Evan B. Goldstein

Hi @Rocketknight1 I misremembered- the error is not after `model.compile()` - compiling a model with a different loss function, added metrics, a custom loss, or custom metrics all compile w/...

Hi @Rocketknight1 , sorry for the delay. Attached below is some code and example image & label pairs all zipped up. Let me know if you prefer another format/delivery mechanism...

thx for that code highlight @Rocketknight1 , super helpful and i understand it now -- i would need a similar upsampling routine. related Q to finding the output resolution -...

Awesome, thanks so much for all the helpful info @Rocketknight1 & @sayakpaul . I can close this issue now as i understand the landscape much better and it seems the...

thanks for pointing this out @cresposito !

correct, we just need to write a custom loss that upsamples the logits to the mask size

the relevant warning message is: > Some layers from the model checkpoint at nvidia/mit-b0 were not used when initializing TFSegformerForSemanticSegmentation: ['classifier'] > - This IS expected if you are initializing...

@dbuscombe-usgs, 2 Qs for you: 1) Have you used a segformer model when running `seg_images_in_folder` ? 2) How have you done prediction with a segformer model?

this might be related to model saving.. see this link -> github. com/ huggingface/ transformers/issues/18582 (with spaces added to avoid a hard link to that issue)

**cautionary tale:** :skull_and_crossbones: :skull_and_crossbones: :skull_and_crossbones: :skull_and_crossbones: ok, to rehash the problem i was having: - Seg_images_in_folder did not perform well with segformer models in some unknown situations... the images were...