Adeel Hassan

I like that idea. Here is an attempt: https://github.com/azavea/raster-vision/pull/1506/commits/0e31373f5b5d9b7b21325466d557c42eb2c61261 Result: https://raster-vision--1506.org.readthedocs.build/en/1506/tutorials/reading_labels.html

Interestingly, the built-in sphinx search (which is used in local builds and PR builds) [lists the correct result as the top result](https://raster-vision--1501.org.readthedocs.build/en/1501/search.html?q=max_windows&check_keywords=yes&area=default). But the actual deployed RTD build [uses ElasticSearch](https://docs.readthedocs.io/en/stable/server-side-search.html),...

Some graphics that might be useful to include in the docs:   

@dependabot rebase

A solution might be to declare these as [optional dependencies](https://setuptools.pypa.io/en/latest/userguide/dependency_management.html#optional-dependencies) and add some exception handling around their imports.

Thanks for the suggestion. This makes sense to me. I suppose we can just drop the row and column of the ignored class from the confusion matrix before computing the...

It might take me some days to get around to this though.

@dependabot rebase

`SemanticSegmentationLabels` does not provide a mechanism for subsetting the extent. If you want to do that, you will need something like the following: ```py import numpy as np from rastervision.core.data.label...

Closing this now. Feel free to reopen if there are more questions.