platform/upstream/nnstreamer.git
4 years ago[C-API] Add element-wise control functions
Sangjung Woo [Tue, 16 Jun 2020 08:10:12 +0000 (17:10 +0900)]
[C-API] Add element-wise control functions

This patch newly adds element-wise control functions to get and set
property values of the target element in NNStreamer pipeline.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[C-API] Rename ml_pipeline_get_element() to ml_pipeline_get_gst_element()
Sangjung Woo [Wed, 17 Jun 2020 10:22:23 +0000 (19:22 +0900)]
[C-API] Rename ml_pipeline_get_element() to ml_pipeline_get_gst_element()

This patch renames ml_pipeline_get_element() to
ml_pipeline_get_gst_element(). It is only used for demo application
internally and the name 'ml_pipeline_get_element' is going to use for
other purposes.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[CAPI] Add ml_error_e into CAPI_ML_FRAMEWORK group accepted/tizen/unified/20200702.141729 submit/tizen/20200702.043208
Sangjung Woo [Wed, 1 Jul 2020 10:29:00 +0000 (19:29 +0900)]
[CAPI] Add ml_error_e into CAPI_ML_FRAMEWORK group

This patch newly adds the ml_error_e Enumeration into the
CAPI_ML_FRAMEWORK doxygen group.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Decoder] Add tensor decoder subplugin flatbuffers
gichan-jang [Wed, 26 Feb 2020 11:30:28 +0000 (20:30 +0900)]
[Decoder] Add tensor decoder subplugin flatbuffers

Add tensor decoder subplugin for flatbuffer

flatbuffers package can be found here :
https://launchpad.net/~nnstreamer/+archive/ubuntu/ppa-build-test/+packages

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Test] Add flatbuffer converter and decode test
gichan-jang [Tue, 26 May 2020 12:16:54 +0000 (21:16 +0900)]
[Test] Add flatbuffer converter and decode test
Add flatbuffer converter and decoder run test

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[TEST] Fix Tensor converter test
gichan-jang [Thu, 25 Jun 2020 11:54:13 +0000 (20:54 +0900)]
[TEST] Fix Tensor converter test

fix tensor converter unit test

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[C-Api] fix mem leak
Jaeyun [Fri, 26 Jun 2020 02:04:20 +0000 (11:04 +0900)]
[C-Api] fix mem leak

1. add g_free to release the memory when appending mem block in buffer.
2. free interator in switch api.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Repo] fix a mem leak accepted/tizen/unified/20200629.143505 submit/tizen/20200626.060219
Hyoung Joo Ahn [Thu, 25 Jun 2020 11:37:02 +0000 (20:37 +0900)]
[Repo] fix a mem leak

fix a mem leak at tensor_repo

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Mux] add the new sync_mode:REFRESH
Hyoung Joo Ahn [Thu, 25 Jun 2020 03:13:58 +0000 (12:13 +0900)]
[Mux] add the new sync_mode:REFRESH

This new sync mode allows PUSHING with only one new buffer of sinkpads of `tensor_mux`. Previously, if `tensor_mux` want to push its buffers to srcpad, all of sink pads had to be filled with new buffers. However, this new mode, REFRESH, push buffers when `tensor_mux` receive a new buffer. For the other sink pads, except the one received the new buffer, will use again the previous one. It decides EOS when any of the sink pads receive it.

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Test] fix mem leak
Jaeyun [Thu, 25 Jun 2020 11:42:18 +0000 (20:42 +0900)]
[Test] fix mem leak

fix possible mem leak in c-api testcases.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Filter] fix a memory leak
Hyoung Joo Ahn [Thu, 25 Jun 2020 10:39:34 +0000 (19:39 +0900)]
[Filter] fix a memory leak

the allocated memory(when `allocate_in_invoke` is FALSE) should be free when `invoke()` returns not 0(OK)

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Tests/Filter/EdgeTPU] Enable test() in the meson build script
Wook Song [Wed, 24 Jun 2020 07:09:21 +0000 (16:09 +0900)]
[Tests/Filter/EdgeTPU] Enable test() in the meson build script

This patch enables test(), which has been commented out, in the meson
build script for the EdgeTPU filter.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Meson] Add missing environment variable, NNSTREAMER_SOURCE_ROOT_PATH
Wook Song [Wed, 24 Jun 2020 06:49:51 +0000 (15:49 +0900)]
[Meson] Add missing environment variable, NNSTREAMER_SOURCE_ROOT_PATH

This patch adds the missing environment variable,
NNSTREAMER_SOURCE_ROOT_PATH, to the test environment object in the
meson build script. This variable is necessary to run the test cases
via 'ninja test'.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Mux/Merge] code clean
Jaeyun [Fri, 19 Jun 2020 06:19:55 +0000 (15:19 +0900)]
[Mux/Merge] code clean

code clean, prepare new sync-mode.
- remove unnecessary param
- reduce code complexity
- change return val (true to push buffer to src pad)

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[CodeClean] remove unnecessary defines
Jaeyun [Tue, 23 Jun 2020 11:17:32 +0000 (20:17 +0900)]
[CodeClean] remove unnecessary defines

code clean, remove unnecessary predefinition to set pad template.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[TEST] Fix nnfw unit test
gichan-jang [Wed, 24 Jun 2020 11:04:43 +0000 (20:04 +0900)]
[TEST] Fix nnfw unit test

When testing the operation of the multi-framework, if tflite is disabled, the test fails.
If tflite is disabled, skip the test.

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Tests] Rename NNSTREAMER_BUILD_ROOT_PATH with NNSTREAMER_SOURCE_ROOT_PATH
Wook Song [Wed, 24 Jun 2020 05:57:51 +0000 (14:57 +0900)]
[Tests] Rename NNSTREAMER_BUILD_ROOT_PATH with NNSTREAMER_SOURCE_ROOT_PATH

In most cases, the environment variable, NNSTREAMER_BUILD_ROOT_PATH,
is used for indicating the root directory of the path to which the
'test_models' directory belongs, which means the source root path. To
avoid confusion, this patch renames it with NNSTREAMER_SOURCE_ROOT_PATH.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Decoder] Add tensor decoder subplugin flatbuffers
gichan-jang [Wed, 26 Feb 2020 11:30:28 +0000 (20:30 +0900)]
[Decoder] Add tensor decoder subplugin flatbuffers

Add tensor decoder subplugin for flatbuffer

flatbuffers package can be found here :
https://launchpad.net/~nnstreamer/+archive/ubuntu/ppa-build-test/+packages

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Test] Add test for imageseg-decoder and tensor_filter profiling
Yongjoo Ahn [Tue, 23 Jun 2020 05:20:38 +0000 (14:20 +0900)]
[Test] Add test for imageseg-decoder and tensor_filter profiling

- Add test cases for newly added option `snpe-deeplab` and `snpe-depth`
- It also check profiling option of tensor_filter

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Conf/MemLeak] Add missing g_free() to nnstreamer_conf.c
Dongju Chae [Thu, 18 Jun 2020 05:43:18 +0000 (14:43 +0900)]
[Conf/MemLeak] Add missing g_free() to nnstreamer_conf.c

This patch adds missing `g_free()` to `nnstreamer_conf.c`

`nnsconf_get_custom_value_string()` allocates `hashkey` in the first
line. However, if a value is found in the first hash lookup or a value
does not exist in anywhere, `hashkey` is not inserted to the hash table,
which results in memory leak.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[RepoSink] ignore sync and preroll
Jaeyun [Fri, 19 Jun 2020 06:45:48 +0000 (15:45 +0900)]
[RepoSink] ignore sync and preroll

repo-sink may not receive a buffer when pipeline constructed.
set sync and async option to ignore this case.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Coverage] Exclude generated files by flatbuf
gichan-jang [Fri, 19 Jun 2020 08:01:00 +0000 (17:01 +0900)]
[Coverage] Exclude generated files by flatbuf

Exclude generated files by flatbuf from coverage report

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Tests/EdgeTPU] Use dummy device mode instead of dummy library
Wook Song [Thu, 18 Jun 2020 02:57:06 +0000 (11:57 +0900)]
[Tests/EdgeTPU] Use dummy device mode instead of dummy library

This patch revises the unit test cases for EdgeTPU to use dummy device
mode instead of the dummy EdgeTPU library.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[C-Api] handle invalid info when creating data handle
Jaeyun [Wed, 17 Jun 2020 03:45:50 +0000 (12:45 +0900)]
[C-Api] handle invalid info when creating data handle

When creating the data handle, return error if given tensors-info is invalid.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years agoTAOS-CI / Enable clang-format for C++ codes
MyungJoo Ham [Fri, 5 Jun 2020 05:51:55 +0000 (14:51 +0900)]
TAOS-CI / Enable clang-format for C++ codes

1. Enable clang TAOS-CI module
2. Add "C++ ONLY" mode for it.

Note that I need to update TAOS-CI to make it accept such options.
This depends on https://github.com/nnstreamer/TAOS-CI/pull/628

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Test/C-Api] add testcase of internal functions
Jaeyun [Tue, 16 Jun 2020 07:35:35 +0000 (16:35 +0900)]
[Test/C-Api] add testcase of internal functions

add simple testcases to handle the case of invalid param.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[C-Api] error code when failed to allocate tensor data
Jaeyun [Tue, 16 Jun 2020 07:26:21 +0000 (16:26 +0900)]
[C-Api] error code when failed to allocate tensor data

Remove unnecessary error code (stream-pipe).
If failed to allocate mem for tensors-data handle, API returns out-of-mem error.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Meson] converter sub-plugin path in test-env
Jaeyun [Mon, 15 Jun 2020 12:47:53 +0000 (21:47 +0900)]
[Meson] converter sub-plugin path in test-env

Update sub-plugin path in test-env, set converter plugin path.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Fix] Remove unncessary arg checking before calling g_strdup()
Dongju Chae [Tue, 16 Jun 2020 11:04:20 +0000 (20:04 +0900)]
[Fix] Remove unncessary arg checking before calling g_strdup()

This patch removes uncessary argument checking before calling g_strdup()

If an argument of g_strdup() is NULL, its return value is also NULL.
So, checking its argument before calling is duplicated and useless.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Repo/Test] make it easier to understand
Hyoung Joo Ahn [Tue, 16 Jun 2020 03:13:41 +0000 (12:13 +0900)]
[Repo/Test] make it easier to understand

applying line change at the complicated pipeline to understand easier.

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Filter/EdgeTPU] Add a 'dummy' device mode for the test purpose
Wook Song [Thu, 11 Jun 2020 07:47:17 +0000 (16:47 +0900)]
[Filter/EdgeTPU] Add a 'dummy' device mode for the test purpose

This patch adds a 'dummy' device mode for the test purpose instead of
using the dummy shared library that pretends to be libedgetpu.so.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Android] do not convert meta in invoke method
Jaeyun [Wed, 10 Jun 2020 04:11:21 +0000 (13:11 +0900)]
[Android] do not convert meta in invoke method

To improve performance, it is necessary to remove metadata conversion in invoke method.
Do not handle dynamic mode in invoke method and update description to change the tensor info.
Also, if output info is changed using set-info method, update cached instance in private data.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android] create info handle in private data
Jaeyun [Wed, 10 Jun 2020 04:03:22 +0000 (13:03 +0900)]
[Android] create info handle in private data

create info handle when allocating private data.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Filter/tflite] Fix typo
junmin-kim [Tue, 16 Jun 2020 05:46:30 +0000 (14:46 +0900)]
[Filter/tflite] Fix typo

Assignment should be ended with semicolon if there is no special reason.
Replace `,` with `;`

Signed-off-by: junmin-kim <junmindd.kim@samsung.com>
4 years ago[Android] constructor with info
Jaeyun [Wed, 10 Jun 2020 09:08:33 +0000 (18:08 +0900)]
[Android] constructor with info

In tensors-data class, update private constructor with tensor-info.
This can remove method call to set metadata in native.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Bugfix/Auto] Fix the bug to access freed memory region in auto fwname
Dongju Chae [Fri, 12 Jun 2020 07:36:37 +0000 (16:36 +0900)]
[Bugfix/Auto] Fix the bug to access freed memory region in auto fwname

This patch fixes the memory bug in auto tensor filter detection.

`gst_tensor_filter_get_available_framework` assumes that `fw_name` and
`prop->fwname` are pointing different memory regions.

However, `_gtfc_setprop_Model` has passed the argument `fw_name` as
`prop->fwname`, which may cause the invalid memory access.

In this case, `fw_name` is being accessed after `prop->fwname` is freed.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Tensor Converter] Add tensor converter subplugin flatbuffer
gichan-jang [Thu, 14 May 2020 02:25:12 +0000 (11:25 +0900)]
[Tensor Converter] Add tensor converter subplugin flatbuffer
Add tensor converter subplugin flatbuffer

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Filter/Custom] copy metadata in registration
Jaeyun [Wed, 10 Jun 2020 10:35:40 +0000 (19:35 +0900)]
[Filter/Custom] copy metadata in registration

In the filter custom-easy, prevent error case when developer tries to register custom-filter with local variables.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Filter/EdgeTPU] Handle the error case when a compiled model is given
Wook Song [Wed, 10 Jun 2020 10:33:42 +0000 (19:33 +0900)]
[Filter/EdgeTPU] Handle the error case when a compiled model is given

This patch is to handle the error case when a model compiled by
edgetpu-compiler is given. In general, a higher version of
tensorflow-lite (> 1.13) could support more various types of the tflite
models.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Converter] reduce complexity
Jaeyun [Fri, 12 Jun 2020 10:30:16 +0000 (19:30 +0900)]
[Converter] reduce complexity

code clean to reduce complexity
1. internal function to push multiple buffers.
2. remove unnecessary args

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android/Doc] build options
Jaeyun [Fri, 12 Jun 2020 08:27:56 +0000 (17:27 +0900)]
[Android/Doc] build options

Update document, build options for android library.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Tests/Meson] Fix the unknown variable error related to python filters
Wook Song [Mon, 15 Jun 2020 05:29:53 +0000 (14:29 +0900)]
[Tests/Meson] Fix the unknown variable error related to python filters

This patch fixes the unknown variable error related to the filter
sub-plugins for python 2 and 3.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Meson] Remove trailing whitespaces
Wook Song [Mon, 15 Jun 2020 05:12:12 +0000 (14:12 +0900)]
[Meson] Remove trailing whitespaces

This is a trivial patch to eliminate trailing whitespaces of the meson
build script in the source root.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Doc] fix wrong file name
Jaeyun [Mon, 15 Jun 2020 05:05:10 +0000 (14:05 +0900)]
[Doc] fix wrong file name

fix wrong file name in converter header file.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Doc] fix invalid document
Jaeyun [Fri, 12 Jun 2020 09:18:50 +0000 (18:18 +0900)]
[Doc] fix invalid document

typo correction and fix document for converter sub-plugin api.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Tests/Filter] Add test case for NNS_custom_easy_unregister()
Sangjung Woo [Thu, 11 Jun 2020 02:12:11 +0000 (11:12 +0900)]
[Tests/Filter] Add test case for NNS_custom_easy_unregister()

This patch newl adds test cases for NNS_custom_easy_unregister().

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Filter/Custom/Easy] Add unregister function
Sangjung Woo [Thu, 11 Jun 2020 01:58:56 +0000 (10:58 +0900)]
[Filter/Custom/Easy] Add unregister function

This patch newly adds 'NNS_custom_easy_unregister()' function to remove
registered function so that pipeline itself can be reused without
changing custom model name.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Debian/control] Add flatbuf-dev build dependent package
gichan-jang [Wed, 10 Jun 2020 05:40:26 +0000 (14:40 +0900)]
[Debian/control] Add flatbuf-dev build dependent package
As flatbuf headers are separated from tflite-dev pkg, flatbuf-dev pkg is added.

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Converter] Change find external converter function
gichan-jang [Wed, 10 Jun 2020 10:23:41 +0000 (19:23 +0900)]
[Converter] Change find external converter function
Change to search for sub-plugins according to the mime type regardless of the name of the library

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Converter] Change prototype of subplugin
gichan-jang [Fri, 29 May 2020 07:09:17 +0000 (16:09 +0900)]
[Converter] Change prototype of subplugin
Change subplugin function prototype of tensor converter

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Converter] Change tensor config to tensors config
gichan-jang [Thu, 28 May 2020 01:45:33 +0000 (10:45 +0900)]
[Converter] Change tensor config to tensors config

Change tensor config to tensors config. Assume single tensor only except subplugin.

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Converter] Prepare to register sub-plugin
gichan-jang [Mon, 18 May 2020 01:37:16 +0000 (10:37 +0900)]
[Converter] Prepare to register sub-plugin

Preprare to register sub-plugin for flatbuffers and protobuf

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Filter/C++] Do not go further when failed to open NN framework
Wook Song [Wed, 10 Jun 2020 07:54:43 +0000 (16:54 +0900)]
[Filter/C++] Do not go further when failed to open NN framework

This patch adds more catch blocks that return negative error code in
cpp_open so that the execution does not go further when it failed to
open NN framework.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Dist/Debian] Provide a sub-package for the Edge TPU extension
Wook Song [Wed, 10 Jun 2020 04:51:24 +0000 (13:51 +0900)]
[Dist/Debian] Provide a sub-package for the Edge TPU extension

This patch provides a sub-package for the Edge TPU extension.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Ext/Filter/Meson] Revise resolving build dependency for Edge TPU
Wook Song [Wed, 10 Jun 2020 02:50:10 +0000 (11:50 +0900)]
[Ext/Filter/Meson] Revise resolving build dependency for Edge TPU

This patch revises the mechanism that resolves the build dependency on
libedgetpu and header files.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Ext/Filter/Meson] Elminate white spaces
Wook Song [Wed, 10 Jun 2020 02:27:56 +0000 (11:27 +0900)]
[Ext/Filter/Meson] Elminate white spaces

This is a trivial fix that eliminates white spaces in the meson build
script for the filter extensions.

Signed-off-by: Wook Song <wook16.song@samsung.com>
4 years ago[Tests] Add 'nosync' option into mux
Sangjung Woo [Wed, 10 Jun 2020 01:16:46 +0000 (10:16 +0900)]
[Tests] Add 'nosync' option into mux

To fix `framerate` issue in mux, it set sync-mode to slowest as default.
However, This LSTM test case is hanging as a side-effect since it uses
both tensor_repo and tensor_mux. This patch adds 'nosync' option into
mux so that the test pipeline starts and finishes without hanging.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[Mux] Set sync-mode to slowest as default
Sangjung Woo [Tue, 9 Jun 2020 05:10:26 +0000 (14:10 +0900)]
[Mux] Set sync-mode to slowest as default

After passing mux / demux, the framerate is extremely bigger than
before. (Issue #2440) This is mainly because sync_mode of tensor_mux
filter is SYNC_NOSYNC. To fix this bug, this patch set sync_mode to
SYNC_SLOWEST as default.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years agoFilter/Custom/Easy: fix doxygen entry
MyungJoo Ham [Tue, 9 Jun 2020 10:29:48 +0000 (19:29 +0900)]
Filter/Custom/Easy: fix doxygen entry

Fix easy-custom-filter API's doxygen entry.
It gets output dimenesion info as an input, not output.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Android] implement on-load in native
Jaeyun [Tue, 9 Jun 2020 06:22:21 +0000 (15:22 +0900)]
[Android] implement on-load in native

register native functions when loading native library.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android] return to static build
Jaeyun [Tue, 9 Jun 2020 04:24:12 +0000 (13:24 +0900)]
[Android] return to static build

To execute model faster, return to static build.
Make all sub-plugins static libs and build one native so file.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Fix][Filter/Prop] Fix performance property (latency) for tensorfilter
Yongjoo Ahn [Fri, 5 Jun 2020 07:15:54 +0000 (16:15 +0900)]
[Fix][Filter/Prop] Fix performance property (latency) for tensorfilter

- Initialize with zero value.

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Decoder/Neon] Support neon accl for SNPE deeplab/depth seg decoder
Dongju Chae [Thu, 4 Jun 2020 11:25:54 +0000 (20:25 +0900)]
[Decoder/Neon] Support neon accl for SNPE deeplab/depth seg decoder

This patch supports neon accl for SNPE deeplab/depth seg decoder

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Decoder/Fix] Revise tensordec-imagesegment impl. and fix minor bugs
Dongju Chae [Thu, 4 Jun 2020 10:53:01 +0000 (19:53 +0900)]
[Decoder/Fix] Revise tensordec-imagesegment impl. and fix minor bugs

This patch revises tensordec-imagesegment impl. and fix minor bugs

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Android] private data to set metadata
Jaeyun [Wed, 3 Jun 2020 05:19:06 +0000 (14:19 +0900)]
[Android] private data to set metadata

Each invoke callback gets/returns tensor data object including metadata.
To improve performance, Cache tensor metadata object and avoid duplicated conversion.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years agoStart development of 1.5.3 as 1.6.0-RC4 accepted/tizen/unified/20200608.144742 submit/tizen/20200608.023449
MyungJoo Ham [Thu, 4 Jun 2020 02:50:24 +0000 (11:50 +0900)]
Start development of 1.5.3 as 1.6.0-RC4

As soon as all major known bugs and SVACE/Coverity issues are fixed,
I'm going to release 1.6.0.

Cheers!

```
1.5.2 -> 1.5.3
- Mediapipe's graphs (NNStreamer pipeline equivalent) may be embedded as an element in NNStreamer pipeline
- Qualcomm SNPE is supported (tensor-filter subplugin)
- Verisilicon Vivante is supported (tensor-filter subplugin)
- NNStreamer --> Protobuf decoder added.
- New tensor-filter subplugin API "v1" released. (with v0 backward compatibility)
- Tensor-filter now accepts C++ classes as a subplugin. Edge-TPU subplugin is re-written as an example.

- Meson script re-worked.
- Semantics of hardware-acceleration options for tensor-filter re-worked.
- API/Android: nnfw-runtime (neurun) and SNPE support
- API/Android: usability update.
- API/Android: less invokation latency. (more optimization coming in next versions)
- API/C: bugfixes, architectural upgrade, latency reduction.
- tensor-filter has latency and throughput performance monitors.
- tensor-sink is by default "sync=false". If appsink or tensor_sink in NNStreamer Pipeline API's pipeline has sync=true, emit warning messages.
- Architectural updates: lower CC, less duplication, removed dependency cycles, less complicated #if statements and blocks.
- Test suite updates: timeout handling, arm-arch error fixes, more test cases (supplying more negative cases), floating-point handling. and a lot more fixes and performance (latency) optimizations.
- Build script updates: cleaned up dependencies and applied "feature" meson feature.
- No more essential class in assert() for optimized binaries.
- Much less assertions. Apply error-handling instead.
- Daily build & test activated and published.
- Rules & policy updated to comply with LF/AI.
- Memory leaks removed: from demux, split, and a few more components.

- Known bug: edge-TPU subplugin is not working in Tizen devices recently.
```

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Android] typo in mk file
Jaeyun [Fri, 5 Jun 2020 13:41:37 +0000 (22:41 +0900)]
[Android] typo in mk file

fix typo in mk file (jni nnstreamer defines)

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years agoTizen/spec: disable mvncsdk/edgeTPU for DA build
MyungJoo Ham [Fri, 5 Jun 2020 05:57:22 +0000 (14:57 +0900)]
Tizen/spec: disable mvncsdk/edgeTPU for DA build

DA requested to remove mvncsdk/edgeTPU from the dependency.
Disable them for their builds.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[NNFW] set backend after loading model
Jaeyun [Thu, 4 Jun 2020 11:16:20 +0000 (20:16 +0900)]
[NNFW] set backend after loading model

recently updated, nnfw set-backend should be called after loading the model.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years agodoc: how to enable vivante tensor filter
Geunsik Lim [Tue, 2 Jun 2020 11:10:18 +0000 (20:10 +0900)]
doc: how to enable vivante tensor filter

This commit is to append how to enable the Vivante tensor filter
based on Vivante NPU Toolkit library on Ubuntu/ARM and Tizen/ARM devices.

Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
4 years agoMovidius Filter: Fixed typos
Geunsik Lim [Thu, 4 Jun 2020 11:28:56 +0000 (20:28 +0900)]
Movidius Filter: Fixed typos

This commit is trivial. Let's fix typos in the Movidius filter file.

Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
4 years ago[Filter/TFLite] remove ineffective codes
Hyoung Joo Ahn [Thu, 4 Jun 2020 08:38:30 +0000 (17:38 +0900)]
[Filter/TFLite] remove ineffective codes

setting allocation type is not effective anymore.

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Decoder] Extend tensordec-imagesegment to support snpe-deeplab/depth
Dongju Chae [Mon, 1 Jun 2020 12:31:08 +0000 (12:31 +0000)]
[Decoder] Extend tensordec-imagesegment to support snpe-deeplab/depth

This patch extends tensordec-imagesegment to support snpe-deeplab/depth.
Also, it include minor code refactoring.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Filter/Prop] Add performance properties (lat/fps) for tensor filter
Dongju Chae [Wed, 3 Jun 2020 04:01:35 +0000 (13:01 +0900)]
[Filter/Prop] Add performance properties (lat/fps) for tensor filter

This patch adds performance properties (latency/throughput) for tensor filter.

Those properties will be used for the overlay information, to show the performance
of tensor filter (e.g., SNPE).

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years agoSPEC: fixed indentation
Geunsik Lim [Thu, 4 Jun 2020 02:25:57 +0000 (11:25 +0900)]
SPEC: fixed indentation

This commit is trivial. It is to keep an indentation consistently.

Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
4 years ago[Android/Custom] drop input buffer
Jaeyun [Wed, 3 Jun 2020 02:40:36 +0000 (11:40 +0900)]
[Android/Custom] drop input buffer

In custom-filter interface, returning null in invoke-callback will drop the input buffer.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[C-Api] function to get pipeline element
Jaeyun [Wed, 3 Jun 2020 05:08:39 +0000 (14:08 +0900)]
[C-Api] function to get pipeline element

Add new function to get the gst-element from pipeline handle.
For native developers, it is necessary to get pipeline element to use gstreamer functions.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Log] add definition log-fatal
Jaeyun [Thu, 4 Jun 2020 02:38:10 +0000 (11:38 +0900)]
[Log] add definition log-fatal

define new nns_logf (common fatal-log definition with nns_log prefix)

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Decoder/protobuf] Add protobuf tensor_decoder sub-plugin submit/tizen/20200604.023022
Yongjoo Ahn [Tue, 24 Mar 2020 08:48:03 +0000 (17:48 +0900)]
[Decoder/protobuf] Add protobuf tensor_decoder sub-plugin

- Add tensor_decoder sub-plugin "protobuf" to support Protocol Buffers

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Filter] bugfix: dangling pointer access.
MyungJoo Ham [Wed, 3 Jun 2020 05:25:59 +0000 (14:25 +0900)]
[Filter] bugfix: dangling pointer access.

gfree should be called when you completed accesing it.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoREADME update: add committer, update hw-accl, ci status
MyungJoo Ham [Wed, 3 Jun 2020 06:26:13 +0000 (15:26 +0900)]
README update: add committer, update hw-accl, ci status

- Add the new committer
- Add the list of hardware accelerators
- Add daily-release status

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years agoDoc update: component-description
MyungJoo Ham [Wed, 3 Jun 2020 06:21:43 +0000 (15:21 +0900)]
Doc update: component-description

There are a few components completed. Update component list

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
4 years ago[Filter] fix a tiny mem leak
Hyoung Joo Ahn [Wed, 3 Jun 2020 10:36:58 +0000 (19:36 +0900)]
[Filter] fix a tiny mem leak

move a few `g_free()` to fix mem leak

Signed-off-by: Hyoung Joo Ahn <hello.ahn@samsung.com>
4 years ago[Build/Fix] Add a missing header to 'nnstreamer_cppplugin_api_filter.hh'
Dongju Chae [Wed, 3 Jun 2020 09:30:08 +0000 (18:30 +0900)]
[Build/Fix] Add a missing header to 'nnstreamer_cppplugin_api_filter.hh'

This patch adds a missing header, string.h, to
'nnstreamer_cppplugin_api_filter.hh', which uses memcpy().

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[Conf] parse sub-plugins name
Jaeyun [Tue, 2 Jun 2020 07:20:26 +0000 (16:20 +0900)]
[Conf] parse sub-plugins name

Basename is unnecessary, set sub-plugins name while loading configuration.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Build/Tests] Fix 'tests/meson.build' to always find 'copy'
Dongju Chae [Tue, 2 Jun 2020 01:26:21 +0000 (10:26 +0900)]
[Build/Tests] Fix 'tests/meson.build' to always find 'copy'

This patch fixes minor build errors when `libpng_dep` does not exist.
`tests/nnstreamer_filter_extensions_common/meson.build` is using this
variable regardless of `libpng_dep`'s existence.

Signed-off-by: Dongju Chae <dongju.chae@samsung.com>
4 years ago[single] Add missing gstTensorFilterFramework v1 support
Parichay Kapoor [Mon, 18 May 2020 18:13:51 +0000 (23:43 +0530)]
[single] Add missing gstTensorFilterFramework v1 support

Add missing support for gstTensorFilterFramework in checking framework availability
for a given backend accelerator/hardware

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Android] cache method id
Jaeyun [Wed, 20 May 2020 11:49:51 +0000 (20:49 +0900)]
[Android] cache method id

1. get required method ids to improve performance when initializing the pipe info.
2. add internal method to handle arraylist.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Filter] release mem block when execution failed
Jaeyun [Fri, 29 May 2020 03:50:18 +0000 (12:50 +0900)]
[Filter] release mem block when execution failed

release in/out mem blocks when failed to invoke nn framework.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android/SNPE] Decouple SNPE env setting from native-api
Yongjoo Ahn [Mon, 1 Jun 2020 07:38:17 +0000 (16:38 +0900)]
[Android/SNPE] Decouple SNPE env setting from native-api

- Move function for android SNPE env setting from nnstreamer-native-api.c to tensor_filter_snpe.cc

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[SNPE] Add custom properties
Yongjoo Ahn [Wed, 27 May 2020 06:19:28 +0000 (15:19 +0900)]
[SNPE] Add custom properties

- Add Debug message
- Add custom prop for snpe runtime and cpu_fallback options
ex) "custom=Runtime:GPU,CPUFallback:true"

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Android] prepare dsp/npu runtime for snpe filter
Yongjoo Ahn [Fri, 15 May 2020 01:15:48 +0000 (10:15 +0900)]
[Android] prepare dsp/npu runtime for snpe filter

- Add required snpe-sdk libs
- Set `jniLibs.srcDirs` to this project explictly contains snpe-sdk libs for dsp/npu
- Make function `nns_set_env` to set env "ADSP_LIBRARY_PATH" (https://developer.qualcomm.com/docs/snpe/dsp_runtime.html)
- Log SNPE version
- Prevent build SNAP and SNPE together in android build script and
  Android.mk

Signed-off-by: Yongjoo Ahn <yongjoo1.ahn@samsung.com>
4 years ago[Aggregator] Remove duplicated function to set caps
gichan-jang [Fri, 29 May 2020 07:50:25 +0000 (16:50 +0900)]
[Aggregator] Remove duplicated function to set caps
Remove duplicated works to set source pad cap when sink event is received

Signed-off-by: gichan-jang <gichan2.jang@samsung.com>
4 years ago[Unittest] Fix a potential bug of unit testcase
Sangjung Woo [Fri, 22 May 2020 05:26:47 +0000 (14:26 +0900)]
[Unittest] Fix a potential bug of unit testcase

This patch fixes the potential bug (i.e. pipeline parsing error) of unit
testcase when using Protobuf v3.8.

Signed-off-by: Sangjung Woo <sangjung.woo@samsung.com>
4 years ago[subplugins] Update subplugins accl support
Parichay Kapoor [Fri, 15 May 2020 10:28:47 +0000 (15:58 +0530)]
[subplugins] Update subplugins accl support

Update subplugin's accelerators support as per updated rules for auto/default

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
4 years ago[Tensor Filter] Added a sub-plugin filter for VIM3/Vivante NPU
Geunsik Lim [Fri, 15 May 2020 05:00:50 +0000 (14:00 +0900)]
[Tensor Filter] Added a sub-plugin filter for VIM3/Vivante NPU

This commit is to append the Vivante sub-plugin based on the tensor filter
of NNStreamer to support the Vivante NPU driver.

**Changelog**

* Version 10:
   * Added license clause
   * Improved statements of the nnsreamer.spec file

* Version 1-9:
   * Fixed memory leak (e.g., g_free)
   * FIxed security (e.g., nnstreamer.nanifest) for Tizen packaging
   * doc: Added how to build and run the Vivante filter on Ubuntu 18.04 (ARM64)
   * Updated the invoke function to support various models
     (e.g., Inception, Yolo, Mobile SSD)
   * Added Tizen package script for Tizen platform
   * Fixed incorrect var type (e.g., unsigned int for num_tensors)
   * Added annotations for code maintenance
   * Added run.sh for Ubuntu18.04 aarch64 + VIM3 board
   * Added convert function for vivante tensor data type
   * Added vsi_nn_GetTensor of OpenVX apis to obtain tensor data (input, output)
   * Added info[i].name with dummy data (e.g., tensor id)
   * Removed deprecated variables and functions
   * Updated the vivante filter with "multi-files" structure for re-usability
     - "model=/opt/vivante/model/inception_v3.nb,
       /opt/vivante/model/libinception_v3.so
   * Updated the existing vivante tensor filter based on dlopen() scheme
   * Added a functions to check the dlopen operation to avoid undefined symbol issue
   * Updated annotaiton (e.g., image size is 299:299)
   * Added Vivante functions (common APIs) in the vivante_open function
   * Added Vivante functions (common APIs) in the vivante_invoke function
   * Added Vivante functions (common APIs) in the vivante_close function
   * Update annotation to avoid misunderstanding
   * Fixed typos (GstTensorsInfo)
   * Add todo list (e.g., glib's time functions)
   * Added vsi_nn_pub.h header file to use libraries of acuity-ovxlib-dev tool
   * Added a skeleton of the Vivate tensor filter
   * Imported the major implementation of the Vivante (inceptionv3 model)
     for Nnstreamer pipeline. libvivante_v3.so will be used by "model=***"
     interface to support a generic approach for re-usability.
   * Added vivante_inception_v3.h header file for .so file of a NN model
   * Added a data structure (e.g., vivante_data) to manage Vivante data

Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
4 years ago[Android] fix warning
Jaeyun [Thu, 21 May 2020 05:32:20 +0000 (14:32 +0900)]
[Android] fix warning

fix build warning, remove unnecessary code.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Merge] reduce complexity
Jaeyun [Mon, 25 May 2020 06:02:44 +0000 (15:02 +0900)]
[Merge] reduce complexity

code clean in tensor-merge element, reduce code complexity.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android/Test] sample for media format
Jaeyun [Wed, 27 May 2020 10:08:37 +0000 (19:08 +0900)]
[Android/Test] sample for media format

test/sample to input media data format into appsrc.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>
4 years ago[Android/Doc] link proper method
Jaeyun [Mon, 25 May 2020 10:53:44 +0000 (19:53 +0900)]
[Android/Doc] link proper method

link proper method, for the interfaces defined in Pipeline/CustomFilter.

Signed-off-by: Jaeyun <jy1210.jung@samsung.com>