Picture Ranking Method, and Terminal

• US 20180268205A1
• Filed: 05/21/2018
• Published: 09/20/2018
• Est. Priority Date: 03/12/2014
• Status: Active Grant

• Current Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Inventors: Jiang, Zhuolin, Yang, Qiang
• Granted Patent: US 10,242,250 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/583 using metadata automatically derived from the content
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06V 20/30 in albums, collections or shared content, e.g. social network photos or video
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation
• CPC Class Codes: G06V 40/172 Classification, e.g. identification
• CPC Class Codes: G06V 40/179 metadata assisted face recognition

Abstract

A picture ranking method and a terminal comprises acquiring pictures stored in a terminal, detecting whether the pictures are first-type pictures, where a first-type picture refers to a picture including a human face, and when the pictures are first-type pictures, ranking the pictures according to a social relation model, or when the pictures are not first-type pictures, ranking the pictures according to a preset rule.

Claims

1. A picture ranking method, comprising:

acquiring pictures stored in a terminal;

detecting whether the pictures are first-type pictures, the first-type picture referring to a picture comprising a human face;

determining contacts corresponding to the pictures according to the human faces in the pictures when the pictures are the first-type pictures;

acquiring, according to a social relation model, social relations between the contacts corresponding to the pictures and a user of the terminal; and

ranking the pictures according to the social relations.

2. The method of claim 1, wherein before acquiring social relations between the contacts corresponding to the pictures and the user of the terminal, the method further comprises training the social relation model using a stored profile picture of the contacts and information about the contacts.

3. The method of claim 2, wherein training the social relation model using the stored profile picture of the contacts and the information about the contacts comprises:

extracting profile picture data of the contacts; and

annotating the profile picture data using the information about the contacts to obtain the social relation model, wherein the information about the contacts comprises identity features of the contacts, the identity features being used to represent a social relation between the contacts and the user of the terminal.

4. The method of claim 1, wherein acquiring the social relations between the contacts corresponding to the pictures and the user of the terminal comprises:

acquiring, using the social relation model, information about the contacts corresponding to the pictures, wherein the information about the contacts corresponding to the pictures comprises identity features of the contacts corresponding to the pictures, the identity features being used to represent a social relation between the contacts corresponding to the pictures and the user of the terminal; and

determining the social relations between the contacts corresponding to the pictures and the user of the terminal according to the identity features of the contacts corresponding to the pictures.

5. The method of claim 1, further comprising ranking the pictures according to a preset rule when the pictures are not the first-type pictures.

6. The method of claim 4, further comprising:

adding the pictures to an unrecognized face group; and

performing clustering ranking on images in the unrecognized face group according to a preset rule when the information about the contacts corresponding to the pictures is not acquired.

7. The method of claim 1, wherein before detecting whether the pictures are first-type pictures, the method further comprises:

extracting a feature value of a stored human face image;

learning the feature value of a human face image using a machine learning method; and

establishing a facial recognition model according to learning the feature value of the human face image.

8. The picture ranking method of claim 7, wherein detecting whether the pictures are first-type pictures comprises:

extracting feature values of the pictures;

performing a matching between the feature values of the pictures and the facial recognition model; and

determining that the pictures as the first-type pictures when a result of the matching is higher than a first preset value.

9. The picture ranking method of claim 7, wherein detecting whether the pictures are first-type pictures comprises:

extracting feature values of the pictures;

performing a matching between the feature values of the pictures and the facial recognition model; and

determining that the pictures are not the first-type pictures when a result of the matching is not higher than a first preset value.

10. A terminal, comprising:

at least one processor; and

a non-transitory computer-readable storage medium coupled to the at least one processor and storing programming instructions, wherein the programming instructions cause the at least one processor to be configured to;

acquire pictures stored in a terminal;

detect whether the pictures are first-type pictures, the first-type picture referring to a picture comprising a human face;

determine contacts corresponding to the pictures according to the human faces in the pictures when the pictures are the first-type pictures;

acquire, according to a social relation model, social relations between the contacts corresponding to the pictures and a user of the terminal; and

rank the pictures according to the social relations

11. The terminal of claim 10, wherein the programming instructions further cause the at least one processor to be configured to train the social relation model using one or more stored profile pictures of the contacts and information about the contacts before acquiring the social relations between the contacts corresponding to the pictures and the user of the terminal.

12. The terminal of claim 10, wherein the programming instructions further cause the at least one processor to be configured to:

extract profile picture data of the contacts; and

annotate the profile picture data using information about the contacts to obtain the social relation model, wherein the information about the contacts comprises identity features of the contacts, the identity features being used to represent a social relation between the contacts and the user of the terminal.

13. The terminal of claim 10, wherein the programming instructions further cause the at least one processor to be configured to:

acquire, using the social relation model, information about the contacts corresponding to the pictures, wherein the information about the contacts corresponding to the pictures comprises identity features of the contacts corresponding to the pictures, the identity features being used to represent a social relation between the contacts corresponding to the pictures and the user of the terminal; and

determine the social relations between the contacts corresponding to the pictures and the user of the terminal according to the identity features of the contacts corresponding to the pictures.

14. The terminal of claim 10, wherein the programming instructions further cause the at least one processor to be configured to rank the pictures according to a preset rule when the pictures are not the first-type pictures.

15. The terminal of claim 13, wherein the programming instructions further cause the at least one processor to be configured to:

add the pictures to an unrecognized face group; and

perform clustering ranking on images in the unrecognized face group according to a preset rule when the information about the contacts corresponding to the pictures is not acquired.

16. The terminal of claim 10, wherein the programming instructions further cause the at least one processor to be configured to:

extract a feature value of a stored human face image;

learn the feature value of a human face image using a machine learning method; and

establish a facial recognition model according to learning the feature value of the human face image.

17. The terminal of claim 16, wherein the programming instructions further cause the at least one processor to be configured to:

extract feature values of the pictures;

perform a matching between the feature values of the pictures and the facial recognition model; and

determine that the pictures are the first-type pictures when a result of the matching is higher than a first preset value.

18. The terminal of claim 16, wherein the programming instructions further cause the at least one processor to be configured to:

extract feature values of the pictures;

perform a matching between the feature values of the pictures and the facial recognition model; and

determine that the pictures are not the first-type pictures when a result of the matching is not higher than a first preset value.