Conversation
|
The tests are failing because the test data is not present in opencv_extra during testing. Is there any way to do that? |
|
Hi! Thanks for contributing! Please choose 3.4 branch as a target one (we will port to master after that) |
|
Please remove the branch locally (or rename) and just create a new branch from 3.4 but with the same name |
|
This patch should go into 3.4 branch first. Please:
Note: no needs to re-open PR, apply changes "inplace". P.S. The same steps are required for corresponding PR from opencv_extra. |
b856781 to
796ee4b
Compare
|
@aryanshomray Friendly reminder. Please pay attention that the patch breaks dnn module tests on CI. |
|
Hi, I am not able to understand the reason for this error since it should be able to access the patch data. |
|
@aryanshomray, an error is Most likely, |
|
@aryanshomray Thanks for the contribution! The test passes now. Please squash the commits here and in extra repo to have clean git history. |
f0ca179 to
1a952c9
Compare
|
|
||
| TEST_P(Test_Darknet_layers, tanh) | ||
| { | ||
| testDarknetLayer("tanh"); |
There was a problem hiding this comment.
@aryanshomray
Could you please update test data to have non-trivial different values in testdata/dnn/darknet/tanh_out.npy file?
Current all 8 values are equal to 1.
BTW, GitHub shows "Empty file." in opencv_extra PR, but it is not correct (GitHub bug)
There was a problem hiding this comment.
Hi, I have made the changes in the data. All 1 values might be because of large input numbers that I had taken for the test.
1a952c9 to
601851c
Compare
Merge with extra: opencv/opencv_extra#847
This PR resolves the issue #19393 and the test data is added in opencv_extra through PR #opencv/opencv_extra/pull/847
Pull Request Readiness Checklist
See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request
Patch to opencv_extra has the same branch name.