Skip to content

dnn : fix illegal memory access in int8 convolution#20800

Merged
opencv-pushbot merged 1 commit intoopencv:masterfrom
jebastin-nadar:fix-int8-conv
Oct 3, 2021
Merged

dnn : fix illegal memory access in int8 convolution#20800
opencv-pushbot merged 1 commit intoopencv:masterfrom
jebastin-nadar:fix-int8-conv

Conversation

@jebastin-nadar
Copy link
Copy Markdown
Contributor

@jebastin-nadar jebastin-nadar commented Oct 3, 2021 •
edited by alalek
Loading

resolves #20799

[ RUN      ] Test_Int8_layers.Convolution2D/0, where GetParam() = OCV/CPU
--28388-- REDIR: 0x7798eb0 (libc.so.6:posix_memalign) redirected to 0x4c33f60 (posix_memalign)
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/tensorflow/tf_importer.cpp (2759) populateNet DNN/TF: parsing model (N/A version info). Number of nodes = 6
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/tensorflow/tf_importer.cpp (2759) populateNet DNN/TF: parsing model (N/A version info). Number of nodes = 3
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/tensorflow/tf_importer.cpp (2759) populateNet DNN/TF: parsing model (N/A version info). Number of nodes = 10
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/tensorflow/tf_importer.cpp (2759) populateNet DNN/TF: parsing model (N/A version info). Number of nodes = 10
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/tensorflow/tf_importer.cpp (2759) populateNet DNN/TF: parsing model (N/A version info). Number of nodes = 13
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/onnx/onnx_importer.cpp (569) populateNet DNN/ONNX: loading ONNX v3 model produced by 'pytorch':0.3. Number of nodes = 1, inputs = 3, outputs = 1
[ INFO:0] global /home/jebastin/opencv_build/opencv/modules/dnn/src/onnx/onnx_importer.cpp (569) populateNet DNN/ONNX: loading ONNX v3 model produced by 'pytorch':0.3. Number of nodes = 2, inputs = 5, outputs = 1
[       OK ] Test_Int8_layers.Convolution2D/0 (6893 ms)

Modified handling of outputMultipliers. Made it similar how biasvec is handled for odd number of filters.

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

  • I agree to contribute to the project under Apache 2 License.
  • To the best of my knowledge, the proposed patch is not based on a code under GPL or other license that is incompatible with OpenCV
  • The PR is proposed to proper branch
  • There is reference to original bug report and related work
  • There is accuracy test, performance test and test data in opencv_extra repository, if applicable
    Patch to opencv_extra has the same branch name.
  • The feature is well documented and sample code can be built with the project CMake
force_builders=Custom Win
buildworker:Custom Win=windows-3
build_image:Custom Win=msvs2019
test_opencl:Custom Win=ON

@jebastin-nadar
Copy link
Copy Markdown
Contributor Author

jebastin-nadar commented Oct 3, 2021

@alalek Should I port the relaxed thresholds for failed int8 tests from #20535 to this PR? I dont think #20535 is getting merged before the release, so it makes sense to fix the pending build failures from #20228 ASAP.

alalek
alalek approved these changes Oct 3, 2021
View reviewed changes
Copy link
Copy Markdown
Member

@alalek alalek left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you!

@opencv-pushbot opencv-pushbot merged commit 603d623 into opencv:master Oct 3, 2021
@alalek alalek mentioned this pull request Oct 15, 2021
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@alalek alalek alalek approved these changes

@vpisarev vpisarev Awaiting requested review from vpisarev

Assignees

@alalek alalek

Labels

bug category: dnn

Projects

None yet

Milestone

4.5.4

Development

Successfully merging this pull request may close these issues.

DNN: Invalid memory access in Int8 code

3 participants

@jebastin-nadar @alalek @opencv-pushbot