Attribute/attribute group

In machine learning, an attribute is a measurable property or characteristic of a data point that the model uses to make predictions. For image data, attributes can describe visual properties like color, texture, shape, or size, as well as metadata like capture conditions, camera settings, or timestamps. Attribute groups bundle related attributes together to organize product or object information.

In annotation workflows, attributes add extra detail beyond the primary label. For instance, a bounding box labeled "vehicle" might have attributes for vehicle type (car, truck, motorcycle), color (red, blue, white), and condition (moving, parked, damaged). These additional tags let models learn finer distinctions without creating separate classes for every combination, keeping the label hierarchy manageable.

Attribute-based annotation is common in retail (product color, size, material), autonomous driving (pedestrian age group, pose, occluded vs. visible), and medical imaging (lesion texture, margin regularity, calcification pattern). Datature Nexus supports attribute groups within annotation projects, letting teams define structured metadata alongside spatial labels.