System and method for detecting user attention
First Claim
1. A voice-controllable system that responds to a spoken command only when a camera has imaged a frontal face at the same time that the command was spoken, comprising:
- a memory for storing image data generated by the camera; and
a processor configured toaccess the image data stored in the memory and process the image data to determine whether or not the frontal face has been imaged;
provide a signal when a determination has been made that the frontal face has been imaged,wherein the processor determines that the frontal face has been imaged by determining whether or not a first image includes a first frontal face, performing a first frontal face detection operation on a subsequent image when the first image does include the first frontal face, and performing a second frontal face detection operation, more computationally intensive than the first frontal face detection operation, on the subsequent image when the first image does not include the first frontal face, andwherein when a multiple of frontal faces are detected in the first image one or more of the detected frontal faces is used as a template to perform a template-matching search for one or more frontal faces in a subsequent image, and when the template-matching search does not detect one or more frontal faces in the subsequent image the second, more computationally intensive, frontal face detection operation is performed on the subsequent image such that (i) the more computationally intensive, frontal face detection operation is performed on the entire subsequent image only after the template-matching search is performed on the entire subsequent image and the template-matching search does not detect one or more frontal faces in the subsequent image, and (ii) the more computationally intensive, frontal face detection operation is not performed on the subsequent image when the template-matching search does detect one or more frontal faces in the subsequent image.
2 Assignments
0 Petitions
Accused Products
Abstract
A system and method for conditioning execution of a control function on a determination of whether or not a person'"'"'s attention is directed toward a predetermined device. The method involves acquiring data concerning the activity of a person who is in the proximity of the device, the data being in the form of one or more temporal samples. One or more of the temporal samples is then analyzed to determine if the person'"'"'s activity during the time of the analyzed samples indicates that the person'"'"'s attention is not directed toward the device. The results of the determination are used to ascertain whether or not the control function should be performed.
65 Citations
9 Claims
-
1. A voice-controllable system that responds to a spoken command only when a camera has imaged a frontal face at the same time that the command was spoken, comprising:
-
a memory for storing image data generated by the camera; and a processor configured to access the image data stored in the memory and process the image data to determine whether or not the frontal face has been imaged; provide a signal when a determination has been made that the frontal face has been imaged, wherein the processor determines that the frontal face has been imaged by determining whether or not a first image includes a first frontal face, performing a first frontal face detection operation on a subsequent image when the first image does include the first frontal face, and performing a second frontal face detection operation, more computationally intensive than the first frontal face detection operation, on the subsequent image when the first image does not include the first frontal face, and wherein when a multiple of frontal faces are detected in the first image one or more of the detected frontal faces is used as a template to perform a template-matching search for one or more frontal faces in a subsequent image, and when the template-matching search does not detect one or more frontal faces in the subsequent image the second, more computationally intensive, frontal face detection operation is performed on the subsequent image such that (i) the more computationally intensive, frontal face detection operation is performed on the entire subsequent image only after the template-matching search is performed on the entire subsequent image and the template-matching search does not detect one or more frontal faces in the subsequent image, and (ii) the more computationally intensive, frontal face detection operation is not performed on the subsequent image when the template-matching search does detect one or more frontal faces in the subsequent image. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
Specification