Changes * Add support for Keras 3 in 79 * TensorFlow 2.16+ has Keras 3.x as a dependency (although it is possible to use legacy Keras 2.x if need-be) * Remove deprecated/unsupported usage of numpy (also thanks to bwmr in 77) * Add support for Python 3.12 * Miscellaneous smaller bug fixes and improvements
**Breaking changes** - Exporting to TensorFlow's "SavedModel" format is no longer supported when using Keras 3
Maintenance release with various smaller changes: - Fix issue when using TensorFlow 2.13.0 - Support Python 3.11 - Re-activate `h5py` version pin when using TensorFlow 1 (i.e. installation via `pip install "csbdeep[tf1]"`) - Update isotropic reconstruction notebook - Fix smaller issues in example notebooks
0.7.3
Maintenance release to remove pin of `h5py` version (see [CSBDeep 0.6.2](https://github.com/CSBDeep/CSBDeep/releases/tag/0.6.2)). (There still *might* be h5py-related issues when using TensorFlow 2.4 or earlier, which is not recommended.)
0.7.2
Bugfix release pinning `protobuf < 3.21` to be compatible with TensorFlow 1.x. (Only if CSBDeep is installed via `pip install "csbdeep[tf1]"`.)
Bugfix release to be compatible with old versions of Keras (cf. [failed legacy tests](https://github.com/CSBDeep/CSBDeep/actions/runs/2360006291)).
0.7.0
Minor changes, but now only supporting Python 3.6 or later (i.e. dropping Python 2.7 and 3.5). - Fix several deprecated uses of other packages (distutils, scipy, numpy, tifffile, keras) - Other minor changes