Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Add mask target generator operator for Mask-RCNN #16268

Merged
merged 3 commits into from
Oct 14, 2019

Conversation

Kh4L
Copy link
Contributor

@Kh4L Kh4L commented Sep 25, 2019

Description

Add mrcnn_target GPU operator.

It computes Mask-RCNN mask targets. It takes the same arguments and produces the same outputs as this Gluon block.

Micro-benchmark:

mrcnn_target operator 374ns vs Reference 5.43μs

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

  • Changes are complete (i.e. I finished coding on this PR)
  • All changes have test coverage:
  • Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
  • Code is well-documented:
  • For user-facing API changes, API doc string has been updated.
  • For new C++ functions in header files, their functionalities and arguments are documented.
  • For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
  • To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

  • Operator, tests, (and when applicable, API doc)

Comments

  • Add type support
  • Need to test Symbol API

Signed-off-by: Serge Panev <spanev@nvidia.com>
Signed-off-by: Serge Panev <spanev@nvidia.com>
@anirudhacharya
Copy link
Member

Micro-benchmark:

mrcnn_target operator 374ns vs Reference 5.43μs

What is the reference in this benchmark?

@Jerryzcn
Copy link
Contributor

Micro-benchmark:

mrcnn_target operator 374ns vs Reference 5.43μs

What is the reference in this benchmark?

Probably the current python version

src/operator/contrib/mrcnn_target.cu Outdated Show resolved Hide resolved
src/operator/contrib/mrcnn_target.cu Outdated Show resolved Hide resolved
src/operator/contrib/mrcnn_target.cu Outdated Show resolved Hide resolved
@Kh4L
Copy link
Contributor Author

Kh4L commented Sep 25, 2019

mrcnn_target operator 374ns vs Reference 5.43μs

@anirudhacharya

@Jerryzcn is right, it is this Gluon code, hybridized.

Signed-off-by: Serge Panev <spanev@nvidia.com>
Copy link
Member

@wkcn wkcn left a comment

Choose a reason for hiding this comment

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

LGTM. Thank you!

@zhreshold zhreshold merged commit 1d4ede3 into apache:master Oct 14, 2019
aaronmarkham pushed a commit to aaronmarkham/incubator-mxnet that referenced this pull request Oct 16, 2019
* Add mask target generator operator for Mask-RCNN

Signed-off-by: Serge Panev <spanev@nvidia.com>

* Disable the unit test for CPU default ctx

Signed-off-by: Serge Panev <spanev@nvidia.com>

* Address PR comments

Signed-off-by: Serge Panev <spanev@nvidia.com>
apeforest pushed a commit that referenced this pull request Nov 14, 2019
* Add mask target generator operator for Mask-RCNN

Signed-off-by: Serge Panev <spanev@nvidia.com>

* Disable the unit test for CPU default ctx

Signed-off-by: Serge Panev <spanev@nvidia.com>

* Address PR comments

Signed-off-by: Serge Panev <spanev@nvidia.com>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants