Notification and Privacy Management of Online Photos and Videos

• US 20150033362A1
• Filed: 02/01/2013
• Published: 01/29/2015
• Est. Priority Date: 02/03/2012
• Status: Active Grant

• Current Assignee: See-out Pty Ltd (Clarivate Plc)
• Original Assignee: See-out Pty Ltd (Clarivate Plc)
• Inventors: Mau, Sandra, Bigdeli, Abbas
• Granted Patent: US 9,514,332 B2
• Time in Patent Office: Days
• US Class Current: 726/27
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/5866 using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
• CPC Class Codes: G06F 21/6245 Protecting personal data, e.g. for financial or medical purposes
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06V 20/30 in albums, collections or shared content, e.g. social network photos or video
• CPC Class Codes: G06V 40/172 Classification, e.g. identification

Abstract

Systems and methods for notification and privacy management of online photos and videos are herein disclosed. According to one embodiment, a computer-implemented method includes detecting a first feature from a first image belonging to an image source. The first image includes a tag. The computer-implemented method further includes determining a first feature location of the first feature, determining a first tag location of the tag, extracting a first feature signature from the first feature based on a relationship between the first feature location and the first tag location, detecting a second feature from a second image belonging to the image source, extracting a second feature signature from the second feature, performing a first comparison between the first feature signature and the second feature signature, and deriving a first similarity score based on a result of the first comparison.

Claims

1. A computer-implemented method, comprising:

detecting a first feature from a first image belonging to an image source, wherein the first image comprises a tag and;

determining a first feature location of the first feature;

determining a first tag location of the tag;

extracting a first feature signature from the first feature based on a relationship between the first feature location and the first tag location;

detecting a second feature from a second image belonging to the image source;

extracting a second feature signature from the second feature;

performing a first comparison between the first feature signature and the second feature signature; and

deriving a first similarity score based on a result of the first comparison.

2. The computer-implemented method of claim 1, wherein the relationship is a first length from the first feature location to the first tag location being less than a second length from a third feature location of a third feature to the first tag location, and wherein the first image comprises the third feature.

3. The computer-implemented method of claim 1, wherein the relationship is a length less than a desired limit, and wherein the length is from the first feature location to the first tag location.

4. The computer-implemented method of claim 1, wherein extracting the first feature signature from the first feature comprises one or more of cropping the first feature, normalizing the first feature, scaling the first feature, filtering the first feature, extracting a low level image feature from the first feature, comparing the low level image feature to a probabilistic model, clustering the first feature signature to a plurality of feature signatures and verifying the first feature signature.

5. The computer-implemented method of claim 1, wherein extracting the second feature signature from the second feature comprises one or more of cropping the second feature, normalizing the second feature, scaling the second feature, filtering the second feature, extracting a low level image feature from the second feature, comparing the low level image feature to a probabilistic model, clustering the second feature signature to a plurality of feature signatures and verifying the second feature signature.

6. The computer-implemented method of claim 1, wherein the first feature signature and second feature signature are probabilistic histograms.

7. The computer-implemented method of claim 1, further comprising providing the first feature to a user for confirmation.

8. The computer-implemented method of claim 1, wherein the image source comprises an online album belonging to a user, an online album belonging to a connection related to the user, an internet and an intranet.

9. The computer-implemented method of claim 1, wherein detecting the first feature and the second feature is based on a detection algorithm.

10. The computer-implemented method of claim 1, wherein extracting the first feature signature and the second feature signature is based on a probabilistic multi-region histogram approach.

11. The computer-implemented method of claim 1, wherein performing the first comparison between the first feature signature and the second feature signature is based on a similarity metric.

12. The computer-implemented method of claim 1, further comprising:

detecting a third feature from a third image belonging to the image source, wherein the third image comprises the tag;

determining a third feature location of the third feature;

determining a third tag location of the tag in the third image;

extracting a third feature signature from the third feature based on a relationship between the third feature location and the third tag location;

performing a second comparison between the second feature signature and the third feature signature to derive a second similarity score; and

deriving a highest rank based on ranking the first similarity score and the second similarity score, wherein the highest rank represents a highest similarity of the first comparison and the second comparison.

13. The computer-implemented method of claim 1, further comprising:

deriving a true accept function based on deriving a plurality of true accept data points, wherein the plurality of true accept data points represents an identity of a test feature is common with an identity of a training feature;

deriving a false accept function based on a plurality of false accept data points, wherein the plurality of false accept data points represents the identity of the test feature is different from the identity of a training feature; and

deriving a threshold function based on the true accept function and the false accept function

14. The computer-implemented method of claim 13, wherein deriving a true accept data point of the plurality of true accept data points comprises:

extracting a set of training feature signatures from a set of training features, wherein identities of training features in the set of training features are known;

extracting a test feature signature from a test feature, wherein the identity of the test feature is known, and wherein the identity of the test feature is common with the identity of a training feature in the set of training features;

deriving a set of match scores by comparing the test feature signature with one or more of training feature signatures in the set of training feature signatures; and

deriving a highest rank based on ranking match scores within the set of match scores, wherein the highest rank represents a match score between the training feature signature and the test feature signature within the set of match scores;

15. The computer-implemented method of claim 13, wherein deriving a false accept data point of the plurality of false accept data points comprises:

extracting a set of training feature signatures from a set of training features, wherein identities of training features in the set of training features are known;

extracting a test feature signature from a test feature, wherein the identity of the test feature is known, and wherein the identity of the test feature is different from the identity of a training feature in the set of training features;

deriving a set of match scores by comparing the test feature signature with one or more of training feature signatures in the set of training feature signatures; and

deriving a highest rank based on ranking match scores within the set of match scores, wherein the highest rank represents a match score between the training feature signature and the test feature signature within the set of match scores;

16. The computer-implemented method of claim 13, further comprising displaying the second image to a user.

17. The computer-implemented method of claim 16, further comprising providing the user with options to manage the second image.

18. The computer-implemented method of claim 17, wherein the options comprise one or more of adjusting the threshold, adding a second tag to the second feature in the second image, removing the second tag from the second feature in the second image, adjusting a privacy setting of the second image belonging to the user, sending a message to an owner of the second image, saving the second image to an online album and sharing the online album with other users.

19. The computer-implemented method of claim 13, further comprising providing a user with options to manage the first image.

20. The computer-implemented method of claim 19, wherein the options comprise one or more of selecting the first image, unselecting the first image and deleting the first image.

21. A computer-implemented method, comprising:

deriving a true accept function based on deriving a plurality of true accept data points, wherein the plurality of true accept data points represents an identity of a test feature is common with an identity of a training feature;

deriving a false accept function based on a plurality of false accept data points, wherein the plurality of false accept data points represents the identity of the test feature is different from the identity of a training feature; and

deriving a threshold function based on the true accept function and the false accept function;

22. The computer-implemented method of claim 21, wherein deriving a true accept data point of the plurality of true accept data points comprises:

extracting a set of training feature signatures from a set of training features, wherein identities of training features in the set of training features are known;

extracting a test feature signature from a test feature, wherein the identity of the test feature is known, and wherein the identity of the test feature is common with the identity of a training feature in the set of training features;

deriving a set of match scores by comparing the test feature signature with one or more of training feature signatures in the set of training feature signatures; and

deriving a highest rank based on ranking match scores within the set of match scores, wherein the highest rank represents a match score between the training feature signature and the test feature signature within the set of match scores;

23. The computer-implemented method of claim 21, wherein deriving a false accept data point of the plurality of false accept data points comprises:

extracting a set of training feature signatures from a set of training features, wherein identities of training features in the set of training features are known;

extracting a test feature signature from a test feature, wherein the identity of the test feature is known, and wherein the identity of the test feature is different from the identity of a training feature in the set of training features;

deriving a set of match scores by comparing the test feature signature with one or more of training feature signatures in the set of training feature signatures; and

deriving a highest rank based on ranking match scores within the set of match scores, wherein the highest rank represents a match score between the training feature signature and the test feature signature within the set of match scores;