Information processing apparatus and rule generation method

1. An information processing apparatus comprising:

a memory; and

a processor coupled to the memory and the processor configured to:

acquire a plurality of sample videos;

identify a position and time at which an attribute appears in each of the plurality of sample videos, a plurality of labels of the attribute being output by each of one or more pre-trained models to which each of the plurality of sample videos is input;

perform a clustering of the plurality of labels of the attribute based on the position and time of the attribute for each of the plurality of sample videos such that a plurality of clusters is classified according to the attribute, respectively, and each of the plurality of clusters is provided with one or more of the plurality of labels of the attribute; and

generate a rule by combining one or more of the plurality of labels of the attribute that are included in a cluster having a highest frequency of appearance among cluster groups obtained for all of the plurality of sample videos.

2. The information processing apparatus according to claim 1 , wherein the processor is further configured to identify, as the position of the attribute, a representative value of coordinates of a center point of an object corresponding to the attribute for frames of a sample video in which the object appears.

3. The information processing apparatus according to claim 1 , wherein the processor is further configured to identify, as the time of the attribute, a set of sample video frames in which an object corresponding to the attribute appears.

4. The information processing apparatus according to claim 1 , wherein the processor is further configured to classify, as a first cluster having the highest frequency, a second cluster among cluster groups obtained from sample videos in which the first cluster does not appear, the second cluster having a degree of element coincidence with the first cluster that is equal to or higher than a predetermined threshold.

5. The information processing apparatus according to claim 1 , wherein the processor is further configured to:

acquire additional sample videos; and

search, among the additional sample videos, a target video including one or more labels of the attribute that is in common with the one or more of the plurality of labels of the attribute combined in the rule.

6. A rule generation method comprising:

acquiring, by a computer, a plurality of sample videos;

identifying a position and time at which an attribute appears in each of the plurality of sample videos, a plurality of labels of the attribute being output by each of one or more pre-trained models to which each of the plurality of sample videos is input;

performing a clustering of the plurality of labels the attribute based on the position and time of the attribute for each of the plurality of sample videos such that a plurality of clusters is classified according to the attribute, respectively, and each of the plurality of clusters is provided with one or more of the plurality of labels of the attribute; and

generating a rule by combining one or more of the plurality of labels of the attribute that are included in a cluster having a highest frequency of appearance among cluster groups obtained for all of the plurality of sample videos.

7. The rule generation method according to claim 6 , further comprising:

identifying, as the position of the attribute, a representative value of coordinates of a center point of an object corresponding to the attribute for frames of a sample video in which the object appears.

8. The rule generation method according to claim 6 , further comprising:

identifying, as the time of the attribute, a set of sample video frames in which an object corresponding to the attribute appears.

9. The rule generation method according to claim 6 , further comprising:

classifying, as a first cluster having the highest frequency, a second cluster among cluster groups obtained from sample videos in which the first cluster does not appear, the second cluster having a degree of element coincidence with the first cluster that is equal to or higher than a predetermined threshold.

10. The rule generation method according to claim 6 , further comprising:

acquiring additional sample videos; and

searching, among the additional sample videos, a target video including one or more labels of the attribute that is in common with the one or more of the plurality of labels of the attribute combined in the rule.

11. A non-transitory computer-readable recording medium having stored therein a program that causes a computer to execute a process, the process comprising:

acquiring a plurality of sample videos;

identifying a position and time at which an attribute appears in each of the plurality of sample videos, a plurality of labels of the attribute being output by each of one or more pre-trained models to which each of the plurality of sample videos is input;

performing a clustering of the plurality of labels of the attribute based on the position and time of the attribute for each of the plurality of sample videos such that a plurality of clusters is classified according to the attribute, respectively, and each of the plurality of clusters is provided with one or more of the plurality of labels of the attribute; and

generating a rule by combining one or more of the plurality of labels of the attribute that are included in a cluster having a highest frequency of appearance among the plurality of cluster groups obtained for all of the plurality of sample videos.

12. The non-transitory computer-readable recording medium according to claim 11 , the process further comprising:

identifying, as the position of the attribute, a representative value of coordinates of a center point of an object corresponding to the attribute for frames of a sample video in which the object appears.

13. The non-transitory computer-readable recording medium according to claim 11 , the process further comprising:

identifying, as the time of the attribute, a set of sample video frames in which an object corresponding to the attribute appears.

14. The non-transitory computer-readable recording medium according to claim 11 , the process further comprising:

classifying, as a first cluster having the highest frequency, a second cluster among cluster groups obtained from sample videos in which the first cluster does not appear, the second cluster having a degree of element coincidence with the first cluster that is equal to or higher than a predetermined threshold.

15. The non-transitory computer-readable recording medium according to claim 11 , the process further comprising:

acquiring additional sample videos; and

searching, among the additional sample videos, a target video including one or more labels of the attribute that is in common with the one or more of the plurality of labels of the attribute combined in the rule.