This makes the Slide Manifest (from slide_etl) the main data frame of record for a project. Additional columns are added for annotations, and the tiles_url column is modified when tiles are generated, tissue is detected, and tiles are saved in the h5 store. Dask is used at the slide and tile level with batches for parallelizing when possible.
Currently, the tile tissue inference doesn't work with Dask, as we need to use a Dask-ML compatible library, i.e., Skorch.
We also need some additional unit tests for the API functions.
This makes the Slide Manifest (from slide_etl) the main data frame of record for a project. Additional columns are added for annotations, and the tiles_url column is modified when tiles are generated, tissue is detected, and tiles are saved in the h5 store. Dask is used at the slide and tile level with batches for parallelizing when possible.
Currently, the tile tissue inference doesn't work with Dask, as we need to use a Dask-ML compatible library, i.e., Skorch.
We also need some additional unit tests for the API functions.