IndexError: Target 255 is out of bounds Lesion Segmentation

I am trying to perform segmentation using fast version 2.6.3 and Kept on getting the cuda runtime error so I switched to cpu and was able to see the error. I have no idea how to convert the mask image to be in the proper range (0 …N). Also realised that there open_mask function is no longer available in the current fastai version.

Any help??


Dont do this. Ask the question, do not @ moderators or multiple people unless for good reason. I understand you are frustrated but this is not polite and is frustrating on us.

1 Like

Sorry for spamming mate, am a new user trying to use fast for my project and it seems like the only thing am doing is debugging all the time and its frustrating.

1 Like

Welcome to SE :wink:

Fastai does have a learning curve (or in this case its really pytorch)

Read through this tutorial i wrote, it discusses preparing your masks Binary Segmentation | walkwithfastai



1 Like