Current limitations for fix:Expand,Shape andConstantFill layers are not supported, since their parameters use constBlobs, which are not accessible during allocation of layers memory

There is a problem with loading ONNX networks with input shapes denoted by some string parameter (e.g. "width", "height"). It seems that OpenCV parser replaces these parameters to zeroes by default.

One of the examples is MaskRCNN in#17914 . I believe that converting from PyTorch to ONNX with "dynamic_axes" will produce the same issues.

Xopencv_extra=named_params_onnxforce_builders=Custom,Custom Win,Custom Macbuild_image:Custom=ubuntu-openvino-2021.1.0:20.04build_image:Custom Win=openvino-2021.1.0build_image:Custom Mac=openvino-2021.1.0test_modules:Custom=dnn,python2,python3,javatest_modules:Custom Win=dnn,python2,python3,javatest_modules:Custom Mac=dnn,python2,python3,javabuildworker:Custom=linux-1# disabled due high memory usage: test_opencl:Custom=ONtest_opencl:Custom=OFFtest_bigdata:Custom=1test_filter:Custom=*

sl-sergei added category: dnn pr: reproducerReproduces some bug, not expected to be merged labels

Aug 20, 2020

sl-sergei mentioned this pull request

Aug 20, 2020

Add testdata for networks with parametrized input dimsopencv/opencv_extra#798

Merged

sl-sergei force-pushed thefix_17914 branch from6683c7c to0c106b9Compare

October 7, 2020 19:11

Copy link

ContributorAuthor

sl-sergei commentedOct 12, 2020

@alalek Can you please check test Layer_Test_Slice.slice_channels_and_batch_17762 ? My changes should not affect this test case at all, but anyway it fails on MYRIAD device

sl-sergei requested a review fromdkurt

October 15, 2020 19:07

Copy link

ContributorAuthor

sl-sergei commentedOct 15, 2020

@dkurt Can you please review the changes? Also, I think I need your help with some tests on Custom machines. For some reason, there are failed cases on MYRIAD/CPU with DLIE, even though some of them are not related to my code (for example, Layer_Test_Slice.slice_channels_and_batch_17762). I can't reproduce it locally on my machine, maybe you have any ideas about failed tests.

sl-sergei changed the title~~WIP: Fix problems with loading ONNX networks with parametrized inputs~~Fix problems with loading ONNX networks with parametrized inputs

Nov 16, 2020

sl-sergei force-pushed thefix_17914 branch 2 times, most recently from2edc42d to00ab83fCompare

November 16, 2020 11:59

Copy link

Member

alalek commentedNov 16, 2020

BTW, avoid using#N in commit messages. This causes creation of false cross-links on GitHub.

Copy link

ContributorAuthor

sl-sergei commentedNov 16, 2020

BTW, avoid using#N in commit messages. This causes creation of false cross-links on GitHub.

thank you!

Copy link

Member

alalek commentedNov 16, 2020•
edited
Loading

Ignore this ABI error. It belongs toexperimental namespace
I will filter out that on CI later (there is a bug in filter code).

vtable ABI issue is hard to resolve (usually it is done throughdynamic_cast, but this requires the closed list of "derived" classes)

sl-sergei force-pushed thefix_17914 branch from4acefab tob1072d0Compare

November 16, 2020 18:39

sl-sergei requested review fromdkurt and removed request fordkurt

November 16, 2020 18:40

Copy link

ContributorAuthor

sl-sergei commentedNov 16, 2020

Ignore this ABI error. It belongs toexperimental namespace
I will filter out it on CI later.
vtable ABI issue is hard to resolve (usually it is done throughdynamic_cast, but this requires the closed list of "derived" classes)

Thanks! Is there some article/wiki page about such details of OpenCV's CI configuration? I am asking just for the future work, so there will be less problems like this

Copy link

Member

alalek commentedNov 16, 2020•
edited
Loading

ABI checks are not intended to be configured externally. It is a part ofBuildbot configuration (updates require restart, GitHub code may be a bit outdated).

https://github.com/opencv/opencv/wiki/CI-configuration