Nick Weir

Hi @bruesba, Thanks for your well-documented comments, we'll have a look at why performance is poor with the pre-trained model and will clarify the documentation. Is that image from the...

Hi @bruesba, That makes sense re: the `DropChannel` error - XD_XD's model was trained on images that were originally RGB+near-IR with the near-IR band dropped to produce a 3-channel image....

Hi @bruesba, Agreed that looks a lot better. If you haven't swapped the blue and red channels in your image, you could try that and see if it helps even...

Hi @Elhamahm, Looks like there's a typo in our documentation indicating that the densenet model works for 3-channel imagery - it's designed for four channel inputs.

You're referring to the pixel masks that get compared to the prediction outputs, correct? The densenet161 model that you selected is designed to take a three-channel mask. When it was...

Sure! Check out [this tutorial](https://solaris.readthedocs.io/en/latest/tutorials/notebooks/api_masks_tutorial.html).

Hey @rbavery, It's definitely really slow at the moment. I think it's possible it's because of the size of the arrays that get generated - for example, if we're looking...

OK sounds good, happy to try this approach out then! Thanks for the suggestion. If we begin using dask here, I might try to re-write some of the CLI at...

Hi @rbavery, I think we might want to just silence these warnings that are passing from GDAL through rasterio and into solaris. When tiling for an ML model we're generally...

@jshermeyer I moved this over to the solaris repo, flagging you here so you know where it is.