I was at my local Vons the other day and had two identical boxes of rice at the self checkout. Scanned the one in my right hand and put the one in my left hand

I was at my local Vons the other day and had two identical boxes of rice at the self checkout. Scanned the one in my right hand and put the one in my left hand in the bagging area, then tried to scan the one in my right hand again (this was all to speed things up just slightly). The screen threw up a message asking me to check on what I had put in the bagging area. Below it was a video of me shot from the ceiling, with bounding boxes drawn around the two boxes of right (basically identifying that the thing I had scanned was not what I had put in the bagging area). I was actually pretty impressed... at my work we do some computer vision work, and we're dealing with fairly high-resolution, static images. This managed to draw perfect bounding boxes on a video, and the whole thing processed in less than a couple of seconds. Seems like if you can do that, there ought to be other ways you can speed things up. Currently for produce with no bar code, I have to type in the name of the produce and then weigh it. They ought to be able to identify it as a zucchini via computer vision and skip the first step, at least.