New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Panoptic segmentation #96
Conversation
You should upload the files to OpenDR's owncloud, under the FTP server material, with the desired structure. When you upload the files, send me an email and I will take care of uploading the files to the FTP server |
87a7bc4
to
ecc59d9
Compare
@vniclas testing is now available again. So, when you are ready, you can use the "test sources" label to run the style checks and "test tools" to run the tool tests. |
Is there a particular reason why the |
The examples and test files have been changed to |
I tried to run the sample example but I'm getting some errors:
It seems that some dependencies are not correctly listed. |
As written in the known issues section of the PR, the dependencies have not been completely sorted out yet since the base repositories assume a manual installation of |
Ok, so the example added in D7.1 it currently doesn't work and cannot be tested. For installation you can simply add a bash script that will be then called by the main installation system. But what does exactly requires the third party dependency? |
First with respect of what? |
fbc562a
to
3386ae0
Compare
I checked all 3rd party tools used in this PR. The minimum common requirement for pytorch is Basically, the problem is that both inplace_abn and mmdet attempt to load |
The documentation where a first set of dependencies is specified is D2.1:
I would say that the issues is rather in these libraries than in the pip installation system. |
897a1e8
to
54e14ea
Compare
Thanks, @stefaniapedrazzi! That helped a lot. Sorry that I was not aware of this specification since it is not mentioned in the repo's wiki and I just joined this project one month ago. I added Do you have any idea why clang fails on ubuntu-20, whereas it passes on ubuntu-18? I formatted the files with clang-format-9 on an ubuntu-20 machine. |
I just checked the code and all the already merged methods are using About the error with clang, it may be that a different version is installed depending on the system. |
This reverts commit 34924c9.
6e652d9
to
f456330
Compare
f456330
to
1c693b1
Compare
@@ -189,7 +189,7 @@ def infer(self, img): | |||
raise UserWarning("No model is loaded, cannot run inference. Load a model first using load().") | |||
self.model.eval() | |||
predict = self.model(image).argmax(dim=1).squeeze().cpu() | |||
heatmap = Heatmap(predict) | |||
heatmap = Heatmap(predict.numpy()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Note: Fixing this in another package. The PR of bisenet
did not implement type checking for the Heatmap
class although only Numpy arrays were expected according to the original comments in the code. Thus, the error was not detected in the original PR.
449d6f9
to
054b8aa
Compare
054b8aa
to
43c0427
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good overall, thank you! Please double check the default parameters, as I did it very quickly.
src/opendr/perception/panoptic_segmentation/efficient_ps/efficient_ps_learner.py
Show resolved
Hide resolved
src/opendr/perception/panoptic_segmentation/efficient_ps/example_usage.py
Outdated
Show resolved
Hide resolved
tests/sources/tools/perception/panoptic_segmentation/efficient_ps/test_efficient_ps.py
Show resolved
Hide resolved
src/opendr/perception/panoptic_segmentation/efficient_ps/example_usage.py
Outdated
Show resolved
Hide resolved
Thanks for your detailed review! I'll fix everything in the course of today. |
Co-authored-by: ad-daniel <44834743+ad-daniel@users.noreply.github.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you!
Hi,
this PR adds the EfficientPS network. The original repo can be found here.
Please also check issue #90.Todos:
/docs/reference
Heatmap
implementation with version proposed in Semantic segmentation #100and updates pending in Data and Target setter/getter fixes #98Install CUDA in GitHub CI--> will not be resolvable since the code requires GPUs. See comment.Known issues:
reason for failing tests: 3rd party dependencies assume an existing pytorch installation since they attempt to loadtorch
in theirsetup.py