- Load manifest of a work/load cropped image around object of interest - Load value from a table (text detection output, faces, etc.) - User action (check yes or no, edit inscription, assign name to person, etc.) - Applications: obtaining inscriptions, naming people, labeling objects