platform/core/ml/nnfw.git
6 years agoSupport ReLU activation (#571)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 06:26:06 +0000 (15:26 +0900)]
Support ReLU activation (#571)

This commit revises Conv2D operation kernel to support ReLU activation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoGather acl kernel operator prototypes (#569)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 11 Apr 2018 06:07:13 +0000 (15:07 +0900)]
Gather acl kernel operator prototypes (#569)

* Gather acl kernel operator prototypes

This will gather acl kernel operator prototypes to root include folder

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* add reshape

6 years agoUpdate convolution manual calculation xls (#562)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 11 Apr 2018 05:44:34 +0000 (14:44 +0900)]
Update convolution manual calculation xls (#562)

This will update xls file with 3x5 input with 2 kernels

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years ago[NNOP] The first version of reshape layer (#566)
김정현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 05:36:37 +0000 (14:36 +0900)]
[NNOP] The first version of reshape layer (#566)

This commit introduces the first version of reshape layer using ACL.
Unit test will be added more in later commits.

Signed-off-by: Junghyun Kim <jh0822.kim@samsung.com>
6 years agoRemove ext libs used to build TF Lite (#541)
서상민/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 05:32:51 +0000 (14:32 +0900)]
Remove ext libs used to build TF Lite (#541)

For issue #491

This patch is to remove external libs that were used to build TF Lite and to use a script that brings
those libraries from the tensorflow directory.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoMove conv kernel to cl folder (#561)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 11 Apr 2018 05:00:51 +0000 (14:00 +0900)]
Move conv kernel to cl folder (#561)

This will move Conv kernel source into cl folder for cl implementations

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agotflite_run: initialize _elapsed in ElapsedTime (#559)
서상민/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 04:38:26 +0000 (13:38 +0900)]
tflite_run: initialize _elapsed in ElapsedTime (#559)

Fixes issue #540

This patch initializes _elapsed in ElapsedTime as 0.0 to set a correct
time value.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoAdd 3x5 2 kernel convolution test (#552)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 11 Apr 2018 04:22:59 +0000 (13:22 +0900)]
Add 3x5 2 kernel convolution test (#552)

* Add 3x5 2 kernel convolution test

This will add 3x5 with 2 kernel convolution test and fix format conversions

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* apply review comments

6 years agoIntroduce interactive docker option (#545)
김정현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 04:03:03 +0000 (13:03 +0900)]
Introduce interactive docker option (#545)

To build and test continuously in the docker environment, run as below:
`$ DOCKER_INTERACTIVE=1 ./docker/Docker_build_cross_arm_ubuntu_without_aclbuild.sh`

Signed-off-by: Junghyun Kim <jh0822.kim@samsung.com>
6 years ago[Pure ACL] Fix build break (#556)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 03:57:38 +0000 (12:57 +0900)]
[Pure ACL] Fix build break (#556)

This commit removes inputs/outputs field from Plan class (as it now has
a reference to Model class which has inputs/outputs) to fix build break.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoFix convolution kernel test failure (#557)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 11 Apr 2018 03:57:16 +0000 (12:57 +0900)]
Fix convolution kernel test failure (#557)

This will fix convolution kernel on acl test

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years ago[NNOP] The first version of fullyConnectedFloat32 (#547)
김정현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 01:55:51 +0000 (10:55 +0900)]
[NNOP] The first version of fullyConnectedFloat32 (#547)

This commit introduces the first version of fullyConnectedFloat32
to integrate ACL with NN runtime.

Signed-off-by: Junghyun Kim <jh0822.kim@samsung.com>
6 years agoRemove functions used in only ExecutionPlan in ExecutionBuilder (#546)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 11 Apr 2018 01:51:24 +0000 (10:51 +0900)]
Remove functions used in only ExecutionPlan in ExecutionBuilder (#546)

This commit removes functions used in only ExecutionPlan in ExecutionBuilder

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoIntroduce CLUniqueTensor (#551)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 01:47:21 +0000 (10:47 +0900)]
Introduce CLUniqueTensor (#551)

This commit introduces CLUniqueTensor which initialize and allocate
buffer in its constructor, and free that buffer in its destructor.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL] Pass NNAPI model to execution (#544)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 01:31:23 +0000 (10:31 +0900)]
[Pure ACL] Pass NNAPI model to execution (#544)

The following functions need to access model information through
ANeuralNetworksExecution object:
 - ANeuralNetworksExecution_setInput
 - ANeuralNetworksExecution_setOutput

This commit renames class Model in internal::arm_compute namespace as
Plan, revises it to include a reference to NNAPI model.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoSet arm_compute::TensorShape for bias correctly (#549)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 00:56:28 +0000 (09:56 +0900)]
Set arm_compute::TensorShape for bias correctly (#549)

This commit revises convFloat32 implementation based on ARM Compute to
correctly set bias shape. This change fixes clBuffer exception during
bias memory allocation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoRemove comment from test_driver.sh (#542)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 00:30:40 +0000 (09:30 +0900)]
Remove comment from test_driver.sh (#542)

6 years agoInvoke NNFW ACL kernels via runtime (#535)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Wed, 11 Apr 2018 00:13:42 +0000 (09:13 +0900)]
Invoke NNFW ACL kernels via runtime (#535)

* Invoke NNFW ACL kernels via runtime

This commit allows users to invoke NNFW ACL Conv2D kernel via runtime.
It also adds link directory for ARM Compute Library.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoEnable RESHAPE of CpuExecutor. (#496)
TANUJ TEKRIWAL/SRI-Bangalore-System SW/./삼성전자 [Tue, 10 Apr 2018 23:40:25 +0000 (05:10 +0530)]
Enable RESHAPE of CpuExecutor. (#496)

Add the RESHAPE operation file and enable the
operation in CpuExecutor.

Signed-off-by: Tanuj Tekriwal <tanuj.tekri@samsung.com>
6 years agoBring Android NN Runtime Generated Tests (#463)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 10 Apr 2018 23:39:34 +0000 (08:39 +0900)]
Bring Android NN Runtime Generated Tests (#463)

C++ Tests generated from python specs are added.
Its build is default off since it requires g++ >= 7.
You can enable it by defining BUILD_ANDROID_NN_RUNTIME_TEST=ON.

Simple tests (float_xxx and add) pass.
For others (conv_quants_xxx, depthwise_xxx, etc), we have to enable
the code from android NN (currently enclosed with comment).

I included generated C++ not python spec file and generator script.
Because i) It tooks about 12 sec in my machine to generate.
       ii) I belive the tests seldom change.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years ago[Pure ACL] Store Network Inputs/Ouputs (#543)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 23:36:03 +0000 (08:36 +0900)]
[Pure ACL] Store Network Inputs/Ouputs (#543)

This commit revises tflite::Model and arm_comput::Model to store the
inputs and outputs of a given network.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoExtract general functions from Conv2D_acl.test.cpp (#539)
김정현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 22:29:29 +0000 (07:29 +0900)]
Extract general functions from Conv2D_acl.test.cpp (#539)

- TestEnvironment -> gtest_env.cpp
- initData() and compareData() -> util.cpp

Signed-off-by: Junghyun Kim <jh0822.kim@samsung.com>
6 years agoUpdate the TensorFlow submodule to the latest commit (#537)
서상민/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 09:12:59 +0000 (18:12 +0900)]
Update the TensorFlow submodule to the latest commit (#537)

This patch updates the TensorFlow submodule to the latest commit in the
nnfw branch.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoFix broken links to tensorflow (#538)
최형규/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 09:12:51 +0000 (18:12 +0900)]
Fix broken links to tensorflow (#538)

- Fix broken links to tensorflow

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
6 years agoFix obs build fail on libkernelacl (#531)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:46:22 +0000 (17:46 +0900)]
Fix obs build fail on libkernelacl (#531)

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years agoReturn correctly on NNAPI test failure (#536)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:41:57 +0000 (17:41 +0900)]
Return correctly on NNAPI test failure (#536)

This commit revises TfLiteInterpMatchApp to return boolean instead of
integer.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL NNRT] Introduce Conv2D Param Constructor (#528)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:41:25 +0000 (17:41 +0900)]
[Pure ACL NNRT] Introduce Conv2D Param Constructor (#528)

This commit introduces a constructor for implicit Conv2D Param class
which initializes parameters using arrays.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoUse N/H/W/C ordering for fromNNShape (#530)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:41:16 +0000 (17:41 +0900)]
Use N/H/W/C ordering for fromNNShape (#530)

NNAPI assumes that N/H/W/C binding for 3D feature map tensor.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agogenerate xml test result for unittest (#527)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:31:52 +0000 (17:31 +0900)]
generate xml test result for unittest (#527)

test results with xml format are generated on report dir.
This files will be used on junit of CI.

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years agoClean includes in NeuralNetworks.cpp (#533)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 10 Apr 2018 08:29:42 +0000 (17:29 +0900)]
Clean includes in NeuralNetworks.cpp (#533)

This commit cleans includes in NeuralNetworks.cpp (except NeuralNetworksOEM.h)

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoRemove unused includes and codes in CompilationBuilder (#529)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 10 Apr 2018 08:29:05 +0000 (17:29 +0900)]
Remove unused includes and codes in CompilationBuilder (#529)

This commit removes unused includes and codes in CompilationBuilder.

- Remove partition codes
- enable VLOG in constructor
- Remove ExecutionPlan codes
- Remove unused includes

Related : #503
Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoRemove unused functions and includes in ModelBuilder (#512)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 10 Apr 2018 08:27:47 +0000 (17:27 +0900)]
Remove unused functions and includes in ModelBuilder (#512)

This commit removes unused functions and includes in ModelBuilder.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoIntroduce nnfw::util::tensor::Object (#505)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 08:10:12 +0000 (17:10 +0900)]
Introduce nnfw::util::tensor::Object (#505)

This commit introduces nnfw::util::tensor::Object, which makes it easy
to generate a random tensor object for testing.

This commit also simplifies nnapi_test and nnapi_unintest_conv_1
implementations using this new helper class.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL NN Runtime] Add Execution Placeholder (#526)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 07:38:48 +0000 (16:38 +0900)]
[Pure ACL NN Runtime] Add Execution Placeholder (#526)

This commit partially implements functions related with ANeuralNetworksExecution
struct.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoIntroduce nnfw_kernels (#517)
최형규/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 07:17:24 +0000 (16:17 +0900)]
Introduce nnfw_kernels (#517)

- Introduce nnfw_kernels to choose kernel for various target at runtime
- Apply nnfw_kernels to convFloat32 used at OperationType::CONV_2D
  - Introduce environment variable NNFW_KERNEL_convFloat32 to select target of convFloat32
  - Tested with convolution_test2.tflite

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
6 years ago[NNAPI Unittest] Add softmax (#471)
최성진/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 10 Apr 2018 06:54:51 +0000 (15:54 +0900)]
[NNAPI Unittest] Add softmax (#471)

This commit adds softmax NNAPI unittest. It is related with #374.

Signed-off-by: SungJin Choi <lotieye.choi@samsung.com>
6 years agoImplement convolution kernel with ACL (#521)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 10 Apr 2018 06:49:30 +0000 (15:49 +0900)]
Implement convolution kernel with ACL (#521)

This will implement convolution kernel with ACL in the simplest way.

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agoEarly version of freezer ( Tensorflow model (code level) to frozen model ) (#419)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 10 Apr 2018 06:47:50 +0000 (15:47 +0900)]
Early version of freezer ( Tensorflow model (code level) to frozen model ) (#419)

This commit introduces a tool to convert Tensorflow model (in code level) to frozen model file.
This could be used in future to generate many frozen model files for test cases.
The following samples to generate frozen model are included.

- relu(wx+b)
- softmax(wx+b)

6 years agoEnable tizen cross/obs build (#482)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 03:28:12 +0000 (12:28 +0900)]
Enable tizen cross/obs build (#482)

1. setup tizen cross/obs build
2. add tizen cross build script

Build assumes that libarmcl-devel, gtest-devel and tensorflow-lite are installed
on tizen's rootfs. Since libarmcl and tensorflow-lite is not present
on public tizen repo, please use private local repo like
`http://10.113.61.134/repo/` for obs build. For cross build, rootfs is
uploaded on `http://npuarchive.mooo.com/archive/nnfw/rootfs/rootfs_arm_tizen.tar.gz`.

6 years ago[NNAPI Unittest] Use random kernel/bias for conv_1 (#524)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 02:43:38 +0000 (11:43 +0900)]
[NNAPI Unittest] Use random kernel/bias for conv_1 (#524)

* [NNAPI Unittest] Use random kernel/bias for conv_1

This commit revises 'nnapi_unittest_conv_1' to use random kernel and
bias values for testing.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL NN Runtime] Implement Compilation (#522)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 02:12:03 +0000 (11:12 +0900)]
[Pure ACL NN Runtime] Implement Compilation (#522)

This commit partially implements ANeuralNetworksCompilation_create with
partially defined ANeuralNetworksCompilation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoUse common test driver for NN API tests (#514)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 01:52:11 +0000 (10:52 +0900)]
Use common test driver for NN API tests (#514)

This commit extracts test driver implemented in 'nnapi_test' as utility,
and uses it to revises 'nnapi_unittest_conv_1'.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL NN Runtime] Support casting operations over Shape (#519)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Tue, 10 Apr 2018 01:49:57 +0000 (10:49 +0900)]
[Pure ACL NN Runtime] Support casting operations over Shape (#519)

This commit introduces various casting operations over operand::Shape
class in pure ACL NN runtime.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoResolve TODO-NNRT (#518)
최형규/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 23:26:58 +0000 (08:26 +0900)]
Resolve TODO-NNRT (#518)

- Resolve TODO-NNRT
  - Remove including <limit> header file

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
6 years agoHandle reshape and depthwise convolution in operator selector (#476)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 14:13:53 +0000 (23:13 +0900)]
Handle reshape and depthwise convolution in operator selector (#476)

Operator selector (select_operator.py) can handle new two operators

- Reshape
- Depthwise convolution

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoClean declaration section in ModelBuilder.h (#511)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 9 Apr 2018 13:21:39 +0000 (22:21 +0900)]
Clean declaration section in ModelBuilder.h (#511)

This commit cleans declaration section in ModelBuilder.h

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoRevert remaining commented LOG codes to original (#506)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 9 Apr 2018 13:20:46 +0000 (22:20 +0900)]
Revert remaining commented LOG codes to original (#506)

This commit reverts remaining commented LOG codes to original.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoEnable static_assert in NeuralNetwors.cpp (#508)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 9 Apr 2018 10:20:03 +0000 (19:20 +0900)]
Enable static_assert in NeuralNetwors.cpp (#508)

This commit enables static_assert in NeuralNetwors.cpp
The assertions of 'OEM Operations' are commented out in current scope now.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoClean up cmake files in directory depends (#507)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 9 Apr 2018 10:19:19 +0000 (19:19 +0900)]
Clean up cmake files in directory depends (#507)

This commit clean up cmake files in directory depends.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agotypo fix: NNFw_OWNER_GID -> NNFW_ (#513)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 07:53:15 +0000 (16:53 +0900)]
typo fix: NNFw_OWNER_GID -> NNFW_ (#513)

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years agofix a typo in nnapi_test (#510)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 9 Apr 2018 07:36:25 +0000 (16:36 +0900)]
fix a typo in nnapi_test (#510)

This commit fixes a typo in nnapi_test.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years ago[Pure ACL NN Runtime] Partailly implement addOperation (#501)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 07:19:53 +0000 (16:19 +0900)]
[Pure ACL NN Runtime] Partailly implement addOperation (#501)

This commit implements ANeuralNetworksModel_addOperation partially for
Conv2D operation with implicit paddings.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[NNAPI Test] Support dynamic tolerance (#509)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 07:19:14 +0000 (16:19 +0900)]
[NNAPI Test] Support dynamic tolerance (#509)

This commit allows 'nnapi_test' to dynamically adjust its tolerance level.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoRename libtensorflow_lite.a to -lite.a (#504)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 07:12:06 +0000 (16:12 +0900)]
Rename libtensorflow_lite.a to -lite.a (#504)

Makefile on tensorflow-lite repo uses -lite.a
To enable obs build, need to follow orignal naming.

6 years agoExtract Tensor IndexFormatter as utility (#489)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 06:52:04 +0000 (15:52 +0900)]
Extract Tensor IndexFormatter as utility (#489)

This commit extracts TensorIndexFormatter (implemented in NNAPI test) as
utility.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoHowTo documentation for docker image of nnfw (#427)
최형규/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 06:50:34 +0000 (15:50 +0900)]
HowTo documentation for docker image of nnfw (#427)

* HowTo documentation for docker image of nnfw

- Add HowToUseDockerImage.md

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
* Add private registery configuration for different docker daemon

- Add private registery configuration for different docker daemon

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
* Apply review feedbacks

- Apply review feedbacks

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
6 years agoIntroduce nnfw kernel Initialize (#502)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 9 Apr 2018 06:49:16 +0000 (15:49 +0900)]
Introduce nnfw kernel Initialize (#502)

This will introduce initialize function for nnfw acl kernels

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agoLabs convolution with acl runtime (#500)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 9 Apr 2018 04:10:49 +0000 (13:10 +0900)]
Labs convolution with acl runtime (#500)

This will update labs convolution test using acl runtime layer

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agoIntroduce tensor::NonIncreasingStride (#499)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 02:28:58 +0000 (11:28 +0900)]
Introduce tensor::NonIncreasingStride (#499)

This commit extracts stride-related code from TensorView in nnfw::support::tflite
as NonIncreasingStride utility class (in nnfw::util::tensor).

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoDocs tizen on xu4 (#498)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 9 Apr 2018 02:26:43 +0000 (11:26 +0900)]
Docs tizen on xu4 (#498)

* Docs tizen on xu4

This will add how-to document about tizen on xu4 from wiki page

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agoIntroduce NNFW_GTEST_LIBS for test programs (#497)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 9 Apr 2018 01:36:32 +0000 (10:36 +0900)]
Introduce NNFW_GTEST_LIBS for test programs (#497)

This will introduce NNFW_GTEST_LIBS variables having gtest libraries for test programs

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years ago[Pure ACL NN Runtime] Implement setOperand functions (#490)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 01:14:18 +0000 (10:14 +0900)]
[Pure ACL NN Runtime] Implement setOperand functions (#490)

This commit implements setOperandValue and  setOperandValueFromMemory
functions for pure ACL NN Runtime.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoIntroduce an image generator (#426)
김정현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 00:51:23 +0000 (09:51 +0900)]
Introduce an image generator (#426)

This commit introduces an image generator:

- How to execute:
`$ ./imagegen.py`
Two output images are generated: `image.bin` and `image.ppm`.

- Changing the size of the output image
Change variable `image_size` at the beginning of `imagegen.py`.

- Changing output image format
If you want to output another image format, you can use as below:
`im.save("image.jpg")`

Signed-off-by: Junghyun Kim <jh0822.kim@samsung.com>
6 years agoIntroduce nnfw::util::tensor::Zipper (#495)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Mon, 9 Apr 2018 00:30:41 +0000 (09:30 +0900)]
Introduce nnfw::util::tensor::Zipper (#495)

This commit introduces Zipper class and related helper methods.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[NNAPI Test] Initialize IFM with float values (#494)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 07:14:26 +0000 (16:14 +0900)]
[NNAPI Test] Initialize IFM with float values (#494)

This commit revises 'nnapi_test' to initialize IFM with float values.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoFix CI build fail due to permission(for pipiline job) (#469)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 07:10:14 +0000 (16:10 +0900)]
Fix CI build fail due to permission(for pipiline job) (#469)

On pipeline job, workspace clean is not a default option. Thus we have
to clean up WS manually on pipeline. Without this patch, cleanup is
not allowed due to permission error. It is based on c93d050642e196b0d667a9dee34117aa330f5fbc.

  1. fix permission error on CI
  2. supress wget output with `-nv`

6 years agoRemove the dependency on hal and hidl from IServiceManager (#444)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 6 Apr 2018 06:58:25 +0000 (15:58 +0900)]
Remove the dependency on hal and hidl from IServiceManager (#444)

This commit removes the dependency on hal and hidl from IServiceManager.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agoAdd kernels implementation placeholder (#448)
박세희/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 6 Apr 2018 06:13:32 +0000 (15:13 +0900)]
Add kernels implementation placeholder (#448)

* Add kernels implementation placeholder

This will add empty kernels placeholder source and initial cmakefiles and empty test source

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
6 years agoWork around the github permission issue
Sangmin Seo [Thu, 5 Apr 2018 05:31:06 +0000 (14:31 +0900)]
Work around the github permission issue

This patch is to work around the access permission issue on github using
submodule.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoAdd tensorflow module
Sangmin Seo [Thu, 5 Apr 2018 02:19:09 +0000 (11:19 +0900)]
Add tensorflow module

For issue #332

This patch adds https://github.sec.samsung.net/RS7-STAR/Tensorflow/tree/nnfw
as a submodule.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoRemove externals/tensorflow
Sangmin Seo [Mon, 2 Apr 2018 06:09:26 +0000 (15:09 +0900)]
Remove externals/tensorflow

For issue #332

As the first step to use git submodule for TensorFlow, this commit first
removes externals/tensorflow.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoRemove unused eq operator over TfLiteTensor (#483)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:49:45 +0000 (11:49 +0900)]
Remove unused eq operator over TfLiteTensor (#483)

equality operator over TfLiteTensor is no longer used.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoPrint out max relative difference (#473)
최형규/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:22:00 +0000 (11:22 +0900)]
Print out max relative difference (#473)

* Print out max non-zero observed relative difference

- Print out max non-zero observed relative difference
  for both MATCHED and UNMATCHED

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
* Apply review feedbacks

- Apply review feedbacks

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
6 years agoIntroduce equality over tensor::Shape (#488)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:21:43 +0000 (11:21 +0900)]
Introduce equality over tensor::Shape (#488)

This commit introduces == operator over tensor::Shape, and use it to
check shape equality over two tensors under comparision.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoIntroduce nnfw::support::tflite::interp::Builder (#487)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:21:10 +0000 (11:21 +0900)]
Introduce nnfw::support::tflite::interp::Builder (#487)

This commit introduces 'nnfw::support::tflite::interp::Builder' as a
common interface for NNAPI test & unittests.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[nnfw_support_tflite] Introduce TensorUtils.h (#486)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:17:29 +0000 (11:17 +0900)]
[nnfw_support_tflite] Introduce TensorUtils.h (#486)

* [nnfw_support_tflite] Introduce TensorUtils.h

This commit introduces TensorUtils.h in nnfw_support_tfilte library.
This header provides various helper functions over TfLiteTensor objects.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
* Remove irrelevant change

6 years ago[Pure ACL Runtime] Implement ANeuralNetworkModel (#477)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:15:44 +0000 (11:15 +0900)]
[Pure ACL Runtime] Implement ANeuralNetworkModel (#477)

This commit partially implements ANeuralNetworkModel in Pure ACL Runtime
to record opearnd types.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoUnify graphdef and graphdump (#453)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 02:15:10 +0000 (11:15 +0900)]
Unify graphdef and graphdump (#453)

This commit unifies graphdef and graphdump projects into tf_model_tools
(to make it easy to understand what they are).

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[NNAPI Unittests] Add 'env' unittest (#442)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 00:52:36 +0000 (09:52 +0900)]
[NNAPI Unittests] Add 'env' unittest (#442)

This commit adds a 'env' unittest in NNAPI unittests module.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoFill feature object with random numbers (#472)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 00:50:54 +0000 (09:50 +0900)]
Fill feature object with random numbers (#472)

This commit renames nnfw::util::feature::RandomObject as nnfw::util::feature::Object,
and revises it to initialize its content with user-defined number generator.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoAdd running unittest in test_driver.sh (#484)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 6 Apr 2018 00:43:00 +0000 (09:43 +0900)]
Add running unittest in test_driver.sh (#484)

Runs unittest binaries in Product/out/unittest when test_driver.sh is
executed.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoExtract operator== over vector as utility (#481)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Fri, 6 Apr 2018 00:01:46 +0000 (09:01 +0900)]
Extract operator== over vector as utility (#481)

This commit extracts operator== (implemented in nnapi_test) into
'util/vector.h' (for future reuse).

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoAdd googletest deps to android_nn_test (#479)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 23:45:59 +0000 (08:45 +0900)]
Add googletest deps to android_nn_test (#479)

This fix issue #475

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years agoAdd gnueabi arm-cross toolchain (#474)
이춘석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 14:30:21 +0000 (23:30 +0900)]
Add gnueabi arm-cross toolchain (#474)

gnueabi cross toolchain is needed to support tizen cross build

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years agoReduce indent level in tflite_run (#464)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 5 Apr 2018 11:51:38 +0000 (20:51 +0900)]
Reduce indent level in tflite_run (#464)

Reduce indent level in tflite_run

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years agoAdd framework tests for mobilenet (#478)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 11:51:14 +0000 (20:51 +0900)]
Add framework tests for mobilenet (#478)

Add framework tests for mobilenet
These tests will not be run as default

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoBug fix in run_test.sh for skip test (#480)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 10:53:54 +0000 (19:53 +0900)]
Bug fix in run_test.sh for skip test (#480)

Fix bug skip all test after some test skipped

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoAdd run_tests.sh in test_driver.sh (#462)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 5 Apr 2018 10:26:47 +0000 (19:26 +0900)]
Add run_tests.sh in test_driver.sh (#462)

run_tests.sh with tflite_run as driver is going to be executed in
test_driver.sh.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoSupport skip test in run_test.sh (#466)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 09:29:56 +0000 (18:29 +0900)]
Support skip test in run_test.sh (#466)

Support skip test in run_test.sh
Can skip by setting STATUS="disabled" in config.sh

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoRemove verbose in run_test.sh (#461)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 08:30:08 +0000 (17:30 +0900)]
Remove verbose in run_test.sh (#461)

Remove verbose in run_test.sh to reduce CI record

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoAdd tools/test_driver on build_test_suite in Makefile (#458)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 5 Apr 2018 07:41:41 +0000 (16:41 +0900)]
Add tools/test_driver on build_test_suite in Makefile (#458)

Adds tools/test_driver on build_test_suite in Makefile. test_driver.sh
is going to be included in the test-suite.tar.gz.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoIntroduce nnfw::support::tflite::TensorView (#456)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 06:17:40 +0000 (15:17 +0900)]
Introduce nnfw::support::tflite::TensorView (#456)

This commit introduces 'nnfw::support::tflite::TensorView' which allows
users to easily access elements in TensorFlow Lite Tensor object.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoAdd test_driver.sh for verification (#439)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 5 Apr 2018 05:43:41 +0000 (14:43 +0900)]
Add test_driver.sh for verification (#439)

* Add test_driver.sh for verification

This commits add test_driver.sh to Product/out/bin for verification.
Additional tests would be appended in this script soon.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoAdd more test scripts for TF Lite models (#422)
서상민/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 04:25:25 +0000 (13:25 +0900)]
Add more test scripts for TF Lite models (#422)

Related issues #84 and #402

This patch adds test scripts for TF Lite models created in #402.

Signed-off-by: Sangmin Seo <sangmin7.seo@samsung.com>
6 years agoSupport Enable/Disable VLOG (#451)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 5 Apr 2018 04:25:06 +0000 (13:25 +0900)]
Support Enable/Disable VLOG (#451)

Introduce environment variable VLOG for Enable/Disable VLOG outputs.
`VLOG` is disabled by default.

Example usage

```
VLOG=1 ./tests/framework/run_test.sh # Enable VLOG
```

6 years agoIntroduce 'nnfw_support_tflite' lib (#452)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 04:14:03 +0000 (13:14 +0900)]
Introduce 'nnfw_support_tflite' lib (#452)

This commit introduces 'nnfw_support_tflite' lib which provides various
helpes for accessing TensorFlow Lite internals.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years ago[Pure ACL NN Runtime] Implement ANeuralNetworksMemory (#424)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 04:08:09 +0000 (13:08 +0900)]
[Pure ACL NN Runtime] Implement ANeuralNetworksMemory (#424)

This commit implements ANeuralNetworksMemory in experimental pure ACL NN
runtime.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoIntroduce IndexIterator (for feature) (#428)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 02:13:17 +0000 (11:13 +0900)]
Introduce IndexIterator (for feature) (#428)

This commit introduces IndexIterator class and relevent helpers in
nnfw::util::feature namespace to make it easy to iterate over feature
elements.

This commit includes updates on NNAPI Conv1 Unittest to demonstrate how
to use these helpers.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
6 years agoPrint shape info in model parser tool (#449)
오형석/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 02:12:00 +0000 (11:12 +0900)]
Print shape info in model parser tool (#449)

* Print shape info in model parser tool

Print shape info in model parser tool to get tensor size

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Update README.md

Update README.md to show shape info

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoSupport update in TfLiteFeatureView (#425)
박종현/동작제어Lab(SR)/Senior Engineer/삼성전자 [Thu, 5 Apr 2018 02:02:43 +0000 (11:02 +0900)]
Support update in TfLiteFeatureView (#425)

This commit introduces a (in-place) update method into TfLiteFeatureView
class.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>