sheight

Sheight is a unitless metric used in visual analysis and computer graphics to represent the perceived height of an object relative to a reference height within a scene. It provides a dimensionless scale that helps compare objects across different camera viewpoints and resolutions.

Definition: The sheight of an object is defined as the ratio of its vertical extent to the vertical extent of the reference height in the same scene. A reference height is chosen by the analyst, often corresponding to a standard object such as a person of typical height. For example, if an object's image height is 300 pixels and the reference height is 600 pixels, its sheight is 0.5.

Calculation and use: In images, sheight can be computed from bounding boxes, masks, or keypoint-based measurements. In 3D environments, it can be derived from world units after projecting to the camera. Because it depends on viewpoint and focal length, sheight reflects perceptual scale rather than absolute size.

Applications and limitations: Sheight is used in data annotation, rendering workflows, and perceptual studies to maintain consistent relative sizing across scenes. It can aid automated labeling and quality control. Limitations include sensitivity to perspective distortion, occlusion, lens effects, and the choice of reference height, which must be explicit and standardized.

History: The term sheight arose in the late 2010s in small academic projects exploring scale perception in computer vision. It has not achieved widespread standardization, and implementations vary by domain.

See also: visual angle, scale, height, bounding box, perspective.