Working with images is interesting. In a previous piece of work, a customer wanted to bring in some semantic segmentation. Semantic Segmentation goes pixel by pixel and assigns probabilities that the pixel belongs to label. You’ll typically see this in street images, with individual cars, people, bikes, etc being segmented out of the overall picture on a pixel level.
This is a companion discussion topic for the original entry at https://benhoff.net/command-line-visual-debugging.html