The patent application describes a method for “Generating Composite Facial Images Using Audio/Video Recording and Communication Devices.” According to the company, this artificial intelligence system receives video recordings from two sources, identifies the persons in the scene, recognizes if they are the same person, and then alerts the user that a person was identified. Put simply, this new method enables the blending of footage from various cameras, applying artificial intelligence to identify persons that could not be otherwise determined by only one of them.
In its application, Amazon explains that sometimes two or more video recording devices “each capture partial facial images (e.g., from different angles and/or vantage points) of the same person.” These “partial facial images can be of limited value in identifying the persons in the images, due to the lack of a complete picture of the person’s face in any one image.” To address this, it would be “advantageous” (Amazon’s phrasing) if the partial facial images “could be combined to form a composite image that shows the entire face (or at least a more complete face) of a person captured in multiple images.”
The problem the patented technology aims to solve is that existing facial surveillance systems “do not provide the ability to identify image data including partial facial images from two or more cameras, or to determine that the partial facial images are of the same person.” The company’s invention purports to “solve this problem by enabling image data from multiple recording and communication devices to be analyzed so that partial facial images of a same person can be identified.” By combining partial facial images, “the identity of the person in the partial facial images can be more effectively determined,” the patent application states.
While the company appears to be contemplating using this system to link “smart” doorbell systems throughout a neighborhood, its patent application acknowledges that apart from the household security context, its system will make it “easier for law enforcement to identify, apprehend, and convict.” An AI developed to make neighborhood watch more efficient could undoubtedly be applied to a municipal camera system, enabling persistent tracking of millions of people who walk the streets of cities like New York, San Francisco, and Boston.
A reasonable person might think Amazon’s development of artificial intelligence for police would have slowed or stopped after over 70 civil rights and civil liberties groups, shareholders, hundreds of employees, and members of Congress raised serious concerns about the technology’s profound impact on civil rights and civil liberties. That reasonable person would be wrong. Instead of pumping the brakes, the patent application reveals Amazon has quietly continued its aggressive work to develop and offer ever more powerful tools for surveillance.
This blog post was written by Emiliano Falcon.
Our IP Address: