You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the walnuts dataset dowloader works well, but the pre-processing dumps most data to only get a central slice, as it was written to do 2D tomo. However, the dataset is 3D, and thus discarding all this data is less than ideal, plus future people may want to do CBCT.
The walnuts dataset pre-procesing needs to change to accept all data. Similarly, it should not flat field correct nor log transform the data, just move it in a nice(er) folder structure into path_of_data/preprocessed/walnuts
Similarly, the data loader should accept inputs to flat field correct etc and should be flexible to request 2d or 3d data (either return some slices, or actually return the full 3D sino).
A lot of the code can be reused from the 2DeteCT dataset.
The text was updated successfully, but these errors were encountered:
Currently, the walnuts dataset dowloader works well, but the pre-processing dumps most data to only get a central slice, as it was written to do 2D tomo. However, the dataset is 3D, and thus discarding all this data is less than ideal, plus future people may want to do CBCT.
The walnuts dataset pre-procesing needs to change to accept all data. Similarly, it should not flat field correct nor log transform the data, just move it in a nice(er) folder structure into
path_of_data/preprocessed/walnuts
Similarly, the data loader should accept inputs to flat field correct etc and should be flexible to request 2d or 3d data (either return some slices, or actually return the full 3D sino).
A lot of the code can be reused from the
2DeteCT
dataset.The text was updated successfully, but these errors were encountered: