platform/core/ml/nnfw.git
5 years agoIntroduce doxygen command (#5213)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 May 2019 03:31:48 +0000 (12:31 +0900)]
Introduce doxygen command (#5213)

* Introduce doxygen command

Doxygen generation command: ./nnfw doxygen

Update doxygen configuration file
- Include codes only build on gbs
- Update path based on repo's root

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Use pushd and pop

5 years agoPackage dependency for peachpy (#5201)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 May 2019 02:02:58 +0000 (11:02 +0900)]
Package dependency for peachpy (#5201)

External peachpy source need for NNPACK
In peachpy source download configuration, it need additional download to execute python setup script - six, enum, opcode
Instead of using setup.py, direct code generation script run with python path setting

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[cker] Depthwise conv quant8 type (#5195)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 May 2019 10:06:29 +0000 (19:06 +0900)]
[cker] Depthwise conv quant8 type (#5195)

Introduce depthwise conv quant8 type kernel in cker
Implement neurun cpu backend for depthwise conv quant8 and enable test

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoEnable ParallelExecutor (#5202)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 17 May 2019 10:05:11 +0000 (19:05 +0900)]
Enable ParallelExecutor (#5202)

Enable ParallelExecutor with option `EXECUTOR=Parallel`.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years ago[cker] Elementwise add kernel (#5203)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 May 2019 10:05:00 +0000 (19:05 +0900)]
[cker] Elementwise add kernel (#5203)

* [cker] Elementwise add kernel

Introduce elementwise add kernel in cker
Implement neurun cpu backend and enable test

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix header

5 years agoIntroduce padding struct (#5210)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 May 2019 10:04:50 +0000 (19:04 +0900)]
Introduce padding struct (#5210)

* Introduce padding struct

Introduce padding struct including padding type and explicit padding value
Use padding type field only

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix bug in debug build

5 years agoRename padding type enum class (#5209)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 May 2019 03:49:01 +0000 (12:49 +0900)]
Rename padding type enum class (#5209)

Rename padding type enum class: Padding -> PaddingType
Prepare introduce padding struct including padding type and value

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoFix dumper bug (#5207)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 May 2019 02:13:57 +0000 (11:13 +0900)]
Fix dumper bug (#5207)

Fix dumper bug for TopKV2 Node

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRemove nnpack installation (#5193)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 May 2019 07:47:16 +0000 (16:47 +0900)]
Remove nnpack installation (#5193)

Remove nnpack related headers and static library installation

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[ACLEx] Introduce CLRNNLayerEx (#5196)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 16 May 2019 07:41:56 +0000 (16:41 +0900)]
[ACLEx] Introduce CLRNNLayerEx (#5196)

This commit introduces CLRNNLayerEx fixing CLRNNLayer of v19.02.

diff --git a/src/runtime/CL/functions/CLRNNLayer.cpp b/src/runtime/CL/functions/CLRNNLayer.cpp
index 63f00ac8e..d89a38316 100644
--- a/src/runtime/CL/functions/CLRNNLayer.cpp
+++ b/src/runtime/CL/functions/CLRNNLayer.cpp
@@ -44,8 +44,8 @@ CLRNNLayer::CLRNNLayer(std::shared_ptr<IMemoryManager> memory_manager)
 Status CLRNNLayer::validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *recurrent_weights, const ITensorInfo *bias, const ITensorInfo *hidden_state,
                             const ITensorInfo *output, const ActivationLayerInfo &info)
 {
-    const int idx_width  = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::WIDTH);
-    const int idx_height = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::HEIGHT);
+    const int idx_width = 0;
+    const int idx_height = 1;
     ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, weights, recurrent_weights, bias, hidden_state, output);
     ARM_COMPUTE_RETURN_ERROR_ON(input->dimension(idx_width) != weights->dimension(idx_width));
     ARM_COMPUTE_RETURN_ERROR_ON(weights->dimension(idx_height) != recurrent_weights->dimension(idx_width));
@@ -72,7 +72,7 @@ void CLRNNLayer::configure(const ICLTensor *input, const ICLTensor *weights, con
     ARM_COMPUTE_ERROR_ON_NULLPTR(input, weights, recurrent_weights, bias, hidden_state, output);
     ARM_COMPUTE_ERROR_THROW_ON(CLRNNLayer::validate(input->info(), weights->info(), recurrent_weights->info(), bias->info(), hidden_state->info(), output->info(), info));

-    const int   idx_height = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::HEIGHT);
+    const int   idx_height = 1;
     TensorShape shape      = compute_rnn_shape(recurrent_weights->info(), hidden_state->info()->dimension(idx_height));

     _is_prepared = false;

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRemove NNAPI Shim include from NNAPI Delegate header (#5200)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 May 2019 07:39:20 +0000 (16:39 +0900)]
Remove NNAPI Shim include from NNAPI Delegate header (#5200)

This header may introduce some name conflicts. Let's remove this
unnecessary include.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
5 years agoUse template for actvation calculation (#5198)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 May 2019 05:03:06 +0000 (14:03 +0900)]
Use template for actvation calculation (#5198)

Change activation implementation to use template
Prepare to support more type (int32, uint32, etc)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoFix PermuteLayer (#5188)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 16 May 2019 00:38:20 +0000 (09:38 +0900)]
Fix PermuteLayer (#5188)

This commit fixes PermuteLayer to consider padding.
  - Fix PermuteLayer to consider padding
  - Fix PermuteLayer to consider layout

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoUpdate Dumper to support implemented operations (#5192)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 15 May 2019 23:59:49 +0000 (08:59 +0900)]
Update Dumper to support implemented operations (#5192)

This commit updates Dumper to support implemented operations now.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoIntroduce ParallelExecutor (#5171)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 15 May 2019 09:43:37 +0000 (18:43 +0900)]
Introduce ParallelExecutor (#5171)

This commit introduces ParallelExecutor without usage.

Draft PR : #4920

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoRemove Obsolescent Functions (atoi) (#5191)
이춘석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 15 May 2019 07:54:54 +0000 (16:54 +0900)]
Remove Obsolescent Functions (atoi) (#5191)

This patch will replace atoi function with strtol.

Reference : https://wiki.sei.cmu.edu/confluence/display/c/MSC24-C.+Do+not+use+deprecated+or+obsolescent+functions

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
5 years ago[hi-perf-cpu] High performance cpu backend (#5064)
Андрей Шедько/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 15 May 2019 07:52:09 +0000 (10:52 +0300)]
[hi-perf-cpu] High performance cpu backend (#5064)

* [hi-perf] High performance cpu backend

initial support for nnpack library

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] NNPACK actually builds correctly.

NNPACK builds, but cpuinfo does not detect CPU yet.

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] Cleanup cmake and test inference

Split nnpack deps download into separate configs.
Added float16 format options to arm builds.
Added nnpack relu test to be run when it is possible

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] Try to fix cmake crossbuild

Try to fix cmake crossbuild

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] Fix x86_64 build

PeachPy on x86_64 needs to generate some files for successful
compilation
Added explicit dependency on Opcodes.py

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] Fix archive type in cmake

Fix archive type in cmake
Fix avx instruction generation for nnpack

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] Review Fixes

file-terminating newlines added, codestyle fixed.

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] More fixes for cmake files

Status reporting improved.
Removed superfluous linkages and includes.

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
* [hi-perf] More review Fixes

Fixed comment, Removed conflicting change

Signed-off-by: Andrei Shedko <a.shedko@samsung.com>
5 years agoIntroduce tensorflow lite file loader directory (#5180)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 14 May 2019 01:10:57 +0000 (10:10 +0900)]
Introduce tensorflow lite file loader directory (#5180)

Introduce tensorflow lite file loader directory in contribs for prototye of nnpackage's model file loader
schema.fbs file is comes from tensorflow r1.13.1
schema_generated.h file is auto-generated by flatc

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoSupport assigning backends to instances of operations (#5167)
Дилшоджон Умронхонович Пошшоев/AI Tools Lab /SRR/Engineer/삼성전자 [Mon, 13 May 2019 01:13:06 +0000 (04:13 +0300)]
Support assigning backends to instances of operations (#5167)

* Support assigning backends to instances of operations

This is needed for HE scheduling

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
* Remove unrelated changes

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
5 years agoEnable some tests for x64 (#5174)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Mon, 13 May 2019 01:12:39 +0000 (10:12 +0900)]
Enable some tests for x64 (#5174)

This commit enables some tests for x64

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRemove unused/unneeded SubGraphContext methods (#5173)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 10 May 2019 06:27:42 +0000 (15:27 +0900)]
Remove unused/unneeded SubGraphContext methods (#5173)

- Remove `releaseAt` which is unused
- Remove `getAt` which is a duplicate

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoChange default layout of acl backend to NHWC (#5159)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 10 May 2019 06:18:58 +0000 (15:18 +0900)]
Change default layout of acl backend to NHWC (#5159)

* Change default layout of acl backend to NHWC

This commit changes default layout of acl backend to NHWC.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
* Enable NCHW test for acl backend

This commit enable NCHW test for acl backend
 - Introduce ACL_DEFAULT_LAYOUT configuration for test by layout

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
* Fix failures of NCHW tests

This commit fixes failures of NCHW tests.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoIntroduce layout info in model operand (#5172)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 10 May 2019 04:39:43 +0000 (13:39 +0900)]
Introduce layout info in model operand (#5172)

Set model operand's shape layout info in operand info to support more layout from model
Default value is NHWC

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoFix compile error. (#5170)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 9 May 2019 09:21:52 +0000 (18:21 +0900)]
Fix compile error. (#5170)

- fix unintended implicit fallthrough in switch case.

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years agoIntroduce asActivationLayerInfo converting activation code to ActivationLayerInfo...
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 9 May 2019 08:07:16 +0000 (17:07 +0900)]
Introduce asActivationLayerInfo converting activation code to ActivationLayerInfo (#5144)

This commit introduces asActivationLayerInfo converting activation code to ActivationLayerInfo.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years ago[tflitefile_tool] Add config option for benchmark tool to save model config file...
윤지영/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 9 May 2019 05:19:21 +0000 (14:19 +0900)]
[tflitefile_tool] Add config option for benchmark tool to save model config file (#5145)

* [tflitefile_tool] Add config option for benchmark tool

This patch adds the `-c` and `--config` options.
This option will print the configuration information of each operator.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
* Save configuration file

This patch allows to save the configuration info about model

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
5 years agoIntroduce ParallelScheduler (#5155)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 9 May 2019 05:09:37 +0000 (14:09 +0900)]
Introduce ParallelScheduler (#5155)

Introduce `ParallelScheduler` along with `Worker` and `ThreadPool`.

Part of #4920

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoOrganizing the files past the date. (#5166)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 9 May 2019 00:56:33 +0000 (09:56 +0900)]
Organizing the files past the date. (#5166)

- Make backup folder `doc/2018`.
- Move the files past the date, if possible.

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years agoUpdate `roadmap.md` (#5148)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 8 May 2019 08:15:25 +0000 (17:15 +0900)]
Update `roadmap.md` (#5148)

* Update `roadmap.md`

- create document sub folder named `2018`
- move the last `roadmap.md` document and `nnfw_architecture` image to 2018 folder.
- create new `roadmap.md` and `nnfw_architecture` for 2019 with updated contests.

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years agoUpdate `workgroups.md` document (#5160)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 8 May 2019 08:10:12 +0000 (17:10 +0900)]
Update `workgroups.md` document (#5160)

* Update `workgroups.md` document

- Update workgroups.md document with latest workgroup configuration.
- Tempolarly rename old file to `2018_workgroups.md` for backup.
  It will moved to `docs/2018` folder after #5148 merged.

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years agoChange ConcatLayer to support NHWC (#5165)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 8 May 2019 05:35:03 +0000 (14:35 +0900)]
Change ConcatLayer to support NHWC (#5165)

This commit ConcatLayer to support NHWC.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRemove pack simple kernel (#5163)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 8 May 2019 05:11:21 +0000 (14:11 +0900)]
Remove pack simple kernel (#5163)

Remove pack simple kernel in libs/ARMComputeEx
Use stack kernel in acl 19.02
Fix generated test to test 2D input (permutation issue)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRun mixed backend for test coverage (#5164)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 8 May 2019 05:10:54 +0000 (14:10 +0900)]
Run mixed backend for test coverage (#5164)

Use mixed backend setting for test coverage run

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago Redo PermuteLayer run as template (#5107)
Дилшоджон Умронхонович Пошшоев/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 8 May 2019 04:36:49 +0000 (07:36 +0300)]
 Redo PermuteLayer run as template (#5107)

* Redo PermuteLayer run as template

PermuteLayer always assumes, that its data type is float. It was causing
an error during HE scheduler implementation. PR 5073

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
* Revert "Not supported yet" to NYI

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
* Merge ctors

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
5 years agoRemove nnapi dependency in neurun (#5153)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 8 May 2019 01:08:08 +0000 (10:08 +0900)]
Remove nnapi dependency in neurun (#5153)

Remove nnapi dependency in neurun core and backend code

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove Subgraph and SubgraphContext to upper dir (#5143)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 7 May 2019 09:29:19 +0000 (18:29 +0900)]
Move Subgraph and SubgraphContext to upper dir (#5143)

Move Subgraph and SubgraphContext from `model::operation::` to
`model::`.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoSupport NHWC layout for ConstantInitializer (#5139)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 7 May 2019 09:27:29 +0000 (18:27 +0900)]
Support NHWC layout for ConstantInitializer (#5139)

This commit supports NHWC layout for ConstantInitializer.
  - Support NHWC layout for ConstantInitializer.
  - Support calculating the offset of cpu tensor up to 4 dimensions and in NHWC.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRemove workaround mixed test (#5156)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 7 May 2019 08:35:32 +0000 (17:35 +0900)]
Remove workaround mixed test (#5156)

Remove workaround for mixed backend test because CI scripts are updated

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoInternal padding type string converter (#5151)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 7 May 2019 04:59:24 +0000 (13:59 +0900)]
Internal padding type string converter (#5151)

Use internal padding type to convert padding type to string instead of NNAPI PaddingCode

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove resolving transpose convolution padding (#5150)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 7 May 2019 04:59:01 +0000 (13:59 +0900)]
Move resolving transpose convolution padding (#5150)

Move resolving transpose convolution operation's padding type: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMerge Lowered and Subgraph section in Graph (#5122)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 7 May 2019 04:44:44 +0000 (13:44 +0900)]
Merge Lowered and Subgraph section in Graph (#5122)

Subgraph is also part of LOWERED phase so we should not distinguish
these members.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoRevise Job class for ParallelExecutor (#5121)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 7 May 2019 04:44:32 +0000 (13:44 +0900)]
Revise Job class for ParallelExecutor (#5121)

- Introduce `Backend` as a member
- Intrdouce `reset` method for next execution

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoIntroduce FullyConnectedReshapingLayer into ARMComputeEX (#5140)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 7 May 2019 04:44:19 +0000 (13:44 +0900)]
Introduce FullyConnectedReshapingLayer into ARMComputeEX (#5140)

This commit introduces FullyConnectedReshapingLayer into ARMComputeEx.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRemove unpack simple kernel (#5135)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 7 May 2019 04:43:26 +0000 (13:43 +0900)]
Remove unpack simple kernel (#5135)

* Remove unpack simple kernel

Remove unpack simple kernel in libs/ARMComputeEx
Use unstack kernel in acl 19.02
Fix generated test to test 3D input (permutation issue)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Update comment, remove change in pack

5 years agoFix Operands::exist to be able to remove operands (#5146)
Дилшоджон Умронхонович Пошшоев/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 7 May 2019 03:33:10 +0000 (06:33 +0300)]
Fix Operands::exist to be able to remove operands (#5146)

If we remove operands after building a graph, existing method doesn't
work. Moreover, complexity of std::unordered_map::find is O(1), so we
don't loose much in time

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
5 years agoSupport NHWC layout for calculating the offset of acl tensor (#5137)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 3 May 2019 09:41:31 +0000 (18:41 +0900)]
Support NHWC layout for calculating the offset of acl tensor (#5137)

This commit supports NHWC layout for calculating the offset of acl tensor.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoMove fullyconnected activation resolve (#5132)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:59:42 +0000 (16:59 +0900)]
Move fullyconnected activation resolve (#5132)

Move fullyconnected operation's activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove maxpool padding & activation resolve (#5120)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:53:13 +0000 (16:53 +0900)]
Move maxpool padding & activation resolve (#5120)

Move maxpool padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove OperationIndex definition to Index.h (#5117)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 3 May 2019 07:52:24 +0000 (16:52 +0900)]
Move OperationIndex definition to Index.h (#5117)

As all index classes other than `OperationIndex` are all in `Index.h`,
this commit moves `OperationIndex` definition to `Index.h` and removes
`OperationIndex.h`.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoMove elementwise op padding & activation resolve (#5129)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:34:00 +0000 (16:34 +0900)]
Move elementwise op padding & activation resolve (#5129)

Move arithmetic elementwise operation padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove conv2d padding & activation resolve (#5124)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:33:40 +0000 (16:33 +0900)]
Move conv2d padding & activation resolve (#5124)

Move convolution padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove averagepool padding & activation resolve (#5123)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:33:20 +0000 (16:33 +0900)]
Move averagepool padding & activation resolve (#5123)

* Move averagepool padding & activation resolve

Move average pool padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix constant check assertion and fix bug

- Fix to set explicit padding type
- Remove assert because asScalar has same assertion

5 years agoMove L2 pooling padding & activation resolve (#5126)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:30:37 +0000 (16:30 +0900)]
Move L2 pooling padding & activation resolve (#5126)

Move l2 pooling padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoMove depthwise-conv padding & activation resolve (#5119)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 07:26:03 +0000 (16:26 +0900)]
Move depthwise-conv padding & activation resolve (#5119)

Move depthwise convolution padding & activation type resolve: backend to frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRuntime test with mixed backends (#5125)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 3 May 2019 06:51:34 +0000 (15:51 +0900)]
Runtime test with mixed backends (#5125)

* Runtime test with mixed backends

Initial version of runtime test with mixed backends (Fixed backend
assignment)

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
* Remove logging

* Fix typo comments and add TODO

5 years ago[dotdump] Show Subgraphs (#5128)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 3 May 2019 06:24:01 +0000 (15:24 +0900)]
[dotdump] Show Subgraphs (#5128)

Show subgraphs using dot's subgraph(cluster) feature.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoChange in GTests of PReLU (#5110)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 3 May 2019 02:04:26 +0000 (11:04 +0900)]
Change in GTests of PReLU (#5110)

* Change in GTests of PReLU

Since prelu_ex_float_1 and prelu_ex_float_2 perform similar tests
This patch adds a replacement test

Signed-off-by: prasannar <prasanna.r@samsung.com>
* Add Quant8 test cases for PReLU op

This patch adds Quant8 test cases for PReLU op

Signed-off-by: prasannar <prasanna.r@samsung.com>
5 years agoMove BackendSet to graph::BackendSet (#5130)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 3 May 2019 00:56:03 +0000 (09:56 +0900)]
Move BackendSet to graph::BackendSet (#5130)

Move `BackendSet` to `graph::BackendSet`. It was in
`graph::operand::BackendSet` which is misleading.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years ago[neurun] Introduce CopySource and CopySink (#5116)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 2 May 2019 05:47:49 +0000 (14:47 +0900)]
[neurun] Introduce CopySource and CopySink (#5116)

This commit Introduces CopySource and CopySink.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoRevise header include in OperationFactory.cc (#5101)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 2 May 2019 05:36:22 +0000 (14:36 +0900)]
Revise header include in OperationFactory.cc (#5101)

- Change to use brackets for core header include
- Include NNAPIConverter.h to prepare to use activation and padding type converter

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRenewed README.md with 2019 plan (#5102)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 2 May 2019 05:35:50 +0000 (14:35 +0900)]
Renewed README.md with 2019 plan (#5102)

* Renewd README.md with 2019 plan

- Update target platform
- Trim the sentence
- Add notification about _nncc_ integration

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
* Fix typo.

- remove duplicated word.

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years ago[neurun] Introduce getting layout into backend (#5030)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 2 May 2019 04:42:05 +0000 (13:42 +0900)]
[neurun] Introduce getting layout into backend (#5030)

This commit introduces getting layout into backend tensors.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
5 years agoActivation builder using internal activation type (#5099)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 2 May 2019 04:08:40 +0000 (13:08 +0900)]
Activation builder using internal activation type (#5099)

Introduce activation builder append method using neurun internal activation type

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoUpdate sample app document (#5104)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 2 May 2019 03:57:26 +0000 (12:57 +0900)]
Update sample app document (#5104)

- Describe runtime with tensorfliw lite is not official support
- Tensorflow -> tensorflow lite
- Update chapter "Build a model": title, etc.

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRemove unused RandomObject.h (#5108)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 2 May 2019 01:14:22 +0000 (10:14 +0900)]
Remove unused RandomObject.h (#5108)

Remove unused RandomObject
Random value generation for tests are defined at libs/tflite/include/tflite/Diff.h

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoSoftware requirement specification for 2019 (#5105)
이성재/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 30 Apr 2019 12:26:24 +0000 (21:26 +0900)]
Software requirement specification for 2019 (#5105)

* Software requirement specification for 2019

- Initial draft of SRS

Signed-off-by: Sung-Jae Lee <sj925.lee@samsung.com>
5 years agoIntroduce SubgraphIndex (#5106)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 09:05:56 +0000 (18:05 +0900)]
Introduce SubgraphIndex (#5106)

`OperationIndex` was used for Subgraph indices, which can be confusing.
This commit introduces `SubgraphIndex` for `Subgraph`s.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoAdd HowtoMakeSampleAppOnNnfw.md (#4129)
Дилшоджон Умронхонович Пошшоев/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 30 Apr 2019 06:12:54 +0000 (09:12 +0300)]
Add HowtoMakeSampleAppOnNnfw.md (#4129)

* Add HowtoMakeSampleAppOnNnfw.md

Related issue: 4128
Add tutorial about How to make a sample app on nnfw

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
* Fix "_nnfw_ API" as "NNAPI"

Fix "_nnfw_ API" as "NNAPI"

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
* Describe how to implement app with nnfw

Prev version was how to implement in the level of nnapi_delegate. This
version is how to create on top of nnapi_delegate

Signed-off-by: Poshshoev Dilshodzhon <d.poshshoev@samsung.com>
5 years agoRemove return in cpu backend kernel (#5096)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Apr 2019 06:00:02 +0000 (15:00 +0900)]
Remove return in cpu backend kernel (#5096)

Remove unused bool type return in cpu backend kernel
Remove bool type return in helper function: always return true
Remove __wur (warning unused return) attribute in helper function

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[dotdump] Remove Operation LowerInfo (#5087)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 05:07:21 +0000 (14:07 +0900)]
[dotdump] Remove Operation LowerInfo (#5087)

As now LowerInfo is per-subgraph(not per-operation), we need to remove
Operation LowerInfo and Backend label from `DotNodeInfo`.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoActivation range calculator using internal type (#5093)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Apr 2019 05:04:18 +0000 (14:04 +0900)]
Activation range calculator using internal type (#5093)

Introduce helper function in cpu backend for activation range calculation
Helper function use neurun internal activation type instead of NNAPI FuseCode type

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoUse nchw::View instead of Reader (#5084)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 03:41:45 +0000 (12:41 +0900)]
Use nchw::View instead of Reader (#5084)

`Reader` is only for tensors of no paddings, while `nchw::View` can
cover padding cases.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoMove graph::Index to util::Index (#5094)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 03:41:23 +0000 (12:41 +0900)]
Move graph::Index to util::Index (#5094)

`graph::Index` is a class for indices in general, so it should not
belong to `graph` namespace.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoRemove Operation::InitParam (#5095)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 03:41:13 +0000 (12:41 +0900)]
Remove Operation::InitParam (#5095)

This was just for creation of operations from NNAPI, so it is not used
any longer.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoIntroduce activation and padding type converter (#5092)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Apr 2019 01:01:03 +0000 (10:01 +0900)]
Introduce activation and padding type converter (#5092)

Introduce nnapi to neurun internal activation and padding type converter

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[neurun] Revise Executor classes to have a proper container for Subgraph (#5035)
김용섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 30 Apr 2019 00:50:16 +0000 (09:50 +0900)]
[neurun] Revise Executor classes to have a proper container for Subgraph (#5035)

* [neurun] Revise Executor classes to have a proper container for Subgraph

- ExecutorBase has no any container for Subgraph
- LinearExecutor has a vector container
- DataflowExecutor has a map container

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
* Extract Element as a file from Linear class

* Revise Executor classes with owned proper containers

5 years agoExecute add operation in interpreter (#5088)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 30 Apr 2019 00:47:22 +0000 (09:47 +0900)]
Execute add operation in interpreter (#5088)

Implement naive interpreter execution module
Fix interpreter gtest's input/output value

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoPrepare interpreter add operation (#5060)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 29 Apr 2019 06:51:27 +0000 (15:51 +0900)]
Prepare interpreter add operation (#5060)

Add methods in ExecEnv to check and get tensor
Restrict ExecEnv to own tensors have buffer space
Assert for add operation

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoIntroduce neurun activation and padding enum class (#5075)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 29 Apr 2019 04:10:00 +0000 (13:10 +0900)]
Introduce neurun activation and padding enum class (#5075)

Introduce neurun activation type enum: none, relu, relu1, relu6
Introduce neurun padding type enum: explicit, same, valid

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago [neurun] Enable Split op in neurun (#5068)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 26 Apr 2019 04:24:19 +0000 (13:24 +0900)]
 [neurun] Enable Split op in neurun  (#5068)

* [neurun] Enable Split op in neurun

This commit enables Split op in neurun for acl_cl

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* Fix nnapi_delegate bug

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix build error by rebase

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoEnable custom oprations test (#5071)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 26 Apr 2019 00:48:56 +0000 (09:48 +0900)]
Enable custom oprations test (#5071)

Enable custom operations framework test again

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[neurun] Rename classes in model::operand (#5062)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 25 Apr 2019 04:40:07 +0000 (13:40 +0900)]
[neurun] Rename classes in model::operand (#5062)

Rename all classes in `model::operand` Many changes are included in this
commit to avoid conflicts many times.

- `operand::Object` -> `Operand`
- `operand::Info` -> `OperandInfo`
- `operand::Data` -> `Data`
- `operand::DataType` -> `DataType`
- `operand::TypeInfo` -> `TypeInfo`
- `operand::Shape` -> `Shape`
- `operand::Index` -> `OperandIndex`
- `operand::IO::Index` -> `IOIndex`
- `operand::IndexMap` -> `OperandIndexMap`
- `operand::IndexSequence` -> `OperandIndexSequence`

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
5 years agoRemove usage field in operand (#5066)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 25 Apr 2019 02:18:24 +0000 (11:18 +0900)]
Remove usage field in operand (#5066)

Remove usage field and related methods in internal operand

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years ago[neurun] Rename classes in model::operation (#5018)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 24 Apr 2019 08:08:11 +0000 (17:08 +0900)]
[neurun] Rename classes in model::operation (#5018)

* [neurun] Rename classes in model::operation

Rename the rest of classes in `model::operation` other than Operation
derivatives and subgraph modules. Many changes are included in this
commit to avoid conflicts many times.

- `operation::NodeVisitor` -> `OpVisitor`
- `operation::OperandConstraint` -> `OperandConstraint`
- `operation::IndexList` -> `OperationIndexList`
- `operation::Set` -> `OperationManager`
- `operation/Op.lst` -> `Op.lst`
- `operation/Node.Include.h` -> `Operation.Include.h`

Part of #4985

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
* Fix nits

* Rename OpXXX to OperationXXX

- OpVisitor.h => OperationVisitor.h
- Op.lst => Operations.lst

* Rebase fix

* Rename OperationManager to Operations

* Fix ifdef guards - OperandConstraint and Operation

5 years agoImplement COPY kernel for PermuteLayer (#5052)
이한종/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 24 Apr 2019 08:00:12 +0000 (17:00 +0900)]
Implement COPY kernel for PermuteLayer (#5052)

* Implement COPY kernel for PermuteLayer

PermuteLayer was supported NHWC to NCHW or vice versa. So it was not
able to handle just COPY case.

For #5044

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
* Handle paddings

5 years agoRemove operand usage check in neurun internal (#5057)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Apr 2019 07:59:51 +0000 (16:59 +0900)]
Remove operand usage check in neurun internal (#5057)

Remove operand usage check assertion

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoCheck operand usage in frontend (#5056)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Apr 2019 05:13:15 +0000 (14:13 +0900)]
Check operand usage in frontend (#5056)

Check operand usage by using operand usage vector in nnapi frontend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRemove make tflite (#5055)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Apr 2019 03:54:42 +0000 (12:54 +0900)]
Remove make tflite (#5055)

Remove make tflite command: not used anymore for svace and coverity check

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoUpdate howto document (#5047)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 24 Apr 2019 00:48:31 +0000 (09:48 +0900)]
Update howto document (#5047)

Update howto document's tflite model link

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoRename hasData to isConstant (#5046)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 10:37:34 +0000 (19:37 +0900)]
Rename hasData to isConstant (#5046)

Rename hasData to isConstant
It is used to check operand is constant

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoFix LD_LIBRARY_PATH setting for model benchmark (#5051)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 10:32:22 +0000 (19:32 +0900)]
Fix LD_LIBRARY_PATH setting for model benchmark (#5051)

Fix LD_LIBRARY_PATH setting for model benchmark to use pure acl

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoFix validation test (#5034)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 10:32:00 +0000 (19:32 +0900)]
Fix validation test (#5034)

Set operand's value for activation to ANEURALNETWORKS_FUSED_NONE
Test model's input operand numbers and size are also changed: [0, 1, 2] -> [0, 1]
By this, we can enable all validation test on acl-cl backend

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoManage operand stack and operation stack for interpretation (#5020)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 08:31:17 +0000 (17:31 +0900)]
Manage operand stack and operation stack for interpretation (#5020)

- Introduce operand ready checker
- Introduce operation stack to save prepared operation
- Manage operand stack and operation stack and show log

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoIntroduce operand usage vectorin frontend (#5039)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 07:59:17 +0000 (16:59 +0900)]
Introduce operand usage vectorin frontend (#5039)

Collect operand usage info in frontend
Clear when model building is finished
Prepare removing nnapi-dependent usage info in neurun model operand

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoIntroduce build option for tflite accuracy tool (#5048)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 07:58:13 +0000 (16:58 +0900)]
Introduce build option for tflite accuracy tool (#5048)

Introduce build option for tflite accuracy tool and set off as default

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoChange default runtime to neurun (#5042)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 07:09:14 +0000 (16:09 +0900)]
Change default runtime to neurun (#5042)

* Change default runtime to neurun

Change default runtime: pacl -> neurun

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Enable loggin runtime default

5 years ago[batch_run] A batch execution tool used to run experiments (#4507)
Saulo Aldighieri Moraes/AI R&D /SRBR/Engineer/삼성전자 [Tue, 23 Apr 2019 07:06:08 +0000 (04:06 -0300)]
[batch_run] A batch execution tool used to run experiments (#4507)

* Related: #4405

The batch execution tool can be used to run experiments. It reads a neural network model from a file and a series of input images from a directory, runs each image through the network, and collect
statistics, such as execution time and accuracy.

Signed-off-by: Saulo A. Moraes <s.moraes@samsung.com>
* Remove boost_ext reference

Build script cleanup, remove unnecessary reference to boost_ext.

Signed-off-by: Saulo A. Moraes <s.moraes@samsung.com>
* Related: #4405

  Code review fixes and tool rename from batch_run to tflite_accuracy.

Signed-off-by: Saulo A. Moraes <s.moraes@samsung.com>
5 years agoRemove operand usage getter methods (#5038)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 23 Apr 2019 04:20:20 +0000 (13:20 +0900)]
Remove operand usage getter methods (#5038)

- Remove unused isModelInput() method
- Remove usage() method. Use hasData() and model's input/output

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
5 years agoIgnore try-run result on cross build (#5031)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 22 Apr 2019 03:50:16 +0000 (12:50 +0900)]
Ignore try-run result on cross build (#5031)

* Ignore try-run result

Ignore try-run result on cross build when finding cmake threads package
Threads package is checked when finding pre-installed tflite package

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix comment

5 years agoChange ARGMAX definition (#5021)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Mon, 22 Apr 2019 02:25:55 +0000 (11:25 +0900)]
Change ARGMAX definition (#5021)

Chagne ARGMAX definition same with current NNAPI
- Allow one axis reduce
- Output rank = input rank - 1

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>