Identify all the objects in images or videos from 80 available common object types. Get the output as annotated images or videos, and as JSON metadata.