In addition to the changes listed below, the codebase structure has evolved:
- All source files are now in keras_cv/src/.
- All API files are now in keras_cv/api/.
- Users can install KerasCV directly from GitHub or local git repository.
This doesn't change how users will interact with KerasCV and doesn't update the public API.
What's Changed
- Version bump master to 0.8.2 by @sampathweb in #2290
- Move keras-cv markdown files to toplevel by @mattdangerw in #2291
- Add
version()
API to unify with Keras and KerasNLP by @grasskin in #2199 - Update random_crop_and_zoom.py by @sachinprasadhs in #2294
- Bug fix for Stable Diffusion model timesteps by @haifeng-jin in #2303
- Add BASNet Segmentation Model (#2006) by @hamidriasat in #2271
- Fix image.shape type by @sachinprasadhs in #2305
- Create workflow for auto assignment of issues and for stale issues by @sachinprasadhs in #2313
- Fix format and Update Vectorized Base to fix tf.data with non-TF backends by @sampathweb in #2323
- Fix CI Test for Basnet OOM and PyCoCo Test Failure for JAX by @sampathweb in #2322
- Fix TensorFlow-GPU flaky failure by @tirthasheshpatel in #2325
- Remove Bazel steps from Contributing Guide by @sampathweb in #2324
- Update version to 0.8.3 by @sampathweb in #2327
- Bump the github-actions group with 3 updates by @dependabot in #2330
- Pin Jax Version in GPU CI by @sampathweb in #2338
- Bump the python group with 5 updates by @dependabot in #2340
- Update auto-assignment.js by @sachinprasadhs in #2337
- Revert back to JAX 0.4.23 by @sampathweb in #2342
- move conftest.py to top level by @sampathweb in #2345
- Fix format of Custom Ops build by @sampathweb in #2346
- Fix Custom Ops Build by @sampathweb in #2347
- Add Waymo folder License to the top by @sampathweb in #2348
- update ops build by @sampathweb in #2349
- Bump the github-actions group with 2 updates by @dependabot in #2343
- Add CLIP to KerasCV by @divyashreepathihalli in #2331
- Migrate VGG16 from legacy to backbone by @pranavvp16 in #2341
- CLIP : Enable CLIP.from_preset() Kaggle tests by @divyashreepathihalli in #2357
- Adds psutil to CI Build Script by @sampathweb in #2355
- Update clip_image_model.py by @divyashreepathihalli in #2359
- Fixed ROI Pooling Output Shape to Consider Multiple ROIs (#2350) by @VarunS1997 in #2360
- Add training scripts for yolov8 and deeplabv3plus by @divyashreepathihalli in #2361
- [Port] OD layers to Keras 3 by @ariG23498 in #2295
- bug fix for num_steps=1 by @haifeng-jin in #2373
- Clip update inputs format by @divyashreepathihalli in #2377
- Fix rendering issue. by @sachinprasadhs in #2375
- CI: psutil fix: Remove install of Python 3.9 by @sampathweb in #2392
- Fix docstring of pyramid keys in
YOLOV8Detector
by @smitlg in #2391 - Export Model APIs by @sachinprasadhs in #2389
- Fix nlp import by @divyashreepathihalli in #2396
- Update CI to use TF 2.16 by @sampathweb in #2378
- Standardize docstring by @sachinprasadhs in #2399
- Update CI to skip deps install from keras_nlp by @sampathweb in #2405
- Add Video Swin Transformer by @innat in #2369
- Update CLIP to a functional model by @divyashreepathihalli in #2393
- Fix import of Keras by @sampathweb in #2420
- Generated API by @sampathweb in #2425
- Add keras-nlp to nightly action by @sampathweb in #2426
- Raise error if tensorflow_text is not found by @sampathweb in #2427
- Export video swin APIs by @sachinprasadhs in #2428
- ver-0.9.0.dev0 by @sampathweb in #2429
- Revert "ver-0.9.0.dev0" by @sampathweb in #2431
New Contributors
- @grasskin made their first contribution in #2199
- @sachinprasadhs made their first contribution in #2294
- @hamidriasat made their first contribution in #2271
- @pranavvp16 made their first contribution in #2341
- @VarunS1997 made their first contribution in #2360
- @smitlg made their first contribution in #2391
Full Changelog: v0.8.2...v0.9.0