Amazon Brings Computer Vision to the Home
by Adam Pease
Amazon has announced a suite of new features coming to its Ring doorbell system. Now, users will be able to receive notifications when a package is present at their doorstep, and down the road, Amazon will be rolling out a series of enhanced recognition features that bring computer vision to the consumer home. This blog reviews the announcement and its implications for the computer vision market.
What Is Amazon Doing with Ring Doorbell?
Amazon’s Ring Doorbell has been an ongoing opportunity for Amazon to deploy some of its new advancements in AI and cloud computing in a consumer context. Historically, Ring has deployed a variety of evolving features that fit into the smart home, especially in the area of security.
In the past, Ring’s security features have centered around its Neighbors app and local footage sharing capabilities, which network with other Ring devices to create broader security coverage. While these features have been attractive for many, others have raised concerns about privacy with Ring devices. With its preexisting focus on security, though, Ring is positioned to pivot further into scene recognition and automate its security features.
Computer Vision Comes to the Home
At-home cameras may be one of the first widespread applications of computer vision in the home. Ring’s newest feature enables users to receive notifications through an object recognition system that scans the doorstep for a new package. More intriguing, however, is the upcoming set of features that give users more tools to set up their own custom alerts.
Users can define areas of the environment seen by the doorbell camera and track motion and change on specific objects. For example, users could draw a bounding box around their garage door, guiding the computer vision algorithm to observe that area of the environment for change. It would be easy for users to set up a notification that pings them every time the garage door opens or closes.
These applications are only the beginning of what we can imagine for computer vision in at-home camera contexts like Ring. More than likely, these advancements will be followed by more pervasive automation of recognition, especially in the context of home security. It is possible to imagine a world in which doorbells not only store security footage, but can be programmed to notify users of strangers in the neighborhood, or employ facial recognition to unlock homes.
The bottom line is that computer vision is being integrated into the home, faster than many expected. It remains to be seen if concerns about privacy will hamper adoption, but computer vision is moving forward quickly. While not everybody has a Ring device, Amazon’s move foreshadows the broader adoption of computer vision as a basic feature in camera systems.