IVGCVSW-4209 Create a public API for the ArmNN Utils

 * Moved the relevant armnnUtils headers to the new location:
   include/armnnUtils
 * Update the header usage throughout the source code

!android-nn-driver:2387

Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I2ba15cebcacafad2b5a1a7b9c3312ffc585e09d6

79 files changed, 266 insertions, 204 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 21d1336578..ad9e9ee91d 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -36,14 +36,16 @@ add_custom_target(AdditionalCMakeFiles SOURCES ${additional_cmake_files})
 
 set(armnnUtils_sources)
 list(APPEND armnnUtils_sources
+    include/armnnUtils/DataLayoutIndexed.hpp
+    include/armnnUtils/Permute.hpp
+    include/armnnUtils/FloatingPointConverter.hpp
+    include/armnnUtils/TensorUtils.hpp
     src/armnnUtils/GraphTopologicalSort.hpp
     src/armnnUtils/Half.hpp
     src/armnnUtils/Logging.hpp
     src/armnnUtils/Logging.cpp
-    src/armnnUtils/Permute.hpp
     src/armnnUtils/Permute.cpp
     src/armnnUtils/DataLayoutIndexed.cpp
-    src/armnnUtils/DataLayoutIndexed.hpp
     src/armnnUtils/DotSerializer.cpp
     src/armnnUtils/DotSerializer.hpp
     src/armnnUtils/HeapProfiling.cpp
@@ -55,7 +57,6 @@ list(APPEND armnnUtils_sources
     src/armnnUtils/CsvReader.cpp
     src/armnnUtils/CsvReader.hpp
     src/armnnUtils/FloatingPointConverter.cpp
-    src/armnnUtils/FloatingPointConverter.hpp
     src/armnnUtils/VerificationHelpers.hpp
     src/armnnUtils/VerificationHelpers.cpp
     src/armnnUtils/ParserHelper.hpp
@@ -65,7 +66,6 @@ list(APPEND armnnUtils_sources
     src/armnnUtils/PrototxtConversions.cpp
     src/armnnUtils/QuantizeHelper.hpp
     src/armnnUtils/TensorIOUtils.hpp
-    src/armnnUtils/TensorUtils.hpp
     src/armnnUtils/TensorUtils.cpp
     )
 
diff --git a/src/armnnUtils/DataLayoutIndexed.hpp b/include/armnnUtils/DataLayoutIndexed.hpp
index 8bd9701a5e..03404bda5d 100644
--- a/src/armnnUtils/DataLayoutIndexed.hpp
+++ b/include/armnnUtils/DataLayoutIndexed.hpp
@@ -1,5 +1,5 @@
 //
-// Copyright © 2017 Arm Ltd. All rights reserved.
+// Copyright © 2019 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
 
diff --git a/src/armnnUtils/FloatingPointConverter.hpp b/include/armnnUtils/FloatingPointConverter.hpp
index 7b201c7ead..cf573a2ee8 100644
--- a/src/armnnUtils/FloatingPointConverter.hpp
+++ b/include/armnnUtils/FloatingPointConverter.hpp
@@ -1,5 +1,5 @@
 //
-// Copyright © 2017 Arm Ltd. All rights reserved.
+// Copyright © 2019 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
 
@@ -9,6 +9,7 @@
 
 namespace armnnUtils
 {
+
 class FloatingPointConverter
 {
 public:
@@ -18,4 +19,5 @@ public:
 
     static void ConvertFloat16To32(const void *srcFloat16Buffer, size_t numElements, float *dstFloat32Buffer);
 };
-} //namespace armnnUtils
+
+} // namespace armnnUtils
diff --git a/src/armnnUtils/Permute.hpp b/include/armnnUtils/Permute.hpp
index b9ef94bc86..1ae00a136b 100644
--- a/src/armnnUtils/Permute.hpp
+++ b/include/armnnUtils/Permute.hpp
@@ -1,7 +1,8 @@
 //
-// Copyright © 2017 Arm Ltd. All rights reserved.
+// Copyright © 2019 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #pragma once
 
 #include <armnn/TensorFwd.hpp>
diff --git a/src/armnnUtils/TensorUtils.hpp b/include/armnnUtils/TensorUtils.hpp
index b67431d757..fbfb8f4e1e 100644
--- a/src/armnnUtils/TensorUtils.hpp
+++ b/include/armnnUtils/TensorUtils.hpp
@@ -1,5 +1,5 @@
 //
-// Copyright © 2017 Arm Ltd. All rights reserved.
+// Copyright © 2019 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
 
diff --git a/src/armnn/NetworkQuantizer.cpp b/src/armnn/NetworkQuantizer.cpp
index 4a8bae52f2..dd2f32414a 100644
--- a/src/armnn/NetworkQuantizer.cpp
+++ b/src/armnn/NetworkQuantizer.cpp
@@ -13,13 +13,14 @@
 #include "QuantizerVisitor.hpp"
 #include "OverrideInputRangeVisitor.hpp"
 
+#include <TensorIOUtils.hpp>
+
 #include <armnn/ILayerVisitor.hpp>
 #include <armnn/INetwork.hpp>
 #include <armnn/Tensor.hpp>
 #include <armnn/Types.hpp>
 
-#include <TensorUtils.hpp>
-#include <TensorIOUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <boost/variant.hpp>
 
diff --git a/src/armnn/layers/ArgMinMaxLayer.cpp b/src/armnn/layers/ArgMinMaxLayer.cpp
index bfd71d519b..35bf272e41 100644
--- a/src/armnn/layers/ArgMinMaxLayer.cpp
+++ b/src/armnn/layers/ArgMinMaxLayer.cpp
@@ -2,13 +2,14 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "ArgMinMaxLayer.hpp"
 
+#include "ArgMinMaxLayer.hpp"
 #include "LayerCloneBase.hpp"
 
-#include <TensorUtils.hpp>
-
 #include <armnn/TypesUtils.hpp>
+
+#include <armnnUtils/TensorUtils.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
@@ -88,4 +89,4 @@ void ArgMinMaxLayer::Accept(ILayerVisitor& visitor) const
     visitor.VisitArgMinMaxLayer(this, GetParameters(), GetName());
 }
 
-} // namespace armnn
\ No newline at end of file
+} // namespace armnn
diff --git a/src/armnn/layers/BatchToSpaceNdLayer.cpp b/src/armnn/layers/BatchToSpaceNdLayer.cpp
index a168fe8bbd..335811186c 100644
--- a/src/armnn/layers/BatchToSpaceNdLayer.cpp
+++ b/src/armnn/layers/BatchToSpaceNdLayer.cpp
@@ -2,20 +2,20 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "BatchToSpaceNdLayer.hpp"
 
+#include "BatchToSpaceNdLayer.hpp"
 #include "LayerCloneBase.hpp"
 #include "LayerWithParameters.hpp"
 #include "BatchToSpaceNdLayer.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 #include <numeric>
 
 using namespace armnnUtils;
diff --git a/src/armnn/layers/Convolution2dLayer.cpp b/src/armnn/layers/Convolution2dLayer.cpp
index 4300d55e1e..b205315b07 100644
--- a/src/armnn/layers/Convolution2dLayer.cpp
+++ b/src/armnn/layers/Convolution2dLayer.cpp
@@ -2,15 +2,18 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "Convolution2dLayer.hpp"
 
+#include "Convolution2dLayer.hpp"
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
+
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
+
 #include <string>
-#include <DataLayoutIndexed.hpp>
 
 using namespace armnnUtils;
 
diff --git a/src/armnn/layers/DepthToSpaceLayer.cpp b/src/armnn/layers/DepthToSpaceLayer.cpp
index de01ca9868..80992fa20f 100644
--- a/src/armnn/layers/DepthToSpaceLayer.cpp
+++ b/src/armnn/layers/DepthToSpaceLayer.cpp
@@ -4,16 +4,15 @@
 //
 
 #include "DepthToSpaceLayer.hpp"
-
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 #include <numeric>
 
 namespace armnn
diff --git a/src/armnn/layers/DepthwiseConvolution2dLayer.cpp b/src/armnn/layers/DepthwiseConvolution2dLayer.cpp
index a50a0f6310..306bce57fc 100644
--- a/src/armnn/layers/DepthwiseConvolution2dLayer.cpp
+++ b/src/armnn/layers/DepthwiseConvolution2dLayer.cpp
@@ -2,16 +2,18 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "DepthwiseConvolution2dLayer.hpp"
 
+#include "DepthwiseConvolution2dLayer.hpp"
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
+
 #include <string>
-#include <DataLayoutIndexed.hpp>
 
 using namespace armnnUtils;
 
diff --git a/src/armnn/layers/PermuteLayer.cpp b/src/armnn/layers/PermuteLayer.cpp
index e6a7a38685..459a755117 100644
--- a/src/armnn/layers/PermuteLayer.cpp
+++ b/src/armnn/layers/PermuteLayer.cpp
@@ -2,16 +2,18 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #include "PermuteLayer.hpp"
 
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
+
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <Permute.hpp>
-
 namespace armnn
 {
 
diff --git a/src/armnn/layers/Pooling2dLayer.cpp b/src/armnn/layers/Pooling2dLayer.cpp
index 7c5be01c35..8f5ccb9215 100644
--- a/src/armnn/layers/Pooling2dLayer.cpp
+++ b/src/armnn/layers/Pooling2dLayer.cpp
@@ -2,17 +2,18 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #include "Pooling2dLayer.hpp"
 
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 using namespace armnnUtils;
 
 namespace armnn
diff --git a/src/armnn/layers/ResizeLayer.cpp b/src/armnn/layers/ResizeLayer.cpp
index 44b4d9df5f..50d5f243ff 100644
--- a/src/armnn/layers/ResizeLayer.cpp
+++ b/src/armnn/layers/ResizeLayer.cpp
@@ -2,17 +2,17 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "ResizeLayer.hpp"
 
+#include "ResizeLayer.hpp"
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 using namespace armnnUtils;
 
 namespace armnn
diff --git a/src/armnn/layers/SpaceToBatchNdLayer.cpp b/src/armnn/layers/SpaceToBatchNdLayer.cpp
index 6d5827e771..5002160ff7 100644
--- a/src/armnn/layers/SpaceToBatchNdLayer.cpp
+++ b/src/armnn/layers/SpaceToBatchNdLayer.cpp
@@ -4,16 +4,15 @@
 //
 
 #include "SpaceToBatchNdLayer.hpp"
-
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 #include <numeric>
 
 using namespace armnnUtils;
diff --git a/src/armnn/layers/SpaceToDepthLayer.cpp b/src/armnn/layers/SpaceToDepthLayer.cpp
index 8a9f1c296c..befb940b4a 100644
--- a/src/armnn/layers/SpaceToDepthLayer.cpp
+++ b/src/armnn/layers/SpaceToDepthLayer.cpp
@@ -4,16 +4,15 @@
 //
 
 #include "SpaceToDepthLayer.hpp"
-
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/WorkloadData.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 #include <numeric>
 
 using namespace armnnUtils;
diff --git a/src/armnn/layers/TransposeConvolution2dLayer.cpp b/src/armnn/layers/TransposeConvolution2dLayer.cpp
index 7bd2f3b9d8..1110e04991 100644
--- a/src/armnn/layers/TransposeConvolution2dLayer.cpp
+++ b/src/armnn/layers/TransposeConvolution2dLayer.cpp
@@ -2,17 +2,17 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
-#include "TransposeConvolution2dLayer.hpp"
 
+#include "TransposeConvolution2dLayer.hpp"
 #include "LayerCloneBase.hpp"
 
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 using namespace armnnUtils;
 
 namespace armnn
diff --git a/src/armnn/optimizations/ConvertConstants.hpp b/src/armnn/optimizations/ConvertConstants.hpp
index bb38ae8811..e8a0e62295 100644
--- a/src/armnn/optimizations/ConvertConstants.hpp
+++ b/src/armnn/optimizations/ConvertConstants.hpp
@@ -7,8 +7,9 @@
 
 #include "Optimization.hpp"
 
+#include <armnnUtils/FloatingPointConverter.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
-#include <FloatingPointConverter.hpp>
 
 #include <Half.hpp>
 
diff --git a/src/armnn/optimizations/MovePermuteUp.hpp b/src/armnn/optimizations/MovePermuteUp.hpp
index 7f4ad1b422..a7a477be84 100644
--- a/src/armnn/optimizations/MovePermuteUp.hpp
+++ b/src/armnn/optimizations/MovePermuteUp.hpp
@@ -5,7 +5,8 @@
 #pragma once
 
 #include "Optimization.hpp"
-#include "Permute.hpp"
+
+#include <armnnUtils/Permute.hpp>
 
 namespace armnn
 {
diff --git a/src/armnn/test/CreateWorkload.hpp b/src/armnn/test/CreateWorkload.hpp
index d6d1ab3c5b..0537877b99 100644
--- a/src/armnn/test/CreateWorkload.hpp
+++ b/src/armnn/test/CreateWorkload.hpp
@@ -6,15 +6,16 @@
 
 #include "TestUtils.hpp"
 
-#include <backendsCommon/WorkloadData.hpp>
-#include <backendsCommon/WorkloadFactory.hpp>
-#include <backendsCommon/CpuTensorHandle.hpp>
-
 #include <Graph.hpp>
-#include <DataLayoutIndexed.hpp>
 #include <Network.hpp>
 #include <ResolveType.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
+#include <backendsCommon/WorkloadData.hpp>
+#include <backendsCommon/WorkloadFactory.hpp>
+#include <backendsCommon/CpuTensorHandle.hpp>
+
 #include <boost/test/unit_test.hpp>
 #include <boost/cast.hpp>
 
diff --git a/src/armnn/test/FloatingPointConverterTest.cpp b/src/armnn/test/FloatingPointConverterTest.cpp
index 733aa3842b..4497ca70a8 100644
--- a/src/armnn/test/FloatingPointConverterTest.cpp
+++ b/src/armnn/test/FloatingPointConverterTest.cpp
@@ -3,7 +3,8 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include "FloatingPointConverter.hpp"
+#include <armnnUtils/FloatingPointConverter.hpp>
+
 #include <Half.hpp>
 
 #include <boost/test/unit_test.hpp>
diff --git a/src/armnn/test/OptimizerTests.cpp b/src/armnn/test/OptimizerTests.cpp
index c0ad9c8927..ba18aa31e8 100644
--- a/src/armnn/test/OptimizerTests.cpp
+++ b/src/armnn/test/OptimizerTests.cpp
@@ -5,11 +5,14 @@
 
 #include "TestUtils.hpp"
 
-#include <armnn/ArmNN.hpp>
 #include <Graph.hpp>
 #include <Optimizer.hpp>
+
+#include <armnn/ArmNN.hpp>
+
+#include <armnnUtils/FloatingPointConverter.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
-#include <FloatingPointConverter.hpp>
 
 #include <boost/test/unit_test.hpp>
 
diff --git a/src/armnnDeserializer/Deserializer.cpp b/src/armnnDeserializer/Deserializer.cpp
index 4cd3cf25d7..3e6d5aa298 100644
--- a/src/armnnDeserializer/Deserializer.cpp
+++ b/src/armnnDeserializer/Deserializer.cpp
@@ -8,8 +8,9 @@
 #include <armnn/ArmNN.hpp>
 #include <armnn/Exceptions.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <ParserHelper.hpp>
-#include <Permute.hpp>
 #include <VerificationHelpers.hpp>
 
 #include <boost/filesystem.hpp>
diff --git a/src/armnnTfLiteParser/TfLiteParser.cpp b/src/armnnTfLiteParser/TfLiteParser.cpp
index 9a20740914..090fbed8b5 100644
--- a/src/armnnTfLiteParser/TfLiteParser.cpp
+++ b/src/armnnTfLiteParser/TfLiteParser.cpp
@@ -2,6 +2,7 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #include "TfLiteParser.hpp"
 
 #include <armnn/ArmNN.hpp>
@@ -10,13 +11,16 @@
 #include <boost/filesystem.hpp>
 
 // armnnUtils:
+#include <armnnUtils/Permute.hpp>
+
 #include <ParserHelper.hpp>
-#include <Permute.hpp>
 #include <VerificationHelpers.hpp>
 
 // The generated code based on the Tf Lite schema:
 #include <schema_generated.h>
 
+#include <flatbuffers/flexbuffers.h>
+
 #include <boost/core/ignore_unused.hpp>
 #include <boost/assert.hpp>
 #include <boost/format.hpp>
@@ -28,7 +32,6 @@
 #include <algorithm>
 #include <limits>
 #include <numeric>
-#include <flatbuffers/flexbuffers.h>
 
 using namespace armnn;
 using armnn::CheckLocation;
diff --git a/src/armnnTfParser/TfParser.cpp b/src/armnnTfParser/TfParser.cpp
index 51423bf6a7..8c68659b95 100755
--- a/src/armnnTfParser/TfParser.cpp
+++ b/src/armnnTfParser/TfParser.cpp
@@ -8,10 +8,11 @@
 #include <armnn/TypesUtils.hpp>
 #include <armnn/Descriptors.hpp>
 
+#include <armnnUtils/Permute.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <GraphTopologicalSort.hpp>
 #include <ParserHelper.hpp>
-#include <Permute.hpp>
-#include <DataLayoutIndexed.hpp>
 
 #include <google/protobuf/io/zero_copy_stream_impl.h>
 #include <google/protobuf/text_format.h>
diff --git a/src/armnnTfParser/test/DepthwiseConvolution2d.cpp b/src/armnnTfParser/test/DepthwiseConvolution2d.cpp
index 464e62fc23..43a7ebc28e 100644
--- a/src/armnnTfParser/test/DepthwiseConvolution2d.cpp
+++ b/src/armnnTfParser/test/DepthwiseConvolution2d.cpp
@@ -3,13 +3,17 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include <boost/test/unit_test.hpp>
-#include "armnnTfParser/ITfParser.hpp"
 #include "ParserPrototxtFixture.hpp"
+
+#include "armnnTfParser/ITfParser.hpp"
+
+#include <armnnUtils/Permute.hpp>
+
+#include <boost/test/unit_test.hpp>
+
 #include <string>
 #include <iostream>
 
-#include <Permute.hpp>
 using namespace armnnUtils;
 using namespace armnn;
 
diff --git a/src/armnnUtils/DataLayoutIndexed.cpp b/src/armnnUtils/DataLayoutIndexed.cpp
index 02f1e816ac..92051b6c83 100644
--- a/src/armnnUtils/DataLayoutIndexed.cpp
+++ b/src/armnnUtils/DataLayoutIndexed.cpp
@@ -3,7 +3,8 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include "DataLayoutIndexed.hpp"
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 using namespace armnn;
 
 namespace armnnUtils
diff --git a/src/armnnUtils/FloatingPointConverter.cpp b/src/armnnUtils/FloatingPointConverter.cpp
index 92409d4eea..3bdde11eb8 100644
--- a/src/armnnUtils/FloatingPointConverter.cpp
+++ b/src/armnnUtils/FloatingPointConverter.cpp
@@ -3,7 +3,7 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include "FloatingPointConverter.hpp"
+#include <armnnUtils/FloatingPointConverter.hpp>
 
 #include "Half.hpp"
 
diff --git a/src/armnnUtils/ParserHelper.cpp b/src/armnnUtils/ParserHelper.cpp
index 2286f8b6ed..990a9b2098 100644
--- a/src/armnnUtils/ParserHelper.cpp
+++ b/src/armnnUtils/ParserHelper.cpp
@@ -6,7 +6,7 @@
 #include "ParserHelper.hpp"
 
 // armnnUtils
-#include "Permute.hpp"
+#include <armnnUtils/Permute.hpp>
 
 #include <boost/format.hpp>
 
diff --git a/src/armnnUtils/Permute.cpp b/src/armnnUtils/Permute.cpp
index 24a8286bba..a758bd9fbf 100644
--- a/src/armnnUtils/Permute.cpp
+++ b/src/armnnUtils/Permute.cpp
@@ -3,10 +3,11 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include "Permute.hpp"
+#include <armnn/Tensor.hpp>
+
+#include <armnnUtils/Permute.hpp>
 
 #include "Half.hpp"
-#include <armnn/Tensor.hpp>
 
 #include <cassert>
 #include <cstring>
diff --git a/src/armnnUtils/TensorUtils.cpp b/src/armnnUtils/TensorUtils.cpp
index 601277491c..c1d1200938 100644
--- a/src/armnnUtils/TensorUtils.cpp
+++ b/src/armnnUtils/TensorUtils.cpp
@@ -3,7 +3,7 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include "TensorUtils.hpp"
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <backendsCommon/ITensorHandle.hpp>
 
diff --git a/src/armnnUtils/test/TensorUtilsTest.cpp b/src/armnnUtils/test/TensorUtilsTest.cpp
index a903b63f83..d24740b762 100644
--- a/src/armnnUtils/test/TensorUtilsTest.cpp
+++ b/src/armnnUtils/test/TensorUtilsTest.cpp
@@ -3,10 +3,10 @@
 // SPDX-License-Identifier: MIT
 //
 
-#include <TensorUtils.hpp>
-
 #include <armnn/Types.hpp>
 
+#include <armnnUtils/TensorUtils.hpp>
+
 #include <boost/test/unit_test.hpp>
 
 using namespace armnn;
diff --git a/src/backends/backendsCommon/WorkloadData.cpp b/src/backends/backendsCommon/WorkloadData.cpp
index 6d5c6e8343..251d3b96b8 100644
--- a/src/backends/backendsCommon/WorkloadData.cpp
+++ b/src/backends/backendsCommon/WorkloadData.cpp
@@ -2,11 +2,13 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #include "WorkloadData.hpp"
 
 #include "CpuTensorHandle.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <algorithm>
 #include <iomanip>
@@ -15,7 +17,6 @@
 
 #include <boost/format.hpp>
 #include <boost/numeric/conversion/cast.hpp>
-#include <TensorUtils.hpp>
 
 using namespace armnnUtils;
 
diff --git a/src/backends/backendsCommon/WorkloadUtils.hpp b/src/backends/backendsCommon/WorkloadUtils.hpp
index d581b5ab9e..6092c5fc90 100644
--- a/src/backends/backendsCommon/WorkloadUtils.hpp
+++ b/src/backends/backendsCommon/WorkloadUtils.hpp
@@ -10,8 +10,9 @@
 
 #include <armnn/Tensor.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <Half.hpp>
-#include <Permute.hpp>
 #include <Profiling.hpp>
 
 #include <boost/cast.hpp>
diff --git a/src/backends/backendsCommon/test/DataLayoutUtils.hpp b/src/backends/backendsCommon/test/DataLayoutUtils.hpp
index f89325829e..9411212f4f 100644
--- a/src/backends/backendsCommon/test/DataLayoutUtils.hpp
+++ b/src/backends/backendsCommon/test/DataLayoutUtils.hpp
@@ -5,11 +5,11 @@
 
 #pragma once
 
-#include <Permute.hpp>
-
 #include <armnn/Tensor.hpp>
 #include <armnn/Types.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 template<typename T>
 void PermuteTensorNchwToNhwc(armnn::TensorInfo& tensorInfo, std::vector<T>& tensorData)
 {
diff --git a/src/backends/backendsCommon/test/InstanceNormalizationEndToEndTestImpl.cpp b/src/backends/backendsCommon/test/InstanceNormalizationEndToEndTestImpl.cpp
index 0ba2a74895..a3dd88c893 100644
--- a/src/backends/backendsCommon/test/InstanceNormalizationEndToEndTestImpl.cpp
+++ b/src/backends/backendsCommon/test/InstanceNormalizationEndToEndTestImpl.cpp
@@ -4,12 +4,11 @@
 //
 
 #include "InstanceNormalizationEndToEndTestImpl.hpp"
-
-#include "DataLayoutIndexed.hpp"
 #include "EndToEndTestImpl.hpp"
 #include "ResolveType.hpp"
 
-#include <Permute.hpp>
+#include <armnnUtils/Permute.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 #include <armnn/INetwork.hpp>
 
@@ -377,4 +376,4 @@ void InstanceNormalizationNchwEndToEndTest2(const std::vector<armnn::BackendId>&
                                   gamma,
                                   beta,
                                   eps);
-}
\ No newline at end of file
+}
diff --git a/src/backends/backendsCommon/test/ResizeEndToEndTestImpl.hpp b/src/backends/backendsCommon/test/ResizeEndToEndTestImpl.hpp
index 1eeb944805..8cc3a26bd8 100644
--- a/src/backends/backendsCommon/test/ResizeEndToEndTestImpl.hpp
+++ b/src/backends/backendsCommon/test/ResizeEndToEndTestImpl.hpp
@@ -8,7 +8,8 @@
 
 #include <armnn/ArmNN.hpp>
 
-#include <Permute.hpp>
+#include <armnnUtils/Permute.hpp>
+
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
diff --git a/src/backends/backendsCommon/test/SpaceToDepthEndToEndTestImpl.cpp b/src/backends/backendsCommon/test/SpaceToDepthEndToEndTestImpl.cpp
index 8eb1c97350..6d1a7b0556 100644
--- a/src/backends/backendsCommon/test/SpaceToDepthEndToEndTestImpl.cpp
+++ b/src/backends/backendsCommon/test/SpaceToDepthEndToEndTestImpl.cpp
@@ -4,15 +4,14 @@
 //
 
 #include "SpaceToDepthEndToEndTestImpl.hpp"
-
 #include "ResolveType.hpp"
-#include "DataLayoutIndexed.hpp"
 #include "EndToEndTestImpl.hpp"
 
-#include <Permute.hpp>
-
 #include <armnn/INetwork.hpp>
 
+#include <armnnUtils/Permute.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/test/DataLayoutUtils.hpp>
 
 #include <test/TestUtils.hpp>
diff --git a/src/backends/backendsCommon/test/TransposeConvolution2dEndToEndTestImpl.hpp b/src/backends/backendsCommon/test/TransposeConvolution2dEndToEndTestImpl.hpp
index 4935a18b51..970c44d0d9 100644
--- a/src/backends/backendsCommon/test/TransposeConvolution2dEndToEndTestImpl.hpp
+++ b/src/backends/backendsCommon/test/TransposeConvolution2dEndToEndTestImpl.hpp
@@ -8,7 +8,8 @@
 
 #include <armnn/ArmNN.hpp>
 
-#include <Permute.hpp>
+#include <armnnUtils/Permute.hpp>
+
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
diff --git a/src/backends/backendsCommon/test/layerTests/BatchNormalizationTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/BatchNormalizationTestImpl.cpp
index 68cda7ccf3..a92ff03753 100644
--- a/src/backends/backendsCommon/test/layerTests/BatchNormalizationTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/BatchNormalizationTestImpl.cpp
@@ -5,12 +5,13 @@
 
 #include "BatchNormalizationTestImpl.hpp"
 
-#include <DataLayoutIndexed.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/IBackendInternal.hpp>
 #include <backendsCommon/WorkloadFactory.hpp>
diff --git a/src/backends/backendsCommon/test/layerTests/ConcatTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/ConcatTestImpl.cpp
index e9932c81ef..53bfc20590 100644
--- a/src/backends/backendsCommon/test/layerTests/ConcatTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/ConcatTestImpl.cpp
@@ -5,12 +5,13 @@
 
 #include "ConcatTestImpl.hpp"
 
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/test/TensorCopyUtils.hpp>
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
 
diff --git a/src/backends/backendsCommon/test/layerTests/ConstantTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/ConstantTestImpl.cpp
index 3f22c31c7d..17908197c0 100644
--- a/src/backends/backendsCommon/test/layerTests/ConstantTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/ConstantTestImpl.cpp
@@ -5,12 +5,13 @@
 
 #include "ConstantTestImpl.hpp"
 
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 
 #include <backendsCommon/test/TensorCopyUtils.hpp>
diff --git a/src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.cpp
index 22e7e29db7..e825fffaea 100644
--- a/src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.cpp
@@ -5,13 +5,14 @@
 
 #include "Conv2dTestImpl.hpp"
 
-#include <DataLayoutIndexed.hpp>
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
-#include <TensorUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 
 #include <backendsCommon/test/DataLayoutUtils.hpp>
diff --git a/src/backends/backendsCommon/test/layerTests/L2NormalizationTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/L2NormalizationTestImpl.cpp
index 569f5af227..d67f7b64aa 100644
--- a/src/backends/backendsCommon/test/layerTests/L2NormalizationTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/L2NormalizationTestImpl.cpp
@@ -5,10 +5,11 @@
 
 #include "L2NormalizationTestImpl.hpp"
 
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
-#include <TensorUtils.hpp>
+
+#include <armnnUtils/TensorUtils.hpp>
+#include <armnnUtils/Permute.hpp>
 
 #include <backendsCommon/test/TensorCopyUtils.hpp>
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
diff --git a/src/backends/backendsCommon/test/layerTests/Pooling2dTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/Pooling2dTestImpl.cpp
index a7fbfefcb0..2012dfd602 100644
--- a/src/backends/backendsCommon/test/layerTests/Pooling2dTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/Pooling2dTestImpl.cpp
@@ -5,13 +5,14 @@
 
 #include "Pooling2dTestImpl.hpp"
 
-#include <armnn/LayerSupport.hpp>
-
-#include <DataLayoutIndexed.hpp>
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
-#include <TensorUtils.hpp>
+
+#include <armnn/LayerSupport.hpp>
+
+#include <armnnUtils/TensorUtils.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+#include <armnnUtils/Permute.hpp>
 
 #include <backendsCommon/WorkloadInfo.hpp>
 
diff --git a/src/backends/backendsCommon/test/layerTests/ResizeTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/ResizeTestImpl.cpp
index 895afe06af..198c60a9ad 100644
--- a/src/backends/backendsCommon/test/layerTests/ResizeTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/ResizeTestImpl.cpp
@@ -5,13 +5,14 @@
 
 #include "ResizeTestImpl.hpp"
 
-#include <DataLayoutIndexed.hpp>
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
-#include <TensorUtils.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/TensorUtils.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/test/DataLayoutUtils.hpp>
 #include <backendsCommon/test/TensorCopyUtils.hpp>
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
diff --git a/src/backends/backendsCommon/test/layerTests/SpaceToBatchNdTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/SpaceToBatchNdTestImpl.cpp
index f815604aaa..ea2130d325 100644
--- a/src/backends/backendsCommon/test/layerTests/SpaceToBatchNdTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/SpaceToBatchNdTestImpl.cpp
@@ -5,12 +5,13 @@
 
 #include "SpaceToBatchNdTestImpl.hpp"
 
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/test/TensorCopyUtils.hpp>
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
 
diff --git a/src/backends/backendsCommon/test/layerTests/SpaceToDepthTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/SpaceToDepthTestImpl.cpp
index 05413230dd..c4b11a7213 100644
--- a/src/backends/backendsCommon/test/layerTests/SpaceToDepthTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/SpaceToDepthTestImpl.cpp
@@ -5,12 +5,13 @@
 
 #include "SpaceToDepthTestImpl.hpp"
 
-#include <Permute.hpp>
 #include <QuantizeHelper.hpp>
 #include <ResolveType.hpp>
 
 #include <armnn/ArmNN.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <backendsCommon/test/TensorCopyUtils.hpp>
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
 
diff --git a/src/backends/backendsCommon/test/layerTests/TransposeConvolution2dTestImpl.cpp b/src/backends/backendsCommon/test/layerTests/TransposeConvolution2dTestImpl.cpp
index 53c1e9d1b7..71b08aa159 100644
--- a/src/backends/backendsCommon/test/layerTests/TransposeConvolution2dTestImpl.cpp
+++ b/src/backends/backendsCommon/test/layerTests/TransposeConvolution2dTestImpl.cpp
@@ -5,10 +5,11 @@
 
 #include "TransposeConvolution2dTestImpl.hpp"
 
+#include <QuantizeHelper.hpp>
+
 #include <armnn/ArmNN.hpp>
 
-#include <Permute.hpp>
-#include <QuantizeHelper.hpp>
+#include <armnnUtils/Permute.hpp>
 
 #include <backendsCommon/CpuTensorHandle.hpp>
 
diff --git a/src/backends/cl/workloads/ClArgMinMaxWorkload.cpp b/src/backends/cl/workloads/ClArgMinMaxWorkload.cpp
index 3270b0a2f3..a79a7b286d 100644
--- a/src/backends/cl/workloads/ClArgMinMaxWorkload.cpp
+++ b/src/backends/cl/workloads/ClArgMinMaxWorkload.cpp
@@ -10,7 +10,7 @@
 
 #include <backendsCommon/CpuTensorHandle.hpp>
 
-#include <TensorUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <cl/ClTensorHandle.hpp>
 #include <cl/ClLayerSupport.hpp>
diff --git a/src/backends/neon/workloads/NeonArgMinMaxWorkload.cpp b/src/backends/neon/workloads/NeonArgMinMaxWorkload.cpp
index 4b43052365..0fa9d43b15 100644
--- a/src/backends/neon/workloads/NeonArgMinMaxWorkload.cpp
+++ b/src/backends/neon/workloads/NeonArgMinMaxWorkload.cpp
@@ -7,8 +7,10 @@
 #include "NeonWorkloadUtils.hpp"
 
 #include <aclCommon/ArmComputeTensorUtils.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
-#include <TensorUtils.hpp>
+
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <arm_compute/runtime/NEON/functions/NEArgMinMaxLayer.h>
 
diff --git a/src/backends/neon/workloads/NeonConvertFp16ToFp32Workload.cpp b/src/backends/neon/workloads/NeonConvertFp16ToFp32Workload.cpp
index a86b1086dc..01f09a56f6 100644
--- a/src/backends/neon/workloads/NeonConvertFp16ToFp32Workload.cpp
+++ b/src/backends/neon/workloads/NeonConvertFp16ToFp32Workload.cpp
@@ -4,9 +4,11 @@
 //
 
 #include "NeonConvertFp16ToFp32Workload.hpp"
-#include <FloatingPointConverter.hpp>
+
+#include <armnnUtils/FloatingPointConverter.hpp>
 
 #include <Half.hpp>
+
 #include <backendsCommon/WorkloadUtils.hpp>
 
 namespace armnn
diff --git a/src/backends/neon/workloads/NeonConvertFp32ToFp16Workload.cpp b/src/backends/neon/workloads/NeonConvertFp32ToFp16Workload.cpp
index 4b6f97bb78..2205b63da1 100644
--- a/src/backends/neon/workloads/NeonConvertFp32ToFp16Workload.cpp
+++ b/src/backends/neon/workloads/NeonConvertFp32ToFp16Workload.cpp
@@ -6,8 +6,10 @@
 #include "NeonConvertFp32ToFp16Workload.hpp"
 
 #include <Half.hpp>
-#include <FloatingPointConverter.hpp>
 #include <Profiling.hpp>
+
+#include <armnnUtils/FloatingPointConverter.hpp>
+
 #include <backendsCommon/WorkloadUtils.hpp>
 
 namespace armnn
diff --git a/src/backends/neon/workloads/NeonDepthwiseConvolutionWorkload.cpp b/src/backends/neon/workloads/NeonDepthwiseConvolutionWorkload.cpp
index f039a6bdbe..e39fe54199 100644
--- a/src/backends/neon/workloads/NeonDepthwiseConvolutionWorkload.cpp
+++ b/src/backends/neon/workloads/NeonDepthwiseConvolutionWorkload.cpp
@@ -7,9 +7,12 @@
 
 #include "NeonWorkloadUtils.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <aclCommon/ArmComputeTensorUtils.hpp>
+
 #include <neon/NeonLayerSupport.hpp>
+
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <backendsCommon/WorkloadUtils.hpp>
 
diff --git a/src/backends/reference/RefLayerSupport.cpp b/src/backends/reference/RefLayerSupport.cpp
index 5a84d8ac78..05684dcbc0 100644
--- a/src/backends/reference/RefLayerSupport.cpp
+++ b/src/backends/reference/RefLayerSupport.cpp
@@ -6,17 +6,20 @@
 #include "RefLayerSupport.hpp"
 #include "RefBackendId.hpp"
 
-#include <DataLayoutIndexed.hpp>
-#include <InternalTypes.hpp>
-#include <LayerSupportCommon.hpp>
-
 #include <armnn/Types.hpp>
 #include <armnn/Descriptors.hpp>
 #include <armnn/BackendRegistry.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
+#include <InternalTypes.hpp>
+#include <LayerSupportCommon.hpp>
+
 #include <backendsCommon/LayerSupportRules.hpp>
+
 #include <backendsCommon/test/WorkloadTestUtils.hpp>
 
+#include <boost/cast.hpp>
 #include <boost/core/ignore_unused.hpp>
 
 #include <vector>
diff --git a/src/backends/reference/workloads/ArgMinMax.cpp b/src/backends/reference/workloads/ArgMinMax.cpp
index 2687a4e8ac..76616f1465 100644
--- a/src/backends/reference/workloads/ArgMinMax.cpp
+++ b/src/backends/reference/workloads/ArgMinMax.cpp
@@ -5,7 +5,7 @@
 
 #include "ArgMinMax.hpp"
 
-#include <TensorUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <boost/numeric/conversion/cast.hpp>
 
diff --git a/src/backends/reference/workloads/BaseIterator.hpp b/src/backends/reference/workloads/BaseIterator.hpp
index 95a31fbdd6..ca5110c2fd 100644
--- a/src/backends/reference/workloads/BaseIterator.hpp
+++ b/src/backends/reference/workloads/BaseIterator.hpp
@@ -5,9 +5,10 @@
 
 #pragma once
 
-#include "FloatingPointConverter.hpp"
-
 #include <armnn/ArmNN.hpp>
+
+#include <armnnUtils/FloatingPointConverter.hpp>
+
 #include <ResolveType.hpp>
 
 #include <boost/assert.hpp>
diff --git a/src/backends/reference/workloads/BatchNormImpl.cpp b/src/backends/reference/workloads/BatchNormImpl.cpp
index b80af8c937..e742c7280f 100644
--- a/src/backends/reference/workloads/BatchNormImpl.cpp
+++ b/src/backends/reference/workloads/BatchNormImpl.cpp
@@ -8,7 +8,7 @@
 
 #include <armnn/Tensor.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 #include <cmath>
 
diff --git a/src/backends/reference/workloads/BatchToSpaceNd.hpp b/src/backends/reference/workloads/BatchToSpaceNd.hpp
index b757d3709c..a375aaae52 100644
--- a/src/backends/reference/workloads/BatchToSpaceNd.hpp
+++ b/src/backends/reference/workloads/BatchToSpaceNd.hpp
@@ -5,15 +5,16 @@
 
 #pragma once
 
-#include <armnn/Types.hpp>
-
-#include <backendsCommon/Workload.hpp>
-#include <backendsCommon/WorkloadData.hpp>
 #include "BaseIterator.hpp"
 #include "Decoders.hpp"
 #include "Encoders.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnn/Types.hpp>
+
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
+#include <backendsCommon/Workload.hpp>
+#include <backendsCommon/WorkloadData.hpp>
 
 namespace armnn
 {
diff --git a/src/backends/reference/workloads/ConvImpl.hpp b/src/backends/reference/workloads/ConvImpl.hpp
index 7dba760d87..562fd3e296 100644
--- a/src/backends/reference/workloads/ConvImpl.hpp
+++ b/src/backends/reference/workloads/ConvImpl.hpp
@@ -13,13 +13,11 @@
 
 #include <armnn/Tensor.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 #include <boost/assert.hpp>
 #include <boost/numeric/conversion/cast.hpp>
 
-#include <DataLayoutIndexed.hpp>
-
 #include <cmath>
 #include <limits>
 
diff --git a/src/backends/reference/workloads/Decoders.hpp b/src/backends/reference/workloads/Decoders.hpp
index dcd498cb63..b9cd7f9573 100644
--- a/src/backends/reference/workloads/Decoders.hpp
+++ b/src/backends/reference/workloads/Decoders.hpp
@@ -6,8 +6,9 @@
 #pragma once
 
 #include "BaseIterator.hpp"
-#include "FloatingPointConverter.hpp"
-#include "TensorUtils.hpp"
+
+#include <armnnUtils/FloatingPointConverter.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <boost/assert.hpp>
 
diff --git a/src/backends/reference/workloads/DepthToSpace.cpp b/src/backends/reference/workloads/DepthToSpace.cpp
index d500e9b100..91ca160ae2 100644
--- a/src/backends/reference/workloads/DepthToSpace.cpp
+++ b/src/backends/reference/workloads/DepthToSpace.cpp
@@ -5,8 +5,8 @@
 
 #include "DepthToSpace.hpp"
 
-#include <DataLayoutIndexed.hpp>
-#include <Permute.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+#include <armnnUtils/Permute.hpp>
 
 #include <boost/assert.hpp>
 
diff --git a/src/backends/reference/workloads/Encoders.hpp b/src/backends/reference/workloads/Encoders.hpp
index 5c0cffa7ca..0d578d68de 100644
--- a/src/backends/reference/workloads/Encoders.hpp
+++ b/src/backends/reference/workloads/Encoders.hpp
@@ -6,7 +6,8 @@
 #pragma once
 
 #include "BaseIterator.hpp"
-#include "TensorUtils.hpp"
+
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <boost/assert.hpp>
 
diff --git a/src/backends/reference/workloads/InstanceNorm.cpp b/src/backends/reference/workloads/InstanceNorm.cpp
index 9d6532fa6e..08c555f0e8 100644
--- a/src/backends/reference/workloads/InstanceNorm.cpp
+++ b/src/backends/reference/workloads/InstanceNorm.cpp
@@ -8,7 +8,7 @@
 
 #include <armnn/Tensor.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 #include <cmath>
 
diff --git a/src/backends/reference/workloads/LogSoftmax.cpp b/src/backends/reference/workloads/LogSoftmax.cpp
index 3fa3dc0d8c..ddf5674fb8 100644
--- a/src/backends/reference/workloads/LogSoftmax.cpp
+++ b/src/backends/reference/workloads/LogSoftmax.cpp
@@ -5,7 +5,7 @@
 
 #include "LogSoftmax.hpp"
 
-#include <TensorUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <cmath>
 
diff --git a/src/backends/reference/workloads/Pooling2d.cpp b/src/backends/reference/workloads/Pooling2d.cpp
index cf83f8ce2b..ea8f4ee7e7 100644
--- a/src/backends/reference/workloads/Pooling2d.cpp
+++ b/src/backends/reference/workloads/Pooling2d.cpp
@@ -4,11 +4,12 @@
 //
 
 #include "Pooling2d.hpp"
-#include "DataLayoutIndexed.hpp"
 
 #include <armnn/Exceptions.hpp>
 #include <armnn/Types.hpp>
 
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <boost/numeric/conversion/cast.hpp>
 
 #include <limits>
diff --git a/src/backends/reference/workloads/RefConvertFp16ToFp32Workload.cpp b/src/backends/reference/workloads/RefConvertFp16ToFp32Workload.cpp
index 886e77a31b..ef813eb69b 100644
--- a/src/backends/reference/workloads/RefConvertFp16ToFp32Workload.cpp
+++ b/src/backends/reference/workloads/RefConvertFp16ToFp32Workload.cpp
@@ -4,9 +4,9 @@
 //
 
 #include "RefConvertFp16ToFp32Workload.hpp"
-
 #include "RefWorkloadUtils.hpp"
-#include "FloatingPointConverter.hpp"
+
+#include <armnnUtils/FloatingPointConverter.hpp>
 
 #include <Half.hpp>
 
diff --git a/src/backends/reference/workloads/RefConvertFp32ToFp16Workload.cpp b/src/backends/reference/workloads/RefConvertFp32ToFp16Workload.cpp
index 33270ad10f..559901f2f2 100644
--- a/src/backends/reference/workloads/RefConvertFp32ToFp16Workload.cpp
+++ b/src/backends/reference/workloads/RefConvertFp32ToFp16Workload.cpp
@@ -4,12 +4,12 @@
 //
 
 #include "RefConvertFp32ToFp16Workload.hpp"
-
-#include "FloatingPointConverter.hpp"
 #include "RefWorkloadUtils.hpp"
 #include "Profiling.hpp"
 
-#include "Half.hpp"
+#include <armnnUtils/FloatingPointConverter.hpp>
+
+#include <Half.hpp>
 
 namespace armnn
 {
diff --git a/src/backends/reference/workloads/RefL2NormalizationWorkload.cpp b/src/backends/reference/workloads/RefL2NormalizationWorkload.cpp
index 3764b9a49a..6fec1abe6f 100644
--- a/src/backends/reference/workloads/RefL2NormalizationWorkload.cpp
+++ b/src/backends/reference/workloads/RefL2NormalizationWorkload.cpp
@@ -4,13 +4,13 @@
 //
 
 #include "RefL2NormalizationWorkload.hpp"
-
 #include "RefWorkloadUtils.hpp"
 #include "Decoders.hpp"
 #include "Encoders.hpp"
-#include "DataLayoutIndexed.hpp"
 
-#include "Profiling.hpp"
+#include <Profiling.hpp>
+
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 #include <boost/numeric/conversion/cast.hpp>
 
@@ -21,80 +21,80 @@ using namespace armnnUtils;
 namespace armnn
 {
 RefL2NormalizationWorkload::RefL2NormalizationWorkload(
-            const L2NormalizationQueueDescriptor& descriptor,
-            const WorkloadInfo& info)
-            : BaseWorkload<L2NormalizationQueueDescriptor>(descriptor, info) {}
+        const L2NormalizationQueueDescriptor& descriptor,
+        const WorkloadInfo& info)
+    : BaseWorkload<L2NormalizationQueueDescriptor>(descriptor, info) {}
 
-    void RefL2NormalizationWorkload::Execute() const
-    {
-        ARMNN_SCOPED_PROFILING_EVENT(Compute::CpuRef, "RefL2NormalizationWorkload_Execute");
+void RefL2NormalizationWorkload::Execute() const
+{
+    ARMNN_SCOPED_PROFILING_EVENT(Compute::CpuRef, "RefL2NormalizationWorkload_Execute");
 
-        const TensorInfo& inputInfo = GetTensorInfo(m_Data.m_Inputs[0]);
-        const TensorInfo& outputInfo = GetTensorInfo(m_Data.m_Outputs[0]);
+    const TensorInfo& inputInfo = GetTensorInfo(m_Data.m_Inputs[0]);
+    const TensorInfo& outputInfo = GetTensorInfo(m_Data.m_Outputs[0]);
 
-        auto inputDecoder  = MakeDecoder<float>(inputInfo, m_Data.m_Inputs[0]->Map());
-        auto outputEncoder = MakeEncoder<float>(outputInfo, m_Data.m_Outputs[0]->Map());
+    auto inputDecoder  = MakeDecoder<float>(inputInfo, m_Data.m_Inputs[0]->Map());
+    auto outputEncoder = MakeEncoder<float>(outputInfo, m_Data.m_Outputs[0]->Map());
 
-        DataLayoutIndexed dataLayout(m_Data.m_Parameters.m_DataLayout);
+    DataLayoutIndexed dataLayout(m_Data.m_Parameters.m_DataLayout);
 
-        const TensorShape& shape = inputInfo.GetShape();
-        unsigned int paddedShapeArray[4];
-        const int idxShift = 4 - boost::numeric_cast<int>(shape.GetNumDimensions());
+    const TensorShape& shape = inputInfo.GetShape();
+    unsigned int paddedShapeArray[4];
+    const int idxShift = 4 - boost::numeric_cast<int>(shape.GetNumDimensions());
 
-        const unsigned int batches = (idxShift == 0) ? shape[0] : 1;
-        paddedShapeArray[0] = batches;
+    const unsigned int batches = (idxShift == 0) ? shape[0] : 1;
+    paddedShapeArray[0] = batches;
 
-        const int channelsIdx = boost::numeric_cast<int>(dataLayout.GetChannelsIndex());
-        const unsigned int channels = (channelsIdx - idxShift >= 0)
-                                      ? shape[boost::numeric_cast<unsigned int>(channelsIdx - idxShift)]
-                                      : 1;
-        paddedShapeArray[channelsIdx] = channels;
+    const int channelsIdx = boost::numeric_cast<int>(dataLayout.GetChannelsIndex());
+    const unsigned int channels = (channelsIdx - idxShift >= 0)
+                                  ? shape[boost::numeric_cast<unsigned int>(channelsIdx - idxShift)]
+                                  : 1;
+    paddedShapeArray[channelsIdx] = channels;
 
-        const int heightIdx = boost::numeric_cast<int>(dataLayout.GetHeightIndex());
-        const unsigned int height = (heightIdx - idxShift >= 0)
-                                    ? shape[boost::numeric_cast<unsigned int>(heightIdx - idxShift)]
-                                    : 1;
-        paddedShapeArray[heightIdx] = height;
+    const int heightIdx = boost::numeric_cast<int>(dataLayout.GetHeightIndex());
+    const unsigned int height = (heightIdx - idxShift >= 0)
+                                ? shape[boost::numeric_cast<unsigned int>(heightIdx - idxShift)]
+                                : 1;
+    paddedShapeArray[heightIdx] = height;
 
-        const int widthIdx = boost::numeric_cast<int>(dataLayout.GetWidthIndex());
-        const unsigned int width = (widthIdx - idxShift >= 0)
-                                   ? shape[boost::numeric_cast<unsigned int>(widthIdx - idxShift)]
-                                   : 1;
-        paddedShapeArray[widthIdx] = width;
+    const int widthIdx = boost::numeric_cast<int>(dataLayout.GetWidthIndex());
+    const unsigned int width = (widthIdx - idxShift >= 0)
+                               ? shape[boost::numeric_cast<unsigned int>(widthIdx - idxShift)]
+                               : 1;
+    paddedShapeArray[widthIdx] = width;
 
-        const TensorShape& paddedShape = TensorShape(4, paddedShapeArray);
+    const TensorShape& paddedShape = TensorShape(4, paddedShapeArray);
 
-        for (unsigned int n = 0; n < batches; ++n)
+    for (unsigned int n = 0; n < batches; ++n)
+    {
+        for (unsigned int c = 0; c < channels; ++c)
         {
-            for (unsigned int c = 0; c < channels; ++c)
+            for (unsigned int h = 0; h < height; ++h)
             {
-                for (unsigned int h = 0; h < height; ++h)
+                for (unsigned int w = 0; w < width; ++w)
                 {
-                    for (unsigned int w = 0; w < width; ++w)
+                    float reduction = 0.0;
+                    for (unsigned int d = 0; d < channels; ++d)
                     {
-                        float reduction = 0.0;
-                        for (unsigned int d = 0; d < channels; ++d)
-                        {
-                            unsigned int inputIndex = dataLayout.GetIndex(paddedShape, n, d, h, w);
+                        unsigned int inputIndex = dataLayout.GetIndex(paddedShape, n, d, h, w);
 
-                            (*inputDecoder)[inputIndex];
-                            const float value = inputDecoder->Get();
-                            reduction += value * value;
-                        }
+                        (*inputDecoder)[inputIndex];
+                        const float value = inputDecoder->Get();
+                        reduction += value * value;
+                    }
 
-                        unsigned int index = dataLayout.GetIndex(paddedShape, n, c, h, w);
+                    unsigned int index = dataLayout.GetIndex(paddedShape, n, c, h, w);
 
-                        float maximum = reduction < m_Data.m_Parameters.m_Eps ? m_Data.m_Parameters.m_Eps : reduction;
+                    float maximum = reduction < m_Data.m_Parameters.m_Eps ? m_Data.m_Parameters.m_Eps : reduction;
 
-                        const float scale = 1.0f / sqrtf(maximum);
+                    const float scale = 1.0f / sqrtf(maximum);
 
-                        (*inputDecoder)[index];
-                        (*outputEncoder)[index];
-                        outputEncoder->Set(inputDecoder->Get() * scale);
-                    }
+                    (*inputDecoder)[index];
+                    (*outputEncoder)[index];
+                    outputEncoder->Set(inputDecoder->Get() * scale);
                 }
             }
         }
     }
+}
 
 } //namespace armnn
diff --git a/src/backends/reference/workloads/RefNormalizationWorkload.cpp b/src/backends/reference/workloads/RefNormalizationWorkload.cpp
index 8ff2d9cf92..0427baf475 100644
--- a/src/backends/reference/workloads/RefNormalizationWorkload.cpp
+++ b/src/backends/reference/workloads/RefNormalizationWorkload.cpp
@@ -4,14 +4,14 @@
 //
 
 #include "RefNormalizationWorkload.hpp"
-
 #include "RefWorkloadUtils.hpp"
 #include "Decoders.hpp"
 #include "Encoders.hpp"
 
 #include <armnn/Tensor.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
+
 #include <Profiling.hpp>
 
 #include <boost/log/trivial.hpp>
diff --git a/src/backends/reference/workloads/RefPermuteWorkload.cpp b/src/backends/reference/workloads/RefPermuteWorkload.cpp
index 4d43b7e560..4e7b76bf0a 100644
--- a/src/backends/reference/workloads/RefPermuteWorkload.cpp
+++ b/src/backends/reference/workloads/RefPermuteWorkload.cpp
@@ -6,7 +6,8 @@
 #include "RefPermuteWorkload.hpp"
 #include "RefWorkloadUtils.hpp"
 
-#include <Permute.hpp>
+#include <armnnUtils/Permute.hpp>
+
 #include <ResolveType.hpp>
 
 namespace armnn
diff --git a/src/backends/reference/workloads/Resize.hpp b/src/backends/reference/workloads/Resize.hpp
index 8bd8999e5d..4c357946d9 100644
--- a/src/backends/reference/workloads/Resize.hpp
+++ b/src/backends/reference/workloads/Resize.hpp
@@ -6,9 +6,10 @@
 #pragma once
 
 #include "BaseIterator.hpp"
+
 #include <armnn/Tensor.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 namespace armnn
 {
@@ -20,4 +21,4 @@ void Resize(Decoder<float>&               in,
             armnnUtils::DataLayoutIndexed dataLayout = DataLayout::NCHW,
             ResizeMethod                  resizeMethod = ResizeMethod::NearestNeighbor);
 
-} //namespace armnn
+} // namespace armnn
diff --git a/src/backends/reference/workloads/Softmax.cpp b/src/backends/reference/workloads/Softmax.cpp
index f745d816c2..5036389a10 100644
--- a/src/backends/reference/workloads/Softmax.cpp
+++ b/src/backends/reference/workloads/Softmax.cpp
@@ -5,7 +5,7 @@
 
 #include "Softmax.hpp"
 
-#include <TensorUtils.hpp>
+#include <armnnUtils/TensorUtils.hpp>
 
 #include <cmath>
 #include <vector>
diff --git a/src/backends/reference/workloads/SpaceToBatchNd.cpp b/src/backends/reference/workloads/SpaceToBatchNd.cpp
index 0bc2396973..b6bab17367 100644
--- a/src/backends/reference/workloads/SpaceToBatchNd.cpp
+++ b/src/backends/reference/workloads/SpaceToBatchNd.cpp
@@ -5,7 +5,7 @@
 
 #include "SpaceToBatchNd.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 using namespace armnnUtils;
 
diff --git a/src/backends/reference/workloads/SpaceToDepth.cpp b/src/backends/reference/workloads/SpaceToDepth.cpp
index 4a4f4183d9..604a9051af 100644
--- a/src/backends/reference/workloads/SpaceToDepth.cpp
+++ b/src/backends/reference/workloads/SpaceToDepth.cpp
@@ -5,7 +5,7 @@
 
 #include "SpaceToDepth.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 using namespace armnnUtils;
 
diff --git a/src/backends/reference/workloads/TensorBufferArrayView.hpp b/src/backends/reference/workloads/TensorBufferArrayView.hpp
index c06407241d..e03c42fe60 100644
--- a/src/backends/reference/workloads/TensorBufferArrayView.hpp
+++ b/src/backends/reference/workloads/TensorBufferArrayView.hpp
@@ -7,9 +7,9 @@
 
 #include <armnn/Tensor.hpp>
 
-#include <boost/assert.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
-#include <DataLayoutIndexed.hpp>
+#include <boost/assert.hpp>
 
 namespace armnn
 {
diff --git a/src/backends/reference/workloads/TransposeConvolution2d.cpp b/src/backends/reference/workloads/TransposeConvolution2d.cpp
index 5662c58809..5698014181 100644
--- a/src/backends/reference/workloads/TransposeConvolution2d.cpp
+++ b/src/backends/reference/workloads/TransposeConvolution2d.cpp
@@ -5,7 +5,7 @@
 
 #include "TransposeConvolution2d.hpp"
 
-#include <DataLayoutIndexed.hpp>
+#include <armnnUtils/DataLayoutIndexed.hpp>
 
 namespace armnn
 {
diff --git a/tests/ImagePreprocessor.cpp b/tests/ImagePreprocessor.cpp
index 74bc943ee8..f0184e466e 100644
--- a/tests/ImagePreprocessor.cpp
+++ b/tests/ImagePreprocessor.cpp
@@ -2,11 +2,14 @@
 // Copyright © 2017 Arm Ltd. All rights reserved.
 // SPDX-License-Identifier: MIT
 //
+
 #include "InferenceTestImage.hpp"
 #include "ImagePreprocessor.hpp"
-#include "Permute.hpp"
+
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <boost/numeric/conversion/cast.hpp>
 #include <boost/assert.hpp>
 #include <boost/format.hpp>
diff --git a/tests/ImageTensorGenerator/ImageTensorGenerator.hpp b/tests/ImageTensorGenerator/ImageTensorGenerator.hpp
index 3e164bc9e0..b1cb5e36f5 100644
--- a/tests/ImageTensorGenerator/ImageTensorGenerator.hpp
+++ b/tests/ImageTensorGenerator/ImageTensorGenerator.hpp
@@ -4,9 +4,11 @@
 //
 
 #include "../InferenceTestImage.hpp"
-#include "Permute.hpp"
+
 #include <armnn/TypesUtils.hpp>
 
+#include <armnnUtils/Permute.hpp>
+
 #include <algorithm>
 #include <fstream>
 #include <iterator>
@@ -178,4 +180,4 @@ template <>
 void WriteImageTensorImpl<uint8_t>(const std::vector<uint8_t>& imageData, std::ofstream& imageTensorFile)
 {
     std::copy(imageData.begin(), imageData.end(), std::ostream_iterator<int>(imageTensorFile, " "));
-}
\ No newline at end of file
+}