Automotive Industry Insights

ASAM OpenLABEL

Description
OpenLABEL V1.0.0 standardizes the annotation format and labeling methods for objects and scenarios. The use of a standardized format will help save costs and resources in converting annotated data. OpenLABEL will be represented in a JSON format and therefore can be easily parsed by tools and applications. OpenLABEL specifies which coordinate systems are being used as a reference for the label.

OpenLABEL also provides methods to label objects in a scene (one point in time, frame) and over a period of scenes, enhancing this with the labeling of actions, intentions, and relations between objects.

The figure shows the concepts related to data annotation for scenario-tagging. ASAM OpenLABEL covers the definition of the annotation schema for scenario-tagging and an ontology for tags. Scenario-tagging use cases focus on raw data that is used in the development, testing, and validation process of ADAS and AD functions, for example test scenarios or simulation scenarios. Often the format of such raw data is OpenSCENARIO, GEOscenario or other domain-specific languages or formats used to describe and store simulation and test scenarios.

Status
The first version of this standard was released in November 2021.