Browse Prior Art Database

Method to identify subjects in a photo without using facial recognition analysis

IP.com Disclosure Number: IPCOM000236679D
Publication Date: 2014-May-08
Document File: 4 page(s) / 97K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system for identifying individuals in a picture when facial recognition software cannot identify the subjects. The system uses an ad-hoc broadcast spectrum to identify users within an image frame without the use of facial recognition analysis, notify a user when an unknown person captures a photo containing that user, and use a confidence metric to correlate a target user's recent activity with a Bluetooth* Low Energy (BLE) device presence in a captured image.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 32% of the total text.

Page 01 of 4

Method to identify subjects in a photo without using facial recognition analysis

People use mobile devices to capture images every day, and often want to share these photos with others, especially those individuals who appear in the picture. Photos are typically shared either through the direct transfer of a file with intended recipients or through a post on a social media portal with tagged individuals. The user must go through a manual process to choose with whom to share the file or to tag the individuals in the photo on a social media portal. In the case of tagging, there is an abundance of prior art for facial recognition solutions to automatically tag individuals in a photo. However, many limitations impede the facial analysis engine from recognizing an individual, including (but not limited to):

• Subject's face is not clearly visible, obscured, or not directed at camera • Insufficient image resolution • Insufficient light conditions • Too many faces to match against
• Identifying the face of a social media user who is not the user's friend

Facial recognition is not sufficient for users to rely upon to automatically identify subjects in an image. The solution addresses identifying individuals in a picture

when facial recognition software cannot identify the subjects.

The novel contribution is a system that enables an application on a mobile device to capture images and simultaneously perform analysis on a local broadcast spectrum to determine whether other mobile device users are nearby. The analysis determines if the location of discovered devices are possibly within the frame of the device's camera, and can store this information as metadata (tagging) along with the recorded image. Privacy controls determine what action may be taken based on device identification, such as user notification, opt-in and opt-out based policies, etc.

The core novelty of the system is the ability to use an ad-hoc broadcast spectrum to:

• Identify users within an image frame • Identify users in an image without the use of facial recognition analysis • Notify a user when an unknown person captures a photo containing the user,
• Use a confidence metric to correlate a target user's recent activity with a Bluetooth* Low Energy (BLE) device presence in a captured image

For example, when a user uses a camera or social media application to take a

picture, the application listens on a broadcast spectrum for the identities and location of participating users within range of the device. When the user takes the picture, users identified in the picture may be notified and can take action (e.g., confirm, delete, refine, etc.) on the metadata.

The image capture application is assigned a universally unique identifier (UUID) that is common to any instance of the application deployed across any device. For example, one social media application might have UUID=0, another has UUID=1, and yet another application UUID=N. Any user that has installed...