Skip to content

Custom layers with nGraph#16628

Merged
alalek merged 2 commits intoopencv:3.4from
dkurt:dnn_ngraph_custom_layers
Feb 26, 2020
Merged

Custom layers with nGraph#16628
alalek merged 2 commits intoopencv:3.4from
dkurt:dnn_ngraph_custom_layers

Conversation

@dkurt
Copy link
Copy Markdown
Member

@dkurt dkurt commented Feb 20, 2020 •
edited by alalek
Loading

Pull Request Readiness Checklist

wrap OpenCV layers which are not supported by Inference Engine backend to nGraph ops. It will let us to remove old API completely.

force_builders=Custom,Custom Win,Custom Mac
build_image:Custom=ubuntu-openvino-2020.1.0:16.04
build_image:Custom Win=openvino-2020.1.0
build_image:Custom Mac=openvino-2020.1.0

test_modules:Custom=dnn,python2,python3,java
test_modules:Custom Win=dnn,python2,python3,java
test_modules:Custom Mac=dnn,python2,python3,java

buildworker:Custom=linux-1
# disabled due high memory usage: test_opencl:Custom=ON
test_opencl:Custom=OFF
test_bigdata:Custom=1
test_filter:Custom=*

@dkurt dkurt requested a review from l-bat February 20, 2020 17:32
@dkurt dkurt force-pushed the dnn_ngraph_custom_layers branch from ff444b3 to 75c886b Compare February 20, 2020 17:32
l-bat
l-bat reviewed Feb 21, 2020
View reviewed changes
dkurt
dkurt commented Feb 21, 2020
View reviewed changes
@dkurt dkurt force-pushed the dnn_ngraph_custom_layers branch from b55416f to aa37b5c Compare February 21, 2020 15:17
alalek
alalek reviewed Feb 21, 2020
View reviewed changes
alalek
alalek reviewed Feb 21, 2020
View reviewed changes
modules/dnn/src/ie_ngraph.cpp
std::map<std::string, InferenceEngine::Parameter> params = {
{"impl", oss.str()},
{"outputs", shapesToStr(outputs)},
{"internals", shapesToStr(internals)}
Copy link
Copy Markdown
Member

@alalek alalek Feb 21, 2020

Choose a reason for hiding this comment

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

Does it really necessary to serialize all of this into strings?
What is about related object lifetime handling?

Copy link
Copy Markdown
Member Author

@dkurt dkurt Feb 24, 2020 •
edited
Loading

Choose a reason for hiding this comment

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

I afraid not at the current state of custom layers at Inference Engine. They still work with limited set of layers parameters: https://github.com/opencv/dldt/blob/b2140c0/inference-engine/src/inference_engine/ie_cnn_layer_builder.cpp#L82-L106

@dkurt dkurt mentioned this pull request Feb 22, 2020
Open
@dkurt dkurt force-pushed the dnn_ngraph_custom_layers branch from aa37b5c to 1ee4fd6 Compare February 24, 2020 17:15
@dkurt dkurt force-pushed the dnn_ngraph_custom_layers branch from 1ee4fd6 to bf1b1ad Compare February 24, 2020 17:19
alalek
alalek approved these changes Feb 26, 2020
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!

@alalek alalek merged commit d8dea78 into opencv:3.4 Feb 26, 2020
@alalek alalek mentioned this pull request Feb 26, 2020
Merged
@dkurt dkurt mentioned this pull request Mar 5, 2020
Merged
2 tasks
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

@mshabunin mshabunin Awaiting requested review from mshabunin

@l-bat l-bat Awaiting requested review from l-bat

Assignees

@alalek alalek

Labels

category: dnn feature

Projects

None yet

Milestone

3.4.10

Development

Successfully merging this pull request may close these issues.

3 participants

@dkurt @alalek @l-bat