MyungJoo Ham [Tue, 12 Nov 2019 11:14:26 +0000 (20:14 +0900)]
Tensor-src-tizensensor: fill minor gst callbacks
- finalize, start, stop, is_seekable filled.
- Internal static functions are partially filled for future direction.
This should follow #1864 and addresses #1143
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Thu, 28 Nov 2019 06:02:07 +0000 (15:02 +0900)]
[Doc] dimension in converter
update docs about the dimension of each media stream.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Tue, 26 Nov 2019 03:45:18 +0000 (12:45 +0900)]
[CodeClean/Test] free log-file str
free log-file name str in clean-dev function.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 21 Nov 2019 01:21:26 +0000 (10:21 +0900)]
[IIO/test/coverity] Resolve coverity issues
Added more checks coherent with in #1859 and #1863 to another location
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Thu, 21 Nov 2019 01:44:21 +0000 (10:44 +0900)]
[tests/coverity] Fix coverity issues related to gtest main
Add try/catch for exceptions thrown by function calls in main bosy when using gtest
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Wook Song [Thu, 7 Nov 2019 06:46:55 +0000 (15:46 +0900)]
[Dist/Tizen] Enable nnstreamer-ncsdk2
This patch enables the tensor filter extension for Intel Movidius Neural
Compute stick SDK2 (NCSDK2) in Tizen.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 7 Nov 2019 04:49:52 +0000 (13:49 +0900)]
[Ext/Filter/Meson/MVNC] Use functions added after 0.44 and 0.47
As the minimum version required is changed to 0.50.0, this patch
modifies the meson build scriptlet for the MVNC tensor_filter extension
to use functions such as warning() and check_header() added after 0.44
and 0.47, respectively, as well.
Signed-off-by: Wook Song <wook16.song@samsung.com>
MyungJoo Ham [Tue, 12 Nov 2019 10:30:27 +0000 (19:30 +0900)]
Tensor-src-tizensensor: add get/set properties
This adds get/set property methods for tizen sensor extension plugin.
This should follow #1857 and addresses #1143
Changes in v2:
- As suggested by @kparichay, state rollbacks are added
- Do not reconfigure if it's not configured.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Tue, 19 Nov 2019 06:19:24 +0000 (15:19 +0900)]
[CAPI] Add enum for incoming NPU hardware
We will be supporting new NPU hardware along with
its userspace libraries.
Prepare enum spaces for them for future development.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Mon, 18 Nov 2019 11:26:59 +0000 (20:26 +0900)]
[Test] dimension in testcases
clearly denotes the dimension of in/out tensor in testcases.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 7 Nov 2019 07:00:03 +0000 (16:00 +0900)]
[tensor_filter/plugin] Update for supporting various accelerators
Added update to support various accelerators for tensor filters
Updated tensorflow-lite nnapi setting based on these accelerators
Added testcases for tensorflow-lite
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Wed, 20 Nov 2019 07:31:51 +0000 (16:31 +0900)]
[spec/meson] Update nnstreamer package to support c++14 language standard
Update nnstreamer package to support c++14 language standard
This is primarily needed to support armnn tensor filter subplugin
armnn headers use c++14 compliant code
V2:
Updated gcc version check for debian
Minimum version of gcc is now 5.4
(this solves the compatibility of gcc with constexpr of c++11 and c++14)
More info in #1817
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Wed, 20 Nov 2019 12:54:44 +0000 (21:54 +0900)]
[Docs] Updated the mirrors for debian build
Updated the mirrors for debian build in the getting-started docs
This now complies with the list of mirrors used by launchpad of nnstreamer
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Wook Song [Thu, 21 Nov 2019 01:28:30 +0000 (10:28 +0900)]
[Tests] Fix Coverity defects in the bmp-to-png tool
This patch fixes unintended-sign-extension defects reported by Coverity
in the bmp-to-png tool.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Jaeyun [Wed, 20 Nov 2019 12:07:58 +0000 (21:07 +0900)]
[Common] change return type
change improper return type to get tensor element size/count.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
MyungJoo Ham [Mon, 18 Nov 2019 00:19:52 +0000 (09:19 +0900)]
[Dist/Tizen] Execute ldconfig after touching .so files
When .so files are added or removed,
ldconfig should be called.
CC: @jaeyun-jung @again4you : Cherry-pick this for 5.5 release.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Tue, 19 Nov 2019 06:14:03 +0000 (15:14 +0900)]
[CAPI] By default a new nnfw is "not available"
Until we implement its supporting code, assume a new
nnfw is NOT SUPPORTED.
This makes sure that it does not reply "supported" for
arbitrary new NNFWs.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Tue, 19 Nov 2019 07:25:42 +0000 (16:25 +0900)]
[Filter/Caffe2] fix invalid data type
1. fix invalid tensor meta while running invoke callback
2. add debug messages for tensor meta info
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Wook Song [Tue, 19 Nov 2019 01:41:02 +0000 (10:41 +0900)]
[Github] Remove an unintended blank in the template file for PR
This patch removes an unintended blank in the self evaluation section of
the template file for PR.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Parichay Kapoor [Mon, 18 Nov 2019 04:44:22 +0000 (13:44 +0900)]
[Test/Src/IIO] Fix coverity defects
Fix coverity defects
1. Related to fd resource leak in a macro with assert failure
2. Add more defect fixes already provided in #1859 and #1863 to another location
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Fri, 8 Nov 2019 10:22:45 +0000 (19:22 +0900)]
[single] Enable flexible input dimension for custom filters
Enable flexible input dimension for custom filters
Add corresponding unittest
Add GstTensorsInfo init bug fix in tensor_filter_single.c
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
MyungJoo Ham [Thu, 7 Nov 2019 09:30:22 +0000 (18:30 +0900)]
New Element: Tensor_SRC_TizenSensor, Init
This element fetches data from Tizen Sensor Framework for NNStreamer.
This commit includes the init parts and shows how it will be built and deployed.
Other function bodies will follow in other commits/PRs.
This addresses #1143
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Parichay Kapoor [Fri, 8 Nov 2019 06:48:08 +0000 (15:48 +0900)]
[nnfw] Fix meson
Fix variables in meson.build for nnfw and nnfw-runtime
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Thu, 14 Nov 2019 08:21:38 +0000 (17:21 +0900)]
[C-Api] fix invalid sink handle
When trying to register sink callbacks to same sink element, it returns null handler.
To fix this, allocate new handle and add it into the handle list.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Wed, 13 Nov 2019 10:27:18 +0000 (19:27 +0900)]
[CodeClean] cleanup sources and fix mem leak in api
1. remove static inline function in header for sam score
2. release handle when called invoke-dynamic
3. return type to validate tensors info
4. add condition to check ncsdk2
5. clean unnecessary code in android native
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
HyoungJoo Ahn [Mon, 11 Nov 2019 09:55:07 +0000 (18:55 +0900)]
[C-API] make other mediatypes available with appsrc
Currently only 'other/tensor' and 'other/tensors' are available for appsrc with C-API. Since there are many mediatypes(video, audio, text, octet) should be used as a src data, and they cannot be used currently, this commit make them available even a partial restriction.
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Wook Song [Tue, 12 Nov 2019 09:10:27 +0000 (18:10 +0900)]
[Tests/Src/IIO] Fix Coverity defects in a macro that generates tests
This patch fixes Coverity defects in GENERATE_TESTS_TO_VERIFY_DATA_WO_TRIGGER,
which is a macro for test generation.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Tue, 12 Nov 2019 08:10:53 +0000 (17:10 +0900)]
[Tests/Src/IIO] Fix defects about resource leak found from Coverity
This patch fixes those resource leak defects which newly caused by the
previous patch set.
Signed-off-by: Wook Song <wook16.song@samsung.com>
MyungJoo Ham [Thu, 7 Nov 2019 01:43:43 +0000 (10:43 +0900)]
[Test/Custom-Easy] A basic unit test case.
A basic unit test (checking if it loads successfully) is added for custom-easy.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Thu, 24 Oct 2019 11:15:49 +0000 (20:15 +0900)]
[Filter/Custom-Easy] Custom-Easy Implementation
Developers may insert a custom filter with a single C function.
In the future, with C++ wrapper, we can insert a LAMBDA function
to a NNStreamer pipeline with a single-line of code.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Sangjung Woo [Mon, 4 Nov 2019 08:02:10 +0000 (17:02 +0900)]
[C-API] Remove unnecessary asterisk on header file
This patch removes an unnecessary asterisk on header file for doxygen.
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
MyungJoo Ham [Tue, 12 Nov 2019 02:05:16 +0000 (11:05 +0900)]
[C-API] Support MVNC with Single API
Let C-API users use Intel NCS.
CC: Wook Song <wook16.song@samsung.com>
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
MyungJoo Ham [Thu, 7 Nov 2019 09:30:22 +0000 (18:30 +0900)]
[Dist] Add Tizen Sensord (sensor-fw) Option
Enable Tizen Sensord (/platform/core/system/sensord.git)
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
HyoungJoo Ahn [Mon, 11 Nov 2019 10:26:14 +0000 (19:26 +0900)]
[Typo] fix typos of tensor_converter.c
fix typos of the `tensor_converter.c` file
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Wook Song [Fri, 8 Nov 2019 15:06:22 +0000 (00:06 +0900)]
[Tests/Source/IIO] Fix defects (CWE-637) found from Coverity
This patch fixes defects related to CWE-637 found from Coverity.
Coverity-scan: CID-263387, CID-263398, CID-263366, CID-263406,
CID-263410, CID-263372, CID-263390, CID-263361,
CID-263414
Signed-off-by: Wook Song <wook16.song@samsung.com>
Geunsik Lim [Fri, 8 Nov 2019 07:26:49 +0000 (16:26 +0900)]
Added a badge to display current coverity defects
This commit is to append a badge to display current defects of Coverity Scan.
**Changelog**
* Added a badge to display the number of the Coverity Scan defects
* Added a badge to display the repository size
* Added a badge to display the Github issues
* Added a badge to display the PRs
Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
Parichay Kapoor [Wed, 6 Nov 2019 10:47:07 +0000 (19:47 +0900)]
[android] Added unittests for android API
Added unittests for android API
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Fri, 25 Oct 2019 10:23:57 +0000 (19:23 +0900)]
[android] Add flexible dimensions support for Android API
Added flexible dimensions support for Android API
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Thu, 7 Nov 2019 02:25:25 +0000 (11:25 +0900)]
[caffe2] Do first run with try/catch
First run after configuring the tensor filter with caffe2 framework is now run with a try/catch
Check #180 for more details
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Mon, 4 Nov 2019 10:12:10 +0000 (19:12 +0900)]
[Android] prepare native libs for c-api
update build script to provide native libraries using nnstreamer c-api.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
MyungJoo Ham [Fri, 8 Nov 2019 03:49:46 +0000 (12:49 +0900)]
[Style] There is a file in DOS format!
Applied dox2unix to a file with DOS text format.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Jaeyun [Wed, 6 Nov 2019 09:13:16 +0000 (18:13 +0900)]
[Test/Api] tc with unavailable nnfw
Add testcase.
If tensorflow is not supported, opening tf model (single api) returns not-supported.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Wed, 30 Oct 2019 10:08:08 +0000 (19:08 +0900)]
[CodeClean] reduce complexity
1. declare function to iterate the elements in pipeline description
2. fix api declaration (params in same line)
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Wed, 6 Nov 2019 08:28:48 +0000 (17:28 +0900)]
[android] Updated README.md
Added updates to README.md for setting up and building nnstreamer API for android
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Tue, 5 Nov 2019 02:20:47 +0000 (11:20 +0900)]
[filter/nnfw] Unittest for invoke
Added unittest for invoke for nnfw tensor filter extension
Added unittest for get input/output dimensions
Corresponding bug fixes are also added
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Fri, 25 Oct 2019 09:41:36 +0000 (18:41 +0900)]
[Api/Single] replace single implementation
Replace the single api without pipeline, remove old version and clear build script.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Tue, 22 Oct 2019 04:49:45 +0000 (13:49 +0900)]
[nnfw] Adding nnfw base test for verification
Added nnfw base testing scenario for verification along with a few bug fixes
NNFW repo has been updated to test with cpu backend
However, that patch has not been uploaded to main tizen repo
So, in order to test this patch, clone nnfw repo and build locally
Also, building nnfw repo installs libbackend_acl_cl.so in /usr/lib in emulator
Delete this file to test nnstreamer with cpu backend with nnfw successfully
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
MyungJoo Ham [Thu, 24 Oct 2019 10:51:31 +0000 (19:51 +0900)]
[Filter/Custom-Easy-C] Prototype of Custom-Easy Tensor Function
This allows to define a custom-filter (C) easily.
This will be especially helpful for C/C++ application writers.
The same approach may be implemented for C#/Java for Tizen.NET and Android app writers.
Change-Id: Id1ff43424283c217c0688e91b919ade99e3cc134
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Sangjung Woo [Thu, 17 Oct 2019 07:10:16 +0000 (16:10 +0900)]
[Doc] Guide: how to write .NET Application
This patch newly adds the guide for writing .NET Application with
Machine Learning APIs.
Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
gichan-jang [Mon, 28 Oct 2019 05:54:08 +0000 (14:54 +0900)]
[Doc] Update document
Update document for planned feature.
Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
HyoungJoo Ahn [Thu, 17 Oct 2019 12:17:57 +0000 (21:17 +0900)]
[CAPI/DOCS] upload the first version of guide doc
this is the first version of guide doc
Signed-off-by: HyoungJoo Ahn <hello.ahn@samsung.com>
Jaeyun [Tue, 29 Oct 2019 02:29:56 +0000 (11:29 +0900)]
[Android/Api] enum for pipeline type
define enum for pipeline and element type, remove string compare.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Mon, 28 Oct 2019 11:04:32 +0000 (20:04 +0900)]
[Api] error code for no-mem
define error code for no-mem case.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Wook Song [Mon, 28 Oct 2019 01:17:53 +0000 (10:17 +0900)]
[Meson] Refactor the logic that checks dependencies on numpy
This patch refactors the logic that checks dependencies on numpy.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Jaeyun [Fri, 25 Oct 2019 04:09:56 +0000 (13:09 +0900)]
[Android] set library name in build script
change the build option 'lite' and add library name with given options.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Thu, 24 Oct 2019 08:23:56 +0000 (17:23 +0900)]
[Coverage] source path for coverage report
fix the cpp filter-subplugin path for test coverage report.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Mon, 21 Oct 2019 10:28:44 +0000 (19:28 +0900)]
[single] More inline functions for flexible dimensions
Added extra inline funcitons to provide easier API for dyanmic input dimensions
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
MyungJoo Ham [Fri, 25 Oct 2019 06:23:08 +0000 (15:23 +0900)]
[Filter/Custom] Fix compiler complaints
Recent TAOS-CI/Android complains:
.../tensor_filter_custom.c: In function 'custom_loadlib':
.../tensor_filter_custom.c:95:15: warning: assignment discards 'const' qualifier from pointer target type
dlsym_error = dlerror ();
^
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Injae Kang [Fri, 25 Oct 2019 11:31:29 +0000 (20:31 +0900)]
[Build] Include pip include path to find numpy
When numpy is installed using pip, meson couldn't find
numpy/arrayobject.h
I included pip include path to find numpy.
This patch resolves #1805.
Signed-off-by: Injae Kang <abcinje@gmail.com>
Parichay Kapoor [Mon, 21 Oct 2019 04:18:14 +0000 (13:18 +0900)]
[single] unittest for flexible dimension with tflite
Added unittest for flexible dimension with tflite
Added some bug fixes which were missed from previous commits due to commit over-write
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Wook Song [Thu, 24 Oct 2019 14:57:18 +0000 (23:57 +0900)]
[C-API] Update comment about tizen_error.h
This patch updates comment about tizen_error.h in the C-API header.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 24 Oct 2019 14:54:50 +0000 (23:54 +0900)]
[API] Remove script and patch files no longer needed
After the mock tinze_error.h is added, the script and patch files for
modification nnstreamer.h at build time are no longer needed. This patch
removes those files as well.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 24 Oct 2019 14:49:10 +0000 (23:49 +0900)]
[API/Android] Modify build script to use the mock tizen_error header
This patch modifies the build script for Android API to use mock
tizen_error.h, recently added.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Mon, 21 Oct 2019 15:10:35 +0000 (00:10 +0900)]
[Tests/Meson] Add missing line continuation
In the meson build script, there are missing line continuations that
break building. This patch fixes this issue.
Related to https://github.com/mesonbuild/meson/issues/4720
See also: https://github.com/mesonbuild/meson/commit/
90c9b868b20b11bb089fc5e0c634d5ed76fea0cb
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Mon, 21 Oct 2019 14:16:06 +0000 (23:16 +0900)]
[Tests/Sink] Format code style using gst-indent
This patch formats code style of unittest_sink that contains legacy
using gst-indent.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Mon, 21 Oct 2019 14:13:08 +0000 (23:13 +0900)]
[Tests/Sink] Add a command-line option to control pipeline's speed
In order to control the speed of tests made of videotestsrc-based single
pipeline this patch adds a command-line option.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 17 Oct 2019 06:48:05 +0000 (15:48 +0900)]
[Tests/Sink] Add jitter between start and stop pipeline
In some low performance machine such as VMs in the cloud, test cases in
unittest_sink are failed since they cannot finish their job until the
pipeline is stopped. To work around this issue, this patch adds jitter
between start and stop pipeline.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 24 Oct 2019 09:02:49 +0000 (18:02 +0900)]
[C-API] Use mock header instead of Tizen dependent one
In order to build C-API for Non-Tizen platforms, nnstreamer.h should be
modified at build time to emulate some Tizen-dependent symbols.
Instead, this patch adds a mock header, tizen_error.h, which emulates
those symbols and includes needed system header. Therefore, it is
possible to build C-API by adding this header to the build tree when
enabling C-API for Non-Tizen platform.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 24 Oct 2019 07:33:12 +0000 (16:33 +0900)]
[C-API] Add a platform-dependent header emulating tizen-error.h
This patch adds a platform-dependent header which emulates tizen-error.h
for Non-Tizen platforms. Note that any files under the platform
directory in API's include are filtered out from Tizen packages.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Wook Song [Thu, 24 Oct 2019 07:12:10 +0000 (16:12 +0900)]
[Dist/Tizen] Add a script that generates tarball for source package
For the purpose of filtering out sources which should not be include in
Tizen packages, this patch adds a script for generating source tarball
at build time.
Signed-off-by: Wook Song <wook16.song@samsung.com>
Parichay Kapoor [Thu, 24 Oct 2019 05:15:04 +0000 (14:15 +0900)]
[single] Fix possible svace issue later
Fix possible svace issue later
https://github.com/nnsuite/nnstreamer/pull/1787#discussion_r337823705
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Wed, 23 Oct 2019 03:39:52 +0000 (12:39 +0900)]
[Meson] update required meson version
Set min meson version 0.50 and cleanup the script.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Thu, 24 Oct 2019 04:47:29 +0000 (13:47 +0900)]
[Coverity] fix coverity issues
Update code to fix invalid condition and free functions in nnstreamer plugins.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 24 Oct 2019 05:38:41 +0000 (14:38 +0900)]
[single] flexible dimension unittest fail fix
Added fix for flexible dimension unittest fail fix
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Mon, 21 Oct 2019 03:00:18 +0000 (12:00 +0900)]
[single] simplify get_tensors_info in single API
Added simplification for get_tensors_info in single API implementation
input and output tensors info is stored in single API
This patch updates to use that information than extracting the info again from element
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Thu, 17 Oct 2019 10:32:16 +0000 (19:32 +0900)]
[single] Implementation of flexible input dimension
Added implementation of flexible input dimension at the API level
Handling flexible input dimension at the tensor filter extension
scope to be done.
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
MyungJoo Ham [Tue, 15 Oct 2019 07:47:03 +0000 (16:47 +0900)]
[Filter] Enable C++ subplugin build
Before adding test cases for C++ filter subplugins,
enable build for Tizen/Ubuntu.
Add enable-cppfilter=true for meson build if you
want to use this feature.
Note that the subplugin is not yet fully implemented
and needs further implementation. Thus, we do not have
testcases, yet.
Changes at v2:
- cppfilter is now default
- added isAllocatedBeforeInvoke() virtual function
Change-Id: I5e1942e0b7c957ceb675bd78496041a888978e65
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Parichay Kapoor [Tue, 22 Oct 2019 04:27:04 +0000 (13:27 +0900)]
[pytorch] Do first run with try/catch
First run after configuring the tensor filter with pytorch framework
is now run with a try/catch.
Check #1809 for more details.
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Parichay Kapoor [Fri, 18 Oct 2019 03:41:53 +0000 (12:41 +0900)]
[TF-Lite] Added setInputDimension
Added implementation for setInputDimension
This can allow support for flexible input dimensions
The support is limited to only changing dimensions
Data type for the input or the number of inputs cannot be changed with current TF-Lite API
The rank for the input is allowed to be changed if the model allows
The input dimension can return error if the model does not allow dimension to be changed
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jaeyun [Tue, 22 Oct 2019 10:52:33 +0000 (19:52 +0900)]
[Decoder/Sub] install img segment
Add definition to install image segment subplugin.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Tue, 22 Oct 2019 07:21:07 +0000 (16:21 +0900)]
[Decoder] update callback description
Update callback description for decoder sub-plugin.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Tue, 22 Oct 2019 08:18:00 +0000 (17:18 +0900)]
[Decoder/Sub] mem error case in img segment
1. handle allocation failure case
2. fix possible mem leak to get the out caps
3. add definition to include img segment in android-libray
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Mon, 21 Oct 2019 06:28:59 +0000 (15:28 +0900)]
[Android] add option to build with min plugins
Add build option to set target abi and core gstreamer plugins.
For example, to build library with core plugins for arm64
./build-android-lib.sh --api_option=core --target_abi=arm64
This only includes gstreamer core plugins, so library size is under 15MB.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
niklasjang [Sun, 20 Oct 2019 13:00:02 +0000 (06:00 -0700)]
[Decoder] Set color to segmented image accoring to label_idx
Each pixel has a probability of 21 labels.
Apply a color of the label with the highest probability.
Signed-off-by: niklasjang <niklasjang@gmail.com>
Signed-off-by: Jihoon Lee <ulla4571@gmail.com>
Jaeyun [Fri, 18 Oct 2019 05:51:34 +0000 (14:51 +0900)]
[Android/Doc] update build guide
Update build guide with prebuilt gstreamer-1.16.1 binaries.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Thu, 17 Oct 2019 07:48:01 +0000 (16:48 +0900)]
[single] Added API and skeleton for flexible input dimension
Adding support for flexible input dimension
Added API and skeleton for setting input dimension via single API
V2:
Added more description for retval of ML_ERROR_NOT_SUPPORTED in documentation
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Jihoon Lee [Sun, 20 Oct 2019 10:58:00 +0000 (19:58 +0900)]
[Decoder/Meson] Meson build for image segmentation decoder
In order to add tensor decoder for image segmentation to build tree,
updates meson build script.
Signed-off-by: niklasjang <niklasjang@gmail.com>
Signed-off-by: Jihoon Lee <ulla4571@gmail.com>
Jihoon Lee [Sun, 20 Oct 2019 10:57:40 +0000 (19:57 +0900)]
[WIP] Tensor decoder for image segmentation
In order to implement tensor decoder for image segmentation,
this patch composes skeleton code for it.
Signed-off-by: niklasjang <niklasjang@gmail.com>
Signed-off-by: Jihoon Lee <ulla4571@gmail.com>
Dongwon Lee [Sun, 20 Oct 2019 07:09:09 +0000 (16:09 +0900)]
[CodeClean] replace meaningless variables
rename a and b to other more meaningful ones.
a -> from, b -> to
Signed-off-by: Dongwon Lee <dongwon1103@naver.com>
Injae Kang [Sat, 19 Oct 2019 16:50:56 +0000 (01:50 +0900)]
[Typo] Fix typo
This patch fixes typo in packaging/nnstreamer.spec
Signed-off-by: Injae Kang <abcinje@gmail.com>
Junsang Mo [Thu, 17 Oct 2019 12:46:27 +0000 (21:46 +0900)]
[CodeClean] combine if statements
combine if statements in tensor_transform.c.
I confirmed it worked well in example_image_classification_tflite.
Signed-off-by: Junsang Mo <junsang.mo@samsung.com>
이원준 [Fri, 18 Oct 2019 08:30:13 +0000 (17:30 +0900)]
Add execution bit of gradlew
Added excution bit of gradlew to prevent permission denied error.
Signed-off-by: 이원준 <dldnjs1013@nate.com>
Jaeyun [Thu, 17 Oct 2019 04:18:33 +0000 (13:18 +0900)]
[Security] handle returned value
Check mem allocation failure case and refactor file-close in IIO.
This PR is related with ssms-1016-0006.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Wed, 16 Oct 2019 04:25:39 +0000 (13:25 +0900)]
[tests] Install models/data with tests
Install models/data with tests correctly when needed
Install tizen-capi-unittest with install-test option
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Geunsik Lim [Fri, 18 Oct 2019 04:33:21 +0000 (13:33 +0900)]
Added the coverity badge
This commit is to append the badge of the Coverity (static analysis tool).
TODO: the Coverity module of TAOS-CI will be soon enabled in the nnstreamer GitHub repository.
Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
Jaeyun [Thu, 17 Oct 2019 04:39:09 +0000 (13:39 +0900)]
[Security] [Converter] handle text format in chain function
get min block size and use memcpy when input mime-type is text.
This PR is related with ssms-1016-0005.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Thu, 17 Oct 2019 05:31:30 +0000 (14:31 +0900)]
[Security] python lib name format
Define macro for file format and remove macro(preprocess defines) in the function call.
This PR is related with ssms-1016-0007.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Jaeyun [Wed, 16 Oct 2019 08:31:31 +0000 (17:31 +0900)]
[Test] update model in tensor-filter
Add testcase to re-open nn model in tensor-filter using gst-harness.
Signed-off-by: Jaeyun Jung <jy1210.jung@samsung.com>
Parichay Kapoor [Wed, 16 Oct 2019 04:08:53 +0000 (13:08 +0900)]
[tests] Locate all models and data together
Locate all the models and data together for easier management
Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
Junsang Mo [Wed, 16 Oct 2019 11:40:21 +0000 (20:40 +0900)]
[CodeClean]remove duplicates in TFLiteCore::getTensorDim
make a variable for interpreter->tensor (tensor_idx)->dims.
replace interpreter->tensor (tensor_idx)->dims with the variable.
Signed-off-by: Junsang Mo <junsang.mo@samsung.com>