Picture ranking method, and terminal

Patent metadata

Field	Value
Publication number	US-9977951-B2
Application number	US-201615255468-A
Country	US
Kind code	B2
Filing date	Sep 2, 2016
Priority date	Mar 12, 2014
Publication date	May 22, 2018
Grant date	May 22, 2018

Abstract

Embodiments of the present disclosure disclose a picture ranking method and a terminal. The picture ranking method comprises acquiring pictures stored in a terminal, detecting whether the pictures are first-type pictures, where the 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.

First claim

What is claimed is:

1. A picture ranking method, comprising:
acquiring, by a processor in a terminal, stored pictures from a memory in the terminal;
detecting, from the stored pictures, first-type pictures, second-type pictures, or both the first-type pictures and the second-type pictures, wherein the first-type pictures refer to pictures that comprise at least one human face, and wherein the second-type pictures refer to pictures that do not comprise a human face;
ranking the first pictures according to identity features of the first contacts when the first-type pictures are detected and information about first contacts corresponding to at least some first pictures of the first-type pictures is acquired, wherein the identity features are acquired using a social relation model, wherein the identity features indicate social relation categories selected for the first contacts, and wherein the social relation categories selected for the first contacts identify different types of social relations between the first contacts and a user of the terminal; and
ranking the second-type pictures according to a first preset rule when the second-type pictures are detected.

2. The method according to claim 1 , wherein, when the first-type pictures are detected and information about second contacts corresponding to at least some second pictures of the first-type pictures is not acquired, the method further comprises:
cluster grouping the second pictures to generate one or more cluster groups of the second pictures based on similarities of human faces in the second pictures; and
ranking the cluster groups according to a second preset rule that is different than the first preset rule.

3. The method according to claim 1 , wherein the first preset rule includes time points or locations associated with the second-type pictures.

4. The method according to claim 1 , wherein before ranking the first pictures according to the identity features, the method further comprises:
acquiring interaction statistics between each of the first contacts and the user;
selecting, based on the interaction statistics, the social relation categories for the first contacts from a plurality of categories that includes a friend category, a relative category, a classmate category, and a customer category; and
training the social relation model using stored profile pictures of the first contacts and the social relation categories selected for the first contacts.

5. The method according to claim 4 , wherein training the social relation model using the stored profile pictures of the first contacts and the social relation categories selected for the first contacts comprises:
extracting profile picture data of the first contacts; and
annotating the profile picture data using the social relation categories selected for the first contacts to obtain the social relation model.

6. The method according to claim 1 , wherein ranking the first pictures according to the identity features further comprises:
acquiring, using the social relation model, information about the first contacts corresponding to the first pictures, wherein the information about the first contacts corresponding to the first pictures comprises the identity features selected for the first contacts;
adding the first pictures to a recognized face group when the information about the first contacts corresponding to the first pictures is successfully acquired; and
ranking the first pictures in the recognized face group according to the identity features.

7. The method according to claim 2 , wherein ranking the cluster groups according to the second preset rule comprises:
adding the second pictures to an unrecognized face group; and
performing clustering ranking on the second pictures in the unrecognized face group according to the second preset rule.

8. The method according to claim 1 , wherein before detecting the first-type pictures, the second-type pictures, or both the first-type pictures and the second-type pictures, the method further comprises:
extracting a feature value of a stored human face image;
learning the feature value of the stored human face image using a machine learning method; and
establishing a facial recognition model.

9. The picture ranking method according to claim 8 , wherein detecting the first-type pictures comprises:
extracting feature values of the stored pictures;
performing a matching between the feature values of the stored pictures and the facial recognition model; and
determining that the first-type pictures are the first-type pictures when a result of the matching is higher than a first preset value.

10. The picture ranking method according to claim 8 , wherein detecting the second-type pictures comprises:
extracting feature values of the stored pictures;
performing a matching between the feature values of the stored pictures and the facial recognition model; and
determining that the second-type pictures are not the first-type pictures when a result of the matching is not higher than a first preset value.

11. A terminal, comprising:
a memory configured to store stored pictures;
a processor coupled to the memory and configured to:
acquire the stored pictures from the memory;
detect, from the stored pictures, first-type pictures, second-type pictures, or both the first-type pictures and the second-type pictures, wherein the first-type pictures refer to pictures that comprise at least one human face, and wherein the second-type pictures refer to pictures that do not comprise a human face;
rank the first pictures according to identity features of the first contacts when the first-type pictures are detected and information about first contacts corresponding to at least some first pictures of the first-type pictures is acquired, wherein the identity features are acquired using a social relation model, wherein the identity features indicate social relation categories selected for the first contacts, and wherein the social relation categories selected for the first contacts identify different types of social relations between the first contacts and a user of the terminal; and
rank the second-type pictures according to a first preset rule when the second-type pictures are detected.

12. The terminal according to claim 11 , wherein the processor is further configured to train the social relation model using stored profile pictures of the first contacts and the social relation categories selected for the first contacts.

13. The terminal according to claim 11 , wherein the processor is further configured to:
extract profile picture data of the first contacts; and
annotate the profile picture data using the social relation categories selected for the first contacts to obtain the social relation model.

14. The terminal according to claim 11 , wherein the processor is further configured to:
acquire, using the social relation model, information about the first contacts corresponding to the first pictures, wherein the information about the first contacts corresponding to the first pictures comprises the identity features of the first contacts;
add the first pictures to a recognized face group, and
rank the first pictures in the recognized face group according to the identity features when the information about the first contacts corresponding to the first pictures is successfully acquired.

15. The terminal according to claim 11 , wherein, when the first-type pictures are detected and information about second contacts corresponding to at least some second pictures of the first-type pictures is not acquired, the processor is further configured

Assignees

• Huawei Tech Co Ltd

Inventors

• Jiang Zhuolin
• Yang Qiang

Classifications

• G06V40/172

  Classification, e.g. identification · CPC title

• G06N20/10 Primary

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• G06K9/66

  Physics · mapped topic

• G06K9/00677

  Physics · mapped topic

• G06K9/00268

  Physics · mapped topic