Locating specific label positions in truth and setting it in predicted within TF\Keras loss function?

Question

I am attempting to create a custom loss function for my semantic segmentation neural network. As part of this loss I wish to set the predicted value for a specific class to alwa