Hello When computing the Hausdorff Distance for multi-class semantic segmentation, if a class X is not detected, some implementations opt to set the Hausdorff Distance to infinity. As a modification, we say that if the class is not detected, set the hausdorff distance for that undetected class to be 1000. Otherwise, we end up averaging Hausdorff Distances with a mix of infinity and non-infinity numbers (which results in infinity average Hausdorff Distance). Just out of curiosity, did anyone else do something similar to this for their average Hausdorff Distance calculation across all classes? Thanks so much for looking into this, and please feel free to ask questions if you have any. Best, Jonathan