This PR aims at addressing #112 (closed).
To do:
-
Support end-to-end reconstruction, from radios read to final NXResult -
Logging -
Numpy backend -
Cuda backend -
Support class that can run the same process on a different chunk -
Support binning -
Support datasets where delta_z * n_x * n_angles > memory -
Update documentation and wiki, delete Component. -
Update CLI scripts
Will be done in another PR:
- Dump memory (ex. sinograms)
- Re-run pipeline from memory dump / intermediate step