Facial recognition framework using deep learning for attended robots
AssigneeUiPath, Inc.
InventorsRazvan-Alexandru Iordache, Daniel-Alexandru Morie
Life of this patent
Estimating the term…
Abstract
A facial recognition framework may be configured for robotic process automation (RPA) to automate a workflow for an application interface. A set of images of a user may be captured after the robot is initiated for the automated workflow. The set of images may be utilized by a deep learning neural network model to identify facial characteristics. The automated workflow may be performed by an attended robot based on successful validation of the user with the facial characteristics and credentials of the user for the attended robot.
Claims18
A computing device comprising:
- a processor configured to operate an attended software robotic automation executor to automate a workflow for an application interface;
- and
- an image sensor configured to capture, after the attended software robotic automation executor is initiated, a set of images configured used to identify facial characteristics of a user via a deep-learning neural network model;
- the processor is configured to receive an execution command of the automation of the workflow by the attended software robotic automation executor on a condition that a captured image of the user is successfully validated based on the identified facial characteristics by achieving a confidence level via an assigned confidence score output by the neural network model, and
- the processor terminating the workflow on a condition that the captured image configures an unsuccessful validation.
A method performed by a computing device, the method comprising:
- operating, by the computing device, an attended software robotic automation executor configured to automate a workflow for an application interface;
- capturing, by the computing device after the attended software robotic automation executor is initiated, a set of images of a user used to identify facial characteristics of the user via a deep-learning neural network model;
- receiving, by the computing device, an execution command of the automation of the workflow by the attended software robotic automation executor;
- and
- processing, by the computing device, on a condition that a captured image of the user is successfully validated based on the identified facial characteristics by achieving a confidence level via an assigned confidence score output by the neural network model, and
- terminating, by the computing device, the workflow on a condition that the captured image of the user configures an unsuccessful validation.
A non-transitory computer readable medium including code stored thereon which when executed by a processor cause the system to be performed by a computing device, the method comprising:
- operating, by the computing device, an attended software robotic automation executor configured to automate a workflow for an application interface;
- capturing, by the computing device after the attended software robotic automation executor is initiated, a set of images of a user used to identify facial characteristics of the user via a deep-learning neural network model;
- receiving, by the computing device, an execution command of the automation of the workflow by the attended software robotic automation executor;
- and
- processing, by the computing device, on a condition that a captured image of the user is successfully validated based on the identified facial characteristics by achieving a confidence level via an assigned confidence score output by the neural network model, and
- terminating, by the computing device, the workflow on a condition that the captured image of the user configures an unsuccessful validation.
Dependencies are inferred from the claim text.