* [x] Rework of layer API. Geometry inputs are now tensors instead of attrib. * [x] Tensorflow 2.0 support * [x] Eager execution * [x] Build process is now based on tensorflow custom-op docker iamge * [x] Batch dimension > 1 support
0.0.4
* [x] Added shape_inference::ShapeHandle; to allow Tensorflow to evaluate the shape of the propived operators * [x] Batch dimension is now handled correctly * [x] All Operators require the batch_size stored in the first dimension of the Tensor ! Batch dimensions is now mandatory * [x] Currently all operators are only supporting a batch_size of 1 ! A Batch_size >1 will lead to an Error. * [x] Preparation to Tensorflow 2.0 and Eager execution are included
0.0.3
* [x] Final license: Apache 2.0
0.0.2
* [x] Critical Bugfix: 3D Layers missing Attributes for gradient registration
0.0.1
* [x] Initial package setup: * [x] 2D parallel, fan and 3D cone beam projectors and back-projectors computing A and A^T on the fly * [x] Tensorflow patch for Tensorflow 1.9 up to Tensorflow 1.12 * [x] BUILD file to include custom kernels into the Tensorflow .whl package * [x] Add license