Summary

  • Meta has submitted a patent for a camera system capable of recognizing faces, tagging individuals, and creating clips based on their actions.
  • This system is designed to be assistant-driven, providing pre-organized highlights of activities upon request via a phone or headset.
  • The application builds on a parent filing from 2022 and is currently labeled as "Ready for Examination," meaning it has not yet been approved or released as a product.

Meta Platforms Technologies, the division responsible for Meta's research and development along with its hardware intellectual property, has published a patent for an innovative camera system that identifies individuals through facial recognition and logs their actions.

The patent details a setup where cameras transmit footage to an AI system that tags the people present, recognizes their actions, and compiles this information into video clips that can be accessed on demand via smartphones or headsets.

This development is not entirely new. Meta has been incorporating camera technology into devices like the Ray-Ban smart glasses, which have faced criticism and are often referred to as “pervert glasses” by many users. These glasses, created in collaboration with EssilorLuxottica and released in 2023, allow users to record video from a first-person perspective and interact with an AI to get information about their surroundings.

The same division responsible for these smart glasses and the Quest headsets has filed this patent. The illustrations depict a person wearing glasses while observing a group, suggesting that facial recognition technology could enhance the device's functionality from merely recording to actively identifying individuals present in the environment.

Operational Workflow

The patent outlines a four-step process. Initially, cameras capture either live or recorded video, which is then analyzed by a facial recognition layer that identifies who is present in the footage—not just that a person is there, but specifically who they are.

Following this, an action-detection layer interprets movements such as picking up objects, exiting a room, or gathering in groups. Finally, the system compiles its findings into individual media files, each marked with the identity of a person, their actions, or both.

What sets this apart from a standard smart doorbell is the assistant-driven retrieval system. The patent describes a conversational interface where a connected device can request specific footage, and the system generates the response using the tagged clips instead of requiring manual searching. Diagrams illustrate both server-side and device-level processing, with a natural language processing component aiding in the clip assembly.

However, the ability to identify and log individuals without their explicit consent raises significant concerns. Democratic senators have previously voiced apprehensions regarding Meta’s use of facial recognition in its smart glasses. As reported by Decrypt, the senators expressed that real-time facial recognition could put individuals at risk of stalking and harassment.

Meta's existing hardware is already at the heart of this debate. Owners of the newly released Ray-Ban glasses have been able to record video of strangers in public without their knowledge, and contractors have been found to review private recordings. A wearable camera that identifies individuals would amplify the disparity between Meta's portrayal of a "helpful assistant" and the potential implications of the technology, which is prompting the public to create countermeasures against surveillance.