Geunsik Lim [Thu, 14 May 2020 01:34:49 +0000 (10:34 +0900)]
indent: Update the gst-indent file
This commit is to update the below contents in the 'gst-indent' file.
**Changelog**
* Fixed typo
* Declared a default shell script (e.g., bash) with Linux shebang line
Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
Yongjoo Ahn [Wed, 13 May 2020 05:54:53 +0000 (14:54 +0900)]
[Android/build] Move tensor-filter common to nnstreamer_common_srcs
- To enable SINGLE-only build for cpp class filter (snpe)
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
Yongjoo Ahn [Fri, 8 May 2020 04:11:53 +0000 (13:11 +0900)]
[Android/SNPE] Support SNPE in Android
- Support SNPE for Android api
- Add simple tests (singleshot, pipeline) using pre-loaded SNPE model
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
Parichay Kapoor [Wed, 13 May 2020 03:33:31 +0000 (09:03 +0530)]
[single] Add implementation to support for hw accelerator
single API supported hw accelerator for the NNFW
this hw was used to check if it is supported by the framework
however this hw was never passed to the framework to use
this patch sovles this issue
resolves #2374
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Mon, 11 May 2020 19:40:08 +0000 (01:10 +0530)]
[single] Added documentation for the thread behavior
Added documentation for the thread behavior in various functions in single API
Also removed ERROR state as it was rendundant after #2360
Resolves #2185
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Tue, 12 May 2020 21:05:02 +0000 (02:35 +0530)]
[tflite] bug fix for tflite unittest
unittest of tflite compared output with grep "exp"
however some expressions could be prefix of different exp
this could lead to false positives
added line end marker
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Mon, 11 May 2020 08:34:15 +0000 (17:34 +0900)]
[Demux/Split] unref incoming buffer
After demux/split the incoming buffer, unref the buffer to decrease ref count.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 7 May 2020 17:58:16 +0000 (23:28 +0530)]
[single] single API to process nnfw transparently once opened
single API should process all the frameworks transparently once the single handle has been opened.
The framework check has been performed when the handle was created and opened.
If the operation is not supported, that should be handled in the nnfw rather than in single API.
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Sangjung Woo [Mon, 11 May 2020 09:53:57 +0000 (18:53 +0900)]
[Android] Stop when first error occurs
When running build-android-lib.sh, it does not stop even though error
occur (e.g. command not found). It makes users hard to find the reason
of the problem. This patch stops first error occurs so users can fix it
easily.
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
Yongjoo Ahn [Thu, 9 Apr 2020 07:58:19 +0000 (16:58 +0900)]
[filter/SNPE] Add tensor_filter sub-plugin for SNPE
- Currently, this only supports CPU runtime
- Use meson feature "feature" : check environment variable `SNPE_ROOT`
to enable the build or not
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
Jaeyun [Mon, 11 May 2020 07:53:29 +0000 (16:53 +0900)]
[Meson/Option] remove unnecessary option
remove unnecessary option for disabling audio elements.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun [Fri, 8 May 2020 11:19:13 +0000 (20:19 +0900)]
[C-Api/CodeClean] validate model file
add function for nnfw auto detection and clean up functions to validate model file.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun [Thu, 7 May 2020 07:47:38 +0000 (16:47 +0900)]
[Test] add simple testcases
add simple testcases for native c-api and internal functions.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 7 May 2020 17:32:24 +0000 (23:02 +0530)]
[single] cache tensor filter class once at open
Existing implementation seeks the tensor filter class based on its type everytime a class is needed.
Rather, cache the tensor filter class increasing its ref at open, and unref at single api close.
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Sangjung Woo [Mon, 11 May 2020 02:11:54 +0000 (11:11 +0900)]
[Subplugin] Fix the typo
This patch fixes the typo in header file.
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
Jihoon Lee [Fri, 24 Apr 2020 08:37:01 +0000 (17:37 +0900)]
Change packaging in regards to feature migration
Change debian and rpm build in regards to feature update option. This
commit just inherits past options leaving possibilities open to
exploit "auto" options.
Signed-off-by: Jihoon Lee <jhoon.it.lee@samsung.com>
Jihoon Lee [Mon, 20 Apr 2020 06:34:05 +0000 (15:34 +0900)]
[Meson] Migrate to feature
1. Feature / feature handler added
2. Add video support as feature (and wipe out disable-video-support)
3. Add extra args support
4. Implement enable-tesnorflow, enable-tensorflow-lite ... as feature
5. Implement enable-orc, ... as feature *
- enable-orc -> orcc-support
- enable-tensorflow-lite -> tflite-support
- enable-tensorflow -> tf-support
- enable-caffe2 -> caffe2-support
- enable-pytorch -> pytorch-support
- enable-movidius-ncsdk2 -> mvncsdk2-support
- enable-python -> python2-support / python3-support
- enable-tflite-nnapi-delegation -> tflite-nnapi-delegation
- enable-nnfw-runtime -> nnfw-runtime-support
- enable-armnn -> armnn-suport
Signed-off-by: Jihoon Lee <jhoon.it.lee@samsung.com>
Jihoon Lee [Mon, 20 Apr 2020 06:34:05 +0000 (15:34 +0900)]
Add feature object handler
1. Feature / feature handler added
2. Add video support as feature (and wipe out disable-video-support) *
3. Add extra args support
4. Implement enable-tesnorflow, enable-tensorflow-lite ... as feature
5. Implement enable-orc, ... as feature
6. Change docs
Signed-off-by: Jihoon Lee <jhoon.it.lee@samsung.com>
Sangjung Woo [Mon, 4 May 2020 07:04:40 +0000 (16:04 +0900)]
[Android/AMC] Set the omitted finalize function
'gst_amc_src_finalize()' exists in the code but the finalize of GObject
class is not set. Because of this reason, gst_amc_src_finalize() function
is not called at all. This patch fixes this bug.
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
MyungJoo Ham [Thu, 7 May 2020 07:28:13 +0000 (16:28 +0900)]
[CAPI] Separate error codes: ml-api-common.h
nntrainer wants to share the error codes for its API set.
Separate "common" header and release it via a sub-pacakge.
CC: @jijoongmoon: You can use this for
https://github.com/nnstreamer/nntrainer/issues/75
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Thu, 7 May 2020 10:24:43 +0000 (19:24 +0900)]
[Tests/Regression] GTEST error is ignored
PR #2338 has incurred a regression in the test suite that
make gbs pass even if there is a failed unit test case in gtest.
This fixes the regression of #2338
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
HyoungJoo Ahn [Thu, 7 May 2020 04:37:50 +0000 (13:37 +0900)]
[Unittest/IIO] fix mem leak of unittest_src_iio
fix mem leaks of few test cases
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Hyoung Joo Ahn [Wed, 6 May 2020 11:49:42 +0000 (20:49 +0900)]
[Mediapipe] upload the new subplugin
this subplugin helps running mediapipe graph through nnstreamer
Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
Hyoung Joo Ahn [Wed, 6 May 2020 11:45:04 +0000 (20:45 +0900)]
[Mediapipe] fix shell script
make the nnstreamer lib path rather than error
Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
Jaeyun [Thu, 7 May 2020 09:41:37 +0000 (18:41 +0900)]
[C-Api] unknown tensor type
do not allow setting unknown tensor type.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Wook Song [Thu, 7 May 2020 08:18:42 +0000 (17:18 +0900)]
[Dist/Tizen] Remove an invalid command line
This patch removes an invalid command line that creates a symlink, *.so.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 7 May 2020 07:57:14 +0000 (16:57 +0900)]
[Dist/Tizen] Provide GST_PLUGIN_PATH to unittest_tizen_sensor
'unittest_tizen_sensor' requires a nnstreamer extension plugin,
'tensor_src_tizensensor', in addition to the other nnstreamer plugins.
In order to resolve this dependency at build time, this patch modifies
the rpm spec for Tizen to provide extra GST_PLUGIN_PATH to
'unittest_tizen_sensor'.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 7 May 2020 07:47:42 +0000 (16:47 +0900)]
[Tests/Script] Do not override GST_PLUGIN_PATH in the test script
This patch removes the code line overrides GST_PLUGIN_PATH from the test
script.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Jaeyun [Wed, 6 May 2020 11:41:48 +0000 (20:41 +0900)]
[Test] add testcases for internal function
add testcases for internal function (validate model files)
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
HyoungJoo Ahn [Thu, 7 May 2020 03:42:07 +0000 (12:42 +0900)]
[Unittest/IIO] add the negative case: base dir
testcase for the logic with invalid base dir
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
HyoungJoo Ahn [Wed, 29 Apr 2020 05:39:27 +0000 (14:39 +0900)]
[Unittest/IIO] add the negative case: frequency
testcase for the logic with invalid frequency
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Sangjung Woo [Thu, 7 May 2020 01:48:14 +0000 (10:48 +0900)]
[Conf] Add omitted comment and fix a typo
* Add omitted parameter comment for nnsconf_get_subplugin_info()
* Fix a typo
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
gichan-jang [Wed, 6 May 2020 09:54:06 +0000 (18:54 +0900)]
[TEST] Fix svace defect : nnfw_runtime, tensordec-imagelabel
Fix memory leak of nnfw runtime and tensor decoder image labeling unit test.
Related SVACE defects : 440263, 440264, 440265, 440266, 440261
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
gichan-jang [Wed, 29 Apr 2020 02:24:08 +0000 (11:24 +0900)]
[README] Add daily build and test badge
Add daily build and test badge (Tizen x64, Tizen X86, Ubuntu x64)
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
Wook Song [Fri, 24 Apr 2020 08:35:51 +0000 (17:35 +0900)]
[Tests/Python] Update test scripts for python extensions
This patch modifies the scripts running the test cases for python
filters to use proper PYTHONPATH.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Jaeyun [Wed, 29 Apr 2020 08:57:40 +0000 (17:57 +0900)]
[Transform] return value when failed to set option
1. return false if failed to set option.
2. add simple testcases for the option string.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun Jung [Thu, 30 Apr 2020 05:59:11 +0000 (14:59 +0900)]
[C-Api] new ref to get tensor name
if tensors-info has name, return new ref when getting the tensor name.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
HyoungJoo Ahn [Tue, 28 Apr 2020 11:52:21 +0000 (20:52 +0900)]
[Unittest/IIO] add a testcase for unusual cases
the added testcase cover various cases not covered previous set.
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Jaeyun [Wed, 29 Apr 2020 10:59:29 +0000 (19:59 +0900)]
[Filter/Test] test for filter-fw registration
add test for filter-fw v1 registration without model file.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Yongjoo Ahn [Mon, 4 May 2020 04:25:40 +0000 (13:25 +0900)]
[Test/tensordec-pose] Fix a test case with arithmetic transform
- tensordec-pose treats incoming array as float, thus add `tensor_transform mode=arithmetic`
- expecting proper line converage results
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
Jaeyun [Tue, 28 Apr 2020 10:12:40 +0000 (19:12 +0900)]
[Test] add tc for element properties
update simple testcases for the element properties.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Yongjoo Ahn [Tue, 28 Apr 2020 12:02:26 +0000 (21:02 +0900)]
[Test/tensordec-pose] Add test case with more buffers
The test with `videotestsrc num_buffers=4` sometimes does not enough to execute some functions. This commit adds a test with more buffers.
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
MyungJoo Ham [Tue, 28 Apr 2020 10:53:42 +0000 (19:53 +0900)]
[Test/CAPI-Util] Add negative cases.
Add more negative test cases to increate test coverage rate.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Tue, 28 Apr 2020 04:44:13 +0000 (13:44 +0900)]
[CodeClean] line comment
code-style : remove single line comments.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
gichan-jang [Tue, 28 Apr 2020 11:31:37 +0000 (20:31 +0900)]
[TEST] Add test case for tizen sensor
Add test case for tizen sensor
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
HyoungJoo Ahn [Wed, 29 Apr 2020 03:44:02 +0000 (12:44 +0900)]
[Unittest] apply EXPECT_STREQ for string comparisons
using EXPECT_STREQ rather than `EXPECT_TRUE( g_str_equal...`
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Jaeyun [Mon, 27 Apr 2020 11:46:51 +0000 (20:46 +0900)]
[Test] add sync option in appsink
1. add sync=false option in model reload test.
2. update pipeline description in docs.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun [Mon, 27 Apr 2020 11:18:53 +0000 (20:18 +0900)]
[C-Api/Pipeline] warning message for sync option
add warning message if sink element does not have sync=false option.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun [Mon, 27 Apr 2020 08:49:01 +0000 (17:49 +0900)]
[CodeClean] fix invalid index
1. fix invalid index when unmapping the mem block
2. unref buffer when failed to get mem
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Jaeyun [Fri, 24 Apr 2020 10:23:12 +0000 (19:23 +0900)]
[Android] prepare SNPE
Prepare sub-plugin SNPE in android.
- add enum (nnfw type) and build option
- update document for the changes
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
MyungJoo Ham [Mon, 27 Apr 2020 09:29:02 +0000 (18:29 +0900)]
[Filter/EdgeTPU] Workaround for #2320
Until #2320 is resolved, adding AUTO/DEFAULT to "supported hw"
list makes it easier to use.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Wed, 22 Apr 2020 05:26:58 +0000 (14:26 +0900)]
[Filter/C++Class] Bugfix: close gets a dlb ptr!
Pointer type mismatch fixed!
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Fri, 17 Apr 2020 05:22:26 +0000 (14:22 +0900)]
[Spec/Deb] Update C++ devel env.
1. Do not separate C env and C++ env. Merge them.
Both Ubuntu and Tizen supports C++ by default.
2. Move /nnstreamer/cpp/nnstreamer_cppplugin_api_filter.hh to
/nnstreamer/nnstreamer_cppplugin_api_filter.hh
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Tue, 31 Mar 2020 10:24:52 +0000 (19:24 +0900)]
[Option] Enable tensor-filter C++ class by default
Turn the C++ class support on by default.
And update spec file to add .hh file.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Tue, 31 Mar 2020 10:15:43 +0000 (19:15 +0900)]
[Filter/Edgetpu] Rewrite as a C++ class
This potentially reduces the code size and
maintanance costs.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Mon, 27 Apr 2020 05:44:53 +0000 (14:44 +0900)]
[Merge] fix return type
1. fix invalid return value.
2. mem-unmap when failed to get mem ptr in input buffer.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
gichan-jang [Mon, 27 Apr 2020 10:15:17 +0000 (19:15 +0900)]
[Test] Fix potential test fail
Sync option is added on appsink pad to prevent potential test fail caused by buffer drop.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
Jaeyun [Mon, 27 Apr 2020 10:10:15 +0000 (19:10 +0900)]
[Sink] default sync false
Set default sync option 'false'.
It may be delayed with tensor_filter element, to invoke neural network model.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
Yongjoo Ahn [Wed, 22 Apr 2020 10:07:02 +0000 (19:07 +0900)]
[Android/Test] add SNAP tests using tensorflow model
- For both single-shot and pipeline, add simple tests to validate SNAP with tensorflow model
- Test with CPU, DSP, NPU runtime on recent Snapdragon (S20) and preloaded model files
Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
MyungJoo Ham [Mon, 20 Apr 2020 07:16:22 +0000 (16:16 +0900)]
Error return instead of assert for external errors.
If an error might occur by external reasons,
(invalid input/app behaviors and inappropriate environments)
it should not simply assert away, but should provide
appropriate error messages and handling.
Refactored everything except for tensor_filter_python.cc,
which will be refactored with exception throwing.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Thu, 23 Apr 2020 09:39:34 +0000 (18:39 +0900)]
[Test/NNFW] separate testcase
1. separate testcase - invalid model and invalid data.
2. fix mem leak, destroy handle before creating new data.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
gichan-jang [Fri, 24 Apr 2020 01:02:34 +0000 (10:02 +0900)]
[Test] Fix tizen capi unittest
Today's Daily Build was a success, but the test fails occasionally due to slow speed of arm.
Timeout occurred in the test and increased it to 10 seconds.
If timeout occurs again, I will investigate another cause.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
HyoungJoo Ahn [Thu, 23 Apr 2020 08:37:15 +0000 (17:37 +0900)]
[Unittest] remove the not effective sleeps
remove or unify `g_usleep() of testcases`
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
MyungJoo Ham [Wed, 22 Apr 2020 11:08:42 +0000 (20:08 +0900)]
[Filter/Python] Don't assert. Provide exceptions.
A subplugin shouldn't be arrogant enough to assert.
Return errors or throw exceptions instead.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Hyoung Joo Ahn [Thu, 23 Apr 2020 10:39:27 +0000 (19:39 +0900)]
[Mediapipe] fix meson script
set `opencv_path` when system architecture is detected.
Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
Wook Song [Tue, 21 Apr 2020 08:34:46 +0000 (17:34 +0900)]
[Dist/Tizen] Use the base option, b_coverage, to track coverage
Instead of manually setting cflags for coverage tracking, let's use
b_coverage which is a meson's base option to do it.
Signed-off-by: Wook Song <wook16.song@samsung.com>
gichan-jang [Thu, 23 Apr 2020 07:15:50 +0000 (16:15 +0900)]
[TEST] Fix tizen sensor unit test
Fix tizen sensor unit test.
- Add macro to wait until start the pipeline.
- Wait until expected buffers are processed
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
Jaeyun [Thu, 23 Apr 2020 06:44:38 +0000 (15:44 +0900)]
[Test/NNFW] ignore timeout in singleshot
increase timeout (30 sec) before invoking the model.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
MyungJoo Ham [Thu, 23 Apr 2020 06:11:40 +0000 (15:11 +0900)]
[TEST/util] Clean up wait_pipeline_process_buffers
1. apply "const" for data_received. It won't update it
2. make it robust against TEST_DEFAULT_SLEEP_TIME values
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
gichan-jang [Thu, 23 Apr 2020 05:23:47 +0000 (14:23 +0900)]
[TEST/util] Add unit test util function
Add unit test util function.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
gichan-jang [Tue, 21 Apr 2020 11:08:28 +0000 (20:08 +0900)]
[Test] Fix nnstreamer repo test
Fix nnstremr repo test fail.
It's not best way to fix the problem, But I can not find any other suitable method in SSAT.
If there is a good alternative, please recommend it.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
Dongju Chae [Tue, 21 Apr 2020 03:04:24 +0000 (12:04 +0900)]
[Coverage/Fix] Disable optimization and remove redundant gcov run
This patch disables code optimization for coverage analysis and remove
redundant gcov run.
In fact, we don't need to run the python script, unittestcoverage.py
which recursively executes gcov. Because lcov internally execute gcov,
the results of the script are never used.
FYI...
- when program is compiled with coverage option, .gcno files are created.
- When such programs are executed, .gcda files are created.
- As lcov input files are .gcda. we don't need to run gcov before lcov.
Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
Jaeyun [Tue, 21 Apr 2020 08:41:02 +0000 (17:41 +0900)]
[Android/Build] flag to handle exception
add cxx-flag in android build, enable exception handling.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
MyungJoo Ham [Tue, 21 Apr 2020 02:28:33 +0000 (11:28 +0900)]
[Src-IIO] CHANNEL property handling fix
1. If g_hash_table_insert returns FALSE,
it simply means that the entry is duplicated.
Just skip it, then.
2. If we cannot parse it, don't change status, but
ignore the whole property values.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Tue, 21 Apr 2020 07:24:49 +0000 (16:24 +0900)]
[C-Api] fix duplicated free
do not call free() when destroying gst-buffer.
TODO : prevent memcpy() when pushing input data into src element.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
gichan-jang [Tue, 21 Apr 2020 06:10:49 +0000 (15:10 +0900)]
[TEST] Fix test case name
Fix tese case name of nnstreamer repo unit test.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
MyungJoo Ham [Tue, 21 Apr 2020 01:59:40 +0000 (10:59 +0900)]
[Tests] Add static to internal-global vairables.
Don't export internal variables.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Sat, 21 Mar 2020 07:53:03 +0000 (16:53 +0900)]
[Filter/C++] Framework rework for better usages
By applying C++ tensor-filter framework to edge-tpu,
I've found that the current draft is not useful enough.
Here comes C++ t-f framework updates tested with the
edgetpu subplugin.
Changes in V2:
- Removed macros. Use static-inline functions instead.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Thu, 16 Apr 2020 08:21:31 +0000 (17:21 +0900)]
[Filter/v1] Bugfix of TF-V1 subplugin API / C++ support
By testing with C++ class tf-subplugin, "edget-tpu",
bugs were found from the recently introduced TF-V1 subplugin API.
- V1 APIs uses void * private_data instead of void ** private_data.
There are codes that do not reflect this.
In order to make C++ class subplugins work smoothly,
they need to know "this" pointer when they were invoked
by common C++ wrapper infrastructure, which is a set of
static functions of the base class.
Thus, add a new argument for them to find out "this" pointer.
CC: @kparichay
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
gichan-jang [Mon, 20 Apr 2020 06:59:20 +0000 (15:59 +0900)]
[Tests] Fix sink unittest fail
Test fails occasionally due to slow speed of arm.
The test was modified to wait until the pipeline processing the buffers
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
MyungJoo Ham [Mon, 20 Apr 2020 07:32:38 +0000 (16:32 +0900)]
[Log] Add fatal-log
Separate "fatal" level from "error".
When we convert "assert" fail to general error handling,
log-fatal might be appropriate.
Fixes #2288
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Tue, 14 Apr 2020 07:48:16 +0000 (16:48 +0900)]
[Dec/Subplugins] Remove duplicated code
Add "utility" internal API to remove
duplicated code.
Chages in V2:
- Refactored imglabel_t data structure
- Suggested by @zhoonit
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Wed, 1 Apr 2020 10:56:48 +0000 (19:56 +0900)]
[LF/AI] Update CONTRIBUTING
To follow what LF/AI requires with the Charter.
It has been drafted at https://wiki.lfai.foundation/display/NNSTREAM/NNStreamer+CONTRIBUTING+draft+of+2020.04
updated 2020-04-16
- Clarified "most senior among available committers", Line 45
(suggeted by @zhoonit)
- s/chairman/chairperson
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Fri, 27 Mar 2020 05:57:47 +0000 (14:57 +0900)]
Comply with LF/AI Community Charter
As attorneys of stakeholders are drafting the community charter,
based on the current draft, I am updating related documents.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
gichan-jang [Fri, 17 Apr 2020 05:11:04 +0000 (14:11 +0900)]
[README] Add daily build result badge in table
Add daily build result in ofiicial release table
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
Jaeyun [Mon, 20 Apr 2020 04:43:26 +0000 (13:43 +0900)]
[CodeClean] return type
1. fix ret type of filter callback.
2. minor update (warning w/Wmissing-declarations)
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
MyungJoo Ham [Sat, 18 Apr 2020 02:39:22 +0000 (11:39 +0900)]
[Bug] Do not call essential ops in assert().
Some systems may use NDEBUG and skip all assert calls.
Do not call essential ops in there.
Fixes #2282
Reported-by: @ndufresne
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Mon, 13 Apr 2020 11:16:20 +0000 (20:16 +0900)]
[Tensor-If] Make it buildable
Add empty functions and build infra for Tensor-If.
Tensor-If is a subclass of GstBaseTransform.
The src-pad and sink-pad are supposed to have the same
capabilities.
It appears that in-place op has no benefit, but
not sure if we need to prohibit it.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Hyoung Joo Ahn [Fri, 17 Apr 2020 05:06:49 +0000 (14:06 +0900)]
[Mediapipe] update meson script
add opencv dependency and check architecture of system
Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
Jaeyun [Fri, 17 Apr 2020 08:47:18 +0000 (17:47 +0900)]
[Android] initialize native code
For C-API developers, add gst-init in native code.
Calling initialize() in java is unnecessary.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
MyungJoo Ham [Tue, 14 Apr 2020 07:28:27 +0000 (16:28 +0900)]
[Converter] Refactored: no more large #if/endif code blocks.
Do not enclose a large code block with #if/#endif.
Separate such files instead.
Changed in v2 (2020-04-16)
- Separated no-a/v and a/v headers
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Fri, 17 Apr 2020 04:10:10 +0000 (13:10 +0900)]
[Android] tf-lite version
change tf-lite version to 1.13 in android api build.
Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
gichan-jang [Fri, 17 Apr 2020 04:13:06 +0000 (13:13 +0900)]
[Tests] Fix tizen capi unit test
Test fails occasionally due to slow speed of arm.
The test was modified to only check if the switch is working.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
gichan-jang [Thu, 16 Apr 2020 11:31:20 +0000 (20:31 +0900)]
[Daily build] Add daily build badge
> Builds and tests are conducted every morning at 7: 30 a.m.
> The results are shown on the first page of the nnstreamer.
> Click the badge to download the log and images for the last seven days.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
MyungJoo Ham [Tue, 14 Apr 2020 07:16:14 +0000 (16:16 +0900)]
[Android/Single] Do not turn off a large code block with #if
Separate the file and build independently instead.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Wook Song [Wed, 8 Apr 2020 05:44:08 +0000 (14:44 +0900)]
[Tests/CAPI] Use constexpr instead of define and ifdef for tflite
This patch experimentally replaces #define and #ifdef used for checking
enable-tensorflow-lite is true with constexpr, which is type-safe
replacement for #define.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Mon, 13 Apr 2020 02:25:20 +0000 (11:25 +0900)]
[Tests/CAPI] Use EXPECT_FLOAT_EQ for the test cases comparing float data
This patch replaces EXPECT_EQ with EXPECT_FLOAT_EQ for the test cases which
compare floating-point data.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Parichay Kapoor [Sat, 29 Feb 2020 10:25:24 +0000 (15:55 +0530)]
[filter] Support tensor filter framework V1 single API
Support tensor filter framework v1 callbacks for the single API
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
MyungJoo Ham [Mon, 13 Apr 2020 11:22:04 +0000 (20:22 +0900)]
Lower MCD, Increase SAM Score
SAM analysis says that nnstreamer.c is causing cyclic dependency
between modules.
Although I think this is false positive, let's move nnstreamer.c
to a subdirectory so that SAM won't recognize nnstreamer.c is
"parent module" of other nnstreamer elements.
Because this is the only modules that SAM recognizes as MCD issue,
this may enhance SAM score greatly.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Wed, 8 Apr 2020 04:08:34 +0000 (13:08 +0900)]
[Tensor-IF] An anchor for further development.
Anyone can jump in and start the development!
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>