Facebook Patent Filing Could Take Advertising to A Bizarre New Level
The social network’s proposed system could detect photographs uploaded to the platform that feature items such as popular snacks or alcohol products.
Facebook has been granted a patent for a system that aims to scan users’ photographs to identify products and brands.
The social network’s proposed system could detect photographs uploaded to the platform that feature items such as popular snacks or alcohol products. These images, the patent shows, would be passed on to brands, which can then be used in ads for other users’ to see.
Examples provided in the patent filing show a Facebook user uploading an image taken at a party where they are holding a bottle of Grey Goose vodka. Facebook could then automatically detect the vodka bottle in the photograph and coordinate with the brand to produce an online ad.
A ‘heat map’ could also be included in the system, based on Facebook’s comments in the patent filing. Hypothetically, this could show where individual users had been photographed with products.
Recommended: Flickr Introduces Photo Theft Detection Tools
While the concept does appear intrusive, the documents suggest that certain restrictions would apply regarding users’ privacy settings.
Ads containing products identified by the system would only be targeted to someone’s Facebook friends if they operated under limited privacy settings – essentially, only their friends could see the image.
However, it did not detail specific details on how the system would work at scale and currently it is unclear whether users would be able to opt out of the system.
The patent filing shows that Facebook is eager to tap into and gather information on what its two billion users do in the real world.
Facebook’s patent does not necessarily mean that the social network is building the feature currently. Companies file patents regularly for ideas which never come to fruition. However, the system appears to mirror previous investments the company has made in this area.
Last year, the firm revealed work on its “Rosetta” tool, which is capable of analysing text in photographs. This system analyses and extracts text from more than one billion Facebook and Instagram images on a daily basis and is even capable of extracting text from video stills.
Text data is fed through a text recognition model that aims to understand the context of the text in each individual image.