* Use reference_wrapper for preprocess bindings
* Update tests
* add bindings to I420_SINGLE_PLANE and I420_THREE_PLANES
* remove init from all classes except PrePostProcessor and add RGBX and BGRX to ColorFormat enum
* update test name
* add test for taking input and output by name
* move ref_wrapper to common.hpp
* add common.hpp include to PrePostProcessor
* update comment on ref_wrapper
* added split_to_concat_multi_inputs test
* [GNA] fixes according to comments
*Deleted commented str
*Updated opset for MatMul
*Added gna_sw_exact mode
* [GNA] Added input with n * 512 size
* [GNA] Changed input gen params
* rebase commit
* Added special passes to remove and add Convert operations in POT
* Update passes
* Implement changes to support new FP16 models
* Apply codestyle patch
* Revert Cast inserting and add data parameter
* Update FastBC rule
Co-authored-by: Malinin, Nikita <nikita.malinin@intel.com>
* Try to fix Windows Azure
* Try static MSVC runtime
* Fixed type
* cnpy
* Rename
* Set CMAKE_MSVC_RUNTIME_LIBRARY
* Fixed issues with oneDNN
* Removed IB
* Minimize a number of threads
* Increase image size to avoid test failures on some platforms
There is an assert 'length >= nlanes' and 'nlanes' value depends on machine architecture
Set 320x320 test image to guarantee that image size >= nlanes
Also increased image size for 'plugin shared tests' to ensure the same (even though tests do not work with legacy preprocessing)
* Descreased to 160x160
* Specify toolchain for samples
* Build samples and its tests even for static build
* Used 3.21 to enable CMAKE_TOOLCHAIN_FILE env var usage
* Fixed cmake download path
* Set CMAKE_TOOLCHAIN_FILE conditionally
* Fix
Co-authored-by: Michael Nosov <mikhail.nosov@intel.com>
* add 2nd batch paddle models in Convert_Model_From_Paddle.md
* narrow down supported paddle model from MobileNet-SSDLite to MobileNetv3-SSD in Convert_Model_From_Paddle.md
* Started to write transformation to fuse Interpolate sequence.
* Some changes.
* Written the transformation to fuse two Interpolate layers.
* Deleted commented code.
* Small fixes.
* Some fixes.
* Started to write tests.
* Small fix.
* Added more tests. Deleted commented code.
* Deleted redundant headers.
* Small fix.
* Fixes in the function can_be_fused(): the last statement was decomposed.
* Added operators == and != for op::v4::Interpolate::InterpolateAttrs.
* Added more checks for nullptr.
* Fixed codestyle.
* Added Interpolate registration.
* Small change.
* Implementation of operator== of InterpolateAttrs was moved into header file.
+ Bugfix for invalid removal of reorder in front of fsv32 conv
+ Remove unnecessary reorder of first conv of bfyx output format
Signed-off-by: Min, Byungil <byungil.min@intel.com>
