Currently, functions from dea_tools.classification access spatial information about datacube-loaded data using the .geobox extension, and geometry tools from datacube.utils.geometry.
To future proof our classification/ML tools, we should update our functions to use replacement functionality from odc.geo. For example, accessing spatial information like:
ds.odc.geobox
And geometry tools like:
from odc.geo.geom import Geometry
from odc.geo.xr import assign_crs
# or simply
ds.odc.assign_crs()
Currently, functions from
dea_tools.classification
access spatial information about datacube-loaded data using the.geobox
extension, and geometry tools fromdatacube.utils.geometry
.In ODC version 1.9 and above, these tools will be deprecated and replaced with
odc.geo
(see https://github.com/opendatacube/datacube-core/pull/1424).To future proof our classification/ML tools, we should update our functions to use replacement functionality from
odc.geo
. For example, accessing spatial information like:And geometry tools like:
https://odc-geo.readthedocs.io/en/latest/api.html
Related issue: https://github.com/GeoscienceAustralia/dea-notebooks/issues/1038