Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Datumaro backlog #1468

Closed
8 of 33 tasks
zhiltsov-max opened this issue Apr 25, 2020 · 1 comment
Closed
8 of 33 tasks

Datumaro backlog #1468

zhiltsov-max opened this issue Apr 25, 2020 · 1 comment
Assignees
@zhiltsov-max

Comments

@zhiltsov-max
Copy link
Contributor

zhiltsov-max commented Apr 25, 2020 •
edited
Loading

Enhancements:

  • Maintain actual documentation
  • Try to use attrs library
  • Typing annotations for API classes
  • Provide handy decorators for extractors, converters, importers
  • Simplify implementation of extractors, converters, importers
  • Simplify dataset import - merge extractors and importers as it is done for converters
  • More advanced support of attributes - default values, type, possible values
  • Image processing transforms (lazy)
  • occluded as a class member for shapes and masks instead of a free attribute
  • Possible, dot access for annotation attributes
  • Annotation properties along with attributes so attributes could save their specific meaning
  • Regularize tests for dataset formats, simplify code
  • Provide examples for dataset formats (Support relative paths in import and export #1463)
  • Error control in import and export (probably using warnings Python package) (related: Output a list of errors when importing and exporting annotations #1529)
  • zip/git/whatever - packed sources
  • Separate image reading from annotation reading?
  • RLE for masks by default
  • Source info from extractors
  • Remove AnnotationType, just use types
  • Optimize image saving in converters by source copying
    • Option to keep source image extension if present during export
  • Support various image formats in TF Detection API format
  • Support tracks? Possibly, as dataset-level annotations or groups.
  • Provide from_iterable & categories constructor for Dataset. Reduce 3 levels of indentation in tests.
  • Apparently, move Dataset closer to TaskData from dm.bindings.
    • Add Dataset-level annotations? (e.g. tracks)
    • Dataset metainfo
  • Reduce noise in annotation files by reducing floating point precision (e.g., down to 2 digits)
  • Support label-less datasets in various formats?
  • Make Dataset class the main citizen, instead of Extractor
  • Adding / removal of image watermarks

CLI:

  • Tests
  • Cache control commands
  • Source configuration
  • Import source as project (extractor detection)
@zhiltsov-max zhiltsov-max self-assigned this Apr 25, 2020
@zhiltsov-max zhiltsov-max added this to the Backlog milestone May 12, 2020
@snyk-bot snyk-bot mentioned this issue Oct 9, 2020
Merged
@zhiltsov-max
Copy link
Contributor Author

Moved to openvinotoolkit/datumaro#13

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@zhiltsov-max zhiltsov-max

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@zhiltsov-max