All images, masks, labels, etc. will need to have a standard shape and dtype to enable using the same pre-processing, and to make it easier to convert to the correct format for models at a later stage. A few core pre-processing functions should be provided in cw_nets.data.transform to enable this, which are used as an early step in any dataset object.
nrweir
commented
5 years ago
All images, masks, labels, etc. will need to have a standard shape and dtype to enable using the same pre-processing, and to make it easier to convert to the correct format for models at a later stage. A few core pre-processing functions should be provided in
cw_nets.data.transformto enable this, which are used as an early step in any dataset object.