CVAT-3D Milestone-6 PCD and Kitti formats

[manasars]

Added feature for supporting import and export of PCD and Kitti format files in CVAT.

• I submit my changes into the develop branch
• I have added description of my changes into CHANGELOG

License

• I submit my code changes under the same MIT License that covers the project.

[zhiltsov-max]

• Fix license headers: remove them from empty __init__.py files. Add correct headers to new code files.
• Fix the formatting - keep 80 chars per line, remove unnecessary empty lines, use +1 level of indentation in line continuations
• Add tests on new formats
• Add docs for new formats ( like here https://github.com/openvinotoolkit/datumaro/tree/develop/docs/formats)
• Add examples of datasets in new formats (like here https://github.com/openvinotoolkit/datumaro/tree/develop/tests/assets)
• Really add a line in CHANGELOG.md

[manasars]

@zhiltsov-max

• Fix license headers: remove them from empty __init__.py files. Add correct headers to new code files.
• Fix the formatting - keep 80 chars per line, remove unnecessary empty lines, use +1 level of indentation in line continuations
• Add tests on new formats
  Working on tests
• Add docs for new formats ( like here https://github.com/openvinotoolkit/datumaro/tree/develop/docs/formats)
• Add examples of datasets in new formats (like here https://github.com/openvinotoolkit/datumaro/tree/develop/tests/assets)
• Really add a line in CHANGELOG.md

[zhiltsov-max]

Can we reduce the size of tests/assets/velodynepoint_dataset/velodyne_points/data/000001.pcd? It is 1+ MB, I suppose we can create a more lightweight alternative for tests.

[manasars]

@zhiltsov-max Updated Tests as well.

[zhiltsov-max]

Can we do something about big files in tests? They increase repository size, they are hard to verify and maintain, and they don't seem to be really used.

[manasars]

Can we do something about big files in tests? They increase repository size, they are hard to verify and maintain, and they don't seem to be really used.

Working on the same

[zhiltsov-max]

Ok, then we merge this PR, once CVAT part is approved.

[zhiltsov-max]

Attributes need to be written.

[zhiltsov-max]

Summary:

• Support writing and reading dataset subsets in the new formats
• Support (fix) writing attributes in the new formats
• Support writing and reading of builtin attributes in new formats (occluded, truncated etc.)
• Update tests with annotations

[nmanovic]

@zhiltsov-max , could you please review the PR once again? If it doesn't have critical issues, please merge it today.

[zhiltsov-max]

Covered by #320. Please update CVAT part correspondingly. Label and tag color information is not required in the supervisely format. Attributes are supported. Cuboid tracks are supported (they require track_id annotation attribute - like other tracks in the KITTI format and object annotation attribute in the supervisely format).

[jayraj-solanki]

Hi @zhiltsov-max,

Label and tag color information is not required in the supervisely format

Label and tag color information is tracked in supervisely.
Also we checked downloading the latest annotations in supervisely, the color parameter is present

Attributes are supported

Implemented in CVAT

Cuboid tracks are supported

Since Tracks is out-of-scope in 3D as per SOW, Cuboid tracks isn't implemented.

Is there any changes required from us?

[zhiltsov-max]

Label and tag color information is tracked in supervisely.
Also we checked downloading the latest annotations in supervisely, the color parameter is present

I know. What I'm saying is that it is not mandatory.

Is there any changes required from us?

Adapt the CVAT part of the M6 PR. Maybe, there will be some gaps - please let me know.