Generic media support

[zhiltsov-max]

Summary

Depends on #538
Related #675

• Added DatasetItem.media to replace dedicated members for each media type
• Added the PointCloud media type
• Added the media_type() method to Extractors
• Added merging for all media types, mixed media types for an item or in the dataset produce an error
• Datasets can't have mixed media types in items. If such situation occurs, an error is raised (checked during dataset caching/iteration)
• Datasets can't change media type using transforms
• Extractors must report their media type with the media_type() method
• Added a new mandatory media_type argument to Dataset.from_iterable. It has a default value of Image for the transition period (to be tracked in Support different media types #675).
• Deprecated DatasetItem.image, .related_images, .point_cloud, save-images and require_images
• Added deprecation messages about annotation classes in components.extractor
• Suppressed Datumaro deprecation messages when using Datumaro from CLI

TODO:

• --save-media in CLI
• [ ] Media type deduction by path (do we need it at all?)
• Update formats to the new interface
• Questions with Dataset domain
  • Is it better to differentiate datasets by domain or media type? By media type
  • Dataset items must use the same media type? Yes
  • How to avoid calling transforms with unsupported media types? Transforms and converters must check input media type
  • How to avoid exporting in formats for datasets with unsupported media types?

How to test

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

[yasakova-anastasia]

@zhiltsov-max, I think PR is ready for review.

[sizov-kirill]

Now these approaches are not alternative, because if user using video_frames format it's not possible to export such dataset in other Image media type format. I think we should to mention it in docs or allow VideoFrame media type for Image formats.

[zhiltsov-max]

@kirill-sizov, VideoFrame is subclass of Image, so it should be working.

[sizov-kirill]

@kirill-sizov, VideoFrame is subclass of Image, so it should be working.

@zhiltsov-max, I meant that for example such code throws an error:

from datumaro.components.dataset import Dataset

dataset = Dataset.import_from('video.mp4', format='video_frames')
dataset.export('./save_dir', 'voc', save_media=True)

datumaro.components.errors.MediaTypeError: Media type is not an image