Add MMDetection COCO format importer

[wonjuleee]

Summary

How to test

Checklist

• I have added unit tests to cover my changes.​
• I have added integration tests to cover my changes.​
• I have added the description of my changes into CHANGELOG.​
• I have updated the documentation accordingly

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) 2023 Intel Corporation
#
# SPDX-License-Identifier: MIT

[codecov]

Codecov Report

Attention: 6 lines in your changes are missing coverage. Please review.

Comparison is base (4b9f55f) 80.08% compared to head (d1f57b8) 80.10%.

Files	Patch %	Lines
src/datumaro/plugins/data_formats/coco/base.py	76.92%	1 Missing and 2 partials ⚠️
src/datumaro/plugins/data_formats/mmdet.py	92.10%	1 Missing and 2 partials ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1213      +/-   ##
===========================================
+ Coverage    80.08%   80.10%   +0.02%     
===========================================
  Files          268      269       +1     
  Lines        29863    29915      +52     
  Branches      5845     5850       +5     
===========================================
+ Hits         23916    23964      +48     
- Misses        4614     4616       +2     
- Partials      1333     1335       +2     

Flag	Coverage Δ
macos-11_Python-3.8	79.22% <88.46%> (+0.01%)	⬆️
ubuntu-20.04_Python-3.8	80.09% <88.46%> (+0.01%)	⬆️
windows-2022_Python-3.8	80.07% <88.46%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[vinnamkim]

LGTM. However, I have a feeling that we need to find a way to scale these COCO variants (like Roboflow) at the code level more efficiently. I guess that they have same file contents but different paths.

[wonjuleee]

LGTM. However, I have a feeling that we need to find a way to scale these COCO variants (like Roboflow) at the code level more efficiently. I guess that they have same file contents but different paths.

Thank you for the comment. I agree with you. We might be able to support similar directory structures through small modifications and relaxation. This is same for other famous public formats such as VOC and Cityscapes within mmdetection document https://mmdetection.readthedocs.io/en/latest/user_guides/dataset_prepare.html. Let me think about this more for support kaggle-like datasets in the same codebase.