JP2016134144A - Attribute estimation device and attribute estimation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate an attribute of a person in a target area at an early stage.SOLUTION: A group attribute estimation unit 150 estimates an attribute 31 of each of groups 30, formed with persons in a target area divided into the plurality of groups 30, on the basis of an attribute of at least a part of persons included in the groups 30. A personal attribute estimation unit 140 estimates an attribute 21 of the person included in each group 30 on the basis of the attribute 31 of the group 30 estimated by the group attribute estimation unit 150.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a technique for estimating an attribute of a person in a target area such as an airport or a shopping mall.

  Attempts have been made to estimate the attributes of people in the target area and provide services according to the estimated attributes. By providing a service according to the attribute, it is possible to provide a service suitable for each person and improve the service level.

  In Patent Literature 1, people in a store are divided into groups, and the gender and age of the people in the store are estimated. Then, for each group, the group is determined from the gender and age estimated for the people belonging to the group. Is described as to whether or not is a parent-child or a couple.

JP 2004-54376 A

Estimating each person's attributes early and accurately contributes to an improvement in service level. However, there are situations where the attributes of some people cannot be estimated. As a result, it is difficult to improve the service level for people whose attributes cannot be estimated.
An object of this invention is to estimate a person's attribute at an early stage.

The attribute estimation device according to the present invention is:
A group attribute estimator that estimates the attributes of each group in which the person in the target area is divided into a plurality of groups, from the attributes of the people included in the group;
A human attribute estimation unit that estimates the attribute of the person included in each group from the attribute of the group estimated by the group attribute estimation unit.

  The group attribute estimation unit estimates the attribute of each group from the attributes of some persons included in the group.

The human attribute estimation unit estimates an attribute of a person included in each group from the attribute of the group estimated by the group attribute estimation unit and the behavior of the person,
The group attribute estimation unit estimates an attribute of the group for each group from an attribute of a person included in the group estimated by the human attribute estimation unit.

  The person's action includes a place where the person stopped.

  The person's action includes the order of the places where the person stopped.

The attributes include basic attributes and additional attributes,
When the attribute estimated for a person is a basic attribute, the human attribute estimation unit changes the basic attribute for the person estimated in the past to the estimated basic attribute, and the attribute estimated for the person is added. If it is an attribute, the estimated additional attribute is added to the additional attribute for the person.

The attribute estimation device further includes:
An image data acquisition unit that acquires a plurality of image data obtained by imaging the target area at different times;
From a plurality of image data acquired by the image data acquisition unit, comprising an action specifying unit for specifying the action of a person in the target area,
The human attribute estimating unit estimates a human attribute from the behavior of the person specified by the behavior specifying unit.

An attribute estimation program according to the present invention includes:
A group attribute estimation process for estimating the attributes of each group in which the person in the target area is divided into a plurality of groups, from the attributes of the persons included in the group;
The computer executes a human attribute estimation process for estimating a person attribute included in each group from the group attribute estimated in the group attribute estimation process.

  In this invention, the attribute of a group is estimated from the action of the person included in the group, and the attribute of the person included in the group is estimated from the attribute of the group. This makes it possible to estimate the attributes of other people included in the group from the actions of some people included in the group. As a result, it becomes possible to estimate a person's attribute at an early stage.

Explanatory drawing of the outline | summary of the attribute estimation apparatus 100 which concerns on Embodiment 1. FIG. 1 is a configuration diagram of an attribute estimation apparatus 100 according to Embodiment 1. FIG. 5 is a flowchart showing the operation of the attribute estimation apparatus 100 according to the first embodiment. The figure which shows the example of the rule 42 which the rule memory | storage part 160 memorize | stored. The figure which shows the example of the information which the estimation information storage part 170 memorize | stored. The figure which shows the example of the information which the estimation information storage part 170 memorize | stored. The figure which shows the example of the rule 42 which the rule memory | storage part 160 memorize | stored. The figure which shows the example of the information which the estimation information storage part 170 memorize | stored. The figure which shows the example of the information which the estimation information storage part 170 memorize | stored. The figure which shows the example of the rule 42 which the rule memory | storage part 160 memorize | stored. The figure which shows the example of the rule 42 which the rule memory | storage part 160 memorize | stored. The figure which shows the hardware structural example of the attribute estimation apparatus 100 which concerns on Embodiment 1,2.

Embodiment 1 FIG.
In the first embodiment, an example in which an airport is the target area 10 will be described.
The target area 10 is not limited to an airport, and may be another area where many people gather, such as a shopping mall.

*** Explanation of the outline ***
Based on FIG. 1, the outline | summary of the attribute estimation apparatus 100 which concerns on Embodiment 1 is demonstrated.
Various people 20 come to the airport. For example, at the airport, a passenger boarding an aircraft, a passenger who came to see off the passenger, a passenger who came to pick up the passenger, a visitor who came to visit the aircraft, a suspicious person with unknown purpose Etc. come.
Some people 20 who have come to the airport are accustomed to the airport, while others are not accustomed to it. Some of the people 20 who came to the airport are from the country where the airport is located, and some are from abroad. Some people 20 who have come to the airport are physically disabled.
Thus, people 20 with various characteristics come to the airport for various purposes. Here, the purpose and characteristics of the person 20 who has come to the airport are referred to as an attribute 21 of the person 20.

  People 20 sometimes come to the airport alone, but often come in groups 30 with their companions. Here, the purpose and characteristics of the group 30 are referred to as an attribute 31 of the group 30.

At the airport, a counter 11 for check-in procedures, a help desk 12 for various services and guidance, a shopping area 13 for stores selling souvenirs, and a restaurant area 14 for restaurants are taken off and landing. There are various places such as the observation area 15 where you can see the aircraft.
The counter 11 and the help desk 12 may be set as a corresponding area by adding, for example, a space included in the front 1 meter and the left and right 1 meter to the actual occupied space. The reason why the rear area is not included in the corresponding area is that the rear area is a space where the staff works, and the person 20 who has come to the airport does not enter. For the shopping area 13, the restaurant area 14 and the like, it is conceivable to set the actual occupied space as the corresponding area, but it may be set by adding a space of an arbitrary size to the occupied space. In an emergency, the area can be set at an arbitrary timing, such as changing the area setting.

The attribute estimation device 100 performs the following estimations (1) to (3).
(1) The attribute estimation device 100 estimates the attribute 21 of the person 20 based on the action of the person 20. Here, the action of the person 20 is a place where the person 20 stops. That is, the attribute estimation apparatus 100 estimates the attribute 21 of the person 20 based on the place where the person 20 stopped.
This is because the place where the person 20 stops is likely to characterize the attribute 21 of the person 20.

  For example, the person 20 who has stopped at the counter 11 is highly likely to be a passenger. The person 20 who has stopped at the help desk 12 is likely to be a person 20 who needs some help. There is a high possibility that the person 20 who stops at the shopping area 13 is a person 20 who is interested in shopping for souvenirs.

(2) The attribute estimation device 100 estimates the attribute 31 of the group 30 based on the actions of at least some of the people 20 included in the group 30. That is, the attribute estimation apparatus 100 estimates the attribute 31 of the group 30 based on the location where at least some of the people 20 included in the group 30 stop.
This is because the place where the person 20 included in the group 30 stops is likely to characterize the attribute 31 of the group 30.

  For example, there is a high possibility that the group 30 having the person 20 who has stopped at the counter 11 is the group 30 of the passengers. There is a high possibility that the group 30 including the person 20 who has stopped at the help desk 12 is a group 30 including the person 20 who needs some help. There is a high possibility that the group 30 having the person 20 who has stopped at the shopping area 13 is the group 30 of the person 20 who is interested in shopping.

(3) The attribute estimation device 100 estimates the attribute 21 of the person 20 included in each group 30 from the attribute 31 of the group 30. That is, the attribute estimation device 100 estimates the attribute 21 of another person 20 included in the group 30 using information on a place where at least some of the persons 20 included in the group 30 stopped.
This is because there is a high possibility that the attribute 31 of the group 30 characterizes the attribute 21 of each person 20 included in the group 30.

For example, as described above, there is a high possibility that the group 30 in which the person 20 who has stopped at the counter 11 is the passenger's group 30. In this case, just because there is a person 20 who stopped at the counter 11, not only all the persons 20 in the group 30 are passengers, but only some of the persons 20 in the group 30 are passengers and the remaining persons. 20 may be a see-off customer. However, there is a high possibility that the group 30 in which the person 20 who stopped at the counter 11 is at least the group 30 including the passengers, and not all the persons 20 included in the group 30 are people 20 who are interested in boarding. There is a high possibility. That is, there is a high possibility that all the people 20 included in the group 30 are people 20 who are interested in the location of the boarding gate of the aircraft, operation information such as delay and cancellation of the aircraft, and the like.
Therefore, it is highly likely that setting a passenger as the attribute 21 of the person 20 included in the group 30 is appropriate.

  In addition, as described above, the group 30 having the person 20 who has stopped at the help desk 12 includes the person 20 who needs some help, such as traveling with a disabled person, a pregnant woman, an elderly person, a baby, or a child alone. There is a high possibility of 30. In this case, the person 20 who stops at the help desk 12 may be the person 20 who needs help, or the person 20 included in the group 30 other than the person 20 who stops at the help desk 12 needs help 20. Can be. In any case, it is highly likely that it is appropriate to set that the attribute 21 of the person 20 included in the group 30 needs help.

In addition, as described above, there is a high possibility that the group 30 in which the person 20 who stops at the shopping area 13 is the group 30 of the person 20 who is interested in shopping. In this case, not only all of the people 20 in the group 30 are interested in shopping but also only some of the people 20 in the group 30 may be interested in shopping. However, because the group 30 includes people 20 who are interested in shopping, other people 20 may also be interested in shopping. Therefore, there is a high possibility that all the people 20 included in the group 30 are people 20 who are interested in shopping.
Therefore, it is highly likely that it is appropriate to set that the attribute 21 of the person 20 included in the group 30 is interested in shopping.

  As described above, the attribute estimation device 100 estimates the attribute 31 of the group 30 from the actions of some of the people 20 included in the group 30 and further estimates the attributes 21 of other people included in the group 30. Therefore, even when some of the people 20 included in the group 30 perform an action that can estimate the attribute 21 and other people 20 do not perform an action that can estimate the attribute 21, the attribute 21 of the other person 20 is estimated. It becomes possible to do. As a result, it becomes possible to estimate the attribute 21 of the person 20 at an early stage.

*** Explanation of configuration ***
Based on FIG. 2, the structure of the attribute estimation apparatus 100 which concerns on Embodiment 1 is demonstrated.
The attribute estimation device 100 is connected to an imaging device 200 installed in an airport via a network. In FIG. 2, only one imaging device 200 is shown, but a plurality of imaging devices 200 may be installed in the airport.
The attribute estimation device 100 includes an image data acquisition unit 110, an action identification unit 120, a group estimation unit 130, a human attribute estimation unit 140, a group attribute estimation unit 150, a rule storage unit 160, and an estimation information storage unit 170. With.

The image data acquisition unit 110 acquires a plurality of image data 40 obtained by imaging the target area 10 at different times.
Here, the image data acquisition unit 110 acquires a plurality of image data 40 obtained by imaging the airport that is the target area 10 at different times by the imaging device 200. Further, the image data 40 acquired by the image data acquisition unit 110 may be continuous image data (moving image data).

The action specifying unit 120 specifies the person 20 in the target area 10 from the plurality of image data 40 acquired by the image data acquisition unit 110 and specifies the action 41 of the specified person 20. The action specifying unit 120 specifies the person 20 by face recognition or the like based on the image data 40, and specifies the action 41 of each person 20 by tracking each specified person 20. The action 41 of the person 20 is a place where the person 20 stops as described above.
For example, when the behavior specifying unit 120 stays in front of the counter 11 or the help desk 12 for a reference time or more, the action specifying unit 120 specifies that the counter 11 or the help desk 12 has been stopped. In addition, when the behavior specifying unit 120 stays in the shopping area 13, the restaurant area 14, or the observation area 15 for a reference time or longer, the behavior specifying unit 120 determines that the shopping area 13, the restaurant area 14, or the observation area 15 has stopped. The reference time used here may be different for each place.

The group estimation unit 130 is a group of people 20 in the target area 10 based on the plurality of image data 40 acquired by the image data acquisition unit 110 and the behavior 41 of the people 20 in the target area 10 specified by the behavior specifying unit 120. 30 is estimated.
For example, the group estimation unit 130 estimates that a person 20 who stays within a reference distance and stays for a reference time or longer belongs to the same group 30. Further, the group estimation unit 130 estimates that a person 20 who has performed a specific joint work such as handing over a package or having a conversation belongs to the same group.

The human attribute estimating unit 140 estimates the attribute 21 of the person 20 in the target area 10 from the action 41 of the person 20 specified by the action specifying unit 120 according to a rule 42 stored in a rule storage unit 160 described later.
Further, the human attribute estimation unit 140 estimates the attribute 21 of the person in consideration of the attribute 31 of the group 30 including the person 20 estimated by the group attribute estimation unit 150 described later.

  The group attribute estimation unit 150 follows the rule 42 stored in the rule storage unit 160, which will be described later, for each group 30 in which the person 20 in the target area 10 is divided into a plurality of groups 30, and the attribute 31 of the group 30 This is estimated from the behavior 41 of at least some of the people 20 included in the group 30.

  The rule storage unit 160 is a storage device that stores a rule 42 for estimating the attribute 21 and the attribute 31.

  The estimated information storage unit 170 is a storage device that stores the contents estimated by the group estimation unit 130, the human attribute estimation unit 140, and the group attribute estimation unit 150.

*** Explanation of operation ***
Based on FIG. 3, the operation of the attribute estimation apparatus 100 according to Embodiment 1 will be described.
The operation of attribute estimation apparatus 100 according to Embodiment 1 corresponds to the attribute estimation method according to Embodiment 1. The operation of the attribute estimation apparatus 100 according to the first embodiment corresponds to the processing of the attribute estimation program according to the first embodiment.

  The attribute estimation apparatus 100 repeatedly executes the processing from S1 to S5 at an arbitrary interval.

  In the image data acquisition process of S1, the image data acquisition unit 110 acquires a plurality of image data 40 captured after the previous image data acquisition process.

  In the action specifying process of S2, the action specifying unit 120 specifies the person 20 in the target area 10 from the plurality of image data 40 acquired in S1, and specifies the action 41 of the specified person 20.

In the group estimation process of S3, the group estimation unit 130 estimates the group 30 of people 20 in the target area 10 from the plurality of image data 40 acquired in S1 and the behavior 41 of the person 20 specified in S2. To do.
The group estimation unit 130 stores the estimated group 30 information in the estimation information storage unit 170.

In the human attribute estimation process of S4, the human attribute estimation unit 140 estimates the attribute 21 of the person 20 in the target area 10 from the behavior 41 of the person 20 in the target area 10 specified in S3. In addition, when the attribute 31 of the group 30 including the target person 20 for estimating the attribute 21 is estimated in S5 described later, the human attribute estimating unit 140, that is, the group 30 including the target person 20 is included. When the attribute 31 is stored in the estimated information storage unit 170, the attribute 21 of the target person 20 is estimated in consideration of the attribute 31 of the group 30.
The human attribute estimation unit 140 stores the estimated attribute 21 in the estimation information storage unit 170.

In the group attribute estimation process of S5, the group attribute estimation unit 150 estimates the attribute 31 of the group 30 for each group 30, and the attribute 21 of the person 20 is estimated for the person 20 included in the group 30 in S4. Is estimated from the attribute 21 of the person 20.
The group attribute estimation unit 150 stores the estimated attribute 31 in the estimation information storage unit 170.

An example of the rule 42 stored by the rule storage unit 160 will be described with reference to FIG.
The rule storage unit 160 stores two rules 42, rule A and rule B.
The rule A is a rule 42 that defines the relationship between the behavior 41 of the person 20 and the attributes 21 and 31. In the rule A, an attribute 21 and an attribute 31 are defined for each place where the person 20 stops.
The rule B is a rule 42 that defines the relationship between the attribute 31 and the attribute 21.

Based on FIG. 5 and FIG. 6, the estimation method of the attribute 21 and the attribute 31 by the rules A and B shown in FIG. 4 will be described.
5 and 6 show examples of information stored in the estimated information storage unit 170. FIG. 5 shows a group 30 including each person 20 and an attribute 21, and FIG. The attribute 31 is shown.
As shown in FIG. 5, the person 20A, the person 20B, and the person 20C are included in the group 30X.
When the person 20A stops at the counter 11, according to the rule A, the attribute 21 of the person 20A is estimated as a passenger. Further, according to the rule A, the attribute 31 of the group 30X including the person 20A is estimated as a passenger. Further, according to the rule B, the attribute 21 of the other person 20B and the person 20C included in the group 30X is estimated as a passenger.
Thereafter, when the person 20B stops at the help desk 12, according to the rule A, it is estimated that the attribute 21 of the person 20B needs help in addition to the passenger. Further, according to the rule A, it is estimated that the attribute 31 of the group 30X including the person 20B needs help in addition to the passenger. Further, according to the rule B, it is estimated that the attribute 21 of the other person 20A and person 20C included in the group 30X needs help in addition to the passenger.

  The estimated information storage unit 170 may store the estimated time together with the estimated attribute 21 and attribute 31.

As described above, the attribute estimation device 100 according to Embodiment 1 estimates the attribute 31 of the group 30 including the person 20 from the action 41 of the person 20 and the other attributes included in the group 30 from the attribute 31 of the group 30. The attribute 21 of the person 20 is estimated.
As a result, even if some of the people 20 included in the group 30 perform an action that can estimate the attribute 21, and other people 20 do not perform an action that can estimate the attribute 21, the attribute 21 of the other person 20 can be changed. It becomes possible to estimate. As a result, it becomes possible to estimate the attribute 21 of the person 20 at an early stage.

In the above description, the human attribute estimation unit 140 estimates the attribute 21 of the person 20 from the action 41 of the person 20 and the attribute 31 of the group 30 in which the person 20 is included. However, the attribute 21 of the person 20 may be estimated in consideration of other information obtained from the image data 40.
The other information includes the type and size of the luggage that the person 20 has, the physique, the face, the posture, the clothes, and the like. When the person 20 has a terminal whose position can be specified, the attribute 21 of the person 20 may be estimated using the position specified by the terminal.

In the above description, it is assumed that the action 41 of the person 20 is a place where the person 20 stops. However, the action 41 of the person 20 may be the order of the place where the person 20 stopped. In other words, the attribute 21 of the person 20 may be specified in consideration of not only where the place has stopped but also in which order the place has stopped.
When the line of sight of the person 20 can be specified from the image data 40, what the person 20 has seen may be included as the action 41 of the person 20. Thereby, the object which the person 20 is interested in can be estimated, and the accuracy of the estimation of the attribute 21 can be increased.
The plurality of captured image data 40 can capture a continuous movement, and can thereby determine the action of the person 20 (standing / sitting, falling down, jerking, etc.).

For example, the attribute estimation device 100 transmits the attribute 21 estimated for each person 20 together with the position information of the person 20 to the control device of the digital signage. Then, the attribute estimation device 100 causes the control device to determine display data based on the attribute 21 and the position information. Thereby, appropriate display data can be displayed, and the familiarity of airport guide information (such as flight information), the advertising effect, and the like can be enhanced.
Further, for example, the attribute estimation apparatus 100 transmits the attribute 21 estimated for each person 20 together with the image and position information of the person 20 to the terminal of the airport staff. For example, when the attribute 21 is a suspicious person, the staff can monitor the person 20 and can prevent crimes from occurring.

Embodiment 2. FIG.
In the first embodiment, the method for estimating the attribute 21 of the person 20 at an early stage has been described. However, in the first embodiment, the rule 42 for estimating the attribute 21 and the attribute 31 has been described using a simple example. In the second embodiment, a rule 42 for estimating the attribute 21 and the attribute 31 will be described.
In the second embodiment, parts different from the first embodiment will be described.

The attribute 21 and the attribute 31 include a basic attribute 51 and an additional attribute 52.
The basic attribute 51 is a basic attribute of the person 20 and the group 30. As for the basic attribute 51, the result once estimated may be changed by the subsequent action 41 of the person 20, the attribute 31 of the group 30, or the like. However, in principle, a plurality of basic attributes 51 are not simultaneously established for one person 20 and one group 30. For example, the purpose of the person 20 coming to the airport corresponds to the basic attribute 51.
The additional attribute 52 is an additional attribute of the person 20 and the group 30. The additional attribute 52 is additionally set when the action 41 of the person 20 or the attribute 31 of the group 30 satisfies the condition. For example, an object, a physical feature, or the like that the person 20 is interested in corresponds to the additional attribute 52. The group additional attribute 52 is added or changed as the additional attribute 52 of the person 20 belonging to the group is added or changed.

  When the attribute 21 estimated for a certain person 20 is the basic attribute 51, the human attribute estimation unit 140 changes the basic attribute 51 for the person 20 estimated in the past to the estimated basic attribute 51. In addition, when the attribute 21 estimated for a certain person 20 is the additional attribute 52, the human attribute estimation unit 140 adds the estimated additional attribute 52 to the additional attribute 52 for the person 20.

  Similarly, when the attribute 31 estimated for a certain group 30 is the basic attribute 51, the group attribute estimation unit 150 changes the basic attribute 51 for the group 30 estimated in the past to the estimated basic attribute 51. . In addition, when the attribute 31 estimated for a certain group 30 is the additional attribute 52, the group attribute estimation unit 150 adds the estimated additional attribute 52 to the additional attribute 52 for the group 30.

Based on FIG. 7, an example of the rule 42 stored in the rule storage unit 160 will be described.
As in FIG. 4, the rule storage unit 160 stores two rules 42, rule A and rule B. However, for the rule A, a type indicating whether the attribute 21 and the attribute 31 are the basic attribute 51 or the additional attribute 52 is set.

Based on FIG. 8 and FIG. 9, the estimation method of the attribute 21 and the attribute 31 by the rules A and B shown in FIG. 7 will be described.
8 and 9 show examples of information stored in the estimated information storage unit 170. FIG. 8 shows the group 30 and the attribute 21 in which each person 20 is included. FIG. The attribute 31 is shown.
First, the group 30X will be described.
As illustrated in FIG. 8, it is assumed that the group estimation unit 130 estimates that the person 20A, the person 20B, and the person 20C are included in the group 30X.
When the person 20A and the person 20B stop at the observation area 15 after arriving at the airport, the basic attribute 51 of the attribute 21 of the person 20A and the person 20B is estimated to be a visitor according to the rule A. Further, according to the rule A, the basic attribute 51 of the attribute 31 of the group 30X including the person 20A and the person 20B is estimated to be a visitor. Further, according to the rule B, it is presumed that the basic attribute 51 of the attribute 21 of the other person 20C included in the group 30X is a visitor.
On the other hand, when the person 20C stops at the shopping area after arriving at the airport, according to the rule A, the attribute 21 of the person 20C is estimated that the additional attribute 52 is interested in shopping. Further, according to the rule A, the attribute 31 of the group 30X including the person 20C is estimated that the additional attribute 52 is interested in shopping. Further, according to the rule B, the attribute 21 of the other person 20A and the person 20B included in the group 30X is estimated that the additional attribute 52 is interested in shopping.

  Thereafter, when the person 20A stops at the counter 11, according to the rule A, the attribute 21 of the person 20A is estimated to have the basic attribute 51 as the passenger. That is, the basic attribute 51 of the attribute 21 of the person 20A is changed from the visitor to the passenger. Further, according to the rule A, the attribute 31 of the group 30X including the person 20A is estimated as a passenger. That is, the basic attribute 51 of the attribute 31 of the group 30 is changed from the visitor to the passenger. Further, according to the rule B, the attributes of the other persons 20B and 20C included in the group 30X are estimated as passengers. That is, the basic attribute 51 of the attribute 21 of the person 20B and the person 20C is changed from the visitor to the passenger.

  Further, when the person 20B subsequently stops at the help desk 12, according to the rule A, it is estimated that the attribute 21 of the person 20B has the additional attribute 52 interested in shopping and needs help. Further, according to the rule A, the attribute 31 of the group 30X including the person 20B is estimated that the additional attribute 52 is interested in shopping and needs help. Further, according to the rule B, the attribute 21 of the other person 20A and person 20C included in the group 30X is estimated that the additional attribute 52 is interested in shopping and needs help.

Next, the group 30Y will be described.
Here, in the rules A and B shown in FIG. 7, the basic attribute 51 described above has a higher priority.
It is assumed that it is not estimated that the person 20D, the person 20E, and the person 20F are in the same group 30Y when they arrive at the airport. Then, when the group 20Y has not been estimated and the person 20D and the person 20F stop by the observation area 15, the basic attribute 51 of the attribute 21 of the person 20D and the person 20F is estimated to be a visitor according to the rule A. The When the person 20E stops at the counter 11, according to the rule A, the attribute 21 of the person 20E is estimated to have the basic attribute 51 as a passenger.
Further, when the person 20D stops at the shopping area, it is estimated that the attribute 21 of the person 20D has the additional attribute 52 interested in shopping according to the rule A. Moreover, when the person 20E stops in the restaurant area 14, it is estimated that the attribute 21 of the person 20E has the additional attribute 52 interested in eating and drinking. Further, when the person 20F stops by the help desk 12, it is estimated that the attribute 21 of the person 20F requires the additional attribute 52 to help.

  Thereafter, it is assumed that the person 20D, the person 20E, and the person 20F are estimated to be the group 30Y. Then, according to the rule B, the passenger who is the basic attribute 51 of the person 20E is estimated as the basic attribute 51 of the group 30Y. Here, the basic attribute 51 of the person 20E, not the basic attribute 51 of the person 20D and the person 20F, is estimated as the basic attribute 51 of the group 30Y in the rule B of the passenger who is the basic attribute 51 of the person 20E. This is because the priority is higher than the visitors who are the basic attributes 51 of the person 20D and the person 20F. Further, the additional attribute 52 of the person 20D, the person 20E, and the person 20F is estimated as the additional attribute 52 of the group 30Y. . Further, according to the rule B, the passenger who is the basic attribute 51 of the person 20E is estimated as the basic attribute 51 of the person 20D and the person 20F. That is, the basic attribute 51 of the attribute 21 of the person 20D and the person 20F is changed from the visitor to the passenger.

  The priority among the basic attributes 51 stored in the rule 42 is given priority, for example, in which the obvious suspicious person has the highest priority, followed by the passenger, the visitor, the suspicious action, and the basic attribute is unknown. Can be determined. The obvious suspicious person may be a person 20 having something that can be a weapon.

In the description based on FIGS. 8 and 9, as shown in FIG. 8, the person 20A, the person 20B, and the person 20C included in the group 30X share the additional attribute of the person 20 included in the same group 30X and are the same. An additional attribute 52 is provided. On the other hand, the person 20D, person 20E, and person 20F included in the group 30Y do not share the additional attribute of the person 20 included in the same group 30Y, and have different additional attributes 52. As described above, whether or not the additional attribute 52 of the person 20 included in the same group 30 is shared within the same group 30 can be arbitrarily determined depending on the purpose. For example, whether or not the additional attribute 52 is shared within the same group 30 may be set depending on a place where the attribute estimation apparatus 100 is used and whether the use time is a busy period.
When the additional attribute 52 of the group 30 is to be completely reflected on the person 20, the additional attribute of the person 20 may be shared by the group 30. On the other hand, considering the possibility that the person 20 is estimated to be in another group 30, when it is desired to give priority to the additional attribute 52 of the person 20, the additional attribute of the person 20 may not be shared by the group 30.

  As described above, the attribute estimation apparatus 100 according to Embodiment 2 divides the attribute 21 and the attribute 31 into the basic attribute 51 and the additional attribute 52, and changes the way of estimation. This makes it possible to estimate more appropriate attribute 21 and attribute 31.

In the above description, only airport users are targeted. However, there are also staff working at the airport. Whether it is a user or a staff can be determined by whether or not they are wearing a uniform and whether or not they are in a staff area where only the staff can enter. Even when the same person changes his / her clothes due to face recognition or the like, the user may not be determined as a user once determined as a staff member.
In the case of a staff member, control may be performed so that the attribute 21 corresponding to a user such as a passenger or a visitor is not estimated, and the attribute 21 prepared for the staff is estimated.

  In the above description, in principle, a plurality of basic attributes 51 are not estimated for one person 20. However, there may be a plurality of types of basic attributes 51, and one basic attribute 51 may be estimated for each type for one person 20.

In the above description, the attribute 21 and the attribute 31 are defined by one condition. However, the attribute 21 and the attribute 31 may be defined by a plurality of conditions.
For example, if the user simply stops by the observation area 15, the visitor is estimated, but if the user stops by the observation area 15 after having been estimated to be a passenger, the visitor may be estimated to be a visitor. Further, for example, when the user stops at the counter 11, it is estimated that the person is a passenger or a see-off person, and then the person 20 who enters the inspection area is assumed to be a passenger, and the person 20 who enters the observation area 15 is assumed to be a see-off person. You may be made to do. In addition, at the arrival airport, the arrival of the passenger may be detected and estimated as the arrival person.

Further, the attribute 21 and the attribute 31 may be defined on the condition of the number of times of visit, the staying time, the elapsed time after the attribute estimation, and the like.
For example, it is estimated that a person 20 who stays at least A time and stays at least one of the toilet, chair, and observation area 15 but does not stop at the counter 11 and the inspection area is a suspicious person. Also good.
Further, for example, even a person 20 estimated to be a passenger or a see-off passenger may be estimated as a suspicious person if he / she stays for a long time thereafter. Similarly, even if it stops at the counter 11 and stays for a long time after that, it may be estimated that it is not a passenger but a visitor. Further, since it is common that the staying time is different between the international terminal and the domestic terminal, the rule 42 may be changed.

For example, a rule 42 as shown in FIG. 10 may be used.
According to the rule 42 shown in FIG. 10, in addition to the counter 11, the basic attribute 51 is estimated to be a passenger or a see-off passenger when the help desk 12, the inspection site entrance, and the mileage registration terminal are visited. In addition, when entering the inspection area, it is estimated that the basic attribute 51 is a passenger. In addition, when the user stops by the shopping area 13, the restaurant area 14, or the observation area 15, the basic attribute 51 is estimated to be a visitor.
At the same time, when the user visits the help desk 12, the additional attribute 52 is presumed to require help, and when the user visits the mileage registration terminal, the card holder is presumed. Also, if you stop by the information desk, it is estimated that you are unfamiliar with the airport as the additional attribute 52. If you stop by the shopping area 13, the restaurant area 14, or the observation area 15, you are interested in the additional attribute 52. Is estimated. Moreover, when it is in a chair, it is estimated that it is waiting.
Further, it is possible to estimate the additional attribute 52 that is hurry from the action of the person 20, for example, the running action, or to estimate the additional attribute 52 that is lost from the action that the person is hurrying. it can. Furthermore, it can be set to estimate the degree of urgency, such as hurrying up when the average speed is higher than the speed at which the person 20 stops.
That is, according to the rule 42 shown in FIG. 10, the additional attribute 52 may be estimated together with the basic attribute 51 by a certain action 41.

Furthermore, the rule 42 shown in FIG. 11 may be used together with the rule 42 shown in FIG. In FIG. 11, a rule 42 based on appearance is shown.
In the rule 42 shown in FIG. 11, when it is specified from the image data 40 that the vehicle is in a wheelchair, it is estimated that the person is a traffic weak person, and when it is specified that the baby is held or the child is taken. Presumed to have children.

In the above description, the airport is the target area 10. However, the target area 10 is not limited to an airport, and may be another area where many people gather, such as a shopping mall.
The rule 42 stored in the rule storage unit 160 needs to be set according to the target area 10. For example, at an airport, the additional attribute 52 is interested in shopping. However, in the case of a shopping mall, there is a high possibility that the person 20 who visited the children's clothing department has come to buy children's clothing. Appropriate.

  In the above description, the attribute estimation apparatus 100 includes the action specifying unit 120 and the group estimation unit 130. However, the action specifying unit 120 and the group estimation unit 130 may be provided not in the attribute estimation device 100 but in an external device such as the imaging device 200. And the attribute estimation apparatus 100 may acquire the action 41 of the person 20 which the action specific | specification part 120 specified from the external apparatus and the information of the group 30 which the group estimation part 130 estimated.

  Furthermore, the action specifying unit 120 specifies the person 20 in the target area 10 based on the audio information other than the image data 40 of the image data acquisition unit 110, the strength of the radio signal of the specific terminal, and the like, and the action 41 of the specified person 20 is determined. May be specified. Further, the group estimation unit 130 may operate so as to estimate the group 30 based on the behavior 41 of the person 20 identified by the behavior identification unit 120.

Based on FIG. 12, a hardware configuration example of the attribute estimation apparatus 100 according to the first and second embodiments will be described.
The attribute estimation apparatus 100 is a computer.
The attribute estimation device 100 includes hardware such as a processor 901, an auxiliary storage device 902, a memory 903, a communication device 904, an input interface 905, and a display interface 906.
The processor 901 is connected to other hardware via the signal line 910, and controls these other hardware.
The input interface 905 is connected to the input device 907 by a cable 911.
The display interface 906 is connected to the display 908 by a cable 912.

The processor 901 is an IC (Integrated Circuit) that performs processing. The processor 901 is, for example, a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or a GPU (Graphics Processing Unit).
The auxiliary storage device 902 is, for example, a ROM (Read Only Memory), a flash memory, or an HDD (Hard Disk Drive).
The memory 903 is, for example, a RAM (Random Access Memory).
The communication device 904 includes a receiver 9041 that receives data and a transmitter 9042 that transmits data. The communication device 904 is, for example, a communication chip or a NIC (Network Interface Card).
The input interface 905 is a port to which the cable 911 of the input device 907 is connected. The input interface 905 is, for example, a USB (Universal Serial Bus) terminal.
The display interface 906 is a port to which the cable 912 of the display 908 is connected. The display interface 906 is, for example, a USB terminal or an HDMI (registered trademark) (High Definition Multimedia Interface) terminal.
The input device 907 is, for example, a mouse, a keyboard, or a touch panel.
The display 908 is, for example, an LCD (Liquid Crystal Display).

The auxiliary storage device 902 includes the image data acquisition unit 110, the action specifying unit 120, the group estimating unit 130, the human attribute estimating unit 140, and the group attribute estimating unit 150 (hereinafter, the image data acquiring unit 110, the action specifying unit 120, A program that realizes the function of the group estimation unit 130, the human attribute estimation unit 140, and the group attribute estimation unit 150 (collectively referred to as “parts”) is stored.
This program is loaded into the memory 903, read into the processor 901, and executed by the processor 901.
Further, the auxiliary storage device 902 also stores an OS (Operating System).
Then, at least a part of the OS is loaded into the memory 903, and the processor 901 executes a program that realizes the function of “unit” while executing the OS.
Although one processor 901 is illustrated in FIG. 12, the attribute estimation apparatus 100 may include a plurality of processors 901. A plurality of processors 901 may execute a program for realizing the function of “unit” in cooperation with each other.
In addition, information, data, signal values, and variable values indicating the processing results of “part”, and information, data, signal values, and variable values stored in the rule storage unit 160 and the estimation information storage unit 170 are stored in the memory 903, The file is stored as a file in the auxiliary storage device 902 or a register or cache memory in the processor 901.

  The “part” may be provided as “circuitry”. Further, “part” may be read as “circuit”, “process”, “procedure”, or “processing”. “Circuit” and “Circuitry” include not only the processor 901 but also other types of processing circuits such as a logic IC or GA (Gate Array) or ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array). It is a concept to include.

  10 target area, 11 counter, 12 help desk, 13 shopping area, 14 restaurant area, 15 viewing area, 20 people, 21 attributes, 30 groups, 31 attributes, 40 image data, 41 actions, 42 rules, 51 basic attributes, 52 Additional attribute, 100 attribute estimation device, 110 image data acquisition unit, 120 action identification unit, 130 group estimation unit, 140 human attribute estimation unit, 150 group attribute estimation unit, 160 rule storage unit, 170 estimation information storage unit, 200 imaging device .

Claims (8)

A group attribute estimator that estimates the attributes of each group in which the person in the target area is divided into a plurality of groups, from the attributes of the people included in the group;
An attribute estimation apparatus comprising: a human attribute estimation unit that estimates an attribute of a person included in each group from an attribute of the group estimated by the group attribute estimation unit. The attribute estimation device according to claim 1, wherein the group attribute estimation unit estimates the attribute of each group from the attributes of some persons included in the group. The human attribute estimation unit estimates an attribute of a person included in each group from the attribute of the group estimated by the group attribute estimation unit and the behavior of the person,
The attribute estimation device according to claim 1, wherein the group attribute estimation unit estimates the attribute of the group from the attributes of the person included in the group estimated by the human attribute estimation unit for each group. The attribute estimation device according to claim 3, wherein the action of the person includes a place where the person stops. The attribute estimation device according to claim 3 or 4, wherein the action of the person includes an order of places where the person stops. The attributes include basic attributes and additional attributes,
When the attribute estimated for a person is a basic attribute, the human attribute estimation unit changes the basic attribute for the person estimated in the past to the estimated basic attribute, and the attribute estimated for the person is added. The attribute estimation device according to any one of claims 1 to 5, wherein when the attribute is an attribute, the estimated additional attribute is added to the additional attribute for the person. The attribute estimation device further includes:
An image data acquisition unit that acquires a plurality of image data obtained by imaging the target area at different times;
From a plurality of image data acquired by the image data acquisition unit, comprising an action specifying unit for specifying the action of a person in the target area,
The attribute estimation device according to claim 3, wherein the human attribute estimation unit estimates a human attribute from the behavior of the person specified by the behavior specifying unit. A group attribute estimation process for estimating the attributes of each group in which the person in the target area is divided into a plurality of groups, from the attributes of the persons included in the group;
The attribute estimation program which makes a computer perform the human attribute estimation process which estimates the attribute of the person contained in each said group from the attribute of the group estimated by the said group attribute estimation process.

Images