platform/core/ml/nnfw.git
4 years agoUpdate tflite_benchmark_model for TF 1.14 (#7645)
이상규/On-Device Lab(SR)/Principal Engineer/삼성전자 [Fri, 20 Sep 2019 05:27:19 +0000 (14:27 +0900)]
Update tflite_benchmark_model for TF 1.14 (#7645)

It fixes `tflite_benchmark_model` build error that happends after
tensorflow is updated to 1.14.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
4 years ago[nncc-cmake] Check external module path existance (#7646)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 20 Sep 2019 04:37:26 +0000 (13:37 +0900)]
[nncc-cmake] Check external module path existance (#7646)

Check external module path existance and download if that is not exists
Same with nnfw

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years agoMake Compatible Shapes for CPU Add (#6540)
이춘석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 20 Sep 2019 04:04:58 +0000 (13:04 +0900)]
Make Compatible Shapes for CPU Add (#6540)

- This commit addes routine for making compatible shapes for CPU Add
  kernel

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
4 years agoImplement NNAPI get output rank function (#7581)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Fri, 20 Sep 2019 03:42:00 +0000 (12:42 +0900)]
Implement NNAPI get output rank function (#7581)

Inplement NNAPI ANeuralNetworksExecution_getOutputOperandRank and wrapper function
Introduce private field and method in Execution class to check execution is finished

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[locomotiv] Layout rule for Matrix domain (#7634)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 20 Sep 2019 03:32:07 +0000 (12:32 +0900)]
[locomotiv] Layout rule for Matrix domain (#7634)

This commit updates layout rule for Matrix domain in README.md

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[moco-tf] Implement Conv2DBackpropInput canonicalization partially (#7637)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Fri, 20 Sep 2019 03:27:32 +0000 (12:27 +0900)]
[moco-tf] Implement Conv2DBackpropInput canonicalization partially (#7637)

This commit implements Conv2DBackpropInput canonicalization, except
padding inference.

4 years ago[locomotiv] Implement MatrixEncode, MatrixDecode and MatMul (#7604)
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Fri, 20 Sep 2019 01:51:15 +0000 (04:51 +0300)]
[locomotiv] Implement MatrixEncode, MatrixDecode and MatMul (#7604)

* Implemented operations on loco interpreter (locomotiv)
* Added tests for this operations

Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
4 years ago[mir_onnx] Support ConvTranspose operator (#7610)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 19 Sep 2019 15:47:42 +0000 (18:47 +0300)]
[mir_onnx] Support ConvTranspose operator (#7610)

Add support for `ConvTranspose` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[nnc] Fix DataFormatSwitcher transformation for DeConv2DOp (#7612)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 19 Sep 2019 14:42:43 +0000 (17:42 +0300)]
[nnc] Fix DataFormatSwitcher transformation for DeConv2DOp (#7612)

Rearrange axes of `output_shape` parameter when switching the data format.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[mir_onnx] Fix shape of the bias term in Conv operator converter (#7611)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Thu, 19 Sep 2019 14:33:05 +0000 (17:33 +0300)]
[mir_onnx] Fix shape of the bias term in Conv operator converter (#7611)

Reshape the bias term into [1, C, 1, 1] shape to make it broadcastable.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[exo-tflite] Header-only CanonicalNodeConverter implementation (#7639)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 08:01:12 +0000 (17:01 +0900)]
[exo-tflite] Header-only CanonicalNodeConverter implementation (#7639)

Let's put CanonicalNodeConverter::run method implementation in the
header.

With this change, clients can inherit CanonicalNodeConverter
without "CanonicalNodeConverter.cpp" update.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[loco] Use GTest_AddTest macro (#7635)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 07:15:41 +0000 (16:15 +0900)]
[loco] Use GTest_AddTest macro (#7635)

This commit simplifies loco's CMakeLists.txt by using GTest_AddTest macro.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[exo-tflite] function to convert locoex padding to tflite padding (#7636)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 06:52:44 +0000 (15:52 +0900)]
[exo-tflite] function to convert locoex padding to tflite padding (#7636)

This function converts locoex padding to tflite padding.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[plier-tf] Enable strict build (#7631)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 06:24:45 +0000 (15:24 +0900)]
[plier-tf] Enable strict build (#7631)

* [plier-tf] Enable strict build

Let's apply nncc common bulid flags (such as Wall).

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

4 years ago[exo-tflite] adding converter for loco::ConstGen (#7632)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 06:06:17 +0000 (15:06 +0900)]
[exo-tflite] adding converter for loco::ConstGen (#7632)

converter and its test for loco::ConstGen (to locoex::TFLConst) was added.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[locop] Add TensorReduce visit function (#7597)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 06:05:55 +0000 (15:05 +0900)]
[locop] Add TensorReduce visit function (#7597)

* [locop] Add TensorReduce visit function

This commit will add `TensorReduce` visit function

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* apply comments

* fix mistake

4 years ago[exo-tflite] test case for FeatureBiasAddConverter (#7583)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 05:22:20 +0000 (14:22 +0900)]
[exo-tflite] test case for FeatureBiasAddConverter (#7583)

* [exo-tflite] test case for FeatureBiasAddConverter

An unit test to check `FeatureBiasAddConverter` is added.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* fix typo and wrong index of constgen->at(1)

4 years ago[moco-tf] Basic structure for Conv2DBackpropInput canonicalize (#7627)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 05:21:42 +0000 (14:21 +0900)]
[moco-tf] Basic structure for Conv2DBackpropInput canonicalize (#7627)

This commit adds basic structure for TFConv2DBackpropInput canonicalize
stage.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[plier-tf] Fix copy_shape (#7615)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 05:14:29 +0000 (14:14 +0900)]
[plier-tf] Fix copy_shape (#7615)

copy_shape is expected to throw an exception if there is any unknown
dimension (-1), but the current implementation never throws an
exception.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[res] UNIT_Conv2DBackpropInput_002 (#7606)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 04:39:47 +0000 (13:39 +0900)]
[res] UNIT_Conv2DBackpropInput_002 (#7606)

* [res] UNIT_Conv2DBackpropInput_002

This commit adds UNIT_Conv2DBackpropInput_002, unit test for
Conv2DBackpropInput with valid padding & non-tight case.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
* Review feedback

4 years ago[res] UNIT_Conv2DBackpropInput_000 (#7605)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 04:25:30 +0000 (13:25 +0900)]
[res] UNIT_Conv2DBackpropInput_000 (#7605)

This commit adds UNIT_Conv2DBackpropInput_000, unit test for
Conv2DBackpropInput with valid padding & tight fit case.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years agoIntroduce T/F Testcase explain script (#7603)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 04:06:26 +0000 (13:06 +0900)]
Introduce T/F Testcase explain script (#7603)

* Introduce T/F Testcase explain script

This "explain" script generates a summary over commited TensorFlow
testcases.

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

4 years ago[moco-tf] Support TFMean Canonicalization (#7594)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 04:05:44 +0000 (13:05 +0900)]
[moco-tf] Support TFMean Canonicalization (#7594)

* [moco-tf] Support TFMean Canonicalization

This commit will enable for supporting `TFMean` canonicalization

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* apply comments

4 years ago[moco-tf] Import Conv2dBackpropInput (#7579)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 03:55:56 +0000 (12:55 +0900)]
[moco-tf] Import Conv2dBackpropInput (#7579)

This commit introduces import stage of Conv2dBackpropInput.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[loco] Shape inference for TransposedConv2D (#7564)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Thu, 19 Sep 2019 03:55:12 +0000 (12:55 +0900)]
[loco] Shape inference for TransposedConv2D (#7564)

This commit introduces shape inference for TransposedConv2D and its test

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[moco-tf] ResolveSquaredDifference to TFOptimizier (#7593)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 03:54:25 +0000 (12:54 +0900)]
[moco-tf] ResolveSquaredDifference to TFOptimizier (#7593)

This will add ResolveSquaredDifference transform to TFOptimizier phase

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[nnfw-cmake] Rename cmake module file for subdirecories (#7617)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 03:53:38 +0000 (12:53 +0900)]
[nnfw-cmake] Rename cmake module file for subdirecories (#7617)

Use same name with nncc

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[nncc-cmake] Support excluding install in add_extdirectory cmake macro (#7590)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 02:13:26 +0000 (11:13 +0900)]
[nncc-cmake] Support excluding install in add_extdirectory cmake macro (#7590)

- Support EXCLUDE_FROM_ALL in add_extdirectory (same with nnfw)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[nncc-cmake] Rename cmake module file for subdirecories (#7588)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 02:09:05 +0000 (11:09 +0900)]
[nncc-cmake] Rename cmake module file for subdirecories (#7588)

Change snake-case type file naming to camel-case (same with other modules)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years agoRemove boolean flag setting in cmake module (#7589)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 02:07:04 +0000 (11:07 +0900)]
Remove boolean flag setting in cmake module (#7589)

Module don't use flag (same with nncc)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[nncc-cmake] Revise ExternalSourceTools status message print (#7601)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Thu, 19 Sep 2019 01:46:28 +0000 (10:46 +0900)]
[nncc-cmake] Revise ExternalSourceTools status message print (#7601)

- Remove remain "--" print and use STATUS mode
- More information when download failed

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[exo-tflite] Enable convert Div Sub (#7585)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 01:32:05 +0000 (10:32 +0900)]
[exo-tflite] Enable convert Div Sub (#7585)

This will enable Canonical Node Convert of Div and Sub node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[exo-tflite] Broadcast conversion for Div, Sub (#7591)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 01:31:42 +0000 (10:31 +0900)]
[exo-tflite] Broadcast conversion for Div, Sub (#7591)

This will enable TensorBroadcastConverter to handle Div and Sub node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[moco-tf] Reduce dup code for copy_shapedata (#7592)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Thu, 19 Sep 2019 01:31:10 +0000 (10:31 +0900)]
[moco-tf] Reduce dup code for copy_shapedata (#7592)

This will update copy_shapedata() to use get_node_shape() to reduce duplicate codes in FixShapeTransform

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years agoIntroduce ONNXTests resource (#7565)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 23:33:22 +0000 (08:33 +0900)]
Introduce ONNXTests resource (#7565)

* Introduce ONNXTests resource

This commit introduces ONNXTests resource which provide simple ONNX
models for testing.

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

4 years ago[mir_onnx] Support Sqrt operator (#7548)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 14:51:59 +0000 (17:51 +0300)]
[mir_onnx] Support Sqrt operator (#7548)

Add support for `Sqrt` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[mir_onnx] Support Tanh operator (#7608)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 14:49:14 +0000 (17:49 +0300)]
[mir_onnx] Support Tanh operator (#7608)

Add support for `Tanh` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[mir_tflite] Support Rsqrt operator (#7546)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 14:43:59 +0000 (17:43 +0300)]
[mir_tflite] Support Rsqrt operator (#7546)

Add support for `Rsqrt` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[mir_onnx] Support Reciprocal operator (#7598)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 13:36:50 +0000 (16:36 +0300)]
[mir_onnx] Support Reciprocal operator (#7598)

Add support for `Reciprocal` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years ago[mir_onnx] Support Sub operator (#7547)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 13:32:49 +0000 (16:32 +0300)]
[mir_onnx] Support Sub operator (#7547)

Add support for `Sub` operator.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years agoIntroduce PermutationOperationPass (#7254)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 11:37:34 +0000 (20:37 +0900)]
Introduce PermutationOperationPass (#7254)

This commit introduces PermutationOperationPass that can change layout of each op

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
4 years ago[exo-tflite] Add TensorReduce visit function (#7596)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 11:11:42 +0000 (20:11 +0900)]
[exo-tflite] Add TensorReduce visit function (#7596)

This commit will add `TensorReduce` visit function

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
4 years ago[nnfw-cmake] StampTools module (#7595)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 10:29:59 +0000 (19:29 +0900)]
[nnfw-cmake] StampTools module (#7595)

Extract StampTools module from ExternalSourceTools module (same with nncc)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[moco-tf] Introduce SqueezeReduceNode transformation (#7580)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 10:16:05 +0000 (19:16 +0900)]
[moco-tf] Introduce SqueezeReduceNode transformation (#7580)

* [moco-tf] Introduce SqueezeReduceNode transformation

If Reduce node such as `TFMean` do not keep dimensions, `TFSqueeze` should be inserted.
This commit will introduce `SqueezeReduceNode` transformation for that.

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* revise comments

* add assert

4 years agoImplement NNAPI synchronous execution function (#7574)
오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 10:09:46 +0000 (19:09 +0900)]
Implement NNAPI synchronous execution function (#7574)

Inplement NNAPI ANeuralNetworksExecution_compute function and wrapper
Wrapper uses neurun's synchronous execution method

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
4 years ago[exo-tflite] Enable Add, Mul unit tests (#7586)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 10:08:01 +0000 (19:08 +0900)]
[exo-tflite] Enable Add, Mul unit tests (#7586)

This will enable Add, Mul unit tests having different input shape

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[exo-tflite] adding FeatureBiasAddConverter into convert.cpp (#7584)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 09:42:02 +0000 (18:42 +0900)]
[exo-tflite] adding FeatureBiasAddConverter into convert.cpp (#7584)

This makes FeatureBiasAddConverter work when a knob is on.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[exo-tflite] enclosing namespace exo (#7577)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 08:07:54 +0000 (17:07 +0900)]
[exo-tflite] enclosing namespace exo (#7577)

This commit enclosed codes with `namespace exo`.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[custom op] Remove questionable nnfw_core->nnfw_api dependency (#7410)
Vladimir Plazun/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 08:03:48 +0000 (11:03 +0300)]
[custom op] Remove questionable nnfw_core->nnfw_api dependency (#7410)

* [custom op] Remove questionable nnfw_core->nnfw_api dependency

Split KernelRegistry into KernelRegistry and KernelBuilder

Signed-off-by: Vladimir Plazun <v.plazun@samsung.com>
* format fix

* make common frontend into static lib

4 years ago[encodump] Use safemain (#7578)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 07:59:05 +0000 (16:59 +0900)]
[encodump] Use safemain (#7578)

Let's reuse safemain's main implemenation.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[exo-tfltie] type and shape inference for TFLConst (#7567)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 07:43:04 +0000 (16:43 +0900)]
[exo-tfltie] type and shape inference for TFLConst (#7567)

This adds type and shape inference for `TFLConst`.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[moco-tf] Type inference for Conv2DBackpropInput (#7573)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 07:41:39 +0000 (16:41 +0900)]
[moco-tf] Type inference for Conv2DBackpropInput (#7573)

This commit adds type inference for Conv2DBackpropInput.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years agoRename to moco-integration-test (#7569)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 07:40:05 +0000 (16:40 +0900)]
Rename to moco-integration-test (#7569)

This commit rename 'mocotest-tf' directory to 'moco-integration-test'.
It also updates README.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[exo-tflite] Fix TFLMul export option (#7566)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 07:36:34 +0000 (16:36 +0900)]
[exo-tflite] Fix TFLMul export option (#7566)

This will fix TFLMul export option that was not corrected from copy-paste
As the option is a dummy, there was no problem with actual execution

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[moco-tf] Summary for TFConv2DBackpropInput (#7575)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 07:34:55 +0000 (16:34 +0900)]
[moco-tf] Summary for TFConv2DBackpropInput (#7575)

This commit introduces summary for TFConv2DBackpropInput.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[loco] Introduce MatrixEncode, MatrixDecode and MatrixMul operations (#7406)
Павел Ильютченко/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 07:34:44 +0000 (10:34 +0300)]
[loco] Introduce MatrixEncode, MatrixDecode and MatrixMul operations (#7406)

* Added MatrixEncode, MatrixDecode and MatrixMul nodes
* Added tests for them
* Defined CanonicalShapeInferenceRule and TypeInference

Signed-off-by: Pavel Iliutchenko <p.iliutchenk@samsung.com>
4 years ago[plier-tf] Hide Implementation Details (#7576)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 07:31:33 +0000 (16:31 +0900)]
[plier-tf] Hide Implementation Details (#7576)

No code outside plier-tf uses membuf and imemstream classes.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[loco] Introduce TensorReduce Operation (#7497)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 06:17:24 +0000 (15:17 +0900)]
[loco] Introduce TensorReduce Operation (#7497)

* [loco] Introduce TensorReduce Operation

This commit will introduce `TensorReduce` operation in loco

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* fix wrong shapeInference

* revise comments

4 years ago[tf2tflite] Enable exo loggers with TF2TFLITE knob (#7561)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 06:10:34 +0000 (15:10 +0900)]
[tf2tflite] Enable exo loggers with TF2TFLITE knob (#7561)

With this commit, users are able to enable exo loggers with
TF2TFLITE_Log_Backend knob.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[moco-tf] Shape inference for Conv2DBackpropInput (#7539)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 06:07:58 +0000 (15:07 +0900)]
[moco-tf] Shape inference for Conv2DBackpropInput (#7539)

This commit implements shape inference for Conv2DBackpropInput.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[exo-tflite] revising README.md (#7529)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 06:05:32 +0000 (15:05 +0900)]
[exo-tflite] revising README.md (#7529)

* [exo-tflite] revising README.md

Steps for adding a new TFL node  was added.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* revised

4 years ago[exo-tflite] removing old graph builder (#7524)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 06:05:07 +0000 (15:05 +0900)]
[exo-tflite] removing old graph builder (#7524)

Old graph builder (`PushPullGraph<..>`) is now replaced with `TestGraph`.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[exo-tflite] knob to convert loco::ConstGen (#7477)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 06:04:47 +0000 (15:04 +0900)]
[exo-tflite] knob to convert loco::ConstGen (#7477)

This adds knob to convert loco::ConstGen.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[exo-exporter] adding an empty visit() for TFLConst in OperationExporter (#7568)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 06:03:19 +0000 (15:03 +0900)]
[exo-exporter] adding an empty visit() for TFLConst in OperationExporter (#7568)

`TFLConst` won't be converted into any operation so empty `visit()` was added into `OperationExporter`.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[exo-tflite] Knob for convert Div, Sub (#7560)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:15:16 +0000 (13:15 +0900)]
[exo-tflite] Knob for convert Div, Sub (#7560)

* [exo-tflite] Knob for convert Div, Sub

This will add knob for converting EltwiseDiv and EltwiseSub

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

4 years ago[moco-tf] Introduce ResolveSquaredDifference transform (#7559)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:14:56 +0000 (13:14 +0900)]
[moco-tf] Introduce ResolveSquaredDifference transform (#7559)

* [moco-tf] Introduce ResolveSquaredDifference transform

This will introduce ResolveSquaredDifference transform that decomposes TFSquaredDifference node into TFSub and TFMul nodes

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
* use node->graph()

4 years ago[exo-tflite] Shape inference for TFLDiv, TFLSub (#7552)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:14:30 +0000 (13:14 +0900)]
[exo-tflite] Shape inference for TFLDiv, TFLSub (#7552)

This will enable shape inference for TFLDiv, TFLSub node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[exo-tflite] Type inference for TFLDiv, TFLSub (#7551)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:14:13 +0000 (13:14 +0900)]
[exo-tflite] Type inference for TFLDiv, TFLSub (#7551)

This will enable type inference for TFLDiv, TFLSub node

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[exo-tflite] Export TFLDiv, TFLSub (#7554)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:13:30 +0000 (13:13 +0900)]
[exo-tflite] Export TFLDiv, TFLSub (#7554)

* [exo-tflite] Export TFLDiv, TFLSub

This will enable export for TFLDiv and TFLSub node

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

4 years ago[exo-tflite] Introduce EltwiseBinaryConverter and refactor (#7533)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 04:12:22 +0000 (13:12 +0900)]
[exo-tflite] Introduce EltwiseBinaryConverter and refactor (#7533)

This will introduce EltwiseBinaryConverter and refactor EltwiseAddConverter and EltwiseMulConverter using this

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[locoex-customop] Remove plier-tf dependency (#7540)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 04:12:02 +0000 (13:12 +0900)]
[locoex-customop] Remove plier-tf dependency (#7540)

No code in locoex-customop uses plier-tf.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[mocotest-tf] Rename target to 'moco_integration_test' (#7541)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 04:11:52 +0000 (13:11 +0900)]
[mocotest-tf] Rename target to 'moco_integration_test' (#7541)

This commit renames target 'moco_test_tf' to 'moco_integration_test'.
Related variable names also match to be 'moco_integration_*'

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[moco-tf] Enable importing Mean (#7543)
남궁석/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 04:11:34 +0000 (13:11 +0900)]
[moco-tf] Enable importing Mean (#7543)

* [moco-tf] Enable importing Mean

This commit will enable importing `Mean` as `TFMean`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
* revise trivials

4 years ago[locop] Use GenericNodeSummaryBuilder as default fallback (#7553)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 04:11:00 +0000 (13:11 +0900)]
[locop] Use GenericNodeSummaryBuilder as default fallback (#7553)

This commit replaces BuiltinNodeSummaryBuild with
GenericNodeSummaryBuilder.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[exo-tflite] Expose Logging Context (#7531)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 03:39:49 +0000 (12:39 +0900)]
[exo-tflite] Expose Logging Context (#7531)

This commit exposes exo-tflite Logging Context to public in order to
allos downstream tools to control exo-tflite logging system.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[tf2tflite] Use tfinfo target (#7532)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 02:44:31 +0000 (11:44 +0900)]
[tf2tflite] Use tfinfo target (#7532)

Let's use tfinfo instead of deprecated nnkit_support_tftestinfo.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[exo-tflite] Turn on knobs for style transfer related (#7530)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 02:40:16 +0000 (11:40 +0900)]
[exo-tflite] Turn on knobs for style transfer related (#7530)

This will turn on knobs for style transfer related switches
- enable TFL dialect
- enable convert Add, Aul to TFL dialect
- enable resolve TensorBroadcast

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[neurun] Added sanity checks for profiling mode (#7412)
Ivan Vagin/AI Tools Lab /SRR/Engineer/삼성전자 [Wed, 18 Sep 2019 01:58:02 +0000 (04:58 +0300)]
[neurun] Added sanity checks for profiling mode (#7412)

* Added sanity checks for profiling mode
* Fixed missing profiling info exception type

Signed-off-by: Ivan Vagin <ivan.vagin@samsung.com>
4 years ago[neurun] check null before accessing session (#7536)
이상규/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 01:38:08 +0000 (10:38 +0900)]
[neurun] check null before accessing session (#7536)

session will be null-checked before dereferencing.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
4 years ago[res] Unit test of RealDiv 001 (#7549)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 01:37:31 +0000 (10:37 +0900)]
[res] Unit test of RealDiv 001 (#7549)

This will add RealDiv unit test 001 for broadcasting

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[moco-tf] Add ResolveSquaredDifference knob (#7558)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Wed, 18 Sep 2019 01:19:22 +0000 (10:19 +0900)]
[moco-tf] Add ResolveSquaredDifference knob (#7558)

This will add ResolveSquaredDifference knob

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years agoMake to support Neg op for acl neon (#7535)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Wed, 18 Sep 2019 01:09:23 +0000 (10:09 +0900)]
Make to support Neg op for acl neon (#7535)

This commit makes to support Neg op for acl neon.
  - Introduce NEElementwiseUnaryKernel and NENegLayer
  - Apply NENegLayer layer for neurun

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
4 years ago[moco-tf] Use tf as prefix (#7489)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Wed, 18 Sep 2019 00:00:57 +0000 (09:00 +0900)]
[moco-tf] Use tf as prefix (#7489)

Let's use "tf" prefix instead of "TF" to be compatiable with canonical
prefix.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[locop] Introduce GenericNodeSummaryBuilder (#7518)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 23:46:10 +0000 (08:46 +0900)]
[locop] Introduce GenericNodeSummaryBuilder (#7518)

This commit introduces GenericNodeSummaryBuilder which uses "dialect"
address and opnum as opname.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[loco] Backward plane inference (#7458)
박천교/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 17 Sep 2019 23:33:09 +0000 (08:33 +0900)]
[loco] Backward plane inference (#7458)

This commit introduces backward plane inference, which would be used for
canonical shape inference rule, especially for TransposedConv2D.

Signed-off-by: Cheongyo Bahk <ch.bahk@samsung.com>
4 years ago[nnc] Remove C generating backend (#7545)
Сергей Баранников/AI Tools Lab /SRR/Engineer/삼성전자 [Tue, 17 Sep 2019 16:52:58 +0000 (19:52 +0300)]
[nnc] Remove C generating backend (#7545)

Remove C generating backend skeleton.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>
4 years agoMake to support Dequantize op for acl neon (#7522)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 17 Sep 2019 11:48:29 +0000 (20:48 +0900)]
Make to support Dequantize op for acl neon (#7522)

This commit makes to support Dequantize op for acl neon.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
4 years agoMake to support ReduceSum op for acl neon (#7520)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 17 Sep 2019 11:47:55 +0000 (20:47 +0900)]
Make to support ReduceSum op for acl neon (#7520)

This commit makes to support ReduceSum op for acl neon

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
4 years ago[exo-tflite] Introducing locoex::TFLConst (#7519)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 11:00:28 +0000 (20:00 +0900)]
[exo-tflite] Introducing locoex::TFLConst (#7519)

locoex::TFLConst is added to represent tensor data. (note that this is not for a specific tflite op)

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
4 years ago[enco] Extract Boilerplate as Helper (#7523)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 10:19:08 +0000 (19:19 +0900)]
[enco] Extract Boilerplate as Helper (#7523)

This commit extracts boilerplate in ConstantFoldingTest as a helper
in order to reduce code duplication.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years agoNull-terminating check for backend conf api (#7516)
이춘석/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 09:53:09 +0000 (18:53 +0900)]
Null-terminating check for backend conf api (#7516)

* Null-terminating check for backend conf api

- Add null terminaing function
- Add check routine

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
* Give name to arg max length in backend conf

- Add macro for max length
- 32 for backend name, 64 for op name

* fix format

* add null check

* add null check to nnfw_set_op_backend

* Check null inside nnfw_session

4 years ago[moco-tf] get_node_shape helper in copy_shapedata (#7528)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 09:26:48 +0000 (18:26 +0900)]
[moco-tf] get_node_shape helper in copy_shapedata (#7528)

This will introduce get_node_shape helper in copy_shapedata of FixShape

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years ago[moco-tf] copy shape for binary input nodes (#7525)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 09:07:17 +0000 (18:07 +0900)]
[moco-tf] copy shape for binary input nodes (#7525)

This will update FixShape to use correct copy_shapedata() for binary input nodes

Signed-off-by: SaeHie Park <saehie.park@samsung.com>
4 years agoRemove "moco" module (#7509)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 08:34:09 +0000 (17:34 +0900)]
Remove "moco" module (#7509)

No compiler module depends on "moco" module.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[plier-tf] Declare dependent library visibility (#7511)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 08:33:36 +0000 (17:33 +0900)]
[plier-tf] Declare dependent library visibility (#7511)

Let's declare the visibility of dependent libraries explicitly.
This change allows us to introduce PRIVATE dependencies.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[plier-tf] Remove unnecessary const qualifier (#7513)
박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 [Tue, 17 Sep 2019 08:33:22 +0000 (17:33 +0900)]
[plier-tf] Remove unnecessary const qualifier (#7513)

"const" qualifier on function return type has no effect.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
4 years ago[logo] Simplifying `BiasEncode - BiasDecode` subgraph (#7481)
윤현식/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 08:33:12 +0000 (17:33 +0900)]
[logo] Simplifying `BiasEncode - BiasDecode` subgraph (#7481)

* [logo] Simplifying `BiasEncode - BiasDecode` subgraph

With this commit, logo will Simplifying `BiasEncode - BiasDecode` subgraph into loco::Forward

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
* make code shorter

* remove 'using namespace loco;'

4 years agoMake to support ArgMax op for acl neon (#7515)
장지섭/On-Device Lab(SR)/Engineer/삼성전자 [Tue, 17 Sep 2019 08:23:58 +0000 (17:23 +0900)]
Make to support ArgMax op for acl neon (#7515)

This commit makes to support ArgMax op for acl neon except int32 type.
  - Introduce NEArgMinMax layer
  - Apply NEArgMinMax layer for neurun

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
4 years ago[tflite2circle] Introduce tflite2circle (#7483)
이상규/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 08:21:01 +0000 (17:21 +0900)]
[tflite2circle] Introduce tflite2circle (#7483)

It introduces `tflite2circle` tool.
Its input is `tflite` and output is `circle`.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
4 years ago[moco-tf] copy_shapedata with two inputs (#7499)
박세희/On-Device Lab(SR)/Principal Engineer/삼성전자 [Tue, 17 Sep 2019 07:59:34 +0000 (16:59 +0900)]
[moco-tf] copy_shapedata with two inputs (#7499)

* [moco-tf] copy_shapedata with two inputs

This will introduce copy_shapedata that can receive two inputs and return shape from broadcast algorithm for binary operation nodes

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