Correct TuningLevel enumeration values

According to the header description - include/armnn/IRuntime.hpp, Each level should have below value.
-----------------------------------------------------------------------------------------------------
"TuningLevel" : int [0..3] (0=UseOnly(default) | 1=RapidTuning | 2=NormalTuning | 3=ExhaustiveTuning)
-----------------------------------------------------------------------------------------------------

And also src/backends/cl/ClBackendContext.cpp uses below enumeration values internally
for the tuning level,
----------------------
enum class TuningLevel
{
    None,
    Rapid,
    Normal,
    Exhaustive
};
----------------------

So this patch corrects TuningLevel enumeration values - which is exposed to user - to be consistent
with ones internally used.

This is upstream patch,
 https://review.mlplatform.org/c/ml/armnn/+/5269

Change-Id: I9a8aeea0115579bfd16cbe01f39183b96329cdfd
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: Fix aarch64 build error

Change-Id: Ica83e9c26bdd588915710ae1efb80f7d3b93add0
Signed-off-by: Inki Dae <inki.dae@samsung.com>

tensorflow: update tflite schema files

Change-Id: Id324bd8939376cd5d78fc8d8afe8ece6b1bed347
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: use internal schema file and copy manifest file

Change-Id: I078dd3bc37737eb6ca57eb94e9485b5e9b77fdc1
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: add xxd binary

Change-Id: I1f7bf666b1e156405ad75c7962c1157b6bdaefff
Signed-off-by: Inki Dae <inki.dae@samsung.com>

Use Tizen specific TF_LITE_SCHEMA_PATH

The schema file, schema.fbs, will be installed in /usr/share/...
in case of Tizen so use the specific TF_LITE_SCHEMA_PATH
so that ARMNN can find schema.fbs in given path.

Change-Id: Idd9b23000d18995fc5df4a45b0510c47231ae0c2
Signed-off-by: Inki Dae <inki.dae@samsung.com>

Fix pc file

Change-Id: I57de41dcc1eab8867d2da23ec82527e6d6cac00e
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: Install xxd binary

ARMNN needs xxd binary to dump a given file with hex type
when compiling.

Change-Id: I0802c3cd5749173d8856781b98573e8585add862
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: modify version to 21.02

Change-Id: Ica624cba77373eac73acb33f2dca311fed603f54
Signed-off-by: Inki Dae <inki.dae@samsung.com>

fix build error

Change-Id: I2df41ef0d73282e477948eb0f57150b9f2727087
Signed-off-by: Inki Dae <inki.dae@samsung.com>

Dist/Tizen: support selective framework and pkgconfig

Add package config for armnn package
Add selective enabling of framework in build
Add flatc while building with flatbuffers
Add tflite header files

Change-Id: I77eecd22cf80eb4c5ccfb2a83399354869dac39b
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>

Dist/Tizen: remove obsolete pkg names

There is no one using that name.
Remove the obsolete name.

Change-Id: Ice9cc4fcace5fc6d08d7db396fee976f3615c467
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

Dist/Tizen: spec file clean-up

1. Do not use "release" postfix for the main library.
2. Do not use -n %{name} prefix. Drop that.
3. Added missing interanl dependencies

Change-Id: Ibb617be8206df149cd2849deef4fbc859484ebfc
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

packaging: enable SHARED_BOOST flag

Change-Id: I444982b9b171a4b341b73bd8085255109c0c2a27
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: clean up spec file

Change-Id: If92810d16ceb38786fed8e66d8c03242ea6188f6
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: install required packages

Change-Id: I8e0677c0afc649fadc2299a52b75316c255e1bb7
Signed-off-by: Inki Dae <inki.dae@samsung.com>

update tensorflow lite schema files

Change-Id: Ibc6639cf2c0ebe9c8d9ad09c8967a2286739d17a
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: update spec file for acl v19.11 support

This patch updates the spec file for acl v19.11 support
and drops TF parser support because it needs latest protobuf,
>= 3.7.1

Change-Id: Ic067b032790a3b60fa6b359bead66c67c7bbde09
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: correct armnn version

Change-Id: If7d441f87cd893866800b2db12052d75a2e6ff8e
Signed-off-by: Inki Dae <inki.dae@samsung.com>

add tensorflow lite parser support

Change-Id: I26b101415ed11a89f8bffdb54e635df74e21110d
Signed-off-by: Inki Dae <inki.dae@samsung.com>

add TF parser support

Change-Id: Iba09005cb6249f7411319fc7265bad500ceb57e2
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: add UnitTests app to rpm package

Change-Id: Ifcc58be6e83cc4478227fab1b1cf9df849dc7dae
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packing: keep same package name for arm and aarch64

Change-Id: I2330fe154b500931462a438405093270600e873b
Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: do not pack debug files

Change-Id: I55aec155f624f19a5c9f64f72a671e0026084103
Signed-off-by: Inki Dae <inki.dae@samsung.com>

Fix a build error

Fixed build error: Found '/home/abuild/rpmbuild/BUILDROOT/armnn-18.05-0.arm' in installed files; aborting
refined this patch
Change-Id: I91534256de2144d86b37d42f85bb367643348bce
Signed-off-by : biao716.wang <biao716.wang@samsung.com>

external: add single-file public domain library

Signed-off-by: Inki Dae <inki.dae@samsung.com>

tests: fix Tizen build

This patch moves install directory of examples
to prevent gbs build from failed.

Change-Id: Ib5ef3d8025c0b4375c21df81b7434c405b07c6e1
Signed-off-by: Inki Dae <inki.dae@samsung.com>

tests: fix InferenceTestImage app build error

Signed-off-by: Inki Dae <inki.dae@samsung.com>

packaging: add spec file for Tizen build

Signed-off-by: Inki Dae <inki.dae@samsung.com>

MLCE-356 'Documentation issue on CrossCompilation guide'

* Add updates for Boost, Caffe and TF parsers
* Update path

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ice9e5393ab47a678b5fe7c1ba4070460ef5b473c
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>

Minor doc updates

* Updating README
* Updating android ndk build guide

Change-Id: I09f5e4fe09d590b6b482800730b78cbc84cee9e9
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>

IVGCVSW-5695 Update supported operators

 * Update supported operators for the delegate, parsers,
   serializer and deserializer

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I33ac99a29d894eec055cd05411014075d78b3168

IVGCVSW-5682 Add Deprecation Notification

 * For Quantizer, Tensorflow Parser and Caffe Parser

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: If30daa54947c417c0d700f11e630546b1745b72e

Fix TransformIterator operator=

These two operator overloads were giving compile errors as they had
misspelled the names of the member variables.

Presumably they are not used in any of the code and not tested anywhere,
and so the compilers being used on the CI decided to not report this
error. I encountered this error using the MSVC compiler, but it may
cause problems on others too.

Change-Id: I200fc98dbd024d13dbf0e2eca2a0dcd2b0bef261
Signed-off-by: Rob Hughes <robert.hughes@arm.com>

Remove unnecessary friend function

IOptimizedNetwork declares a friend function (CreatePreCompiledWorkloadTest)
that is not necessary as that function does not use any of the non-public
members of the class. I suspect this is leftover after a refactor,
as the function in question uses GetGraphForTesting() to access the
internals instead.

This friend declaration was causing me a compilation error as the namespace
of the declared friend function did not match the namespace of the definition.
Removing it solves the compilation error and tidies up the code.

Change-Id: I4aa893cb5108cd3b72ca355f725fe7a0cd053a7e
Signed-off-by: Rob Hughes <robert.hughes@arm.com>

IVGCVSW-5394 TfLiteDelegate: Implement the Lstm operator

 * Add LSTM operator

Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: If8c667685fa1176738ffe2e6d08b1c684e7ee6b2

IVGCVSW-5654 Fix failing float 16 delegate tests on android

 * add template specialization declaration to header to resolve clang issue

Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: Id8104d6feda19ed3f5f0cb018d719b1677772d9b

IVGCVSW-5734 Building armnn failed in our Backends CI job

 * CreateTestNetwork() and CreateTransposeTestNetwork should only be
   defined if Ref enabled; the same as the test its called in otherwise
   it's defined but not called.

Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: I7bc522a7884d216d1a8e51afd7cec7a66e4e2947

IVGCVSW-5665 'Basic NN Driver support for next OS Version'

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: If69dbc230e2136def8a697ab25b9c4cb5cb8c731

IVGCVSW-4873 Implement Pimpl Idiom for INetwork and IOptimizedNetwork

!android-nn-driver:5042

Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Ia1ce8b839e81b46428ba0f78463e085e5906958d
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Signed-off-by: Finn Williams <Finn.Williams@arm.com>

Remove the undefined behaviour warning from InstallationViaAptRepository.md

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Ia85dea61a6c63cf6d2ec8c63d4ce7494e66fa1cb

Update pinned version of ACL to 21.02 release branch

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I1671a68b97815a224c89c3abe2f3944d97446c09

IVGCVSW-5694 Update Major, Minor release versions

   * Update Arm NN version in Version.hpp and all guides
   * Change Project Version in Doxyfile

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I05ab20a1ae5d0376a641df68a5877f97e7609052

IVGCVSW-5675 Implement Pimpl Idiom for IProfiler (lower priority)

Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: If716f5f4e9b5433586b8a939d326830482da2f74

IVGCVSW-5686 Add GpuAcc MLGO tuning file configuration argument

Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I3f320499c379162f9d1b00cc8816bd144cd7eee4

IVGCVSW-5679 Failure to run the build instructions to build ArmNN using android ndk

   * Provide FLATBUFFERS_INCLUDE_PATH in GlobalConfig.cmake if ARMCOMPUTECL is enabled

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Iabd5bb9b8a63970df9b7f485d07f8c1ee1a78560

IVGCVSW-5386 TfLiteDelegate: Add Strided Slice operator

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Icd87b1c54e1a5de84893882da30840a9097f6d84

IVGCVSW-5648 Adding serializer support for m_DimensionsSpecificity

The field m_DimensionsSpecificity in TensorShape was not being serialized
and deserialized following implementation of type 1 dynamic tensors.

* Update schema.
* Add to Serializer and Deserializer.

Signed-off-by: Colm Donelan <Colm.Donelan@arm.com>
Change-Id: I7ddbdaf54c8f4b988c6cb300f90ba848a94bdad0

IVGCVSW-5685 Add CpuAcc specific configuration option numberOfThreads

 * Added ability to set number of threads used in CpuAcc backend
 * Enabled number-of-threads option in ExecuteNetwork
 * Added TfLiteDelegate ModelOptions test
 * Added unsigned int type to BackendOptions.hpp

Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Ia576d4f45cbe5df3654bc730bb5ebd5181d82b5a

IVGCVSW-4893 Refactor ILayerVisitor using unified interface strategy.

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: Id7bc8255a8e3f9e5aac65d510bec8a559bf37246

IVGCVSW-5717 Update JsonPrinterTestImpl to match latest ACL

 * Updated JsonPrinterTestImpl to use CpuLogitsDLogSoftmaxKernel_#
 * Update ACL pin to: 8958167c8d609566a40c9e618158ebcbdcd3e3bb

Signed-off-by: David Monahan <david.monahan@arm.com>
Change-Id: Ic48d6cc321d8f4cc15c488244f0d69508fad2d1c

MLCE-347 Bug fixes in Reduce: QueueDescriptor.validate and init REDUCE_MIN

* Allow input tensors of any rank in ReduceQueueDescriptor::validate
* Fix VTS tests failing for REDUCE_MIN due to initialization

Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Id8fba1662ade4e0a967093fe5a53b275847f2393

MLCE-360 'ReduceLayer InferOutputShape Issue'

* Updated ParseReduce() function in TfLiteParser to read correct axis data
* Remove unused m_TargetWidth and m_TargetHight from ReduceDescriptor
* Updated the ArmNN Serializer Schema

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I98b6e00ccba1d8ea8c845cb1ae28840e42339629

IVGCVSW-5392 TfLiteDelegate: Implement the Batch/Space operators

 * Added support for BATCH_TO_SPACE_ND and SPACE_TO_BATCH_ND

Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Ib29333f3e2bc1a18bcc63533da4369b63c45ab90

IVGCVSW-5400 'TfLiteDelegate: FLOOR operator support'

* Added FLOOR operator support to Arm NN TfLiteDelegate

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I986ce8c5a825f509e0f8b3d257fd5b60834c322f

IVGCVSW-5679 Fix Android NDK OOB build

 * Specific OOB builds will fail if flatc is not available.
 * flatc command is now only run if found.

Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Iab6e082fe35dc3c3dfb78a29e1630c9edbb80be6

Update ACL pin to 7e20e29904c98adae5a91c6492fd78da88b7a9bf

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Ica5d673e9f6a192b0cfd28c9ad5a5143aecefba7

MLCE-347 'REDUCE_MIN, REDUCE_MAX, REDUCE_SUM Support'

* Added TfLiteParser support for REDUCE_MIN and REDUCE_MAX operators
* Added ACL workloads support for REDUCE_MIN, REDUCE_MAX, and REDUCE_SUM operators
* Added TfLite Delegate support for REDUCE_MIN, REDUCE_MAX, and REDUCE_SUM operators

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I8085d59946bfd4ab78a59a61f899031ae53371a8

IVGCVSW-4901 Add semantic versioning to Parsers and TfLite Delegate

 * Added Version.hpp to all Parsers
 * Added Version.hpp to TfLite Delegate
 * Updated CMakeLists to use new versions
 * Added GetVersion method to parsers and TfLite Delegate

Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: If29e1e6d9e615f9095ec1c01ad47acfff40b1dd5

IVGCVSW-4873 Implement Pimpl Idiom for IRuntime

Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I52448938735b2aa678c47e0f3061c87fa0c693b1

Update ACL pin to dda6914c6e923187c2ca2c3bfd71677e9c9e5c68

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I0e29496698b5e7934c2b1c18740527cd9eb23ecd

IVGCVSW-5605 Doxygen: Update parser section

 * Removes support.md files from all parsers. Lists of supported
   operators are now kept in doxygen only

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I137e03fdd9f41751624bdd0dd25e2db5ef4ef94f

IVGCVSW-5605 Doxygen: Add remaining guides to doxygen

 * Integrates the debian package installation guide
 * Integrates the AndroidNDK guide
 * Integrates the cross compile guide

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I2243b15ee3f8e976789494199f12f3cc0fd5f00d

ArmnnConfig.cmake install to correct directory

Signed-off-by: Keith Mok <ek9852@gmail.com>
Change-Id: I9b1037275379bdbf452589357576cfa763b7c128

IVGCVSW-4143 Add semantic versioning to Arm NN

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Ibe1200c6d7be9e04821b4a829c83833d44542ad5

Update ACL pin to cf87f509fc23d02c56569f794a3fb59e1b8be277

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I20b69cc042ae5828584b2ecfd2f9694021d7693c

MLCE-326 'Support Dilation in Conv2D in ONNX and Tensorflow Parsers'

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I4a0f07b1e8f80aff0d29405def1f33bde7944e31

Remove LFS configuration

 * Remove LFS configuration
 * Add mock test models for external delegate

Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I9f5b427a995aff22391548cd45821cb2afc8a474

IVGCVSW-5592 Implement Pimpl Idiom for Caffe and Onnx Parsers

Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I760dc4f33c0f87113cda2fa924da70f2e8c19025

IVGCVSW-5592 Implement Pimpl Idiom for Tf and TfLite Parsers

Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I4a82aca4a2c47b3c598b91bc0075c09397be728a

Update ACL pin to 0841ca085301e8ddbc9

Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I8b27541ec11ee907b80f6e2b4ff9ce3d76320076

Fix LFS issue

Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I6b3321cf0d51c06b742432995ea486c987806afc

backends/reference: Add ReduceSum operation support

This patch addes ReduceSum operation support for reference backend,
which computes the sum of elements across dimensions of a tensor.

Changelog v1:
- Fix file header descriptions.

Chagelog v2:
- Fix line limit issue.
- Fix type conversion issue.

Changelog v3:
- Remove tabs.
- Modify newly added file headers.

Changelog v4:
- Symbol on header isn't allowed so drop it from newly added file headers.

Changelog v5:
- Remove tabs, fix the use of brackets and align lines correctly.

Changelog v6:
- Add serializer and deserializer support.

Changelog v7:
- Fix build error add missed code.

Changelog v8:
- Rename ReduceSumDecriptor to ReduceDescriptor
    - Update m_KeepDims field data type to bool on ReduceDescriptor
    - Add ReduceOperation field to ReduceDescriptor

- Rename ReduceSumLayer to ReduceLayer
    - Update ReduceLayer to use ReduceDescriptor
    - Update ReduceLayer::ValidateTensorShapesFromInputs() function

- Rename RefReduceSumWokload to RefReduceWorkload
    - Update workload to use ReduceDescriptor
    - Update workload to use Decoders and Encoders

- Remove ReduceSum.hpp and ReduceSum.cpp
- Added Reduce.hpp and Reduce.cpp
     - Move Mean.cpp (which is implementing REDUCE_MEAN) functionality to Reduce.cpp
     - Update RefMeanWorkload to call Reduce function with ReduceOperation::Mean argument

- Remove Mean.hpp and Mean.cpp
- Update the Serializer/Deserializer ArmnnSchema.fbs for ReduceLayer, ReduceDescriptor, and ReduceOperation
- Update Serializer and Deserializer for serializing/parsing ReduceLayer
- Added TfLiter parser Sum test for REDUCE_SUM operator
- Make corresponding changes on front-end and Ref backend to support REDUCE_SUM operator

Changelog v9:
- Fixed build errors.

Change-Id: I8c8e034f3df73f9565b3c18eff51ecca6c542195
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>

IVGCVSW-5605 Doxygen: Add contribution guide sections

 * Adds a page for contribution guidelines (Uses md guide)
 * Adds a section for contribution guides (Uses md guide)
 * Minor beautification

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Iafd7e3abf9da660ec21a215afada61a27ce928fc

IVGCVSW-5605 Doxygen: Add the TfLite Delegate to Doxygen docu

 * Moves TensorFlowLiteDelegateSupport.md to doxygen
 * Renames a few .dox files to represent the structure of the documentation
 * Adds sections for build and use guides
 * Integrates markdown guides for the delegate into doxygen

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I68b7a29239be88c0617da9bdfce5323ebf26f5bc

IVGCVSW-5619 Enable OptimizerOptions for the python external delegate

 * Add reduce-fp32-to-fp16, reduce-fp32-to-bf16, debug-data, memory-import
options to external delegate
 * Simplify DelegateOptions
 * Add test mock models
 * Unit tests
 * Configure lfs to manage tflite files

Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I1e4db468862ba03d4cb031347bc307cf940b3cb1

Update ACL pin to ea857273d8b4a94fb7f1e63ce9068a60259fb9d3

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Id43ff3eb59f8c93cfe9fc4bcd76b8496512e2859

IVGCVSW-5605 Doxygen: Add PyArmNN to Doxygen docu

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I43082950c55375f62ecd30236d1c475e4ca89faf

IVGCVSW-5605 Doxygen: Move "OtherTools" to "Software Tools"

 * Uses README.md files to reduce duplications

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I2cfcb6cf02ac712656cb8c782c70d002df642b97

IVGCVSW-5605 Doxygen: Move Converter and Quantizer to Software Tools

 * Using README.md files directly to reduce duplications

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I6bd6f6326e608939ac244cc32b329e32bb4802f7

IVGCVSW-5605 Doxygen: Move Serializer and Deserializer to Software Tools

 * Not using the README.md files directly because it doesn't seem to
   be possible to get the content of readme.md combined with the content
   of the support.md files in one page

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I1e5073761f3af8a8d498385f3f7586334e7b29ec

IVGCVSW-5605 Doxygen: Integrate parsers into Software Tools section

 * Creates new section in doxygen called "Software Tools"
 * Moves the Parser page into Software Tools

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I19643074226d7c2e73e560e3abb52f519af1a7b2

IVGCVSW-5605 Doxygen: Use readme.md as mainpage in doxygen

 * Gives the readme file an update
 * Removes introduction.dox
 * Adds FAQ to doxygen

Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Ibb67e7f2cac7e55556295eb7851c616561b17042

IVGCVSW-5593 Implement Pimpl Idiom for serialization classes

Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I01c7bd314219e58f71505dcb787d606dbded914a

IVGCVSW-5374 Provide Android Build for Delegate

Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: Ie8c03153f82434d73572278eedb8a4b5b89ed837

IVGCVSW-5484 Add Network loading time to InferenceModel

 * Added output log to capture time taken to load network into runtime.
 * This time is cut down when loading a cached network.

Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I043c177f17d01df35fbe0752ec5d77e350749164

IVGCVSW-4874 Provide LayerSupportHandle to frontend users

 * Add test for new IsBackendRegistered member function of Handle
 * Move deprecated messages to new frontend API of LayerSupportHandle
 * Update delegate to use dot operator for IsXXXLayerSupported

Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: I70d7166e207a10e4b3583a827ca0dda2169bcba1
!android-nn-driver:4940

IVGCVSW-5374 Provide Android Build for Delegate

 * Remove double brackets for scalars
 * Add in extra location  in cmake for tf lite dynamic library

Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I31ef9a955d4677faabedc0068ae2cfbfe570b58c

Update ACL pin to 4cfab1886549d7582e3867c51278132e5d37681b

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Ica17c75dfab52d9b8fada56658b95794d4e58536

Remove incorrect use of std::vector.reserve

Change-Id: Ic45d4c837eaad8b9f19288d4f3b6e8caa3eaf319
Signed-off-by: Jim Flynn <jim.flynn@arm.com>

IVGCVSW-5391 'ArmNN TfLiteDelegate: Implement the Space/Depth operators'

* Added DEPTH_TO_SPACE and SPACE_TO_DEPTH operators support

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I2595f759181bd7339127e7b114b850b534210dd5

IVGCVSW-5390 'TfLiteDelegate: Implement the Normalisation operators'

* Added L2_NORMALIZATION and LOCAL_RESPONSE_NORMALIZATION opertor support

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ic9e66879cf6469fa8761fb1c9dd5950771f629b0

IVGCVSW-5619 Add OptimizerOptions and NetworkProperties to ArmNN Delegate

 * Add OptimizerOptions, NetworkProperties, DebugCallbackFunction
to DelegateOptions
 * Enable OptimizerOptions when the network is being optimized
 * Enable NetworkProperties when loading network
 * Enable DebugCallbackFunction
 * Add error message when loading network
 * Log warning instead of error when operator is not supported but
could fallback to another backend
 * Improve uint16_t CompareData
 * Unit tests

Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I353035afb442774bfeb1c62570a90755c2ceaf38

Update ACL pin to 6a4ebe1305b544aec1ba0bfc67ed65d94fcc8c2e

* Pass in new window parameter to the schedule_op function

Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I8579d2d6e55ab9888471bd780628df0f73438498

IVGCVSW-5525 Handle Neon optionality on 32 bit linux platforms

 * Add neon detection for linux using HWCAPs
 * Add test to check for backend throwing BackendUnavailable exception

Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: Ib74aeb06abe5f88f21ecdd1edb2a1cd20ee2019d

MLCE-335 'DEPTH_TO_SPACE and GATHER operator support in TfLiteParser'

* Added GATHER operator support to TfLiteParser
* Added DEPTH_TO_SPACE operator support to TfLiteParser

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Id5d3b54e2d850eb9f19417029efbeb73a3029e69

IVGCVSW-5571 Expose the TfLite Delegate to the TfLite python API

 * Implemented external delegate adaptor interface for TfLite
 * Activated armnn logging for delegate
 * Added logging info to indicate if gpu tuning is turned on
 * Added pytests to ensure functionality of the external delegate adaptor
 * Included the delegate directory into doxygen
 * Added documentation on how to use the external delegate in python

Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Id3b4588fb0b9ac7e3f47ba2c19feead7beb58e18

IVGCVSW-5616 Don't fuse activation if quantization parameters are different

Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I6504e922113aa9e397f53e570ebcf47e1f133945

MLCE-336 'TfLite parser and ExecuteNetwork issues'

* Report the error thrown

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: If5e5160f43c11a284b919d76b04d9c4d53e9c56c

IVGCVSW-5399 'TfLiteDelegate: Implement the ArgMinMax operators'

* Added ARG_MIN and ARG_MAX support to armnn_delegate

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ia000c4b64378e28320164edd4df2902ca13dcda6