Feature groups organize image measurements into qualitative families based on human icon perception. This view is intentionally concise: it shows the study-level family and the features grouped under it.
Browse the same seven representative features used by Feature Groups and compare icons with low, medium, and high values.
Inspect image-embedding clusters for the same seven-family composite sample. Measured-feature contributions describe which visual features best separate the AI clusters; they are explanations after clustering, not inputs to the AI model.