Release 0.3.1

[chuneuny-emily]

Summary

This PR fix

• CVS-84321 Export Annotated Dataset - lack of image's depth value for VOC dataset
• CVS-86848 Annotations inside another annotation are duplicated during import for VOC format
• CVS-85120 Cannot export detection->classification dataset

How to test

• I performed the existing unit tests locally, and there was no regression on my changes.
• I performed e2e tests manually, and the issues above were not reproducible anymore.

Checklist

• I submit my changes into the develop branch
• I have added description of my changes into CHANGELOG
• I have updated the documentation accordingly
• I have added tests to cover my changes
• I have linked related issues

License

• I submit my code changes under the same MIT License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below)

# Copyright (C) 2021 Intel Corporation
#
# SPDX-License-Identifier: MIT

[goodsong81]

LGTM.

[zhiltsov-max]

In many cases VOC export is going to take much more time now, because the images have to be loaded. For width and height we often have metainfo in other formats.

[chuneuny-emily]

The images with coco or other formats are not loaded, so item.media.data is None.
It is run when there are the loaded images. If not, it keeps "".

Any suggestion to set depth?

[zhiltsov-max]

item.media.data is None only when the image file is not available, but we have metainfo. If you wanted to avoid extra loading, you need to use has_data

[chuneuny-emily]

I see. Then I will update as using has_data you suggested soon. Do you think it should be included in v0.3.1?

[zhiltsov-max]

I can't answer on this question, release logic is out of my responsibility now.