:recycle: Refactor to change tensors from channel last to channel first - Githubissues

developmentseed / chabud2023

Change detection for Burned area Delineation (ChaBuD) ECML/PKDD 2023 challenge

Other

5 stars 1 forks source link

:recycle: Refactor to change tensors from channel last to channel first #6

Closed weiji14 closed 1 year ago

weiji14 commented 1 year ago

What I am changing

Reordering the chip dimensions to be (12, 512, 512) instead of (512, 512, 12).

How I did it

Modified the _datatree_to_chip function while the tensor is still in an xarray.Dataset format.
Note that only the pre_fire and post_fire tensors are reshaped to (12, 512, 512). The mask tensor remains (512, 512).

How you can test it

Run python -m pytest chabud/tests/ to run the unit tests that capture this change.

Related Issues

Also added a README.md to the chabud/ folder describing the contents.

Patches #4