|
This set me to imagining: every item in the store is dropped into a box before being put on the shelf. The box contains cameras on every interior face to snap a set of 360° views of the item. The UPC code is scanned and entered into inventory. The item goes on the shelf. If sold, the sale removes that item from inventory. Each night after store closing, a Store View bot roams the aisles taking pictures and updating for the next day. You browse to the store on Google Street View, go inside with Store View, and if you see an item you like, you draw a box around it and ask the image search algorithm to look for a match based on appearance. It finds the item, tells you how much it costs, and gives you the option to buy it. Stores would get some basic frequency (annual?) of scan free, more frequent scans or keeping a bot on premises to scan every night would cost more. Integration of the online store would have a fee attached to every purchase. Prices and discounts could be updated daily and there could be different incentives for online vs. in-store purchases. As data storage and computational power get cheaper and cheaper, and image recognition algorithms get more sophisticated, this would seem to be a potential outcome. A package of algorithms could even be marketed as a store manager: moving stock that has sat on the shelf too long with discounts, predicting what will require reorders soonest, integration with price comparison engines to analyze competitiveness, etc. Once imagery is dissectable and searchable in the same way that language is, things could get very interesting very fast. |