platform/upstream/nnstreamer.git
4 years ago[Spec] Update version 1.2.0 to 1.3.0 of rpm spec
Yongjoo Ahn [Thu, 16 Jan 2020 00:17:23 +0000 (09:17 +0900)]
[Spec] Update version 1.2.0 to 1.3.0 of rpm spec

Resolve version mismatch of .spec with current dev version 1.3.0

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Tests/FilterExtCommon] Fix test case failure
Wook Song [Tue, 14 Jan 2020 02:35:11 +0000 (11:35 +0900)]
[Tests/FilterExtCommon] Fix test case failure

This patch fixes failure in the test cases generated from
unittest_tizen_template.cc.in. This is a bug of [1]

[1] cbb65b01278b39789e69c6de888b3e0e78b3e5db

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Spec] cleanup build option
Jaeyun [Fri, 10 Jan 2020 09:34:31 +0000 (18:34 +0900)]
[Spec] cleanup build option

move to proper build option and cleanup unnecessary option.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[CodeClean] remove unnecessary code
Jaeyun [Fri, 10 Jan 2020 09:26:17 +0000 (18:26 +0900)]
[CodeClean] remove unnecessary code

Now single API does not use the pipeline. Remove unnecessary function to get caps from tensor-info.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[common] Added proper GstCaps framerate comparison
Parichay Kapoor [Thu, 9 Jan 2020 12:19:03 +0000 (21:19 +0900)]
[common] Added proper GstCaps framerate comparison

Added proper comparison of GstCaps framerate (1/2 == 2/4)
Added corresponding unittest cases

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Tests/FilterExtCommon] Generate test cases using meson
Wook Song [Fri, 10 Jan 2020 11:08:01 +0000 (20:08 +0900)]
[Tests/FilterExtCommon] Generate test cases using meson

In order to generate test cases from the template file to which the most
recent modification has been applied, this patch changes the approach
for the test case generation from the existing bash script to the meson
build script.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[CodeClean] fix coverity issue
Jaeyun [Mon, 13 Jan 2020 04:47:51 +0000 (13:47 +0900)]
[CodeClean] fix coverity issue

check exception case in testcases.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Caffe2] Change how headers are included
HyoungJoo Ahn [Mon, 13 Jan 2020 05:17:15 +0000 (14:17 +0900)]
[Caffe2] Change how headers are included

The headers are not in this project. Though it does not occur errors, to prevent misunderstanding, it needs to be fixed.

Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
4 years agoDocumentation Update of 2020.01
MyungJoo Ham [Fri, 10 Jan 2020 04:45:15 +0000 (13:45 +0900)]
Documentation Update of 2020.01

We have been progressed a lot. Update the documentation accordingly.

V2: fixed typos

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Ext/Filter/TFLite] Enable a TensorFilter's property, verify_model_path
Wook Song [Fri, 10 Jan 2020 07:03:39 +0000 (16:03 +0900)]
[Ext/Filter/TFLite] Enable a TensorFilter's property, verify_model_path

This patch enables a GstTensorFilter's property, verify_model_path,
which is newly added.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Filter/Common/Reload] Define the behavior for verify_model_path
Wook Song [Fri, 10 Jan 2020 04:57:42 +0000 (13:57 +0900)]
[Filter/Common/Reload] Define the behavior for verify_model_path

This patch defines the behaviour of reloading model files when the
verify_model_path is set to TRUE.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Tests/Script] Make the run_unittests_binaries script find executables
Wook Song [Fri, 10 Jan 2020 10:57:26 +0000 (19:57 +0900)]
[Tests/Script] Make the run_unittests_binaries script find executables

This patch makes the run_unittests_binaries script in ./packaging find
only executable files.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years agoREADME: update reviewers and project status
MyungJoo Ham [Thu, 9 Jan 2020 08:09:34 +0000 (17:09 +0900)]
README: update reviewers and project status

- Added the two new reviewers
- Added more info on the project status

V2: typo-fix and added an omitted item.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Tests/FilterExtCommon] hotfix: Fix build errors, dangling-else
Wook Song [Fri, 10 Jan 2020 10:11:55 +0000 (19:11 +0900)]
[Tests/FilterExtCommon] hotfix: Fix build errors, dangling-else

This patch fixes build errors related to 'dangling-else' (error:
suggest explicit braces to avoid ambiguous 'else') in
unittest_tizen_template.cc.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[tflite/bug] Set accelerator bug solve
Parichay Kapoor [Thu, 9 Jan 2020 23:32:09 +0000 (08:32 +0900)]
[tflite/bug] Set accelerator bug solve

Bug: Not setting accelerator ignored the use_nnapi configuration
This is resolved

Enable setting accelerator unittest non-tizen platforms as well

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[CodeClean] change file extension
Jaeyun [Thu, 9 Jan 2020 07:22:25 +0000 (16:22 +0900)]
[CodeClean] change file extension

1. change file extension to .cc in testcases
2. change path to generate ext-test files

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Android] assertion to load conf file
Jaeyun [Thu, 9 Jan 2020 07:12:59 +0000 (16:12 +0900)]
[Android] assertion to load conf file

Now Android functions do not use the conf file, remove assertion to run Android library.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Android/Test] testcases for updated API
Jaeyun [Tue, 31 Dec 2019 07:02:16 +0000 (16:02 +0900)]
[Android/Test] testcases for updated API

Add testcase for SNAP and updated Java API.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Api/Android] support SNAP in Android API
Jaeyun [Tue, 31 Dec 2019 04:34:12 +0000 (13:34 +0900)]
[Api/Android] support SNAP in Android API

In Java APIs, handle multiple model files and custom option to support SNAP.
Define enum for supported NNFW and implement native functions.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Api/CodeClean] function to create single handle
Jaeyun [Fri, 27 Dec 2019 04:21:47 +0000 (13:21 +0900)]
[Api/CodeClean] function to create single handle

To reduce complexity, add function to create and initialize the single-shot handle.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[C-Api] handle multi files and custom option
Jaeyun [Tue, 24 Dec 2019 11:36:10 +0000 (20:36 +0900)]
[C-Api] handle multi files and custom option

define struct to handle multiple model files and custom option string.
Prepare SNAP single-shot API.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Api/CodeClean] macro to check nnfw
Jaeyun [Tue, 24 Dec 2019 11:28:56 +0000 (20:28 +0900)]
[Api/CodeClean] macro to check nnfw

1. define macro to check nnfw is available
2. update comment and code clean

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[python/ext] Added basic unittests for python
Parichay Kapoor [Tue, 24 Dec 2019 05:00:18 +0000 (14:00 +0900)]
[python/ext] Added basic unittests for python

- Python initialize and finalize is done once for the module rather than doing it per object
Doing initialization per object causes errors when importing modules
- Replaced Assert with appropriate error handing for python extension of tensor filter
- Added basic unittests for python which covers the failure cases

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Filter] Add an option to centralize checking the path of model files
Wook Song [Mon, 6 Jan 2020 07:43:22 +0000 (16:43 +0900)]
[Filter] Add an option to centralize checking the path of model files

Currently, each sub-plugin of the tensor filter should check the
validity of the path of the model files. In general, it leads to
unnecessary code duplication. Moreover, sometimes those are not checked
at all. This patch adds an option to centralize checking the path of model
files so that each sub-plugin is in charge of the validity of model files
only when this option is set to FALSE.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago [TEST] Unit test case for NNFW ML-API
gichan-jang [Tue, 7 Jan 2020 08:16:28 +0000 (17:16 +0900)]
 [TEST] Unit test case for NNFW ML-API
Unit tests are added to verify NNFW with ML-API.
Detail of added test case is :
Test multidimensional input/output us supported in NNFW

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[nnfw] Support setInputDimension with nnfw
Parichay Kapoor [Fri, 3 Jan 2020 04:49:33 +0000 (13:49 +0900)]
[nnfw] Support setInputDimension with nnfw

With updated nnfw API, support setInputDimension with nnfw
Although the API is defined, its not yet implemeted interally
So, the unittest is added, but disabled

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years agoCODEOWNERS: add new members
MyungJoo Ham [Thu, 9 Jan 2020 06:26:13 +0000 (15:26 +0900)]
CODEOWNERS: add new members

Notify Gichan and Yongjoo for any changes in the repo.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoTest: filter/edgetpu unit testing
MyungJoo Ham [Mon, 16 Dec 2019 11:19:54 +0000 (20:19 +0900)]
Test: filter/edgetpu unit testing

Edge-TPU subplugin unit test cases.
(1 positive and 2 negative cases)

This requires edgetpu package update in dashboard.tizen.org
(We are waiting for it)

V2: added test model condition for edgetpu testcase.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[CONF] override NNSTREAMER_CONF_FILE with env-var conf in non-Tizen
Dongju Chae [Wed, 8 Jan 2020 05:12:44 +0000 (14:12 +0900)]
[CONF] override NNSTREAMER_CONF_FILE with env-var conf in non-Tizen

This commit overrides NNSTREAMER_CONF_FILE with env-var conf in
non-Tizen system. So, env-var conf has a higher priority than others.

With this feature, some unittests using a conf file from env-var has
passed even if the default conf file (/etc/nnstremer.ini) exists.
Note that Tizen does not allow to set configurations from env-var.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years agoDist/Tizen: show unit test result summary
MyungJoo Ham [Wed, 18 Dec 2019 04:24:50 +0000 (13:24 +0900)]
Dist/Tizen: show unit test result summary

Aggregate all gtest and SSAT results at the end of testing.

After this, unittest-enabled gbs will show results like:

```
[  117s] GTest (total 232 cases)
[  117s]   Passed: 232 / Failed: 0 / Ignored: 0
[  117s] SSAT (total 513 cases)
[  117s]   Passed: 497 / Failed: 0 / Ignored: 16
[  117s] Grand Total: 745 cases
[  117s]   Passed: 729 / Failed: 0 / Ignored: 16
```

Changes in V2:
- made sure to call close for opened f
- use os.path.splitext
Chnaged in V3:
- with makes exceptions with fail. Handle them.

Suggested TODO in later PRs:
- Count negative test cases.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Filter/OpenVino] Fill the mandatory callback, invoke
Wook Song [Fri, 3 Jan 2020 05:39:31 +0000 (14:39 +0900)]
[Filter/OpenVino] Fill the mandatory callback, invoke

This patch adds a method to run inference for given input tensors using
InferenceEngine on the acceleration device. It also implements the
mandatory callback of GstTensorFilterFramework, invoke, with such method.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[filter/ext] Added basic unittests for all filter extensions
Parichay Kapoor [Mon, 23 Dec 2019 10:29:41 +0000 (19:29 +0900)]
[filter/ext] Added basic unittests for all filter extensions

Added basic unittests for all the the filter extensions
This creates failure cases of unittests for the interface defined with GstTensorFilterFramework
for most of the extensions, and performs basic unittests
Updated the unittest list in packaging files as well

Also added a script file to run all the test cases inside a folder

V2:
Split test cases into successful and negative ones

TODO: Support python extensions in this as well

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[single/hw] Check hw support for framework
Parichay Kapoor [Thu, 19 Dec 2019 03:07:38 +0000 (12:07 +0900)]
[single/hw] Check hw support for framework

Add implementation to check hw support for framework while checking
for its availability.
Added unittests for this functionality with tensorflow-lite

V2:
check hardware support performed while parsing for available hardwares
Split availability test cases into positive and negative

V3:
checkHWSupport->checkAvailability

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Android] add clone method
Jaeyun [Thu, 2 Jan 2020 10:15:48 +0000 (19:15 +0900)]
[Android] add clone method

Add clone method in tensors info class.
This is to prevent meta-info change after allocating data from passed information.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Android] define enumeration for data type and state
Jaeyun [Mon, 23 Dec 2019 10:43:27 +0000 (19:43 +0900)]
[Android] define enumeration for data type and state

Define enumeration for tensor data type and pipeline state (sync with C# API)

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Android] tensors-data class with metadata
Jaeyun [Fri, 22 Nov 2019 08:22:27 +0000 (17:22 +0900)]
[Android] tensors-data class with metadata

1. Add tensors info in TensorsData class.
2. Handle error case when failed to create pipe info in native functions.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[PythonFilter] Add pragma option to fix compile time error for gcc v9.6 accepted/tizen/unified/20200109.065529 submit/tizen/20200108.082636
Sangjung Woo [Wed, 8 Jan 2020 07:10:48 +0000 (16:10 +0900)]
[PythonFilter] Add pragma option to fix compile time error for gcc v9.6

This patch newly adds pragma option to fix compile time error for gcc
v9.6.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Spec] Add python-filter option
Sangjung Woo [Wed, 8 Jan 2020 05:41:12 +0000 (14:41 +0900)]
[Spec] Add python-filter option

This patch newly adds the python-filter option which could support to be
disabled.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Filter/OpenVino] Handle the case when the sub-plugin is re-opened
Wook Song [Thu, 2 Jan 2020 09:28:35 +0000 (18:28 +0900)]
[Filter/OpenVino] Handle the case when the sub-plugin is re-opened

This patch adds a mechanism to handle the case when this sub-plugin is
re-opened.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Filter/OpenVino] Improve provided model(s) parsing
Wook Song [Thu, 2 Jan 2020 08:51:18 +0000 (17:51 +0900)]
[Filter/OpenVino] Improve provided model(s) parsing

This patch improves the mechanism to parse the provided model(s).
The model could be given in the following forms:
 - a path of the .bin or .xml file
 - a path of the prefix of the .bin and .xml file
 - a set of paths of the both .bin and .xml files.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Ext/Filter/Meson] Make meson support the CPU extensions of OpenVino
Wook Song [Thu, 2 Jan 2020 07:39:48 +0000 (16:39 +0900)]
[Ext/Filter/Meson] Make meson support the CPU extensions of OpenVino

This patch updates the meson build script for tensor filter sub-plugins
to support the CPU extensions of the OpenVino toolkit.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Filter/OpenVino] Load a given network into the acceleration device
Wook Song [Thu, 2 Jan 2020 07:28:35 +0000 (16:28 +0900)]
[Filter/OpenVino] Load a given network into the acceleration device

After negotiated, the given neural network should be loaded into the
acceleration device (note that, currently, CPU is the only supported
one).This patch is a draft of the implementation of that mechanism.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Cpp/Ext] Renames file extensions of test source codes written in C++
Dongju Chae [Tue, 7 Jan 2020 10:31:14 +0000 (19:31 +0900)]
[Cpp/Ext] Renames file extensions of test source codes written in C++

This commit renames file extensions of test source codes written in C++
(i.e, *.cpp -> *.cc), which hurt the consistency.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[armnn] object classification unittest + bug solve
Parichay Kapoor [Tue, 31 Dec 2019 07:39:24 +0000 (16:39 +0900)]
[armnn] object classification unittest + bug solve

- Add object classification unittest using tflite quantized mobilenet
- Add orange.png raw format image orange.raw
- As armnn treats uint8 datatype of tflite as quant asymm8 (arc/armnnTfLiteParser/TfLiteParser.cpp:310),
quant asymm8 is interpreted as uint8 in nnstreamer
- Updated current default HW backend to CpuAcc (based on neon) as CpuRef takes over
5 min to complete the mobilenet unittest

Bug solve:
The dimension of data received by armnn are inversed to match the desired order

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[C-API] Set omitted enum value in pipeline buffer policy
Sangjung Woo [Mon, 6 Jan 2020 07:30:16 +0000 (16:30 +0900)]
[C-API] Set omitted enum value in pipeline buffer policy

This patch explicitly set the enum value in pipeline buffer policy as
below.
* ML_PIPELINE_BUF_POLICY_AUTO_FREE: 0
* ML_PIPELINE_BUF_POLICY_DO_NOT_FREE: 1

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[nnfw] Interpret quant8 asymm as uint8
Parichay Kapoor [Tue, 31 Dec 2019 11:37:25 +0000 (20:37 +0900)]
[nnfw] Interpret quant8 asymm as uint8

Interpret quant8 asymm as uint8
Add unittest with object classification

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Ext/Filter/Meson] Update meson.build for the OpenVino sub-plugin
Wook Song [Thu, 2 Jan 2020 07:14:58 +0000 (16:14 +0900)]
[Ext/Filter/Meson] Update meson.build for the OpenVino sub-plugin

This patch updates the meson build script for the tensor filter
sub-plugins to support OpenVino.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Documentation] Modify the meson version required to build up-to-date
Wook Song [Mon, 6 Jan 2020 08:08:21 +0000 (17:08 +0900)]
[Documentation] Modify the meson version required to build up-to-date

This patch modifies the meson version required to build up-to-date, 0.50.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Api] add SNAP in nnfw enum type
Jaeyun [Tue, 17 Dec 2019 06:45:21 +0000 (15:45 +0900)]
[Api] add SNAP in nnfw enum type

Define SNAP in nnfw type, only for Android.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Filter/Custom] unnecessary log
Jaeyun [Mon, 6 Jan 2020 04:31:30 +0000 (13:31 +0900)]
[Filter/Custom] unnecessary log

prevent unnecessary logs if custom-filter does not define set/get dim callbacks.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[CodeClean] fix coverity
Jaeyun [Mon, 6 Jan 2020 03:44:14 +0000 (12:44 +0900)]
[CodeClean] fix coverity

fix some coverity issues and sam score

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[Filter/OpenVino] Add skeleton code of a sub-plugin for OpenVino support
Wook Song [Sun, 29 Dec 2019 21:40:08 +0000 (06:40 +0900)]
[Filter/OpenVino] Add skeleton code of a sub-plugin for OpenVino support

This patch adds skeleton code of a tensor filter sub-plugin for OpenVino
Toolkit (dldt) support.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[armnn] Added unittests for armnn
Parichay Kapoor [Tue, 17 Dec 2019 04:51:24 +0000 (13:51 +0900)]
[armnn] Added unittests for armnn

Added unittests for armnn tensor filter extension

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[armnn] Added tensor filter extension of armnn
Parichay Kapoor [Thu, 14 Nov 2019 08:25:06 +0000 (17:25 +0900)]
[armnn] Added tensor filter extension of armnn

Added tensor filter extension of armnn

V2: Updated the static library install paths for armnn and nnfw subplugins
V3: Updated error handling
V4: Remove Wno-sign-compare

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[filter/ext] Replace asserts with error handling
Parichay Kapoor [Mon, 23 Dec 2019 08:24:55 +0000 (17:24 +0900)]
[filter/ext] Replace asserts with error handling

Replace asserts with appropriate error handling in tensor filter extensions
This will allows test cases for failure scenarios and handle erronous inputs
via single-shot API gracefully with error return than shutting down with assert failure

V2:
Applied for custom filters as well

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[spec] Minor fixes and cleanup
Parichay Kapoor [Wed, 18 Dec 2019 05:42:55 +0000 (14:42 +0900)]
[spec] Minor fixes and cleanup

Added minor fixes:
- nnstreamer-capi.manifest to be not used non-tizen packages
- remove nnfw enable variable as nnfw is always enabled

Cleanup
- remove extra tizen conditions
- Move similar blocks together for better understanding

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[plugin/hw] Add more HW accelerator and update implementation
Parichay Kapoor [Tue, 17 Dec 2019 06:02:42 +0000 (15:02 +0900)]
[plugin/hw] Add more HW accelerator and update implementation

Add TPU as a backend accelerator property option in list of hardware
accelerators
Move default accelerator to avoid changing default accelerator enum value
everytime a new backend device is added

V2:
- Updated the regex to be formed at run-time using a list of supported
accelerators.
- Accelerators are also stored as GEnumValues to do conversion from
string to int value and vice versa easily
- More accelerators are added in heirarchical order. srcn and neon
moved to npu and cpu heirarchy correspondingly.
- Added more accelerators to be supported in future.

V3:
Update backend accelerator in tizen c-api

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Build] remove unnecessary flag
Jaeyun [Thu, 2 Jan 2020 03:47:36 +0000 (12:47 +0900)]
[Build] remove unnecessary flag

remove remained flag 'no-sign-compare'

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[FIX/LOG] Fix a wrong error message for custom-easy tensor filter
Dongju Chae [Mon, 16 Dec 2019 11:19:29 +0000 (20:19 +0900)]
[FIX/LOG] Fix a wrong error message for custom-easy tensor filter

This commit fixes the wrong log message for custom-easy tensor filter.
For this subplugin, we don't need to check whether the model path exists.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[TEST/RELOAD] Test runtime model reloads during tflite inferences
Dongju Chae [Wed, 18 Dec 2019 03:50:04 +0000 (12:50 +0900)]
[TEST/RELOAD] Test runtime model reloads during tflite inferences

This commit adds a testcase to test asynchronous model updates during
tflite inferences. It repeatedly switches tflite models between
mobilenet_v1 and mobilenet_v2. It expects the same index for input img.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[build] Remove -wno-sign-compare
Parichay Kapoor [Mon, 30 Dec 2019 11:56:11 +0000 (20:56 +0900)]
[build] Remove -wno-sign-compare

Remove -Wno-sign-compare
Apply corresponding fixes

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Converter] reduce complexity
Jaeyun [Thu, 26 Dec 2019 02:20:19 +0000 (11:20 +0900)]
[Converter] reduce complexity

define function to get possible media caps from downstream element.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[tflite] Backend unittest should always pass
Parichay Kapoor [Thu, 19 Dec 2019 04:19:52 +0000 (13:19 +0900)]
[tflite] Backend unittest should always pass

Tensorflow-lite unittest which tests for appropriate backend parsing and testing
should always pass.
Removed accelerator input validity check at tensor_filter.c as it is checked in
each filter again.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Android/SNAP] build script for SNAP
Jaeyun [Fri, 20 Dec 2019 06:38:20 +0000 (15:38 +0900)]
[Android/SNAP] build script for SNAP

Update script to build Android library including SNAP.
Before running with option --enable_snap=yes, developer should define SNAP_DIRECTORY (dir path to SNAP sub-plugin and prebuilt library)

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[nnfw] Reverse the order of dimension
Parichay Kapoor [Fri, 27 Dec 2019 13:50:12 +0000 (22:50 +0900)]
[nnfw] Reverse the order of dimension

Reverse the order of dimensions with nnstreamer as
nnfw stores dimension in accordance with tflite

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Filter/TF] remove unnecessary code
Jaeyun [Wed, 18 Dec 2019 08:29:48 +0000 (17:29 +0900)]
[Filter/TF] remove unnecessary code

remove unnecessary code (duplicated with tf_destroyNotify())

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years agoDist/Tizen: add missing manifest for source-tizen-sensor
MyungJoo Ham [Wed, 18 Dec 2019 06:06:16 +0000 (15:06 +0900)]
Dist/Tizen: add missing manifest for source-tizen-sensor

Because it has an executable (shared library), we need
SMACK manifest information.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[nnfw] Set accelerator for nnfw sandbox/helloahn/gbs_test
Parichay Kapoor [Mon, 16 Dec 2019 02:24:29 +0000 (11:24 +0900)]
[nnfw] Set accelerator for nnfw

Allow setting of accelerator for nnfw
Parsing accelerator moved to tensor_filter_common. Currently, used in
tflite and nnfw, and use directly for other filters like this.

V2:
Added note that PROP_ACCELERATOR is supported in a few filter extensions only.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Refactor/TF] Refactor tf tensor-filter subplugin to remove .c/.cc wrapper
Dongju Chae [Mon, 16 Dec 2019 05:24:58 +0000 (14:24 +0900)]
[Refactor/TF] Refactor tf tensor-filter subplugin to remove .c/.cc wrapper

This commit refactors tf tensor-filter subplugin to remove .c/.cc wrapper.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Refactor/Python] Refactor python sub-plugin to remove c/cc wrapper
Dongju Chae [Mon, 16 Dec 2019 07:42:07 +0000 (16:42 +0900)]
[Refactor/Python] Refactor python sub-plugin to remove c/cc wrapper

This commit refactors python tensor-filter sub-plugin to remove c/cc
wrapper.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years agoFilter: support EdgeTPU with tensorflow-lite interface
MyungJoo Ham [Wed, 11 Dec 2019 01:48:09 +0000 (10:48 +0900)]
Filter: support EdgeTPU with tensorflow-lite interface

Support EdgeTPU with non-NNAPI tensorflow-lite interface (delegation).

Tested with Ubuntu 18.04 + Coral Edge-TPU + Mobile-SSD TF Lite model.

TODO (later PRs): How to test with w/ GTest in CI?

Fixes #1661

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Refactor/TFLITE] Refactor tflite tensor-filter subplugin to remove .c/.cc wrapper
Dongju Chae [Mon, 16 Dec 2019 04:01:14 +0000 (13:01 +0900)]
[Refactor/TFLITE] Refactor tflite tensor-filter subplugin to remove .c/.cc wrapper

This commit refactors tflite tensor-filter subplugin to remove .c/.cc
wrapper.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[spec] nnfw now packaged only for tizen
Parichay Kapoor [Mon, 16 Dec 2019 02:19:49 +0000 (11:19 +0900)]
[spec] nnfw now packaged only for tizen

nnfw package is now created and corresponding tests performed
only for tizen in the spec file.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Refactor/Caffe2] Refactor caffe2 tensor-filter subplugin to remove .c/.cc wrapper
Dongju Chae [Mon, 16 Dec 2019 06:34:42 +0000 (15:34 +0900)]
[Refactor/Caffe2] Refactor caffe2 tensor-filter subplugin to remove .c/.cc wrapper

This commit refactors caffe2 tensor-filter subplugin to remove .c/.cc
wrapper.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Refactor/PyTorch] Refactor pytorch subplugin to remove .c/.cc wrapper
Dongju Chae [Mon, 16 Dec 2019 06:53:34 +0000 (15:53 +0900)]
[Refactor/PyTorch] Refactor pytorch subplugin to remove .c/.cc wrapper

This commit refactors pytorch tensor-filter subplugin to remove
.c/.cc wrapper.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Android] header files for sub-plugin
Jaeyun [Mon, 16 Dec 2019 06:46:23 +0000 (15:46 +0900)]
[Android] header files for sub-plugin

provide headers for plugin writer in android native build.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[TEST/RELOAD] Add negative unittests for tflite model reload
Dongju Chae [Fri, 13 Dec 2019 06:52:06 +0000 (15:52 +0900)]
[TEST/RELOAD] Add negative unittests for tflite model reload

This commit adds some negative unittests for tflite model reload.

1. Update model file name (a.tflite --> b.tflite) but b.tflite not exists
2. Update model file name (a.tflite --> b.tflite) but b.tflite has different dimensions
3. Update without file name changes but the file is gone (deleted)
4. Update without file name changes but the file has now different dimensions

Note that testing reload in runtime will be done in another commit.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[TEST/RELOAD] Add unittest to test set_property of model/is-updatable
Dongju Chae [Fri, 13 Dec 2019 04:52:44 +0000 (13:52 +0900)]
[TEST/RELOAD] Add unittest to test set_property of model/is-updatable

This commit adds unittest to test set_property of model/is-updatable.

It tests reloadModel() of tflite subplugin but it's not asynchronous.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Test/Model] Add a tflite model for reload tests, mobilenet_v2
Dongju Chae [Fri, 13 Dec 2019 04:51:26 +0000 (13:51 +0900)]
[Test/Model] Add a tflite model for reload tests, mobilenet_v2

This commit adds a tflite model for reload tests, mobilenet_v2.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years agoRefactor/Conf: avoid using hardcoded categories
MyungJoo Ham [Mon, 9 Dec 2019 10:01:33 +0000 (19:01 +0900)]
Refactor/Conf: avoid using hardcoded categories

To handle paths/files of each subplugin category,
use loops and enums instead of hardcoded variables.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoTest/Fix Compiler Warning (else ambiguity)
MyungJoo Ham [Fri, 13 Dec 2019 06:59:00 +0000 (15:59 +0900)]
Test/Fix Compiler Warning (else ambiguity)

Probably because of GTest macros, having if statement around
EXPECT_* appears to be not a good approach if it is not bracketed.

Fix compiler warnings from such.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoRelease of 1.3.0, the development version
MyungJoo Ham [Wed, 11 Dec 2019 04:12:22 +0000 (13:12 +0900)]
Release of 1.3.0, the development version

When 1.3.x is completed, it will be released as 1.4.0 and
1.5.x development will be started.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoRelease of nnstreamer 1.2.0 accepted/tizen/unified/20191213.115126 submit/tizen/20191213.070732 v1.2.0
MyungJoo Ham [Wed, 11 Dec 2019 04:08:55 +0000 (13:08 +0900)]
Release of nnstreamer 1.2.0

New features are coming with 1.2.0

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoPolicy Update: Test Tizen Features in ARM, always.
MyungJoo Ham [Wed, 4 Dec 2019 11:21:50 +0000 (20:21 +0900)]
Policy Update: Test Tizen Features in ARM, always.

Because TAOS-CI skips unittests for ARM/ARM64 because of the latency,
we need to enable unittests for ARM-only tests unconditionally
if the test-target features are ARM-only.

Especially, this will help preventing regressions from other packages
in Tizen by enabling unit-tests in Tizen build (build.tizen.org) for
those selected features.

Changes in v2:
- Updated indentation (suggested by dongju.chae@samsung.com)

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[TFLITE/RELOAD] Revise tflite sub-plugin to support model reloading
Dongju Chae [Tue, 10 Dec 2019 11:34:15 +0000 (20:34 +0900)]
[TFLITE/RELOAD] Revise tflite sub-plugin to support model reloading

This commit revises tflite sub-plugin to support model reloading.

It implements new interpreter class with its own mutex, which wraps
existing codes. Also, it's designed to support interpreter swapping
to hide the overhead of model reloading.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[TensorSink/UnitTest] Add a test option for sync property in TensorSink
Dongju Chae [Fri, 13 Dec 2019 03:10:13 +0000 (12:10 +0900)]
[TensorSink/UnitTest] Add a test option for sync property in TensorSink

This commit adds a test option, need_sync, for sync property in TensorSink.

Basically, GstBaseSink has a 'sync' property enabled by default, and
TensorSink also does. In this case, if the frame is later than max-lateness,
the sink will drop the buffer without calling the render method.

It can be a problem if some unittests (e.g., 'signals') expects the exact
number of received data. In slow system, if a signal callback is too slow,
some buffer can be dropped.

To fix this problem, this patch adds a test option for sync property and
set 'sync' property as FALSE by default to disable the buffer dropping
feature for 'signals' unittest.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Android] common headers for gstreamer
Jaeyun [Thu, 12 Dec 2019 08:51:47 +0000 (17:51 +0900)]
[Android] common headers for gstreamer

define common header path for gstreamer.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years agoConverters: support external subplugins
MyungJoo Ham [Fri, 6 Dec 2019 02:38:00 +0000 (11:38 +0900)]
Converters: support external subplugins

Allow to add external converters in run-time as subplugins.

Policies to be determined:
- Build & integration policies for external converters.
- Would we allow a few "basic" external converters available by default? (without dlopen)
- Tizen feature keys and privileges along with "NOT SUPPORTED" handling.
- Which profiles would use which of external converters?

TODO:
- Implement multiple external subplugins for unit tests

Changes in V2:
- Typo fixed (comments)

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoUnitTest: Tensor-src-tizensensor / Initial test cases
MyungJoo Ham [Tue, 26 Nov 2019 09:42:06 +0000 (18:42 +0900)]
UnitTest: Tensor-src-tizensensor / Initial test cases

1. Unit tests are added.
2. dummy_sensor.c/h offers dummy Tizen sensor framework for unit tests,
where we cannot activate Tizen sensor framework daemon.
3. Bugs were found with the unit tests and they are fixed.
    - g_hash_table misuses
    - property configuration errors
    - frequency (framerate) handling
    - Clean up and configured-status handling
    - Mutex deadlock
    - Timestamp handling

Changes v1->v2:
- Use G_USEC_PER_SEC for readability, suggested by dongju.chae@samsung.com
Changes v2->v3:
- Added error-clear in testcase, suggested by dongju.chae@samsung.com
- Removed unnecessary code (residue of internal tests), suggested by jy1210.jung@samsung.com
Changes v3->v4:
- Added free ops for pipelines, removed duplicated codes, suggested by jy1210.jung@samsung.com

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Filter] property description
Jaeyun [Thu, 12 Dec 2019 08:57:30 +0000 (17:57 +0900)]
[Filter] property description

Remove unnecessary space in the description.
(Using gst-inspect, the property model shows unnecessary space)

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[nnfw] nnfw fix for aarch64
Parichay Kapoor [Thu, 12 Dec 2019 05:55:42 +0000 (14:55 +0900)]
[nnfw] nnfw fix for aarch64

This PR adds temporary fix for aarch64 enabling successful build and
testing of nnfw plugin for nnstreamer

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[CONF] Remove unnecessary g_key_file_free()
Dongju Chae [Thu, 12 Dec 2019 04:30:19 +0000 (13:30 +0900)]
[CONF] Remove unnecessary g_key_file_free()

This commit removes unnecessary g_key_file_free() because g_autoptr
already provides cleanup process. This bug was detected by valgrind.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years agoDoc/Contributing: update code quality standards
MyungJoo Ham [Thu, 12 Dec 2019 06:51:48 +0000 (15:51 +0900)]
Doc/Contributing: update code quality standards

Let's describe the standards we are already enforcing.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Android] add asset dir for native library
Jaeyun [Mon, 9 Dec 2019 12:47:41 +0000 (21:47 +0900)]
[Android] add asset dir for native library

Add asset dir for native library.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[nnfw/single/test] Add tests for nnfw with single API
Parichay Kapoor [Fri, 8 Nov 2019 07:30:51 +0000 (16:30 +0900)]
[nnfw/single/test] Add tests for nnfw with single API

Add invoke test for nnfw with single API

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[nnfw/single] Enable nnfw with single API
Parichay Kapoor [Fri, 8 Nov 2019 07:29:13 +0000 (16:29 +0900)]
[nnfw/single] Enable nnfw with single API

- Enable nnfw with single API
- nnfw and tensorflow-lite both support tflite extensions
Added meson option to set default behavior which prioritizes tensorflow-lite

V2:
- Add check for availibilty of nnfw after assigning nnfw in case of ML_NNFW_TYPE_ANY

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[nnfw] update nnfw model dir path to model file path
Parichay Kapoor [Fri, 8 Nov 2019 07:26:01 +0000 (16:26 +0900)]
[nnfw] update nnfw model dir path to model file path

Update the nnfw framework input model format to work with file path
nnfw framework takes in model directory path
Directory path is extracted from file path
Metadata is verified in the extracted directory path

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[C-Api/Single] check invalid tensor info
Jaeyun [Wed, 4 Dec 2019 04:16:03 +0000 (13:16 +0900)]
[C-Api/Single] check invalid tensor info

Check input/output tensor info when opening the nn model.
If given info is not matched with model, try to change input info for dynamic mode.

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[C-Api] update util functions
Jaeyun [Fri, 15 Nov 2019 06:28:32 +0000 (15:28 +0900)]
[C-Api] update util functions

1. add new function to compare tensors metadata (type, dimension)
2. refactor function to create tensors data handle

Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
4 years ago[FIX] Add a missing g_clear_error() in unittest_sink.cpp
Dongju Chae [Wed, 11 Dec 2019 01:15:27 +0000 (10:15 +0900)]
[FIX] Add a missing g_clear_error() in unittest_sink.cpp

This commit adds a missing g_clear_error() in unittest_sink.cpp.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>