The tool's current ability to mark an image as depicting a given item would be nicely complemented by the possibility of drawing one or more bounding boxes around where the item appears in the image, similar to how the Commons' Image-Annotator gadget works.
This could then result in "relative position within image" (P2677) qualifiers to the "depicts" statement in the image's structured data, as seen e.g. here.
The tool's current ability to mark an image as depicting a given item would be nicely complemented by the possibility of drawing one or more bounding boxes around where the item appears in the image, similar to how the Commons' Image-Annotator gadget works.
This could then result in "relative position within image" (P2677) qualifiers to the "depicts" statement in the image's structured data, as seen e.g. here.