage.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEIntegralImage.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEIntegralImageKernel.h"
 #include "arm_compute/core/Types.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEIntegralImage::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEIntegralImageKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEIntegralImageKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
     _border_handler.configure(output, _kernel->border_size(), BorderMode::CONSTANT, 0);
diff --git a/src/runtime/NEON/functions/NELaplacianPyramid.cpp b/src/runtime/NEON/functions/NELaplacianPyramid.cpp
index 8232c79f2d..ec93759869 100644
--- a/src/runtime/NEON/functions/NELaplacianPyramid.cpp
+++ b/src/runtime/NEON/functions/NELaplacianPyramid.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NELaplacianPyramid.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IPyramid.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
@@ -33,6 +32,7 @@
 #include "arm_compute/runtime/NEON/functions/NEGaussian5x5.h"
 #include "arm_compute/runtime/NEON/functions/NEGaussianPyramid.h"
 #include "arm_compute/runtime/Tensor.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -86,8 +86,8 @@ void NELaplacianPyramid::configure(const ITensor *input, IPyramid *pyramid, ITen
     // Create Gaussian Pyramid function
     _gaussian_pyr_function.configure(input, &_gauss_pyr, border_mode, constant_border_value);
 
-    _convf = arm_compute::cpp14::make_unique<NEGaussian5x5[]>(_num_levels);
-    _subf  = arm_compute::cpp14::make_unique<NEArithmeticSubtraction[]>(_num_levels);
+    _convf = arm_compute::support::cpp14::make_unique<NEGaussian5x5[]>(_num_levels);
+    _subf  = arm_compute::support::cpp14::make_unique<NEArithmeticSubtraction[]>(_num_levels);
 
     for(unsigned int i = 0; i < _num_levels; ++i)
     {
diff --git a/src/runtime/NEON/functions/NELaplacianReconstruct.cpp b/src/runtime/NEON/functions/NELaplacianReconstruct.cpp
index 36ac4a74d1..8366e2bfec 100644
--- a/src/runtime/NEON/functions/NELaplacianReconstruct.cpp
+++ b/src/runtime/NEON/functions/NELaplacianReconstruct.cpp
@@ -24,11 +24,11 @@
 #include "arm_compute/runtime/NEON/functions/NELaplacianReconstruct.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/IPyramid.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <cstddef>
 
@@ -61,8 +61,8 @@ void NELaplacianReconstruct::configure(const IPyramid *pyramid, const ITensor *i
     _tmp_pyr.init(pyramid_info);
 
     // Allocate add and scale functions. Level 0 does not need to be scaled.
-    _addf   = arm_compute::cpp14::make_unique<NEArithmeticAddition[]>(num_levels);
-    _scalef = arm_compute::cpp14::make_unique<NEScale[]>(num_levels - 1);
+    _addf   = arm_compute::support::cpp14::make_unique<NEArithmeticAddition[]>(num_levels);
+    _scalef = arm_compute::support::cpp14::make_unique<NEScale[]>(num_levels - 1);
 
     const size_t last_level = num_levels - 1;
 
diff --git a/src/runtime/NEON/functions/NEMagnitude.cpp b/src/runtime/NEON/functions/NEMagnitude.cpp
index 9390ca2b6a..7877995d6b 100644
--- a/src/runtime/NEON/functions/NEMagnitude.cpp
+++ b/src/runtime/NEON/functions/NEMagnitude.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEMagnitude.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEMagnitudePhaseKernel.h"
 #include "arm_compute/core/Types.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -35,13 +35,13 @@ void NEMagnitude::configure(const ITensor *input1, const ITensor *input2, ITenso
 {
     if(use_fp16)
     {
-        auto k = arm_compute::cpp14::make_unique<NEMagnitudePhaseFP16Kernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
+        auto k = arm_compute::support::cpp14::make_unique<NEMagnitudePhaseFP16Kernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
         k->configure(input1, input2, output, nullptr);
         _kernel = std::move(k);
     }
     else
     {
-        auto k = arm_compute::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
+        auto k = arm_compute::support::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
         k->configure(input1, input2, output, nullptr);
         _kernel = std::move(k);
     }
diff --git a/src/runtime/NEON/functions/NEMedian3x3.cpp b/src/runtime/NEON/functions/NEMedian3x3.cpp
index aa7cc97081..627e5fb2a7 100644
--- a/src/runtime/NEON/functions/NEMedian3x3.cpp
+++ b/src/runtime/NEON/functions/NEMedian3x3.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEMedian3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEMedian3x3Kernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEMedian3x3::configure(ITensor *input, ITensor *output, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEMedian3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEMedian3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NENonLinearFilter.cpp b/src/runtime/NEON/functions/NENonLinearFilter.cpp
index 01aea3b671..57bd4e7aff 100644
--- a/src/runtime/NEON/functions/NENonLinearFilter.cpp
+++ b/src/runtime/NEON/functions/NENonLinearFilter.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NENonLinearFilter.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NENonLinearFilterKernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -35,7 +35,7 @@ void NENonLinearFilter::configure(ITensor *input, ITensor *output, NonLinearFilt
                                   BorderMode border_mode,
                                   uint8_t    constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NENonLinearFilterKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NENonLinearFilterKernel>();
     k->configure(input, output, function, mask_size, pattern, mask, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NENonMaximaSuppression3x3.cpp b/src/runtime/NEON/functions/NENonMaximaSuppression3x3.cpp
index e92d567d52..3b59820d09 100644
--- a/src/runtime/NEON/functions/NENonMaximaSuppression3x3.cpp
+++ b/src/runtime/NEON/functions/NENonMaximaSuppression3x3.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NENonMaximaSuppression3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NENonMaximaSuppression3x3::configure(ITensor *input, ITensor *output, BorderMode border_mode)
 {
-    auto k = arm_compute::cpp14::make_unique<NENonMaximaSuppression3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NENonMaximaSuppression3x3Kernel>();
     k->configure(input, output, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
 
diff --git a/src/runtime/NEON/functions/NEOpticalFlow.cpp b/src/runtime/NEON/functions/NEOpticalFlow.cpp
index 49135e442c..d851e9a62b 100644
--- a/src/runtime/NEON/functions/NEOpticalFlow.cpp
+++ b/src/runtime/NEON/functions/NEOpticalFlow.cpp
@@ -24,7 +24,6 @@
 #include "arm_compute/runtime/NEON/functions/NEOpticalFlow.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NELKTrackerKernel.h"
 #include "arm_compute/core/TensorInfo.h"
@@ -34,6 +33,7 @@
 #include "arm_compute/runtime/Pyramid.h"
 #include "arm_compute/runtime/Tensor.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
@@ -65,10 +65,10 @@ void NEOpticalFlow::configure(const Pyramid *old_pyramid, const Pyramid *new_pyr
 
     const float pyr_scale = old_pyramid->info()->scale();
 
-    _func_scharr    = arm_compute::cpp14::make_unique<NEScharr3x3[]>(_num_levels);
-    _kernel_tracker = arm_compute::cpp14::make_unique<NELKTrackerKernel[]>(_num_levels);
-    _scharr_gx      = arm_compute::cpp14::make_unique<Tensor[]>(_num_levels);
-    _scharr_gy      = arm_compute::cpp14::make_unique<Tensor[]>(_num_levels);
+    _func_scharr    = arm_compute::support::cpp14::make_unique<NEScharr3x3[]>(_num_levels);
+    _kernel_tracker = arm_compute::support::cpp14::make_unique<NELKTrackerKernel[]>(_num_levels);
+    _scharr_gx      = arm_compute::support::cpp14::make_unique<Tensor[]>(_num_levels);
+    _scharr_gy      = arm_compute::support::cpp14::make_unique<Tensor[]>(_num_levels);
 
     _old_points_internal = LKInternalKeypointArray(old_points->num_values());
     _new_points_internal = LKInternalKeypointArray(old_points->num_values());
diff --git a/src/runtime/NEON/functions/NEPhase.cpp b/src/runtime/NEON/functions/NEPhase.cpp
index 7683f461d3..436d22f3af 100644
--- a/src/runtime/NEON/functions/NEPhase.cpp
+++ b/src/runtime/NEON/functions/NEPhase.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEPhase.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEMagnitudePhaseKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEPhase::configure(const ITensor *input1, const ITensor *input2, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
+    auto k = arm_compute::support::cpp14::make_unique<NEMagnitudePhaseKernel<MagnitudeType::L2NORM, PhaseType::SIGNED>>();
     k->configure(input1, input2, nullptr, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp b/src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp
index 056d33b370..2e2ea11e86 100644
--- a/src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp
+++ b/src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEPixelWiseMultiplicationKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEPixelWiseMultiplication::configure(const ITensor *input1, const ITensor *input2, ITensor *output, float scale, ConvertPolicy overflow_policy, RoundingPolicy rounding_policy)
 {
-    auto k = arm_compute::cpp14::make_unique<NEPixelWiseMultiplicationKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEPixelWiseMultiplicationKernel>();
     k->configure(input1, input2, output, scale, overflow_policy, rounding_policy);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEPoolingLayer.cpp b/src/runtime/NEON/functions/NEPoolingLayer.cpp
index 6f0cc4f160..4c4e11f7db 100644
--- a/src/runtime/NEON/functions/NEPoolingLayer.cpp
+++ b/src/runtime/NEON/functions/NEPoolingLayer.cpp
@@ -23,15 +23,15 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEPoolingLayer.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEPoolingLayerKernel.h"
+#include "support/ToolchainSupport.h"
 
 using namespace arm_compute;
 
 void NEPoolingLayer::configure(ITensor *input, ITensor *output, const PoolingLayerInfo &pool_info)
 {
     // Configure pooling kernel
-    auto k = arm_compute::cpp14::make_unique<NEPoolingLayerKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEPoolingLayerKernel>();
     k->configure(input, output, pool_info);
     _kernel = std::move(k);
 
diff --git a/src/runtime/NEON/functions/NERemap.cpp b/src/runtime/NEON/functions/NERemap.cpp
index 9f06fb699c..882e93beae 100644
--- a/src/runtime/NEON/functions/NERemap.cpp
+++ b/src/runtime/NEON/functions/NERemap.cpp
@@ -24,13 +24,13 @@
 #include "arm_compute/runtime/NEON/functions/NERemap.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/ITensor.h"
 #include "arm_compute/core/NEON/kernels/NERemapKernel.h"
 #include "arm_compute/core/PixelValue.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Validate.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -44,7 +44,7 @@ void NERemap::configure(ITensor *input, const ITensor *map_x, const ITensor *map
     ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(map_y, 1, DataType::F32);
     ARM_COMPUTE_ERROR_ON_MSG(policy == InterpolationPolicy::AREA, "Area interpolation is not supported");
 
-    auto k = arm_compute::cpp14::make_unique<NERemapKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NERemapKernel>();
 
     k->configure(input, map_x, map_y, output, policy);
 
diff --git a/src/runtime/NEON/functions/NEScale.cpp b/src/runtime/NEON/functions/NEScale.cpp
index b70f626df0..0a551e937d 100644
--- a/src/runtime/NEON/functions/NEScale.cpp
+++ b/src/runtime/NEON/functions/NEScale.cpp
@@ -32,6 +32,7 @@
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/Window.h"
 #include "arm_compute/runtime/TensorAllocator.h"
+#include "support/ToolchainSupport.h"
 
 #include <cmath>
 #include <cstddef>
@@ -116,7 +117,7 @@ void NEScale::configure(ITensor *input, ITensor *output, InterpolationPolicy pol
         policy = InterpolationPolicy::NEAREST_NEIGHBOR;
     }
 
-    auto k = arm_compute::cpp14::make_unique<NEScaleKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEScaleKernel>();
 
     // Check if the border mode is UNDEFINED
     const bool border_undefined = border_mode == BorderMode::UNDEFINED;
diff --git a/src/runtime/NEON/functions/NEScharr3x3.cpp b/src/runtime/NEON/functions/NEScharr3x3.cpp
index 04b3f14ce7..ba9985e1d2 100644
--- a/src/runtime/NEON/functions/NEScharr3x3.cpp
+++ b/src/runtime/NEON/functions/NEScharr3x3.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEScharr3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEScharr3x3Kernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NEScharr3x3::configure(ITensor *input, ITensor *output_x, ITensor *output_y, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NEScharr3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEScharr3x3Kernel>();
     k->configure(input, output_x, output_y, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NESobel3x3.cpp b/src/runtime/NEON/functions/NESobel3x3.cpp
index 3b46fd78c1..753b1f600f 100644
--- a/src/runtime/NEON/functions/NESobel3x3.cpp
+++ b/src/runtime/NEON/functions/NESobel3x3.cpp
@@ -23,9 +23,9 @@
  */
 #include "arm_compute/runtime/NEON/functions/NESobel3x3.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NESobel3x3Kernel.h"
 #include "arm_compute/core/PixelValue.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -33,7 +33,7 @@ using namespace arm_compute;
 
 void NESobel3x3::configure(ITensor *input, ITensor *output_x, ITensor *output_y, BorderMode border_mode, uint8_t constant_border_value)
 {
-    auto k = arm_compute::cpp14::make_unique<NESobel3x3Kernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NESobel3x3Kernel>();
     k->configure(input, output_x, output_y, border_mode == BorderMode::UNDEFINED);
     _kernel = std::move(k);
     _border_handler.configure(input, _kernel->border_size(), border_mode, PixelValue(constant_border_value));
diff --git a/src/runtime/NEON/functions/NETableLookup.cpp b/src/runtime/NEON/functions/NETableLookup.cpp
index ebb8a0ac9b..cae117a13e 100644
--- a/src/runtime/NEON/functions/NETableLookup.cpp
+++ b/src/runtime/NEON/functions/NETableLookup.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NETableLookup.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NETableLookupKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NETableLookup::configure(const ITensor *input, const ILut *lut, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NETableLookupKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NETableLookupKernel>();
     k->configure(input, lut, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEThreshold.cpp b/src/runtime/NEON/functions/NEThreshold.cpp
index 93dc124880..37883e5387 100644
--- a/src/runtime/NEON/functions/NEThreshold.cpp
+++ b/src/runtime/NEON/functions/NEThreshold.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NEThreshold.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEThresholdKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NEThreshold::configure(const ITensor *input, ITensor *output, uint8_t threshold, uint8_t false_value, uint8_t true_value, ThresholdType type, uint8_t upper)
 {
-    auto k = arm_compute::cpp14::make_unique<NEThresholdKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NEThresholdKernel>();
     k->configure(input, output, threshold, false_value, true_value, type, upper);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NETranspose.cpp b/src/runtime/NEON/functions/NETranspose.cpp
index 53ac9c5ee3..eb81e02205 100644
--- a/src/runtime/NEON/functions/NETranspose.cpp
+++ b/src/runtime/NEON/functions/NETranspose.cpp
@@ -23,8 +23,8 @@
  */
 #include "arm_compute/runtime/NEON/functions/NETranspose.h"
 
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NETransposeKernel.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -32,7 +32,7 @@ using namespace arm_compute;
 
 void NETranspose::configure(const ITensor *input, ITensor *output)
 {
-    auto k = arm_compute::cpp14::make_unique<NETransposeKernel>();
+    auto k = arm_compute::support::cpp14::make_unique<NETransposeKernel>();
     k->configure(input, output);
     _kernel = std::move(k);
 }
diff --git a/src/runtime/NEON/functions/NEWarpAffine.cpp b/src/runtime/NEON/functions/NEWarpAffine.cpp
index 24fb16f9e3..889d827ed0 100644
--- a/src/runtime/NEON/functions/NEWarpAffine.cpp
+++ b/src/runtime/NEON/functions/NEWarpAffine.cpp
@@ -24,8 +24,9 @@
 #include "arm_compute/runtime/NEON/functions/NEWarpAffine.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEWarpKernel.h"
+#include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -41,14 +42,14 @@ void NEWarpAffine::configure(ITensor *input, ITensor *output, const float *matri
     {
         case InterpolationPolicy::NEAREST_NEIGHBOR:
         {
-            auto k = arm_compute::cpp14::make_unique<NEWarpAffineKernel<InterpolationPolicy::NEAREST_NEIGHBOR>>();
+            auto k = arm_compute::support::cpp14::make_unique<NEWarpAffineKernel<InterpolationPolicy::NEAREST_NEIGHBOR>>();
             k->configure(input, output, matrix, border_mode, constant_border_value);
             _kernel = std::move(k);
             break;
         }
         case InterpolationPolicy::BILINEAR:
         {
-            auto k = arm_compute::cpp14::make_unique<NEWarpAffineKernel<InterpolationPolicy::BILINEAR>>();
+            auto k = arm_compute::support::cpp14::make_unique<NEWarpAffineKernel<InterpolationPolicy::BILINEAR>>();
             k->configure(input, output, matrix, border_mode, constant_border_value);
             _kernel = std::move(k);
             break;
diff --git a/src/runtime/NEON/functions/NEWarpPerspective.cpp b/src/runtime/NEON/functions/NEWarpPerspective.cpp
index 84b2df5bfa..ed5d6a0d27 100644
--- a/src/runtime/NEON/functions/NEWarpPerspective.cpp
+++ b/src/runtime/NEON/functions/NEWarpPerspective.cpp
@@ -24,8 +24,9 @@
 #include "arm_compute/runtime/NEON/functions/NEWarpPerspective.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/NEON/kernels/NEWarpKernel.h"
+#include "arm_compute/core/Validate.h"
+#include "support/ToolchainSupport.h"
 
 #include <utility>
 
@@ -41,14 +42,14 @@ void NEWarpPerspective::configure(ITensor *input, ITensor *output, const float *
     {
         case InterpolationPolicy::NEAREST_NEIGHBOR:
         {
-            auto k = arm_compute::cpp14::make_unique<NEWarpPerspectiveKernel<InterpolationPolicy::NEAREST_NEIGHBOR>>();
+            auto k = arm_compute::support::cpp14::make_unique<NEWarpPerspectiveKernel<InterpolationPolicy::NEAREST_NEIGHBOR>>();
             k->configure(input, output, matrix, border_mode, constant_border_value);
             _kernel = std::move(k);
             break;
         }
         case InterpolationPolicy::BILINEAR:
         {
-            auto k = arm_compute::cpp14::make_unique<NEWarpPerspectiveKernel<InterpolationPolicy::BILINEAR>>();
+            auto k = arm_compute::support::cpp14::make_unique<NEWarpPerspectiveKernel<InterpolationPolicy::BILINEAR>>();
             k->configure(input, output, matrix, border_mode, constant_border_value);
             _kernel = std::move(k);
             break;
diff --git a/src/runtime/Pyramid.cpp b/src/runtime/Pyramid.cpp
index f1b6c93b50..ebd65702a6 100644
--- a/src/runtime/Pyramid.cpp
+++ b/src/runtime/Pyramid.cpp
@@ -24,10 +24,10 @@
 #include "arm_compute/runtime/Pyramid.h"
 
 #include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/PyramidInfo.h"
 #include "arm_compute/core/TensorInfo.h"
 #include "arm_compute/core/TensorShape.h"
+#include "support/ToolchainSupport.h"
 
 #include <cmath>
 
@@ -46,7 +46,7 @@ void Pyramid::init_auto_padding(const PyramidInfo &info)
 void Pyramid::internal_init(const PyramidInfo &info, bool auto_padding)
 {
     _info    = info;
-    _pyramid = arm_compute::cpp14::make_unique<Tensor[]>(_info.num_levels());
+    _pyramid = arm_compute::support::cpp14::make_unique<Tensor[]>(_info.num_levels());
 
     size_t      w            = _info.width();
     size_t      h            = _info.height();