masksvalues
Masksvalues is a term used in data processing and computer vision to denote the numeric values associated with masks, which are binary indicators selecting subsets of data such as pixels, voxels, or data samples. The concept combines a mask, which specifies a region of interest, with a value or set of values that quantify the importance, confidence, or weight of that region.
In practice, a masksvalues object is typically represented as an array or tensor aligned with the domain
Binaries vs probabilistic: binary masks designate coverage simply; masksvalues allow soft influence. They are used in
Generation and normalization: masksvalues may be assigned manually, computed from model predictions (as confidence scores or
Examples include applying a spatially varying blur where masksvalues weight the blur strength, or using a predicted
Related concepts include masks, attention maps, and soft masks in segmentation. The term is primarily used in