Human flow analysis method, human flow analysis apparatus, and human flow analysis system

1. A human flow analysis method performed by a processor of a human flow analysis apparatus, the human flow analysis method comprising:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space;

extracting, based on the acquired movement information, at least two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the two persons are extracted from the plurality of persons;

identifying, based on the acquired movement information, a positional relationship between the extracted two persons with respect to a direction of travel of the extracted two persons;

identifying an association which the extracted at least two persons have with each other by using the identified positional relationship between the extracted two persons with respect to the direction of travel of the extracted two persons;

determining, based on the identified association, whether to group the extracted at least two persons together; and

predicting a behavior of the extracted at least two persons who have been determined to be grouped together.

2. The human flow analysis method according to claim 1 ,

wherein the extracting includes calculating, for all combinations of two persons included in the plurality of persons, a distance between the two persons, and extracting, as the at least two persons, each of all combinations of two persons corresponding to a distance less than a predetermined threshold among all calculated distances.

3. The human flow analysis method according to claim 1 ,

wherein, when an angle formed between a straight line connecting top parts of heads of the extracted two persons, and the direction of travel of the extracted two persons is within a predetermined range centered at 90 degrees, the identifying of the association identifies that the extracted two persons have an association with each other that represents friends, a couple, or a family.

4. The human flow analysis method according to claim 1 ,

wherein, when an angle formed between a straight line connecting top parts of heads of the extracted two persons, and the direction of travel of the extracted two persons is less than or equal to a predetermined angle, the identifying of the association identifies that the extracted two persons have an association with each other that represents a person in wheelchair and a caregiver.

5. The human flow analysis method according to claim 1 , wherein the extracting extracts the two persons among the plurality of persons, who have successively passed through a gate, the gate allowing a predetermined number of persons to pass through the gate at a time.

6. A human flow analysis method performed by a processor of a human flow analysis apparatus, the human flow analysis method comprising:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space;

acquiring an appearance image of the plurality of persons;

extracting, based on the acquired movement information, at least two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the at least two persons are extracted from the plurality of persons;

discriminating respective ages and sexes of the extracted at least two persons from respective facial images of the extracted at least two persons included in the acquired appearance image;

identifying an association which the extracted at least two persons have with each other, by using the respective ages and sexes of the extracted at least two persons that have been discriminated;

determining, based on the identified association, whether to group the extracted at least two persons together; and

predicting a behavior of the extracted at least two persons who have been determined to be grouped together.

7. The human flow analysis method according to claim 1 , further comprising

acquiring gaze information, the gaze information representing gaze directions of the plurality of persons,

wherein the identifying identifies the association by using the gaze directions of the at least two persons included in the acquired gaze information.

8. The human flow analysis method according to claim 7 , wherein the identifying identifies the association when the gaze directions of the extracted two persons point toward an identical object.

9. The human flow analysis method according to claim 8 , wherein the identical object comprises a belonging of one of the extracted two persons.

10. The human flow analysis method according to claim 8 ,

wherein the identical object comprises a physical object that presents public guidance information.

11. The human flow analysis method according to claim 7 ,

wherein the acquiring of the gaze information includes acquiring the gaze information in which facial orientations of the plurality of persons detected from an appearance image of the plurality of persons are assumed to be the gaze directions of the plurality of persons.

12. The human flow analysis method according to claim 1 , further comprising

acquiring audio information of speech uttered by each of the plurality of persons,

wherein the identifying identifies the association by using, among the acquired audio information, the audio information of the extracted two persons having a conversation.

13. The human flow analysis method according to claim 12 ,

wherein the identifying identifies the association by extracting, from the audio information of the extracted two persons having a conversation, a proper noun, or a keyword that identifies kinship.

14. The human flow analysis method according to claim 1 ,

wherein the predicting includes predicting a direction of movement on a group-by-group basis for each grouping of the extracted two persons, by using the identified association of the extracted two persons and the movement information of the extracted two persons.

15. The human flow analysis method according to claim 1 ,

wherein the predicting includes predicting a direction of movement on a group-by-group basis for each grouping of the extracted two persons, by using the identified association of the extracted two persons and area type information, the area type information representing a type of association that is associated with an area existing within the predetermined space.

16. The human flow analysis method according to claim 1 , further comprising

recalculating a distance between the extracted two persons previously determined to be grouped together after the extracted two persons are grouped together, and ungrouping the extracted two persons who have been grouped together when the recalculated distance exceeds a predetermined distance.

17. A human flow analysis apparatus comprising:

a processor; and

a memory that stores a program which causes the processor to execute:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space;

extracting, based on the acquired movement information, two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the two persons are extracted from the plurality of persons;

identifying, based on the acquired movement information, a positional relationship between the extracted two persons with respect to a direction of travel of the extracted two persons;

identifying an association which the extracted two persons have with each other, by using the identified positional relationship between the extracted two persons with respect to the direction of travel of the extracted two persons;

determining, based on the identified association, whether to group the extracted two persons together; and

predicting a behavior of the extracted two persons who have been determined to be grouped together.

18. A human flow analysis system comprising:

a human flow analysis apparatus; and

a terminal apparatus communicatively connected to the human flow analysis apparatus via a network,

wherein the human flow analysis apparatus includes:

a processor;

a memory that stores a program which causes the processor to execute:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space,

extracting, based on the acquired movement information, two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the two persons are extracted from the plurality of persons,

identifying, based on the acquired movement information, a positional relationship between the extracted two persons with respect to a direction of travel of the extracted two persons;

identifying an association which the extracted two persons have with each other, by using the identified positional relationship between the extracted two persons with respect to the direction of travel of the extracted two persons,

determining, based on the identified association, whether to group the extracted two persons together,

predicting a behavior of the extracted two persons who have been determined to be grouped together; and

a transmitter that transmits a behavior prediction result to the terminal apparatus, the behavior prediction result representing the predicted behavior of the extracted two persons, and

wherein the terminal apparatus includes:

a receiver that receives the behavior prediction result transmitted from the transmitter of the human flow analysis apparatus, and

a display that displays the received behavior prediction.

19. A human flow analysis apparatus comprising:

a processor; and

a memory that stores a program which causes the processor to execute:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space;

acquiring an appearance image of the plurality of persons;

extracting, based on the acquired movement information, at least two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the at least two persons are extracted from the plurality of persons;

discriminating respective ages and sexes of the extracted at least two persons from respective facial images of the extracted at least two persons included in the acquired appearance image;

identifying an association which the extracted at least two persons have with each other, by using the respective ages and sexes of the extracted at least two persons that have been discriminated;

determining, based on the identified association, whether to group the extracted at least two persons together; and

predicting a behavior of the extracted at least two persons who have been determined to be grouped together.

20. A human flow analysis system comprising:

a human flow analysis apparatus; and

a terminal apparatus communicatively connected to the human flow analysis apparatus via a network,

wherein the human flow analysis apparatus includes:

a processor;

a memory that stores a program which causes the processor to execute:

acquiring movement information, the movement information representing a history of movement within a predetermined space by a plurality of persons moving within the predetermined space,

acquiring an appearance image of the plurality of persons;

extracting, based on the acquired movement information, at least two persons moving while keeping a distance from each other less than or equal to a predetermined distance for a predetermined period of time, wherein the at least two persons are extracted from the plurality of persons,

discriminating respective ages and sexes of the extracted at least two persons from respective facial images of the extracted at least two persons included in the acquired appearance image,

identifying an association which the extracted at least two persons have with each other, by using the respective ages and sexes of the extracted at least two persons that have been discriminated,

determining, based on the identified association, whether to group the extracted at least two persons together, and

predicting a behavior of the extracted at least two persons who have been determined to be grouped together; and

a transmitter that transmits a behavior prediction result to the terminal apparatus, the behavior prediction result representing the predicted behavior of the extracted at least two persons, and

wherein the terminal apparatus includes:

a receiver that receives the behavior prediction result transmitted from the transmitter of the human flow analysis apparatus, and

a display that displays the received behavior prediction.