Releases: keras-team/keras-cv
Releases · keras-team/keras-cv
v0.2.3
v0.2.2
v0.2.1
v0.2.0
v0.1.1
What's Changed
- Fix COCOMeanAveragePrecision ragged support by @LukeWood in #332
- Add casts to compute_dtype in coco metrics by @LukeWood in #335
- Add FMix Preprocessing Layer by @quantumalaviya in #314
- Update roadmap and readme by @LukeWood in #329
- add Color Jitter preprocessing layers by @innat in #158
- Fix link in ROADMAP.md by @zhiqwang in #340
- Gaussian final by @artu1999 in #339
- Fixed dropblock 2d serialization issue. by @sebastian-sz in #346
- fixed wrong parameter issue to RandomShear class by @pranavjadhav001 in #354
- Add version_check to KerasCV by @LukeWood in #350
- benchmarks for num_cutouts by @parikshit14 in #186
- Add new unit tests for ground truth counting by @LukeWood in #362
- Expose BaseImageAugmentationLayer as a user facing API. by @LukeWood in #363
- Move serialization_test.py to central location. by @sebastian-sz in #365
- Feature templates by @LukeWood in #390
- minor doc string change in hue KPL by @kartik4949 in #392
- Update dropblock test by @LukeWood in #397
- Move LICENSE into root path by @zhiqwang in #394
- Fix tests for Dropblock2d. by @sebastian-sz in #399
- standard_deviation->stddev by @LukeWood in #415
- Remove tf-dep by @LukeWood in #417
- Update random augmentation pipeline to work in graph mode. by @LukeWood in #421
- Workaround tf.function issue by @LukeWood in #428
- Add unit tests to get_random_policy() behavior by @LukeWood in #426
- Fix the wrong comment about Autograph bug by @bhack in #430
New Contributors
- @zhiqwang made their first contribution in #340
- @artu1999 made their first contribution in #339
- @pranavjadhav001 made their first contribution in #354
- @parikshit14 made their first contribution in #186
- @kartik4949 made their first contribution in #392
Full Changelog: v0.1.0...v0.1.1
Contributors
bhack, zhiqwang, and 8 other contributors
Assets 2
v0.1.0
This is KerasCV's v0.1.0 release!
This is the first official release of KerasCV. Please note there is no backwards compatibility contract until KerasCV reaches v1.0.0.
In this release KerasCV introduces:
- 19 new image data augmentation layers
- a Keras native RandAugment implementation
- fully in graph COCO metrics for bounding box detection!
Keep an eye out on keras.io in the upcoming weeks for our new API documentation and guides.
Thank you to all of the wonderful contributors that have made this release possible!
New Contributors
- @qlzh727 made their first contribution in #10
- @LukeWood made their first contribution in #11
- @chjort made their first contribution in #46
- @sebastian-sz made their first contribution in #106
- @innat made their first contribution in #92
- @quantumalaviya made their first contribution in #178
- @winstoncooke made their first contribution in #253
- @bhack made their first contribution in #267
What's Changed
- Remove the existing content from Keras CV project by @qlzh727 in #10
- Create operational resources by @LukeWood in #11
- Initial implementation of COCORecall metric by @LukeWood in #13
- Update the README with some basic information by @LukeWood in #30
- Add MixUp & CutMix preprocessing layers by @LukeWood in #24
- Fix test name for MixUp layer by @chjort in #46
- Implement an unvectorized version of the COCORecall metric by @LukeWood in #45
- Update the bounding box utility to be more readable by @qlzh727 in #51
- Update the pad_bbox_batch_to_shape by @qlzh727 in #56
- Implementation of Cutout layer and RandomErasing layer by @chjort in #47
- Batch fill rectangle by @chjort in #65
- Initial commit for the keras-cv roadmap/community guide line by @qlzh727 in #61
- Removes label smoothing from CutMix and MixUp by @LukeWood in #70
- Add a contributing guide by @LukeWood in #67
- fill_utils tests by @chjort in #75
- Add CALL_FOR_CONTRIBUTIONS, refactor README by @LukeWood in #79
- refactor arguments of fill_utils.rectangle_masks by @chjort in #98
- Remove the
rateargument from all KPLs. by @LukeWood in #103 - add missing get_config methods to layers by @chjort in #107
- Prepare to land COCORecall, refactoring, RaggedTensor inputs, etc by @LukeWood in #99
- Update bbox_test.py by @LukeWood in #116
- Added Solarize preprocessing layer. by @sebastian-sz in #106
- Add API design guidelines for preprocessing layers by @LukeWood in #129
- add grid_mask preprocessing layers by @innat in #92
- Introduces an Equalize preprocessing layer by @LukeWood in #91
- Add ToGray preprocessing layers by @innat in #132
- Update gridmask and cutout arg linting by @LukeWood in #140
- Introduces the COCOMeanAveragePrecision metric by @LukeWood in #139
- %s/bbox/bounding_box by @LukeWood in #145
- Add benchmarking script, update API_DESIGN.md to reflect the results. by @LukeWood in #161
- Support all 1.0 sample weights in COCO metrics by @LukeWood in #168
- Add
value_rangeutilities. by @LukeWood in #144 - Introduce DropBlock2D regularization layer. by @sebastian-sz in #166
- Add performance benchmarking scripts for COCO Metrics by @LukeWood in #147
- Adding num_cutouts to RandomCutout by @quantumalaviya in #178
- Create RandomShear layer using the BaseImageAugmentationLayer by @LukeWood in #180
- Update the preprocessing init to include all the core keras image KPLs by @qlzh727 in #185
- Add ChannelShuffle preprocessing layers by @innat in #124
- Implement check to skip value range transform by @LukeWood in #198
- Creates an autocontrast KPL by @LukeWood in #194
- Introduces Posterization preprocessing layer. by @sebastian-sz in #136
- Implement the Sharpness KPL by @LukeWood in #197
- Fix overridden method in RandomShear and RandomSharpness by @chjort in #204
- Implement ColorDegeneration KPL by @LukeWood in #205
- Disable loop unrolling for COCOMaP by @LukeWood in #216
- Update solarization in preparation for use in RandAugment by @LukeWood in #213
- Migrate Grayscale to use BaseImageAugmentationLayer by @quantumalaviya in #215
- Update image KPL for the latest API update from BaseImageAugmentLayer by @qlzh727 in #219
- Add RandomSaturation KPL by @qlzh727 in #221
- Add RandomHue KPL by @qlzh727 in #231
- Register all the keras cv layers with keras serializable. by @qlzh727 in #235
- Update Equalization to work with BaseImageAugmentationLayer, support value_range by @LukeWood in #234
- Random cutouts to baseimgaug by @LukeWood in #236
- Fix typo in CONTRIBUTING.md by @winstoncooke in #253
- Migrate channel_shuffle to use BaseImageAugmentationLayer by @quantumalaviya in #218
- GridMask implementation with BaseImageAugmentationLayer by @chjort in #159
- Update all KPLs to make value_range required by @LukeWood in #249
- Update cutmix and mixup call methods by @LukeWood in #263
- Fix minor coco utility function by @LukeWood in #246
- Devcontainer by @bhack in #267
- Vectorize Equalization by @quantumalaviya in #201
- Implement an API to support custom Factor sampling by @LukeWood in #238
- Update CutMix to use the BaseImageAugmentationLayer as base. by @qlzh727 in #271
- Update MixUp to use keras BaseImageAugmentationLayer by @qlzh727 in #273
- Fixes formatting for outputs of cut mix and mix up by @LukeWood in #274
- Update RandomHue for v1.0 by @LukeWood in #269
- Update solarization for v1 by @LukeWood in #275
- Add "seed" param to all the random augmentation layers by @qlzh727 in #276
- second pass on seed by @LukeWood in #280
- Creates the RandomAugmentationPipeline Layer by @LukeWood in #281
- Update gridmask for v1 by @LukeWood in #287
- Bounding box support in MixUp augmentation by @quantumalaviya in #282
- Refactor with_labels_test by @LukeWood in #240
- add RGBShift preprocessing layers by @innat in #128
- Remove layers that don't belong in RandAugment from RandAugment by @LukeWood in #294
Full Changelog: https://github.com/keras-team/keras-cv/commits/v0.1.0
Contributors
bhack, qlzh727, and 6 other contributors