US2013158963A1PendingUtilityA1
Systems and methods for improving visual attention models
Est. expiryJul 7, 2030(~4 yrs left)· nominal 20-yr term from priority
G06V 10/451G06F 30/20G06T 7/11G06T 2207/30201G06T 2207/10024G06F 17/5009
45
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
Systems and methods for improving visual attention models use effectiveness assessment from an environment as feedback to improve visual attention models. The effectiveness assessment uses data indicative of a particular behavior, which is related to visual attention allocation, received from the environment to assess relative effectiveness of the environment on influencing the particular behavior.
Claims
exact text as granted — not AI-modified1 . A method of evaluating a visual attention model, comprising:
receiving visual representations of at least a portion of two environments that differ from each other on a visual dimension; receiving output generated by applying the visual attention model on the visual representations of the at least a portion of the two environments; assessing the relative effectiveness of the two environments on influencing a particular human behavior based on data indicative of the particular human behavior received from the two environments, wherein the particular human behavior is inferentially related to attention allocation; and comparing the assessed relative effectiveness to the output generated by the visual attention model.
2 . The method of claim 1 , further comprising modifying the visual attention model when the assessed relative effectiveness is inconsistent with the output generated by the visual attention model.
3 . The method of claim 1 , wherein the particular human behavior is indirectly inferentially related to attention allocation.
4 . The method of claim 1 , wherein the data indicative of the particular human behavior comprises at least one of point-of-sale data and motion sensor data.
5 . The method of claim 1 , wherein the output generated by the visual attention model comprises at least one of a saliency map, relative saliency numbers, sequence numbers, probability of a region being attended within a given time period, and length of time to which a region is attended.
6 . The method of claim 1 , further comprising selecting two environments with similar output generated by the visual attention model on the visual representations of the at least of a portion of two environments.
7 . (canceled)
8 . The method of claim 2 , wherein the visual dimension is represented in the visual attention model, and wherein modifying the visual attention model comprises modifying a parameter of the visual attention model related to the visual dimension.
9 . The method of claim 8 , wherein modifying a parameter of the visual attention model comprises modifying a weight factor for the visual dimension represented in the visual attention model.
10 - 14 . (canceled)
15 . A visual attention model improvement system, comprising:
a module for receiving visual representations of at least a portion of two environments that differ from each other on a visual dimension; a module for receiving output generated by applying the visual attention model on the visual representations of the at least a portion of the two environments; a module for assessing the relative effectiveness of the two environments on influencing a particular human behavior based on data indicative of the particular human behavior received from the two environments, wherein the particular human behavior is inferentially related to attention allocation; and a processing unit for comparing the assessed relative effectiveness to the output generated by the visual attention model.
16 . The system of claim 15 , wherein the processing unit modifies the visual attention model when the assessed relative effectiveness is inconsistent with the output generated by the visual attention model.
17 . The system of claim 15 , wherein the particular human behavior is indirectly inferentially related to attention allocation.
18 - 20 . (canceled)
21 . The system of claim 16 , wherein the visual dimension is represented in the visual attention model.
22 . The system of claim 21 , wherein modifying the visual attention model comprises modifying a parameter of the visual attention model related to the visual dimension.
23 . The system of claim 22 , wherein modifying a parameter of the visual attention model comprises modifying a weight factor for the visual dimension represented in the visual attention model.
24 . The system of claim 16 , wherein modifying the visual attention model comprises adding a parameter related to the visual dimension to the visual attention model.
25 . The system of claim 15 , wherein the visual dimension comprises at least one of color, luminance, orientation, font, edges, motion, faces, intensity, distance from fovea, spatial bias, prior-knowledge influence, and task-based influence.
26 - 28 . (canceled)
29 . A visual attention model improvement system, comprising:
a module for receiving visual representation of at least a portion of an environment; a module for receiving output generated by applying the visual attention model on the visual representation of the at least a portion of the environment; a module for assessing the relative effectiveness of the environment on influencing the particular human behavior based on data indicative of a particular human behavior received from the environment, wherein the particular human behavior is inferentially related to attention allocation; and a processing unit for comparing the assessed relative effectiveness to the output generated by the visual attention model.
30 . The system of claim 29 , wherein the processing unit modifies the visual attention model when the assessed relative effectiveness is inconsistent with the output generated by the visual attention model.
31 . The system of claim 29 , wherein the environment is an environment having a sign.
32 . (canceled)Join the waitlist — get patent alerts
Track US2013158963A1 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.