Google develops method of identifying individuals with web search and facial recognition
Google received a patent in July for a system which searches images from social networks and applies facial recognition to identify individuals, which could allow people to “Reverse Image Search” the identity of others, CB Insights reports.
The patent, “Facial Recognition with Social Network Aiding,” was filed in 2015, and includes communication, social networking, calendar, and collaborative apps as potential sources of data. The system attempts to match a query image based on visual similarity and relevance to the searcher’s social networks, and presents its best guess, possible along with other details, like the identified individual’s occupation or membership in groups.
The patent also notes that the system may have to include an opt-in process for people to authorize the sharing of their information in search results. “Reverse Image Search” can currently be used to identify public figures, such as politicians, by uploading a photo. The filing is vague about potential uses for the technology, and describes a range of open-ended embodiments.
Google researchers recently developed a method for improving the performance of facial recognition on mobile devices.