Lewis Fishgold

Using chunks of shape (4000, 4000) isn't very different than what we were using originally (10_000, 10_000), so I'm not surprised the results are the same. After reading https://docs.dask.org/en/stable/array-chunks.html#reshaping I...

> Thanks for looking into this. I think defining a new compute environment and adding it to the existing job queue makes sense. I agree. It looks like the `ComputeEnvironmentOrder`...

Are you thinking that there will be different notebooks for the different checkbox points you listed? I was thinking we would have notebooks that covered whole workflows for specific datasets/tasks...

There are some datasets like this on MLHub that we could access through STAC such as https://mlhub.earth/data/microsoft_chesapeake

I'd like to show how you can take an off-the-shelf PyTorch model (perhaps from TorchGeo) and then use RV to make predictions on `Scene` objects. Is that something that is...

> Yes, but it's tied up in the `Learner` as well. You need to create a `Learner` with `Learner(model=your_model, training=False)` and then do basically this: https://github.com/raster-foundry/raster-foundry/blob/develop/app-hitl/hitl/src/hitl/rv/predict.py Hmm, ok. I'll think...

I was just thinking about doing something similar. I just learned about this library, but it sounds like a good approach to me.

Other ways of speeding this up: * Read in whole image to RAM instead of doing windowed reads off of disk? * Test if the post-processing step in object detection...

That's script was just used to generate some data that is used in the integration test, but isn't actually run as part of the test itself. But you're right that...

I agree that it seems like a good idea to upgrade Ubuntu and the native dependencies, and remove the unused dependencies. I think it should be straightforward to remove or...