How to access class labels and probabilities in Semantic Segmentation Inference

I followed the SageMaker semantic segmentation example on GitHub, https://github.com/awslabs/amazon-sagemaker-examples/blob/master/introduction_to_amazon_algorithms/semantic_segmentation_pascalvoc/semantic_segmentation_pascalvoc.ipynb. However, in both inference cases the image mask is being plotted. I do not want the image masks. I need to access class labels and probabilities and show my application users classes and probabilities.

Currently, this is not explained in the tutorial. Any recommendations on how this might be solved?

Start with the linked semantic_segmentation_pascalvoc.ipynb and inspect both inference cases, focusing on the prediction data currently used for mask plotting. Confirm how class labels and probabilities can be accessed, then update the tutorial so users can display them and verify the documented behavior in both cases.

Written by the indexing model from the issue text.