This process involves two essential activities: data exploration and engineering.

Data exploration: The Data team explores the classes of interest to recognize and document them, and then they set up an efficient workflow that will align with the required annotation type and the annotation tool.

Engineering: Some projects may require customization of the annotation tools depending on the input data used. The Data team has vast experience with tool customization. One benefit of tool customization is increasing the annotation speed. Tool customization could be as simple as a few scripts for data transformation. Other customizations may involve multiple steps in the engineering workflow.