Now, we are using PNG sequences as default file format. However, I believe the HDF5 format would be better on the long term as this allows chunked reads and writes (useful for block-based processing). This means:
[ ] Implement a default read_volume and write_volume function
[ ] Replace every image IO function in the plugin
[ ] Implement a read_hdf5 function in neuralnets that automatically performs chunking
Now, we are using PNG sequences as default file format. However, I believe the HDF5 format would be better on the long term as this allows chunked reads and writes (useful for block-based processing). This means:
read_volumeandwrite_volumefunctionread_hdf5function in neuralnets that automatically performs chunking