* - Fix Split(18) requiring num_outputs.
- Calculate `sizes` in Resize instead of using the simpler `scales`
- ORT implementation does not round correctly when applying scales
- Update center crop to use float so we are more accurate in choosing the crop area.
- Fix minor issue with Debug step by only adding values that are altered to the renaming graph inputs.
- Update unit tests expected output due to the change in Resize using sizes instead of scales.
- Crop e2e example input so before/after image covers same area.
* Simplify.
CenteredCrop doesn't need to use float as it's dividing by 2 (so using float + floor gives the same result).
Remove Resize impl using scales - we most likely will never go back to it.
Address PR comments
Update doc
* security warning of sentencepiece
* update cmake option policy
* option set
* revert abseil change
* wrong patch name
* upgrade protobuf to 3.20.2, align with ORT 1.14
* weird patch issue
* code refinement
* typo
* Update version.txt
* undo version change
* doc refinement
* one more missing
* is ==
* disaable some case running under ort 1.13
* Update cmake/externals/sentencepieceproject.cmake
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
* Update cmake/externals/sentencepieceproject.cmake
Co-authored-by: Scott McKay <skottmckay@gmail.com>
* Update cmake/externals/sentencepieceproject.cmake
Co-authored-by: Scott McKay <skottmckay@gmail.com>
* Update cmake/externals/sentencepieceproject.cmake
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
* review update
* set default warning level 3 to meet some requirement
---------
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
Co-authored-by: Scott McKay <skottmckay@gmail.com>
* re-enable Python pipelines
* deprecate the direct setup.py call
* pipeline fixing
* run cmake from visual studio
* remove the self check
* support different ort versions
* remove ort 1.9 and add 1.13
* ci pipeline fixing
* fix the test with latest pytorch
* code refinement
* torch version detection
* make package python version more consistent.
* Combined ORT + ort-ext build for Android fails due to protobuf mismatch. Move the setting of SPM_USE_BUILTIN_PROTOBUF to before the populate to fix.
* Set CACHE INTERNAL as well to avoid issue with CMP0077.
* Move the pre/post processing scripts into the python module.
Update usage/examples.
* Use better version parsing.
* Update tests, docs,
* Address PR comments.
Remove global Settings and pass onnx opset around directly where needed. Make PrePostProcessor the owner of the checker context.
* switch to amd64 Python package in CI pipeline
* try to fix binskim errors
* try new cibuildwheel parameter
* fix some binskim issues
* fix some binskim issues
* try switch to x64 again
* undo setup.py change
* because of the quotes?
* explicit AMD64?
* output the debug info
* try fixing a setuptool issue
* cleanup
* refinement
* Add BertTokenizer to iOS package ops config.
* Also register tokenizer ops in com.microsoft.extensions.
Co-authored-by: Wenbing Li <10278425+wenbingl@users.noreply.github.com>
* Added initial CustomOp template generator
* Added arg parsing, error check, and modularized methods
* Added test and models for the same plus refactored code in template generator
* Fixed model path issue
* Final changes
* Removed ONNX models and added model generation in test scripts, changed custom ops to random math ops
authored-by: Sayan Shaw <sayanshaw@microsoft.com>
- Generate AAR with associated artifacts (javadoc.jar, sources.jar, pom).
- Fix JAR/AAR package descriptions.
- Remove `gradle -b build-android.gradle` invocation from CMake. This creates the AAR and it is not necessary to run it for each Android ABI build.
- Support building on Windows.
* align python package version with version.txt
* Update setup.py
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
* remove a line
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
* Add python based build infrastructure to simplify developer builds for various platforms. Majority was copied from the ORT build script so usage is consistent with that.
Left the existing build.bat/build.sh but ideally the CI can be updated to use the new infrastructure so things are more consistent.
Updated gradle to 7.5.1 and Android gradle tools to 7.3.0.
Validated Windows and cross-compiling Android on Windows including builds with explicitly selected ops.
WASM and iOS builds aren't tested yet and might need minor tweaks.
* Update build.py to require Python 3.7, remove git submodule sync, reorder options.
* Use 'cmake -E remove' to remove file.
* Enable specifying the ORT version to fetch
* Add ability to enable Java bindings.
Co-authored-by: Wenbing Li <10278425+wenbingl@users.noreply.github.com>
Co-authored-by: edgchen1 <18449977+edgchen1@users.noreply.github.com>
* Separate ops in operators/cv2 that do and do not require codecs so they're easier to include/exclude from a build.
Remove jpeg2000 from opencv file formats. It costs 1MB and is (afaict) not a common format.
Add ability to enable/disable cv2 ops to gen_selectedops.py.
* Remove super resolution pre/post process ops that are no longer needed.
* Replace super resolution e2e tutorial
Co-authored-by: Wenbing Li <10278425+wenbingl@users.noreply.github.com>
* - Fix incorrect weights matrix for conversion from YCbCr to BGR
- Fix Resize when input is HWC or CHW by converting to NHWC/NCHW so bilinear interpolation is inferred. trilinear is inferred otherwise.
- Not aware of a use case where we'd want to use trilinear - if one exists the temporary addition of the batch dim would need to be made explicit.
- Allow Resize to use antialias if ONNX opset is 18 or higher
- Skip adding unnecessary Mul if FloatToImageBytes multiplier is 1.0.
- Make Debug step available at the top level
Validated opset 18 Resize vs PT for super resolution and the results are now a really good match.
* Update test data to use image that shows diffs better.
Make super resolution output format configurable
Address PR comments.
* Adjust test for diffs on macos
* wip
* integrate app center test command to CI
* add app center cli install
* update ci test
* update based on new app
* minor update yaml file
* add new line at the end of file
* update app center token and comments
* Initial CLIP tokenizer implementation
* Moved common code from CLIP and GPT2 tokenizers into separate file
* add the new file into cmake file list.
* Fix ustring reference issue
* merge changes from main branch
* more merge actions
* Minor changes
Co-authored-by: Sayan Shaw <sayanshaw@microsoft.com>
Co-authored-by: Wenbing Li <10278425+wenbingl@users.noreply.github.com>
Co-authored-by: Wenbing Li <wenbingl@outlook.com>
* Using the header files from the ONNXRuntime package
* Update includes/onnxruntime_customop.hpp
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
* fix the build break.
* one more fixing
* wired top project
* ort 1.9.0 used
* switch to 1.10.0 package.
* change the vmimage to latest
* URL issue
* cmake policy
* ignore onnxruntime.dll native scan
* update the Onebranch exclusedPaths
* fixing some build tool issues
* update again
* typo
* undo of ORT dll removal
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
Initial changes for adding an iOS package.
Add infrastructure to build iOS static framework, xcframework, and CocoaPods package.
Add test project that uses package.
Add package build and test to CI pipeline.
* update the main doc and add a developer doc
* add it back
* fix some typo
* Update README.md
Co-authored-by: Nat Kershaw (MSFT) <nakersha@microsoft.com>
Co-authored-by: Nat Kershaw (MSFT) <nakersha@microsoft.com>
* Initial changes for supporting mobilenet and superresolution.
- Script to update model with pre/post processing
- custom ops for decode/encode
- user just has to provide jpg or png bytes
- superresolution can return the updated image in jpg or png
- models for testing
Updated cmake setup to enable building of the vision pre/post processing ops
- opencv2 is treated as an internal dependency rather than the mechansim for selecting which operators to include.
* Add extra check in decode.
* an android test app for extensions AAR package
* add the pipeline
* fxing the Android CI pipeline
* fix the build issus on macOS
* more fixings
* more fixings
* switch to jdk 11
* gradlew path issue
* update the command lines
* split the test task
* better name