-
Notifications
You must be signed in to change notification settings - Fork 3.1k
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.
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
Create cvat_sdk.datasets, a framework-agnostic version of cvat_sdk.pytorch #6428
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## develop #6428 +/- ##
===========================================
+ Coverage 81.70% 81.73% +0.03%
===========================================
Files 334 337 +3
Lines 38446 38501 +55
Branches 3540 3540
===========================================
+ Hits 31411 31470 +59
+ Misses 7035 7031 -4
|
…torch The new `TaskDataset` class provides conveniences like per-frame annotations, bulk data downloading, and caching without forcing a dependency on PyTorch (and somewhat awkwardly conforming to the PyTorch dataset interface). It also provides a few extra niceties, like easy access to labels and original frame numbers. Note that it's called `TaskDataset` rather than `TaskVisionDataset`, as my plan is to keep it domain-agnostic. The `MediaElement` class is extensible, and we can add, for example, support for point clouds, by adding another `load_*` method. There is currently no `ProjectDataset` equivalent, although one could (and probably should) be added later. If we add one, we should probably also add a `task_id` field to `Sample`.
5 tasks
zhiltsov-max
approved these changes
Jul 18, 2023
Merged
bsekachev
added a commit
that referenced
this pull request
Jul 27, 2023
## \[2.5.2\] - 2023-07-27 ### Added - We've added support for multi-line text attributes (<#6458>) - You can now set a default attribute value for SELECT, RADIO types on UI (<#6474>) - \[SDK\] `cvat_sdk.datasets`, is now available, providing a framework-agnostic alternative to `cvat_sdk.pytorch` (<#6428>) - We've introduced analytics for Jobs, Tasks, and Project (<#6371>) ### Changed - \[Helm\] In Helm, we've added a configurable default storage option to the chart (<#6137>) ### Removed - \[Helm\] In Helm, we've eliminated the obligatory use of hardcoded traefik ingress (<#6137>) ### Fixed - Fixed an issue with calculating the number of objects on the annotation view when frames are deleted (<#6493>) - \[SDK\] In SDK, we've fixed the issue with creating attributes with blank default values (<#6454>) - \[SDK\] We've corrected a problem in SDK where it was altering input data in models (<#6455>) - Fixed exporting of hash for shapes and tags in a specific corner case (<#6517>) - Resolved the issue where 3D jobs couldn't be opened in validation mode (<#6507>) - Fixed SAM plugin (403 code for workers in organizations) (<#6514>) - Fixed the issue where initial frame from query parameter was not opening specific frame in a job (<#6506>) - Corrected the issue with the removal of the first keyframe (<#6494>) - Fixed the display of project previews on small screens and updated stylelint & rules (<#6551>) - Implemented server-side validation for attribute specifications (<#6447>) - \[API\] Fixed API issue related to file downloading failures for filenames with special characters (<#6492>) - \[Helm\] In Helm, we've resolved an issue with multiple caches in the same RWX volume, which was preventing db migration from starting (<#6137>) --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Anastasia Yasakova <yasakova.an@gmail.com> Co-authored-by: yasakova-anastasia <anastasia@cvat.ai> Co-authored-by: Roman Donchenko <roman@cvat.ai> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Boris Sekachev <boris.sekachev@yandex.ru> Co-authored-by: Maxim Zhiltsov <zhiltsov.max35@gmail.com> Co-authored-by: Kirill Sizov <kirill.sizov@cvat.ai> Co-authored-by: Nikita Manovich <nikita@cvat.ai> Co-authored-by: Mariia Acoca <39969264+mdacoca@users.noreply.github.com> Co-authored-by: Kirill Lakhov <kirill.9992@gmail.com> Co-authored-by: Michael Kirpichev <mkirpic+github@gmail.com> Co-authored-by: Michael Kirpichev <m.kirpichev@haut.ai> Co-authored-by: Boris Sekachev <boris@cvat.ai>
PMazarovich
pushed a commit
to PMazarovich/cvat
that referenced
this pull request
Aug 15, 2023
## \[2.5.2\] - 2023-07-27 ### Added - We've added support for multi-line text attributes (<cvat-ai#6458>) - You can now set a default attribute value for SELECT, RADIO types on UI (<cvat-ai#6474>) - \[SDK\] `cvat_sdk.datasets`, is now available, providing a framework-agnostic alternative to `cvat_sdk.pytorch` (<cvat-ai#6428>) - We've introduced analytics for Jobs, Tasks, and Project (<cvat-ai#6371>) ### Changed - \[Helm\] In Helm, we've added a configurable default storage option to the chart (<cvat-ai#6137>) ### Removed - \[Helm\] In Helm, we've eliminated the obligatory use of hardcoded traefik ingress (<cvat-ai#6137>) ### Fixed - Fixed an issue with calculating the number of objects on the annotation view when frames are deleted (<cvat-ai#6493>) - \[SDK\] In SDK, we've fixed the issue with creating attributes with blank default values (<cvat-ai#6454>) - \[SDK\] We've corrected a problem in SDK where it was altering input data in models (<cvat-ai#6455>) - Fixed exporting of hash for shapes and tags in a specific corner case (<cvat-ai#6517>) - Resolved the issue where 3D jobs couldn't be opened in validation mode (<cvat-ai#6507>) - Fixed SAM plugin (403 code for workers in organizations) (<cvat-ai#6514>) - Fixed the issue where initial frame from query parameter was not opening specific frame in a job (<cvat-ai#6506>) - Corrected the issue with the removal of the first keyframe (<cvat-ai#6494>) - Fixed the display of project previews on small screens and updated stylelint & rules (<cvat-ai#6551>) - Implemented server-side validation for attribute specifications (<cvat-ai#6447>) - \[API\] Fixed API issue related to file downloading failures for filenames with special characters (<cvat-ai#6492>) - \[Helm\] In Helm, we've resolved an issue with multiple caches in the same RWX volume, which was preventing db migration from starting (<cvat-ai#6137>) --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Anastasia Yasakova <yasakova.an@gmail.com> Co-authored-by: yasakova-anastasia <anastasia@cvat.ai> Co-authored-by: Roman Donchenko <roman@cvat.ai> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Boris Sekachev <boris.sekachev@yandex.ru> Co-authored-by: Maxim Zhiltsov <zhiltsov.max35@gmail.com> Co-authored-by: Kirill Sizov <kirill.sizov@cvat.ai> Co-authored-by: Nikita Manovich <nikita@cvat.ai> Co-authored-by: Mariia Acoca <39969264+mdacoca@users.noreply.github.com> Co-authored-by: Kirill Lakhov <kirill.9992@gmail.com> Co-authored-by: Michael Kirpichev <mkirpic+github@gmail.com> Co-authored-by: Michael Kirpichev <m.kirpichev@haut.ai> Co-authored-by: Boris Sekachev <boris@cvat.ai>
mikhail-treskin
pushed a commit
to retailnext/cvat
that referenced
this pull request
Oct 25, 2023
…torch (cvat-ai#6428) The new `TaskDataset` class provides conveniences like per-frame annotations, bulk data downloading, and caching without forcing a dependency on PyTorch (and somewhat awkwardly conforming to the PyTorch dataset interface). It also provides a few extra niceties, like easy access to labels and original frame numbers. Note that it's called `TaskDataset` rather than `TaskVisionDataset`, as my plan is to keep it domain-agnostic. The `MediaElement` class is extensible, and we can add, for example, support for point clouds, by adding another `load_*` method. There is currently no `ProjectDataset` equivalent, although one could (and probably should) be added later. If we add one, we should probably also add a `task_id` field to `Sample`.
mikhail-treskin
pushed a commit
to retailnext/cvat
that referenced
this pull request
Oct 25, 2023
- We've added support for multi-line text attributes (<cvat-ai#6458>) - You can now set a default attribute value for SELECT, RADIO types on UI (<cvat-ai#6474>) - \[SDK\] `cvat_sdk.datasets`, is now available, providing a framework-agnostic alternative to `cvat_sdk.pytorch` (<cvat-ai#6428>) - We've introduced analytics for Jobs, Tasks, and Project (<cvat-ai#6371>) - \[Helm\] In Helm, we've added a configurable default storage option to the chart (<cvat-ai#6137>) - \[Helm\] In Helm, we've eliminated the obligatory use of hardcoded traefik ingress (<cvat-ai#6137>) - Fixed an issue with calculating the number of objects on the annotation view when frames are deleted (<cvat-ai#6493>) - \[SDK\] In SDK, we've fixed the issue with creating attributes with blank default values (<cvat-ai#6454>) - \[SDK\] We've corrected a problem in SDK where it was altering input data in models (<cvat-ai#6455>) - Fixed exporting of hash for shapes and tags in a specific corner case (<cvat-ai#6517>) - Resolved the issue where 3D jobs couldn't be opened in validation mode (<cvat-ai#6507>) - Fixed SAM plugin (403 code for workers in organizations) (<cvat-ai#6514>) - Fixed the issue where initial frame from query parameter was not opening specific frame in a job (<cvat-ai#6506>) - Corrected the issue with the removal of the first keyframe (<cvat-ai#6494>) - Fixed the display of project previews on small screens and updated stylelint & rules (<cvat-ai#6551>) - Implemented server-side validation for attribute specifications (<cvat-ai#6447>) - \[API\] Fixed API issue related to file downloading failures for filenames with special characters (<cvat-ai#6492>) - \[Helm\] In Helm, we've resolved an issue with multiple caches in the same RWX volume, which was preventing db migration from starting (<cvat-ai#6137>) --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Anastasia Yasakova <yasakova.an@gmail.com> Co-authored-by: yasakova-anastasia <anastasia@cvat.ai> Co-authored-by: Roman Donchenko <roman@cvat.ai> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Boris Sekachev <boris.sekachev@yandex.ru> Co-authored-by: Maxim Zhiltsov <zhiltsov.max35@gmail.com> Co-authored-by: Kirill Sizov <kirill.sizov@cvat.ai> Co-authored-by: Nikita Manovich <nikita@cvat.ai> Co-authored-by: Mariia Acoca <39969264+mdacoca@users.noreply.github.com> Co-authored-by: Kirill Lakhov <kirill.9992@gmail.com> Co-authored-by: Michael Kirpichev <mkirpic+github@gmail.com> Co-authored-by: Michael Kirpichev <m.kirpichev@haut.ai> Co-authored-by: Boris Sekachev <boris@cvat.ai>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation and context
The new
TaskDataset
class provides conveniences like per-frame annotations, bulk data downloading, and caching without forcing a dependency on PyTorch (and somewhat awkwardly conforming to the PyTorch dataset interface). It also provides a few extra niceties, like easy access to labels and original frame numbers.Note that it's called
TaskDataset
rather thanTaskVisionDataset
, as my plan is to keep it domain-agnostic. TheMediaElement
class is extensible, and we can add, for example, support for point clouds, by adding anotherload_*
method.There is currently no
ProjectDataset
equivalent, although one could (and probably should) be added later. If we add one, we should probably also add atask_id
field toSample
.How has this been tested?
SDK unit tests.
Checklist
develop
branch[ ] I have linked related issues (see GitHub docs)[ ] I have increased versions of npm packages if it is necessary(cvat-canvas,
cvat-core,
cvat-data and
cvat-ui)
License
Feel free to contact the maintainers if that's a concern.