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

Creates the RandomAugmentationPipeline Layer #281

Merged
merged 111 commits into from
Apr 8, 2022
Merged

Creates the RandomAugmentationPipeline Layer #281

merged 111 commits into from
Apr 8, 2022

Conversation

LukeWood
Copy link
Contributor

@LukeWood LukeWood commented Apr 8, 2022

Example from the demo:

Figure_1

@LukeWood
Autocontrast label support
f612638
@LukeWood
Update with_labels test
9ec83cf
@LukeWood
Update with_labels test
ba483c7
@LukeWood
Fix no-op label augs
d6b1f05
@LukeWood
reformat
a1cf315
@LukeWood
Add CutMix to with_labels_test
c28fa76
@LukeWood
Begin migrating to Factor pattern
905beee
@LukeWood
Update layers and tests to pass
d3d100c
@LukeWood
reformat
0f76a17
@LukeWood
Update serialization_test to pass
d5c6445
@LukeWood
Update factor to pass lint
ca8dc24
@LukeWood
Rename Factor => FactorSampler
cd90ad1
@LukeWood
Update docstrings
25ef717
@LukeWood
Update factor -> factor_sampler
97f50fa 
Replace sample() -> __call__()
@LukeWood
add self
d45af5a
@LukeWood
Update docs to suggest keras_cv.FactorSampler over keras_cv.core
159afb8
@LukeWood
Update docstrings to not recommend keras_cv.core
f602ebe
@LukeWood
Update to use stateless_random
66f9839
@LukeWood
Revert stateless random change
202c01c
@LukeWood
Update serialization_test to pass
7488de9
@LukeWood
Update factor to pass lint
bf3cf66
@LukeWood
Add preprocessing util
6e89c0f
@LukeWood
implement random color degeneration layer
15e5227
@LukeWood
format
0292857
@LukeWood
license header + shell lint check for it
a3a991c
@LukeWood
Docstring update
95d45b2
@LukeWood
Add test based on luma value
9ff2304
@LukeWood
much better color degeneration tests
3a9b855
@LukeWood
run isort
65676af
@LukeWood
Update layers.py
962fdaf
@LukeWood LukeWood mentioned this pull request Apr 8, 2022
Closed
@LukeWood
Copy link
Contributor Author

LukeWood commented Apr 8, 2022

RandomPolicy

RandomPolicy is nice. RandomlySampledPolicy maybe?

@LukeWood
Copy link
Contributor Author

LukeWood commented Apr 8, 2022

CC @mattdangerw because he may have some good naming advice

qlzh727
qlzh727 requested changes Apr 8, 2022
View reviewed changes
Copy link
Member

@qlzh727 qlzh727 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the PR.

keras_cv/layers/preprocessing/rand_augment.py Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Show resolved Hide resolved
keras_cv/layers/preprocessing/random_augmentation_pipeline.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/random_augmentation_pipeline.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/random_augmentation_pipeline.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/random_augmentation_pipeline.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/random_augmentation_pipeline.py Outdated Show resolved Hide resolved
fchollet
fchollet reviewed Apr 8, 2022
View reviewed changes
Copy link
Member

@fchollet fchollet left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the update -- it looks very nice now!

keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Outdated Show resolved Hide resolved
keras_cv/layers/preprocessing/rand_augment.py Show resolved Hide resolved
fchollet
fchollet approved these changes Apr 8, 2022
View reviewed changes
Copy link
Member

@fchollet fchollet left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, thanks!

@LukeWood LukeWood merged commit ca01ef1 into master Apr 8, 2022
@LukeWood LukeWood deleted the pipeline branch April 14, 2022 18:51
kartik4949 pushed a commit to kartik4949/keras-cv that referenced this pull request Apr 21, 2022
@LukeWood @kartik4949
Creates the RandomAugmentationPipeline Layer (keras-team#281)
52ac60e 
* Autocontrast label support

* Update with_labels test

* Update with_labels test

* Fix no-op label augs

* reformat

* Add CutMix to with_labels_test

* Begin migrating to Factor pattern

* Update layers and tests to pass

* reformat

* Update serialization_test to pass

* Update factor to pass lint

* Rename `Factor` => `FactorSampler`

* Update docstrings

* Update factor -> factor_sampler

Replace sample() -> __call__()

* add self

* Update docs to suggest keras_cv.FactorSampler over keras_cv.core

* Update docstrings to not recommend keras_cv.core

* Update to use stateless_random

* Revert stateless random change

* Update serialization_test to pass

* Update factor to pass lint

* Add preprocessing util

* implement random color degeneration layer

* format

* license header + shell lint check for it

* Docstring update

* Add test based on luma value

* much better color degeneration tests

* run isort

* Update layers.py

* fix isort params

* remove randaugment

* begin randaugment

* begin working on randaugment layer

* holding commit

* Update randaugment

* Update rand_augment class

* Udpate example

* rebase randaugment

* Implement `NormalFactor`.

* Add some unit tests for RandAugment

* Update docstring, update TODOs

* Add label test

* Run format

* Fix lint issues

* Support value range

* Fix uint8/int32 test cases

* Add test cases for value range

* re-tune cutout policy

* Add seed parameter

* Factor sampler rename

* Update to require value ranges, use new factor api

* Update compare function

* fix serialization_test

* Fix broken rebase

* Add shape and dtype arg

* Revert spurious change

* Revert spurious changes

* Remove inaccurate docstring

* Update sample() calls

* Revert spurious preprocessing change

* Revert spurious preprocessing change

* Update rand_augment docstring

* RandAugment failing due to sub-layers not handling labels

* Rebased on top of add no-op branch

* make randaugment subclass BaseImageAugmentationLayer

* Reformat

* Support kwargs

* remove unused import

* Add extra magnitude check

* Reformat

* tf.range bug?

* Disable auto vectorize

* Update RandAugment policies to support solarize sampling

* Fix lint faulure

* Update test case

* Update rand augment demo and mildly tweak policy

* Update RandAugment per feedback

* update serialization test

* Update stdev param name

* Update seed

* Update augmentated_sample -> augmented_sample

* add rate to test coverage

* add @tf.keras.utils.register_keras_serializable(package="keras_cv")

* Add seed to constructor

* Remove augmentation pipeline

* Additional refactoring

* add extra newline

* Update to compile, update example

* create demo of pipeline

* Update to work with RandAugment

* Update serialization test to include pipeline

* Add custom layer to count invocations

* Test pipeline test

* format

* Update unit test

* What in the world is going on

* Update to make test failures even more clear

* Fix random_augmentation_pipeline

* add seed back to random hue - weird

* make unit tests pass

* fix random augmentation pipeline demo

* Update magnitude stddev

* Wrote docstring

* Fix typo

* Update serialization, address scott comments

* Francois comments

* add clipping information

* update docstring

* Update docstring

* Add f string
LukeWood added a commit to kartik4949/keras-cv that referenced this pull request Apr 29, 2022
@LukeWood
Creates the RandomAugmentationPipeline Layer (keras-team#281)
45842e4 
* Autocontrast label support

* Update with_labels test

* Update with_labels test

* Fix no-op label augs

* reformat

* Add CutMix to with_labels_test

* Begin migrating to Factor pattern

* Update layers and tests to pass

* reformat

* Update serialization_test to pass

* Update factor to pass lint

* Rename `Factor` => `FactorSampler`

* Update docstrings

* Update factor -> factor_sampler

Replace sample() -> __call__()

* add self

* Update docs to suggest keras_cv.FactorSampler over keras_cv.core

* Update docstrings to not recommend keras_cv.core

* Update to use stateless_random

* Revert stateless random change

* Update serialization_test to pass

* Update factor to pass lint

* Add preprocessing util

* implement random color degeneration layer

* format

* license header + shell lint check for it

* Docstring update

* Add test based on luma value

* much better color degeneration tests

* run isort

* Update layers.py

* fix isort params

* remove randaugment

* begin randaugment

* begin working on randaugment layer

* holding commit

* Update randaugment

* Update rand_augment class

* Udpate example

* rebase randaugment

* Implement `NormalFactor`.

* Add some unit tests for RandAugment

* Update docstring, update TODOs

* Add label test

* Run format

* Fix lint issues

* Support value range

* Fix uint8/int32 test cases

* Add test cases for value range

* re-tune cutout policy

* Add seed parameter

* Factor sampler rename

* Update to require value ranges, use new factor api

* Update compare function

* fix serialization_test

* Fix broken rebase

* Add shape and dtype arg

* Revert spurious change

* Revert spurious changes

* Remove inaccurate docstring

* Update sample() calls

* Revert spurious preprocessing change

* Revert spurious preprocessing change

* Update rand_augment docstring

* RandAugment failing due to sub-layers not handling labels

* Rebased on top of add no-op branch

* make randaugment subclass BaseImageAugmentationLayer

* Reformat

* Support kwargs

* remove unused import

* Add extra magnitude check

* Reformat

* tf.range bug?

* Disable auto vectorize

* Update RandAugment policies to support solarize sampling

* Fix lint faulure

* Update test case

* Update rand augment demo and mildly tweak policy

* Update RandAugment per feedback

* update serialization test

* Update stdev param name

* Update seed

* Update augmentated_sample -> augmented_sample

* add rate to test coverage

* add @tf.keras.utils.register_keras_serializable(package="keras_cv")

* Add seed to constructor

* Remove augmentation pipeline

* Additional refactoring

* add extra newline

* Update to compile, update example

* create demo of pipeline

* Update to work with RandAugment

* Update serialization test to include pipeline

* Add custom layer to count invocations

* Test pipeline test

* format

* Update unit test

* What in the world is going on

* Update to make test failures even more clear

* Fix random_augmentation_pipeline

* add seed back to random hue - weird

* make unit tests pass

* fix random augmentation pipeline demo

* Update magnitude stddev

* Wrote docstring

* Fix typo

* Update serialization, address scott comments

* Francois comments

* add clipping information

* update docstring

* Update docstring

* Add f string
ianstenbit pushed a commit to ianstenbit/keras-cv that referenced this pull request Aug 6, 2022
Creates the RandomAugmentationPipeline Layer (keras-team#281)
d563911 
* Autocontrast label support

* Update with_labels test

* Update with_labels test

* Fix no-op label augs

* reformat

* Add CutMix to with_labels_test

* Begin migrating to Factor pattern

* Update layers and tests to pass

* reformat

* Update serialization_test to pass

* Update factor to pass lint

* Rename `Factor` => `FactorSampler`

* Update docstrings

* Update factor -> factor_sampler

Replace sample() -> __call__()

* add self

* Update docs to suggest keras_cv.FactorSampler over keras_cv.core

* Update docstrings to not recommend keras_cv.core

* Update to use stateless_random

* Revert stateless random change

* Update serialization_test to pass

* Update factor to pass lint

* Add preprocessing util

* implement random color degeneration layer

* format

* license header + shell lint check for it

* Docstring update

* Add test based on luma value

* much better color degeneration tests

* run isort

* Update layers.py

* fix isort params

* remove randaugment

* begin randaugment

* begin working on randaugment layer

* holding commit

* Update randaugment

* Update rand_augment class

* Udpate example

* rebase randaugment

* Implement `NormalFactor`.

* Add some unit tests for RandAugment

* Update docstring, update TODOs

* Add label test

* Run format

* Fix lint issues

* Support value range

* Fix uint8/int32 test cases

* Add test cases for value range

* re-tune cutout policy

* Add seed parameter

* Factor sampler rename

* Update to require value ranges, use new factor api

* Update compare function

* fix serialization_test

* Fix broken rebase

* Add shape and dtype arg

* Revert spurious change

* Revert spurious changes

* Remove inaccurate docstring

* Update sample() calls

* Revert spurious preprocessing change

* Revert spurious preprocessing change

* Update rand_augment docstring

* RandAugment failing due to sub-layers not handling labels

* Rebased on top of add no-op branch

* make randaugment subclass BaseImageAugmentationLayer

* Reformat

* Support kwargs

* remove unused import

* Add extra magnitude check

* Reformat

* tf.range bug?

* Disable auto vectorize

* Update RandAugment policies to support solarize sampling

* Fix lint faulure

* Update test case

* Update rand augment demo and mildly tweak policy

* Update RandAugment per feedback

* update serialization test

* Update stdev param name

* Update seed

* Update augmentated_sample -> augmented_sample

* add rate to test coverage

* add @tf.keras.utils.register_keras_serializable(package="keras_cv")

* Add seed to constructor

* Remove augmentation pipeline

* Additional refactoring

* add extra newline

* Update to compile, update example

* create demo of pipeline

* Update to work with RandAugment

* Update serialization test to include pipeline

* Add custom layer to count invocations

* Test pipeline test

* format

* Update unit test

* What in the world is going on

* Update to make test failures even more clear

* Fix random_augmentation_pipeline

* add seed back to random hue - weird

* make unit tests pass

* fix random augmentation pipeline demo

* Update magnitude stddev

* Wrote docstring

* Fix typo

* Update serialization, address scott comments

* Francois comments

* add clipping information

* update docstring

* Update docstring

* Add f string
adhadse pushed a commit to adhadse/keras-cv that referenced this pull request Sep 17, 2022
@LukeWood
Creates the RandomAugmentationPipeline Layer (keras-team#281)
22e3e19 
* Autocontrast label support

* Update with_labels test

* Update with_labels test

* Fix no-op label augs

* reformat

* Add CutMix to with_labels_test

* Begin migrating to Factor pattern

* Update layers and tests to pass

* reformat

* Update serialization_test to pass

* Update factor to pass lint

* Rename `Factor` => `FactorSampler`

* Update docstrings

* Update factor -> factor_sampler

Replace sample() -> __call__()

* add self

* Update docs to suggest keras_cv.FactorSampler over keras_cv.core

* Update docstrings to not recommend keras_cv.core

* Update to use stateless_random

* Revert stateless random change

* Update serialization_test to pass

* Update factor to pass lint

* Add preprocessing util

* implement random color degeneration layer

* format

* license header + shell lint check for it

* Docstring update

* Add test based on luma value

* much better color degeneration tests

* run isort

* Update layers.py

* fix isort params

* remove randaugment

* begin randaugment

* begin working on randaugment layer

* holding commit

* Update randaugment

* Update rand_augment class

* Udpate example

* rebase randaugment

* Implement `NormalFactor`.

* Add some unit tests for RandAugment

* Update docstring, update TODOs

* Add label test

* Run format

* Fix lint issues

* Support value range

* Fix uint8/int32 test cases

* Add test cases for value range

* re-tune cutout policy

* Add seed parameter

* Factor sampler rename

* Update to require value ranges, use new factor api

* Update compare function

* fix serialization_test

* Fix broken rebase

* Add shape and dtype arg

* Revert spurious change

* Revert spurious changes

* Remove inaccurate docstring

* Update sample() calls

* Revert spurious preprocessing change

* Revert spurious preprocessing change

* Update rand_augment docstring

* RandAugment failing due to sub-layers not handling labels

* Rebased on top of add no-op branch

* make randaugment subclass BaseImageAugmentationLayer

* Reformat

* Support kwargs

* remove unused import

* Add extra magnitude check

* Reformat

* tf.range bug?

* Disable auto vectorize

* Update RandAugment policies to support solarize sampling

* Fix lint faulure

* Update test case

* Update rand augment demo and mildly tweak policy

* Update RandAugment per feedback

* update serialization test

* Update stdev param name

* Update seed

* Update augmentated_sample -> augmented_sample

* add rate to test coverage

* add @tf.keras.utils.register_keras_serializable(package="keras_cv")

* Add seed to constructor

* Remove augmentation pipeline

* Additional refactoring

* add extra newline

* Update to compile, update example

* create demo of pipeline

* Update to work with RandAugment

* Update serialization test to include pipeline

* Add custom layer to count invocations

* Test pipeline test

* format

* Update unit test

* What in the world is going on

* Update to make test failures even more clear

* Fix random_augmentation_pipeline

* add seed back to random hue - weird

* make unit tests pass

* fix random augmentation pipeline demo

* Update magnitude stddev

* Wrote docstring

* Fix typo

* Update serialization, address scott comments

* Francois comments

* add clipping information

* update docstring

* Update docstring

* Add f string
freedomtan pushed a commit to freedomtan/keras-cv that referenced this pull request Jul 20, 2023
@mattdangerw
Add layers.serialize and layers.deserialize (keras-team#281)
2327273
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@bhack bhack bhack left review comments

@fchollet fchollet fchollet approved these changes

@mattdangerw mattdangerw Awaiting requested review from mattdangerw

@qlzh727 qlzh727 Awaiting requested review from qlzh727

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@LukeWood @innat @bhack @fchollet @qlzh727