Skip to content

DNN(IE): use HAVE_DNN_IE_NN_BUILDER_2019 for NN Builder API code#16694

Merged
opencv-pushbot merged 1 commit intoopencv:masterfrom
alalek:dnn_disable_nn_builder_api
Mar 6, 2020
Merged

DNN(IE): use HAVE_DNN_IE_NN_BUILDER_2019 for NN Builder API code#16694
opencv-pushbot merged 1 commit intoopencv:masterfrom
alalek:dnn_disable_nn_builder_api

Conversation

@alalek
Copy link
Copy Markdown
Member

@alalek alalek commented Feb 28, 2020 •
edited
Loading

Avoid using of deprecated IE code.

TODO:

  • CMake flags to control building with NN Builder API
  • Backport to 3.4
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=*

#buildworker:Custom=linux-4
#docker_image:Custom=ubuntu-cuda:18.04

@alalek
dnn(IE): use HAVE_DNN_IE_NN_BUILDER_2019 for NN Builder API code
29d2144 
- CMake option: OPENCV_DNN_IE_NN_BUILDER_2019
@alalek alalek mentioned this pull request Mar 3, 2020
Merged
@alalek
Copy link
Copy Markdown
Member Author

alalek commented Mar 5, 2020

Something is broken here.

There are two failed tests:

  • Test_Model.Detection_normalized/3, where GetParam() = NGRAPH/MYRIAD (failed for both IE NN Builder API ON/OFF)
[ RUN      ] Test_Model.Detection_normalized/3, where GetParam() = NGRAPH/MYRIAD
Unmatched prediction: class 15 score 0.995117 box [217 x 211 from (5, 9)]
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:130: Failure
Value of: matched
  Actual: false
Expected: true
Unmatched reference: class 15 score 0.999222 box [227 x 222 from (0, 4)]
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:140: Failure
Expected: (refScores[i]) <= (confThreshold), actual: 0.999222 vs 1.17549e-38
[  FAILED  ] Test_Model.Detection_normalized/3, where GetParam() = NGRAPH/MYRIAD (608 ms)
  • Test_Torch_nets.ENet_accuracy/2, where GetParam() = NGRAPH/CPU (IE NN Builder API = ON):
[ RUN      ] Test_Torch_nets.ENet_accuracy/2, where GetParam() = NGRAPH/CPU
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:70: Failure
Expected: (normL1) <= (l1), actual: nan vs 0.00044
/build/precommit_custom_linux/opencv/modules/dnn/test/test_torch_importer.cpp:390: Failure
Expected equality of these values:
  countNonZero(refMask != testMask)
    Which is: 207440
  0
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:70: Failure
Expected: (normL1) <= (l1), actual: nan vs 0.00044
/build/precommit_custom_linux/opencv/modules/dnn/test/test_torch_importer.cpp:390: Failure
Expected equality of these values:
  countNonZero(refMask != testMask)
    Which is: 207440
  0
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:70: Failure
Expected: (normL1) <= (l1), actual: nan vs 0.00044
/build/precommit_custom_linux/opencv/modules/dnn/test/test_torch_importer.cpp:390: Failure
Expected equality of these values:
  countNonZero(refMask != testMask)
    Which is: 207440
  0
/build/precommit_custom_linux/opencv/modules/dnn/test/test_common.impl.hpp:70: Failure
Expected: (normL1) <= (l1), actual: nan vs 0.00044
/build/precommit_custom_linux/opencv/modules/dnn/test/test_torch_importer.cpp:390: Failure
Expected equality of these values:
  countNonZero(refMask != testMask)
    Which is: 207440
  0
[  FAILED  ] Test_Torch_nets.ENet_accuracy/2, where GetParam() = NGRAPH/CPU (1092 ms)

@dkurt
Copy link
Copy Markdown
Member

dkurt commented Mar 5, 2020

I've seen Test_Model.Detection_normalized/3 failed before.

Test_Torch_nets.ENet_accuracy/2 must be related to #16628 and especially bf1b1ad which solves similar error when the feature was introduced.

@alalek
Copy link
Copy Markdown
Member Author

alalek commented Mar 6, 2020

OK.

This looks strange for me because there are no similar DNN failures on current nightly master build.

Looks like ENet test is flaky too (passed again).

@alalek
Copy link
Copy Markdown
Member Author

alalek commented Mar 6, 2020

👍

@opencv-pushbot opencv-pushbot merged commit 6271192 into opencv:master Mar 6, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@dkurt dkurt Awaiting requested review from dkurt

@mshabunin mshabunin Awaiting requested review from mshabunin

Assignees

@alalek alalek

Labels

category: dnn port/backport done Label for maintainers. Authors of PR can ignore this

Projects

None yet

Milestone

4.3.0

Development

Successfully merging this pull request may close these issues.

3 participants

@alalek @dkurt @opencv-pushbot