summary |
shortlog | log |
commit |
commitdiff |
tree
first ⋅ prev ⋅ next
Inki Dae [Fri, 28 Feb 2020 05:59:12 +0000 (14:59 +0900)]
Modify Load callback interface to pure virtual function
Change-Id: I94555409bb13caf4ea4c71ea2ccc40643d77cd51
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Mon, 24 Feb 2020 05:44:59 +0000 (14:44 +0900)]
Drop GetInferenceResult function
GetInferenceResult function isn't used anymore becuase upper
framework(Inference class in case of Tizen) is already aware of
output tensor buffer so drop this.
Change-Id: I16a65f4cb1c63cafe3dedf252481210978bb1baf
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Tue, 18 Feb 2020 01:54:41 +0000 (10:54 +0900)]
Add the use of user desired input and output layers
Change-Id: I140ad98452f0ab2c3878aa05a0927f4553639d6e
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Fri, 14 Feb 2020 07:19:29 +0000 (16:19 +0900)]
Update adapataion layer API descriptions
Change-Id: Ibea8f1596775f63d2912a74e60714af5c7911d34
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Fri, 14 Feb 2020 04:26:05 +0000 (13:26 +0900)]
Pass output tensor buffers to backend engine for the inference
Change-Id: I3c5ad1499daa1953874441bef42e62accf2bedc0
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Thu, 13 Feb 2020 07:35:11 +0000 (16:35 +0900)]
Pass input tensor buffers coming from Inference layer to backend engine
This patch makes InferenceEngineCommon layer to pass input tensor
buffers allocated by Inference layer to a backend engine for
the inference. As for this, it adds new several callbacks, and
drops unnecessary ones.
Change-Id: I03fd9aa9ce73cb2c1092a5a50d4147ff90e4462e
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Wed, 12 Feb 2020 03:03:57 +0000 (12:03 +0900)]
Remove InferenceEngineVision relevant things
Change-Id: Iad41e0c01f776a787c95bca3252b9c7e0c04fd1a
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Wed, 12 Feb 2020 02:05:33 +0000 (11:05 +0900)]
vision: Drop all OpenCV dependencies
This patch drops OpenCV dependent code. The dropped code will
be moved to Inference layer. This is a step to remove
inference-engine-vision layer.
Change-Id: I3fc29071ec7f7a0e4b0dec72ed7568bcc638462e
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Tue, 11 Feb 2020 06:25:53 +0000 (15:25 +0900)]
Pass model file format to backend engine
This patch passes model file format coming from Inference layer
to backend engine. As for this, it adds enumeration values
which indicate various model file format.
Change-Id: I11b141399498c73cda6a34350a6f6fd3eac74a17
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Mon, 10 Feb 2020 06:28:23 +0000 (15:28 +0900)]
Implement GetBackendCapacity callback
Change-Id: I4f30d077f85a54d09bd97b22a353f564d34a9a33
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Fri, 7 Feb 2020 06:42:03 +0000 (15:42 +0900)]
Consider one more inference target devices
Change-Id: I3183273117f73a36fe87d8dd6ca00c5d9720a02a
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Wed, 5 Feb 2020 09:32:25 +0000 (18:32 +0900)]
Drop SetInputTensorParam/SetOutputTensorParam callbacks
Change-Id: I7348440bc489490e1ce42c7fe908453b825327d8
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Wed, 5 Feb 2020 09:04:30 +0000 (18:04 +0900)]
Add a comment
Change-Id: I61c00dec619c4637cafded6dee08f5788e0181ad
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Wed, 5 Feb 2020 07:11:48 +0000 (16:11 +0900)]
Code refactoring to Inference initialization process
This patch changes Load callback and Constructor with
new version. And it replaces Init/DeInit callbacks with
BindBackend/UnbindBackend callbacks to clarify.
Change-Id: Ic4c39db96387232936ff066dc9fd81026932667d
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Tue, 4 Feb 2020 11:08:25 +0000 (20:08 +0900)]
use a new constructor
Change-Id: I3af15df073a735440effc53807e4af62abe091c1
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Mon, 3 Feb 2020 08:08:32 +0000 (17:08 +0900)]
vision: add initial code of new adaptation layer API
Change-Id: Id3b708ca882354eef56ef58e77b155938c12ec72
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Fri, 31 Jan 2020 06:42:07 +0000 (15:42 +0900)]
add new adaptation layer API support
Change-Id: I87cd2baa78f7afe8b35b22497001588505166373
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Inki Dae [Fri, 31 Jan 2020 01:56:36 +0000 (10:56 +0900)]
Add ARMNN backend support
This patch adds a enumeration value for ARMNN backend support.
Change-Id: I0e5e091203c7aac62a3f95147d6eeb4d735fcfdb
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Tae-Young Chung [Tue, 7 Apr 2020 03:27:38 +0000 (12:27 +0900)]
Keep opencv.pc file
Change-Id: I65b5d0dc7e373aa5ef4d7cae8daf225bbf47e011
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Thu, 2 Jan 2020 08:32:57 +0000 (17:32 +0900)]
Migration with OpenCV-4.2.0
Change-Id: Ie5960bb38c613ec13fa795c6ee3ce48fca3ad1d8
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Hyunsoo Park [Thu, 7 Nov 2019 06:12:33 +0000 (15:12 +0900)]
Fix wrong PC NAME and add library link.
PC NAME of common package is set wrong because of misspellings.
And stdc++fs was not linked on common package.
So i changed these to right way.
Change-Id: I410fe525dbf2f2c677c044329b603928a8f1772a
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Hyunsoo Park [Tue, 15 Oct 2019 03:07:18 +0000 (12:07 +0900)]
Add initializer in constructor for local variable.
New class variable was added, but it doesn't be initialize. So i add init codes.
Change-Id: I47d9f9299395af8b73898859e176c65ade014bcb
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Hyunsoo Park [Thu, 10 Oct 2019 08:08:43 +0000 (17:08 +0900)]
Remove SetLibraryPath api on pure interface.
It is depends on each engine. So i remove this api on pure interface.
Change-Id: I55254ad54cd1fe84373b876da1a96d2a18d2e243
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Hyunsoo Park [Tue, 8 Oct 2019 05:08:13 +0000 (14:08 +0900)]
Remove unreachable code.
Change-Id: Ia28161f5a1d84c10389f1eaaadf53c1cb164d58a
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Hyunsoo Park [Fri, 4 Oct 2019 08:29:19 +0000 (17:29 +0900)]
Add 'SetLibraryPath' api and INI parser.
It is needed when user want to set full path of so file.
Change-Id: I397f9750733454fa5b75c064a61034987bb41f22
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Tae-Young Chung [Thu, 10 Oct 2019 02:39:55 +0000 (11:39 +0900)]
Fix bugs
1. The number of channels should be applied while allocating memory to mInputBuffer.
2. The length of top_results should be less than its' size, not mOutputNumbers.
Change-Id: I0aa01da1a4230803554ab16835966599221b7cbe
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Tue, 24 Sep 2019 07:49:29 +0000 (16:49 +0900)]
Fix wrong type cast
Change-Id: Ieaf42f74bba1eb7f2c832362b519cffb7071dfe9
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Wed, 18 Sep 2019 08:30:35 +0000 (17:30 +0900)]
Reimplement class InferenceEngineVision using InferenceEngineCommon
This commit removes the IInferenceEngineVision and reimlement InferenceEngienVision
as commit title.
In detail,
InferenceEngineCommon provides general inference operation such as
set parameters, load model.
Thus, InferenceEngineVision can be implemented by InferenceEngineCommon,
instead of IInferenceEngineVision.
Change-Id: I7a43bc6ff97e5713c1ac7ae705ff52100f5e4977
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Hyunsoo Park [Fri, 6 Sep 2019 06:29:04 +0000 (15:29 +0900)]
Sets and Checks NULL value to pointer variable.
C++ delete method just release allocated memory. It doesn't set NULL value to variable.
So i make variable to NULL after delete process.
Change-Id: Ia63a08c47b2e41263e7195aaca62bb67903d8c11
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Hyunsoo Park [Thu, 22 Aug 2019 11:21:30 +0000 (20:21 +0900)]
Implementation of inheritance relationship.
IInferenceEngineCommon class should be parent of IInferenceEngineVision class.
So i implemented inheritance relationship between two classes.
Change-Id: I9a4e572dbba9d4e8fd118ec73305f46a912015fa
Signed-off-by: Hyunsoo Park <hance.park@samsung.com>
Tae-Young Chung [Wed, 24 Jul 2019 07:55:31 +0000 (16:55 +0900)]
Change the parameter of SetOutputTensorParamNodes() from std::string to std::vector<std::string>
Now, multiple output tensors are supported.
You can set the output tensors' names to get multiple output tensors
Note that a batch is not supported in current inference, so an input tensor should be one string
Change-Id: I01a3a8fa7f15ce329fbab187f6d72f583f116c2a
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Thu, 11 Jul 2019 01:19:06 +0000 (10:19 +0900)]
Add INFERENCE_TARGET_CUSTOM and change constructor to get parameter of string type
Change-Id: Ib2f1f159911e0dc0d0f1abbbf5807b354b24576b
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Thu, 4 Jul 2019 12:20:48 +0000 (21:20 +0900)]
Fix build break
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tae-Young Chung [Thu, 4 Jul 2019 10:26:17 +0000 (19:26 +0900)]
Initial codes
1. Provides inference-engine-interface-common for general purpose inference
2. Provides inference-engine-interface-vision for vision specific inference
Signed-off-by: Tae-Young Chung <ty83.chung@samsung.com>
Tizen Infrastructure [Mon, 1 Jul 2019 01:49:06 +0000 (01:49 +0000)]
Initial empty repository
projects / platform / core / multimedia / inference-engine-interface.git / log