JP2023148447A - Advertising effectiveness estimation server, advertising effectiveness estimation method, and advertising effectiveness estimation program - Google Patents

Abstract

To more accurately estimate advertising effectiveness of advertisements outside a sensing range of a camera.SOLUTION: An advertising effectiveness estimation server includes: an acquisition unit that acquires a captured video captured by at least one camera capturing a capturing area including at least a part of a viewing area where advertising media can be viewed and first detection information regarding a person detected from the captured video; a model creation unit that creates an expanded estimation model that estimates advertising effectiveness of the advertising media in the viewing area based on the first detection information and second detection information regarding a person in the viewing area; and an advertising effectiveness estimation unit that estimates and outputs the advertising effectiveness using the expanded estimation model and the first detection information.SELECTED DRAWING: Figure 1

Description

The present disclosure relates to an advertising effectiveness estimation server, an advertising effectiveness estimation method, and an advertising effectiveness estimation program.

Patent Document 1 discloses a system that generates viewing information. The system includes first viewing information regarding one or more first images based on the first advertisement, and second viewing information regarding one or more second images based on the second advertisement selected based on the attribute information. , generates. The first image is a captured image of one or more people in a position that may be affected by the first advertisement. The second image is a captured image of one or more people in a position that may be affected by the second advertisement. Viewing information is generated including information on people who have one or more specific attribute information among the people at each advertising influence position.

Japanese Patent Application Publication No. 2020-144895

The present disclosure has been devised in view of the conventional circumstances described above, and provides an advertising effectiveness estimation server, an advertising effectiveness estimation method, and an advertising effectiveness estimation program that more accurately estimate the advertising effectiveness of advertisements outside the detection range of a camera. The purpose is to

The present disclosure includes a captured video captured by at least one camera that captures an imaging area including at least a part of a viewing area where an advertising medium can be viewed, and first detection information regarding a person detected from the captured video. and a model that creates an expanded estimation model that estimates advertising effectiveness of the advertising medium in the viewing area based on the first detection information and second detection information regarding the person in the viewing area. An advertising effectiveness estimation server is provided, comprising: a creation unit; and an advertising effectiveness estimation unit that estimates and outputs the advertising effectiveness using the expanded estimation model and the first detection information.

The present disclosure also provides an advertising effectiveness estimation method executed by one or more computers capable of communicating with at least one camera that captures an imaging area that includes at least a portion of a viewing area where advertising media can be viewed. , acquiring a captured video captured by the camera and first detection information regarding a person detected from the captured video, and combining the first detection information and second detection information regarding the person in the viewing area. an expanded estimation model for estimating the advertising effectiveness of the advertising medium in the viewing area based on the above information, and uses the created expanded estimation model and the first detection information to estimate and output the advertising effectiveness. We provide a method for estimating advertising effectiveness.

The present disclosure also provides an advertising effectiveness estimation program executed by one or more computers capable of communicating with at least one camera that captures an imaging area that includes at least a portion of a viewing area where advertising media can be viewed. the step of acquiring a captured video captured by the camera and first detection information regarding a person detected from the captured video; and second detection regarding the first detection information and a person in the viewing area. creating an expanded estimation model for estimating the advertising effectiveness of the advertising medium in the viewing area based on the information; and using the created expanded estimation model and the first detection information to estimate the advertising effectiveness. To provide an advertising effectiveness estimation program for realizing the steps of estimating and outputting.

According to the present disclosure, it is possible to estimate the advertising effect of an advertisement outside the detection range of the camera with higher accuracy.

A block diagram showing an example of the overall configuration of an advertising effectiveness estimation system according to Embodiment 1. Diagram explaining an example of defining a viewer A diagram illustrating the first detection area and the second detection area when the camera is placed near the advertising medium. A diagram illustrating the first detection area and the second detection area when the camera is not placed near the advertising medium. A block diagram illustrating an example of a functional configuration of a control unit of a server according to Embodiment 1. Flowchart showing an example of the operation procedure of the advertising effectiveness estimation system according to Embodiment 1 Flowchart showing an example of the procedure for creating an expanded estimation model of the advertising effectiveness estimation system according to the first embodiment Diagram illustrating an example of an estimation result screen that compares the actual value and estimated value of advertising effectiveness Diagram illustrating an example of an attribute-based estimation result screen for advertising effectiveness Diagram illustrating an example of the advertising effect estimation result screen by station and day of the week

EMBODIMENT OF THE INVENTION Hereinafter, each embodiment which specifically disclosed the structure and operation of the advertising effectiveness estimation server, the advertising effectiveness estimation method, and the advertising effectiveness estimation program according to this disclosure will be described in detail with reference to the drawings as appropriate. However, more detailed explanation than necessary may be omitted. For example, detailed explanations of well-known matters or redundant explanations of substantially the same configurations may be omitted. This is to avoid unnecessary redundancy in the following description and to facilitate understanding by those skilled in the art. The accompanying drawings and the following description are provided to enable those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter recited in the claims.

(Embodiment 1)
First, with reference to FIG. 1, the overall configuration of an advertising effectiveness estimation system 100 according to Embodiment 1 will be described. FIG. 1 is a block diagram showing an example of the overall configuration of an advertising effectiveness estimation system 100 according to the first embodiment.

The advertising effectiveness estimation system 100 calculates the advertising effectiveness of advertising media installed at a predetermined location by calculating the number of people who came into contact with at least one installed advertising medium (hereinafter referred to as the "number of people who can contact the advertisement"), The number of people who viewed the advertisement (hereinafter referred to as "number of advertisement viewers"), the number of people who watched the advertising medium (hereinafter referred to as "number of advertisement viewers"), the time spent gazing at the advertising medium (hereinafter referred to as "ad viewing time") This is a system that can estimate the following:

Note that the above-mentioned estimation targets such as the number of people who can contact the advertisement, the number of people who view the advertisement, the number of people who gaze at the advertisement, or the advertisement gaze time are merely examples, and are not limited thereto. The estimation target indicates advertising effectiveness and may be arbitrarily defined and set by the user. Furthermore, the estimation target may be used as an index or element for estimating advertising effectiveness. In other words, the advertising effectiveness estimation system 100 estimates a predetermined advertising effectiveness (for example, the visibility rate obtained by dividing the number of people who can contact the advertisement by the number of people who view the advertisement) using the estimation results of two or more estimation targets. You may.

The advertising effectiveness estimation system 100 uses one or more cameras to detect at least a part of the range where the advertising medium can be contacted or viewed (for example, first detection areas AR11 and AR21 shown in FIGS. 3 and 4). and estimate the number of people who can contact the advertisement, the number of people who view the advertisement, the number of people who look at the advertisement, etc. for at least one advertising medium based on the person information and number of people appearing in the captured image (captured video). Create a possible expanded estimation model. At this time, the person information and the number of people appearing in the captured image (captured video) are generated by visual measurement by the operator. Here, personal information and the number of people are generated from captured images (captured videos), but this is not limited to captured images (captured videos), and it is possible to contact advertising media at the site or use advertising media. A person within a visible range may be visually measured. Further, although human information is not essential for learning, by using human information such as gender and age, it is possible to estimate not only the number of people but also trends in the characteristics of the people. The advertising effectiveness estimation system 100 estimates (estimates) advertising effectiveness for each advertising medium using the created expanded estimation model, and outputs the estimated advertising effectiveness (estimation result).

The advertising effectiveness estimation system 100 includes one or more cameras C1, . . . , a terminal device P1, a network NW, a server S1, and a display device MN. Note that if the functions realized by the terminal device P1 can be realized by the cameras C1, . . . , the terminal device P1 may be omitted. Alternatively, the functions realized by the terminal device P1 may be integrated into the server S1.

Each of the one or more cameras C1,... captures an image of at least a portion of a range where at least one advertising medium can be contacted or visible (for example, the second detection area shown in FIGS. 3 and 4). is installed at a possible position, and images an imaging range (for example, the first detection area shown in FIGS. 3 and 4) that includes at least a part of the visible range of any advertising medium. Each of the cameras C1, . . . includes at least a lens (not shown) and an image sensor (not shown). Each of the cameras C1, . . . may use a lens capable of capturing a wide range of images, such as a wide-angle lens or a fisheye lens. The image sensor is a solid-state imaging device such as a CCD (Charged-Coupled Device) or a CMOS (Complementary Metal-Oxide-Semiconductor), and converts an optical image formed on an imaging surface into an electrical signal. Each of the cameras C1, . . . transmits a captured image (captured video) to the terminal device P1 or the server S1. The captured images (captured videos) captured by the cameras C1, . ), etc., and may be transmitted from another terminal device or PC to the terminal device P1 via the network NW at a necessary timing.

The terminal device P1 is connected to each of the cameras C1, . . . and the server S1 via the network NW by wireless communication or wired communication. The wireless communication referred to here is, for example, communication via a wireless LAN (Local Area Network) such as Wi-Fi (registered trademark). The terminal device P1 detects a person appearing in the captured video transmitted from each of the cameras C1, . . . and transmits detection information of the detected person to the server S1. The terminal device P1 includes a communication section 10, a control section 11, a memory 14, and an input section 16.

The communication unit 10 is connected to each of the one or more cameras C1, . . . and the network NW so as to enable wireless or wired communication, and executes data transmission and reception. The communication unit 10 outputs captured images transmitted from each of the one or more cameras C1, . . . to the control unit 11. Further, the communication unit 10 associates the detection information detected by the control unit 11 with the captured video, and transmits the associated information to the server S1 via the network NW.

The control unit 11 is configured using, for example, a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or an FPGA (Field Programmable Gate Array), and controls the operation of each part of the terminal device P1. The control unit 11 cooperates with the memory 14, refers to programs and data held in the memory 14, and executes the programs to implement the functions of each unit. Note that each unit referred to here is the person detection unit 12.

In addition, the control unit 11 monitors operation information (death and death information) of each of the cameras C1, ... and each part of the terminal device P1, and monitors the operation information (death and death information) of each of the cameras C1, ... and the terminal device P1, etc.) or manage the usage status of the advertising effectiveness estimation system 100 concluded with the operators of the cameras C1, . . . , etc.). The control unit 11 regularly or continuously monitors life and death information or manages the usage contract status of the advertising effectiveness estimation system 100, and transmits the monitoring results or the usage contract status of the advertising effectiveness estimation system 100 to the communication unit 10. and send it to the server S1.

The person detection unit 12 performs image analysis processing on the captured video (that is, each of the plurality of captured images forming the captured video) to detect a person. The person detecting unit 12 associates the detected detection information with the captured video, outputs the associated information to the communication unit 10, and causes the communication unit 10 to transmit the information to the server S1. The person detection unit 12 includes a face detection unit 13A and a person detection unit 13B. Note that the information on the person for each captured video may be received from the input unit 16 instead of the person detection unit 12. In this case, the information on the person for each captured video received from the input unit 16 is input by checking the captured video manually, and is sent to the expanded estimation model creation unit 22. Note that the person information for each captured video may be acquired from a terminal device other than the terminal device P1 (for example, a terminal device used by a user, etc.) and output to the expanded estimation model creation section 22.

Here, with reference to each of FIGS. 2 to 4, a description will be given of persons corresponding to the number of people who can contact an advertisement, the number of people who view an advertisement, and the number of people who look at an advertisement as examples of viewers who indicate advertising effectiveness. It should be noted that the definitions of the number of people who can contact an advertisement, the number of people who view an advertisement, and the number of people who watch an advertisement described below are examples, and it goes without saying that the definitions are not limited thereto.

FIG. 2 is a diagram illustrating an example of the definition of a viewer. FIG. 3 is a diagram illustrating the first detection area AR11 and the second detection area AR12 when the camera C11 is placed near the advertising medium AD11. FIG. 4 is a diagram illustrating the first detection area AR21 and the second detection area AR22 when the camera C21 is not placed near the advertising medium AD21. It goes without saying that each of the first detection area AR11 and the second detection area AR12 shown in FIGS. 3 and 4 is an example, and is not limited thereto.

First detection areas AR11 and AR21, which are examples of imaging areas, indicate detection ranges (ranges in which a person can be detected) of the cameras C11 and C12, respectively. Second detection areas AR12 and AR22, which are examples of visible areas, are outside the detection ranges of cameras C11 and C21, respectively, and indicate ranges in which a passing person can touch or visually recognize the advertising medium. Furthermore, detection areas AR10 and AR20, which are an example of the imaging area, are areas where the server S1 executes the enlarged estimation of advertising media, and include first detection areas AR11 and AR21, and second detection areas AR12 and AR22. The range includes.

The number of people who can come into contact with an advertisement indicates the number of people who pass within 10 meters of the advertisement surfaces AD12 and AD22 on which advertisements are displayed in the advertisement media AD11 and AD21.

The number of advertisement viewers is defined as a person passing within 10 meters from the advertising surfaces AD12, AD22 on which advertisements are displayed on the advertising media AD11, AD21, and within 120° of this person's visual field. Indicates the number of people who have been determined to have AD21.

The number of people watching an advertisement is defined as a person who passes within 5 meters from the advertising surface AD12, AD22 on which the advertisement is displayed on the advertising medium AD11, AD21, and the advertising medium is within 70° of the person's close visual field. It shows the number of people who were determined to be included in AD11 and AD21.

The face detection unit 13A performs image analysis processing on the captured image, and detects the faces of the people appearing in the captured image who correspond to the respective viewers for each advertising medium. The face detection unit 13A detects the direction of the detected person's face (hereinafter referred to as “face angle”), the direction of the line of sight (hereinafter referred to as “line of sight angle”), and the like. The face detection unit 13A also detects attribute information (eg, gender, age, etc.) of the detected person. Note that the viewer here includes the number of people who can contact the advertisement, the number of people who view the advertisement, the number of people who watch the advertisement, etc., but is not limited thereto, and may be arbitrarily defined and set by the operator.

The person detection unit 13B executes image analysis processing on the captured image, detects people appearing in the captured image, and counts the number of detected people. The person detection unit 13B also detects walking line information for each detected person based on the time-series changes in the position information of the person detected from the captured video.

Note that the walking line information referred to here is information indicating time-series changes in the positional information of a person detected from a captured video, and is information in which the positional information of a person is arranged in chronological order at a predetermined time or every predetermined frame. or the direction in which the person passes (for example, from right to left, from left to right, from the back) within each detection range (i.e., first detection area) of cameras C1,... It may also be information about the direction (toward the front, etc.).

The person detection unit 12 collects detection information detected by each of the face detection unit 13A and the person detection unit 13B (for example, the number of detected persons, face angle, line of sight angle, attribute information, position information, walking line information, etc.), actual measurement information input by the worker using the input unit 16 (e.g., number of people, face angle, line of sight angle, attribute information, position information, walking line information, etc.), and captured video. are associated with each other, output to the communication unit 10, and transmitted to the server S1. Details of the actual measurement information will be described separately later. Note that the actual measurement information is not essential, and may be omitted when the expanded estimation model has been created in advance by the expanded estimation model creation section 22.

Here, the actual measurement information is information actually measured by the worker operating the terminal device P1, and corresponds to the detection information detected by each of the face detection unit 13A and the person detection unit 13B of the person detection unit 12. It is. The worker uses the captured image (captured video) of the range targeted for expansion estimation of advertising media (for example, detection areas AR10 and AR20 shown in Figures 3 and 4) to visually check the advertising on-site. The number of people who correspond to the viewers for each medium, the face angle, line of sight angle, attribute information (e.g., gender, age, etc.) of these people, or the walking line information for each person, etc. are actually measured. Information is input into the input section 16. In this embodiment, the display section 15 is a part of the terminal device P1, but it may be provided as a separate device or may be omitted.

Note that the actual measurement information may be information corresponding to detection information detected by each of the face detection section 13A and the human detection section 13B. For example, when the face detection unit 13A and the person detection unit 13B each detect the number of people detected in the captured video and the face angle or line of sight angle of each person, the worker All you have to do is input the number of people and the angle of each person's face or line of sight.

In addition, when there are multiple cameras, the person detection unit 12 uses detection information, captured images, and identification information that can identify each of the cameras C1, . . . (for example, a camera ID assigned to each camera). ), installation area information, serial number, etc.) may be further associated and transmitted to the server S1.

The memory 14 includes, for example, a RAM (Random Access Memory) as a work memory used when executing each process of the control unit 11, and a ROM (Read Only Memory) that stores programs and data that define the operation of the control unit 11. and has. Data or information generated or acquired by the control unit 11 is temporarily stored in the RAM. A program that defines the operation of the control unit 11 is written in the ROM.

The memory 14 stores information on the installation positions and detection ranges (imaging ranges) of the cameras C1, . Information about people detected from images (number of people, etc.) is stored. The position information may be two-dimensional position information or three-dimensional position information. By storing three-dimensional position information, even if advertising media exist in overlapping positions, it is possible to identify each advertising medium based on the difference in height.

The display unit 15 is configured using a display such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence). The display unit 15 displays images captured by the cameras C1, . . . output from the control unit 11. Note that the display unit 15 may be connected as a separate external device.

The input unit 16 is a worker interface configured using, for example, a touch panel, a keyboard, a mouse, or the like. The input unit 16 converts the received operator operation of the terminal device P1 into an electrical signal (control command) and outputs it to the control unit 11. The input unit 16 receives input of actual measurement information (for example, the number of detected persons corresponding to each viewer, face angle, line of sight angle, attribute information, position information, walking line information, etc.) actually measured by the worker. . The input unit 16 outputs the received actual measurement information to the person detection unit 12 of the control unit 11. Note that the input unit 16 may be a touch panel configured integrally with the display unit 15.

The server S1 is connected to the terminal device P1 and the display device MN via the network NW so that data can be transmitted and received, respectively. The server S1 uses the detection information transmitted from the terminal device P1 and the captured video to estimate the number of viewers of at least one advertising medium in a predetermined viewing range including inside and outside the detection range of each of the cameras C1,... Create an expanded estimation model for The server S1 uses the created expanded estimation model to estimate the advertising effect (viewer) of at least one advertising medium in a predetermined viewing range including inside and outside the detection range of each of the cameras C1,... The expanded estimation result is generated and output to the display device MN for display. The server S1 includes a communication section 20, a control section 21, a memory 25, and an expanded estimation result database DB.

The communication unit 20, which is an example of an acquisition unit, is connected to the terminal device P1 and the display device MN via the network NW so as to be able to transmit and receive data, respectively. The communication unit 20 outputs captured images (captured videos) transmitted from each of the one or more cameras C1, . 21. At this time, the captured image (captured video) may be captured in real time or recorded. The communication unit 20 also displays each estimation result screen showing the expanded estimation results generated by the control unit 21 (for example, the estimation result screen SC1 shown in FIG. 8, the estimation result screen SC2 by person attribute shown in FIG. 9, or the estimation result screen SC2 shown in FIG. The station/day-of-week estimation result screen SC3 shown in FIG.

The control unit 21 is configured using, for example, a CPU, a DSP, or an FPGA, and controls the operation of each unit of the server S1. The control unit 21 cooperates with the memory 25, refers to programs and data held in the memory 25, and executes the programs to implement the functions of each unit. Note that each section referred to here is the expanded estimation model creation section 22 and the visual recognition effect measurement section 23, respectively.

The control unit 21 also detects viewers who exhibit advertising effectiveness, and various times (such as advertisement viewing time, ) may be generated to make it measurable. Learning to generate training data may be performed using one or more statistical classification techniques. Examples of statistical classification techniques include multiple regression analysis, linear classifiers, support vector machines, quadratic classifiers, kernel density estimation, and decision trees ( Decision Trees, Artificial Neural Networks, Bayesian Techniques and/or Networks, Hidden Markov Models ), Binary Classifiers, Multi-Class Classifiers ( Multi-Class Classifiers), Clustering Technique, Random Forest Technique, Logistic Regression Technique, Linear Regression ( Examples include Linear Regression Technique, Gradient Boosting Technique, and the like. However, the statistical classification techniques used are not limited to these. Here, the teacher data (learning data) is generated based on the information sent from the terminal device P1, but the teacher data (learning data) is generated based on the information directly sent from a device other than the terminal device P1 (for example, a client PC, etc.). Good too.

The expanded estimation model creation unit 22 generates one or more expanded estimation models that estimate advertising effects for at least one advertising medium (for example, the number of viewers, contact time with the advertising medium by viewers, viewing time, gaze time, etc.). create.

The visibility effect measurement unit 23 uses the viewer criteria stored in the memory 25 (specifically, the number of people who can contact the advertisement, the number of people who view the advertisement, or the number of people who look at the advertisement) and the number of people who watch the advertisement transmitted from the terminal device P1. The number of viewers for each advertising medium (specifically, the number of people who can contact the advertisement, the number of people who view the advertisement, or the number of people who gaze at the advertisement) for each advertising medium is calculated based on the person information of the person appearing in the captured video. This calculation result will be referred to as viewer information.

Here, with reference to FIG. 5, the functional configuration of the control unit 21 of the server S1 will be described. FIG. 5 is a block diagram illustrating an example of the internal configuration of the control unit 21 of the server S1 according to the first embodiment.

The expanded estimation model creation unit 22, which is an example of a model creation unit, uses one or more cameras to detect a range in which advertising media can be contacted or visible (for example, a second detection area shown in FIGS. 3 and 4). ), and the actual measurement information regarding the person appearing in the captured image (captured video) captured in the imaged area (for example, the first detection area shown in FIGS. 3 and 4) that includes at least a part of the visible range of the advertising medium. Estimating the advertising effect (for example, the number of viewers, the contact time to the advertising medium by viewers, the viewing time, the gaze time, etc.) for at least one advertising medium based on the detection information of the captured image (captured video)1. The above expanded estimation model will be created. The expanded estimation model referred to here is, for example, a model for calculating the number of people who can contact an advertisement 22A, a model for calculating the number of people who view an advertisement 22B, a model for calculating the number of people who watch an advertisement 22C, a model for calculating the number of hours of advertisement viewing 22D, or the like. The expanded estimation model created here is based on detection information corresponding to each captured image (captured video) of cameras C1,... ) (that is, the number of people who can contact the advertisement, the number of people who view the advertisement, the number of people who look at the advertisement, etc.) is interpolated.

The expanded estimation model created by the expanded estimation model creation section 22 will be explained. The number of people who can be exposed to an advertisement calculation model 22A is a model that estimates the number of people who can be exposed to an advertisement of at least one advertising medium using captured images (imaged images) captured by the cameras C1, . The number of advertisement viewers calculation model 22B is a model that estimates the number of advertisement viewers of at least one advertising medium using captured videos (captured images) captured by the cameras C1, . The number of advertisement viewers calculation model 22C is a model that estimates the number of advertisement viewers of at least one advertising medium using captured videos (captured images) captured by the cameras C1, . Further, the advertisement gaze time calculation model 22D is a model that estimates the gaze time of at least one advertisement medium by an advertisement viewer using captured videos (captured images) captured by the cameras C1, .

Note that the advertising gaze time referred to here is the time during which the person is gazing at the advertising medium, and is determined to be staying within the first detection area based on the person's walking line information. It may be the length of time the person stays in the room, or it may be the time during which the person's face angle and line of sight angle are facing the direction of the advertising medium.

The expanded estimation model creation unit 22 uses the actual measurement information and detection information transmitted from the terminal device P1 to create a model for calculating the number of people who can be exposed to an advertisement 22A, a model for calculating the number of people who view the advertisement 22B, and a model for calculating the number of people who watch the advertisement. 22C and an advertisement viewing time calculation model 22D are respectively created. The expanded estimation model creation unit 22 associates the various created expanded estimation models with the detection information used to create the expanded estimation models, and outputs the resulting correspondence to the visibility effect measurement unit 23 .

Further, the expanded estimation model creation unit 22 may output various expanded estimation models in association with each of the actual measurement information and detection information used to create the various expanded estimation models. Thereby, the visibility effect measurement unit 23 (described later) performs an estimation that compares the actual measurement value based on the actual measurement information acquired (input) via the input unit 16 and the expanded estimation result (that is, the estimated value) based on the detected information. A result screen SC1 (see FIG. 8) can be generated and output.

The creation process of various expanded estimation models will be explained below.

(Creation process of calculation model for number of people who can be contacted by advertisement)
The expanded estimation model creation unit 22 images a range where the advertising medium transmitted from the terminal device P1 can be contacted or visible (for example, the first and second detection areas shown in FIGS. 3 and 4). actual measurement information of the captured image (captured video), and a captured image (captured video) that captures an imaging range (for example, the first detection area shown in FIGS. 3 and 4) that includes at least a part of the visible range of the advertising medium. A correction coefficient for correcting the estimation result is calculated based on the detection information of , and the viewer standard of the people who can be exposed to the advertisement, and a model 22A for calculating the number of people who can be exposed to the advertisement is created. The expanded estimation model creation unit 22 associates the created model 22A for calculating the number of people who can be contacted with an advertisement with the detection information used to create the model 22A for calculating the number of people who can be contacted with an advertisement, and It is output to the possible number calculation unit 24A.

In the process of creating the model 22A for calculating the number of people who can be exposed to advertisements, the completed model 22A for calculating the number of people who can be exposed to advertisements may be associated with environmental information about the installation location and stored in the memory 25. Thereby, the server S1 can apply a similar advertisement contactable number calculation model 22A based on the environmental information of the installation location even in a case where actual measurement information does not exist.

The viewer standard here is a standard for detecting a predetermined viewer based on the definition of the viewer who is the estimation target (see FIG. 2). For example, the standard for calculating the number of people who can be contacted by an advertisement is a standard that defines detection conditions for detecting people who correspond to the number of people who can be contacted by an advertisement from each of actual measurement information and detection information, or captured video. , arbitrarily set by the operator. For example, in the example shown in FIG. 2, the standard for calculating the number of people who can come into contact with an advertisement is "the total number of people who passed within 10 meters of the advertisement surface."

The environmental correction coefficient referred to here is a coefficient that is set according to the environmental information of the installation location where the advertising medium is installed, and performs the weighting process of the advertising effect estimated by the created expanded estimation model. This is the coefficient for For example, the environmental correction coefficient is calculated from the relationship between actually measured information and detected information. In addition, the number of users using the installation location where the advertising media is installed, the amount of traffic passing through the installation location (number of people), the advertising area of the advertising media, the proportion of people's walking lines in each direction at the installation location, the advertising media coefficients set based on the environmental information of the visible area (area), the installation location of advertising media (e.g., stations, buildings, passages, roads, etc.), or based on heuristic analysis using these environmental information. A determined coefficient or the like may be provided. For example, if the advertising medium is installed inside a station, the environmental correction coefficient may be calculated based on the area within the station where the advertising medium can be seen by station users, the number of people using this station per day, or a predetermined amount of The determination may be made using information such as the number of users and the flow of people during the time period.

Note that, if the advertising contactable number calculation model 22A corresponding to the installation location of the advertising medium is not set, the expanded estimation model creation unit 22 generates a model that has environmental information similar to the environmental information of the installation location of this advertising medium. It is determined whether there is a place to install the advertising media. As for the environmental information, for example, the scale of the facility, the total number of users, the flow of people or the line of flow of people, the shape of the area in front of the advertisement, the number of residents, the number of visitors, etc. can be considered as criteria. When the expanded estimation model creation unit 22 determines that there is an installation location of another advertising medium that has environmental information similar to the environmental information of the installation location of the advertising medium, the expanded estimation model creation unit 22 calculates the number of people who can contact the advertisement corresponding to this other installation location. The calculation model 22A may be referred to and selected as the calculation model 22A for calculating the number of people who can be contacted by an advertisement of the advertising medium whose advertising effectiveness is to be estimated. In addition, the number of people who can come into contact with an advertisement model 22A is applied to locations where the detection information or viewer information of a captured image (captured video) that captures an imaging range that includes at least a part of the visible range of advertising media is similar. You may choose

(Creation process of calculation model for number of advertisement viewers)
The expanded estimation model creation unit 22 also generates actual measurement information of a range where the advertising medium transmitted from the terminal device P1 can be contacted or visible (for example, detection areas AR10 and AR20 shown in FIGS. 3 and 4). , detection information of a captured image (captured video) captured in an imaging range that includes at least a part of the visible range of advertising media (for example, the first detection area shown in FIGS. 3 and 4), and detection information of a captured image (captured video) that includes at least a part of the visible range of advertising media, and An advertisement viewer number calculation model 22B is created based on the user criteria. The expanded estimation model creation unit 22 associates the created advertisement viewer number calculation model 22B with the detection information used to create the advertisement viewer number calculation model 22B, and calculates the number of advertisement viewers of the expanded estimation unit 24. It is output to the calculation section 24B.

Here, the viewer standard for calculating the number of advertisement viewers is a standard that defines detection conditions for detecting people who correspond to the number of advertisement viewers from actual measurement information and detection information, or captured video. It is set arbitrarily by the operator. For example, in the example shown in FIG. 2, the standard for calculating the number of advertisement viewers is "the number of people within 10 meters from the advertisement surface and within the field of view (120 degrees)".

(Creating model for calculating the number of people watching advertisements)
The expanded estimation model creation unit 22 also generates actual measurement information of a range where the advertising medium transmitted from the terminal device P1 can be contacted or visible (for example, detection areas AR10 and AR20 shown in FIGS. 3 and 4). , detection information of a captured image (captured video) that captures an imaging range that includes at least a part of the visible range of advertising media (for example, the first detection area shown in FIGS. 3 and 4), and detection information of a captured image (captured video) that includes at least a part of the visible range of advertising media, and A model 22C for calculating the number of people watching advertisements is created based on the user criteria. The expanded estimation model creation unit 22 associates the created advertising viewer number calculation model 22C with the detection information used to create the advertising viewer number calculation model 22C, and calculates the number of advertising viewers from the expanded estimation unit 24. It is output to the calculation section 24C.

Here, the standard for calculating the number of people watching advertisements is the standard for defining detection conditions for detecting people corresponding to the number of people watching advertisements from actual measurement information and detection information, or from captured images. , arbitrarily set by the operator. For example, in the example shown in FIG. 2, the standard for calculating the number of people gazing at an advertisement is "the number of people within 10 meters from the advertisement surface and within the close visual field (70 degrees)".

(Creation process of advertisement gaze time calculation model)
The expanded estimation model creation unit 22 also generates actual measurement information of a range where the advertising medium transmitted from the terminal device P1 can be contacted or visible (for example, detection areas AR10 and AR20 shown in FIGS. 3 and 4). , detection information of a captured image (captured video) captured in an imaging range that includes at least a part of the visible range of advertising media (for example, the first detection area shown in FIGS. 3 and 4), and calculation of advertisement gaze time. An advertisement viewing time calculation model 22D is created based on the viewer criteria for viewing the advertisement. The expanded estimation model creation unit 22 associates the created advertising viewing time calculation model 22D with the detection information used to create the advertising viewing time calculation model 22D, and calculates the advertising viewing time of the expanded estimation unit 24. It is output to the calculation section 24D.

Here, the calculation formula for calculating advertisement gaze time is to detect a person corresponding to the number of advertisement gazers from each of the actual measurement information and detection information or from the captured video, and to detect the detected person from the first detection area. It is an arithmetic expression that defines the measurement conditions for measuring (calculating) the length of time a person stays in an advertising medium, or the time that a detected person's face angle and line of sight angle are facing the advertising medium. Set. In addition, the calculation formula for calculating advertisement gaze time is such that the time spent in the measured first detection area, the face angle, or the time when the line of sight angle is facing the advertising medium is determined by the predetermined time (e.g., If it is determined that it is longer than 3 seconds, 5 seconds, etc., it will be measured as this person's advertisement gaze time, and if it is determined that it is less than a predetermined time, it will not be measured as this person's advertisement gaze time. May be set.

As an example of the machine learning method described above, the expanded estimation model creation unit 22 sets, for example, information extracted from actual measurement information as an objective variable and information extracted from detected information as an explanatory variable, and performs multiple regression. Machine learning may be performed to detect a person who is a viewer by performing an analysis.

In this embodiment, when creating these expanded estimation models (specifically, the model for calculating the number of people who can be exposed to advertisements, the model for calculating the number of people who view advertisements, the model for calculating the number of people who watch advertisements, and the model for calculating the number of hours of advertisement viewing), Although the information transmitted from the device P1 is used, past numerical values stored in the memory 25 or the expanded estimation result database DB may also be used. In addition, in this embodiment, an enlarged model is created based on the actual measurement information, detection information, and viewer standards, but the visibility effect measurement unit 23 calculates viewer information for each of the actual measurement information and detection information in advance. However, it may be created based on the information of each viewer.

The visibility effect measurement unit 23, which is an example of the advertising effect estimation unit, stores the advertising effect (information such as the number of viewers and the number of hours) estimated by the expansion estimation unit 24 in the expansion estimation result database DB. The visibility effect measurement unit 23 also displays each estimation result screen (for example, the estimation result screen SC1 shown in FIG. 8, the estimation result screen SC1 shown in FIG. A personal attribute-specific estimation result screen SC2 shown in FIG. 10 or a station/day-of-week estimation result screen SC3 shown in FIG. 10) is generated and transmitted to the display device MN for output.

The visibility effect measurement unit 23 realizes the respective functions of a calculation unit 24A for the number of people who can come into contact with an advertisement, a calculation unit 24B for the number of people who view an advertisement, a calculation unit 24C for the number of people who watch an advertisement, and a calculation unit 24D for the number of hours of advertisement viewing. Calculate the number of people or various times.

The number of people who can be exposed to advertisements calculation section 24A uses the number of people who can be exposed to advertisements calculation model 22A outputted from the expanded estimation model creation section 22 and the detection information used to create the number of people who can be exposed to advertisements calculation model 22A. , calculate the number of people who can be exposed to the advertisement. For example, the number of people who can be exposed to an advertisement calculation unit 24A calculates the number of people who can be exposed to an advertisement based on the number of people who can be exposed to an advertisement calculation model 22A and the number of people detected within the first detection area.

The advertisement viewer number calculation unit 24B uses the advertisement viewer number calculation model 22B output from the expanded estimation model creation unit 22 and the detection information used to create the advertisement viewer number calculation model 22B. Calculate the number of people.

For example, the advertisement viewer number calculation unit 24B calculates the number of advertisement viewers using the advertisement viewer number calculation model 22B and the face angle or gaze angle of a person detected within the first detection area as a reference direction. Model 22B, the number of people detected for whom the advertising medium is located within a 120° field of view centered on the reference direction (that is, the advertising medium is located within 120° of the detected person's field of view), and the advertising surface of the advertising medium. The number of advertisement viewers may be calculated using walking line information of a person passing through an area within 120 degrees with respect to the center point of . Note that the correction coefficient may be determined by heuristic analysis based on walking line information of a person passing through an area of 120° from the center of the advertising surface of the advertising medium.

The advertisement gazer number calculation unit 24C uses the advertisement gazer number calculation model 22C output from the expanded estimation model creation unit 22 and the detection information used to create the advertisement gazer number calculation model 22C. Calculate the number of people.

For example, the advertisement gazer number calculation unit 24C uses the advertisement gazer number calculation model 22C and the face angle or line of sight angle of a person detected within the first detection area as a reference direction, and uses the advertisement gazer number calculation model 22C as a reference direction. The advertising medium is located within a 70° close field of view (that is, the advertising medium is located within a 70° close field of view of the detected person). The number of gazers may be calculated, or the advertisement gazer number calculation model 22C, the number of people detected for whom the advertising medium is located within a 70° close field of view centered on the reference direction, and the center of the advertising surface of the advertising medium may be calculated. The number of people gazing at an advertisement may be calculated using walking line information of a person passing within a 70° close field of view based on the point. Note that the correction coefficient may be determined by multiple regression analysis of detection information of a person whose advertising medium is located within a 70° close field of view, or may be determined by a multiple regression analysis of detection information of a person who is located within a 70° close field of view, or may be determined by a multiple regression analysis of detection information of a person who is located within a 70° field of view of the advertising medium. It may be determined by heuristic analysis based on walking line information.

The advertisement gaze time calculation unit 24D uses the advertisement gaze time calculation model 22D output from the expanded estimation model creation unit 22 and the detection information used to create the advertisement gaze time calculation model 22D. Calculate (measure) time.

The memory 25 as an example of a storage unit includes, for example, a RAM as a work memory used when executing each process of the control unit 21, and a ROM that stores programs and data that define the operation of the control unit 21. Data or information generated or acquired by the control unit 21 is temporarily stored in the RAM. A program that defines the operation of the control unit 21 is written in the ROM. The memory 25 stores each of the various expanded estimation models created by the expanded estimation model creation section 22 for each advertising medium. The memory 25 also stores viewer criteria used to create various expanded estimation models, information on the created expanded estimation models, and the like. Note that part or all of this information may be stored in the memory 14 of the terminal device P1 or an external memory.

The expanded estimation result database DB includes a storage device including a semiconductor memory such as a RAM and a ROM, and a storage device such as an SSD (Solid State Drive) or an HDD. The expanded estimation result database DB stores expanded estimation results estimated by the visibility effect measurement unit 23.

The display device MN is connected to the server S1 so as to be able to transmit and receive data. Note that the display device MN shown in FIG. 1 is an example connected to the server S1 via the network NW so as to be able to communicate wirelessly, but it is not a display that is connected to the server S1 so as to be capable of wired communication. It's okay. The display device MN is a terminal device such as a PC, a notebook PC, a tablet terminal, or a smartphone, which has a display such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence). The display device MN displays each estimation result screen transmitted from the server S1 (for example, the estimation result screen SC1 shown in FIG. 8, the estimation result screen SC2 by person attribute shown in FIG. 9, or the estimation result by station/day of the week shown in FIG. 10). screen SC3, etc.) is output (displayed).

The display device MN is also capable of accepting user operations, and based on the user operations, the display device MN also performs a layout change operation of the estimation result screen transmitted from the server S1 (for example, changing the graph type, changing the layout displayed on the estimation result screen). number of advertising media, etc.), and selection operations related to display targets displayed on the estimation result screen (for example, number of viewers, time, attribute information of a person, day of the week, location of advertising media, etc.).

Note that in the above description, an example will be described in which the estimation result screen is generated by the server S1 and output (displayed) by the display device MN, but the estimation result screen is not limited to this. For example, the estimation result screen may be generated by the display device MN. In such a case, the server S1 stores the expanded estimation results, information about the advertising media that has been expanded and estimated, information on the location where the advertising media is installed, and information used in the expanded estimation process (for example, captured video, actual measurement information, detected information, etc.). The display device MN may be able to generate the estimation result screen by transmitting data used to generate the estimation result screen, such as information, etc., to the display device MN.

Next, with reference to FIG. 6, the operating procedure of the advertising effectiveness estimation system 100 will be described. FIG. 6 is a flowchart illustrating an example of the operation procedure of the advertising effectiveness estimation system 100 according to the first embodiment. Note that the processing of steps St1 to St2 shown in FIG. 6 may be executed by each of the cameras C1, . . . . Further, the number of cameras may be at least one.

Each of the cameras C1, . (For example, each of the first detection areas AR11 and AR21 shown in FIGS. 3 and 4) is imaged. Each of the cameras C1, . . . transmits captured images to the terminal device P1.

Note that the advertising effectiveness estimation system 100 may image a first detection area that includes at least a part of the second detection area of each of the plurality of advertising media that is the estimation target range using one camera, or Each of the plurality of cameras may image the first detection area that includes at least a portion of the second detection area of one advertising medium, which is the estimation target range. Further, the advertising effectiveness estimation system 100 may image a first detection area including the entire second detection area of one or more advertising media, which is an estimation target range, using each of one or more cameras.

In addition, when estimating the number of people viewing advertisements, the number of people gazing at advertisements, or the length of time they gazed at advertisements, each of the cameras C1, ... should be installed at a position where it is possible to image a person's face (for example, around the advertising medium, etc.). Although it is desirable that the

The control unit 11 of the terminal device P1 acquires captured images transmitted from each of the one or more cameras C1, . . . (St1). The person detection unit 12 executes a person detection process from the acquired captured video, and collects detection information (for example, the number of people detected for each viewer, face angle, line of sight angle, attribute information, position information, walking distance, etc.). action line information, etc.) is acquired (St2). The input unit 16 also receives input of actual measurement information by the operator. The control unit 11 acquires the actual measurement information accepted by the input unit 16 (St2). Note that when the expanded estimation model has been created in advance by the expanded estimation model creation unit 22, the process of acquiring actual measurement information is not essential and may be omitted. The control unit 11 associates the acquired actual measurement information with the detection information and transmits them to the server S1.

The control unit 21 of the server S1 acquires the actual measurement information and detection information transmitted from the terminal device P1, and executes an expanded estimation model creation process using the acquired actual measurement information and detection information (St3). . Specifically, at least one of the actual measurement information regarding the person in the range where the advertising medium can be contacted or visible (for example, detection areas AR10 and AR20 shown in FIGS. 3 and 4) and the visible range of the advertising medium is This is detection information based on a captured image (captured video) that captures an imaging range (for example, the first detection area shown in FIGS. 3 and 4) including the area. The control unit 21 executes an expanded estimation model creation process using the detection information acquired through the person detection process and the detection information acquired from the terminal device P1.

The control unit 21 of the server S1 uses the created expanded estimation models (for example, a model for calculating the number of people who can contact an advertisement 22A, a model for calculating the number of people who view an advertisement 22B, a model for calculating the number of people who watch an advertisement 22C, a model for calculating the number of hours of advertisement viewing 22D, etc.). ) and the acquired detection information, an expanded estimation process for each advertising medium is executed (St4). The control unit 21 generates each estimation result screen for each advertising medium based on the number of viewers, viewing time, etc. obtained by the expanded estimation process, and transmits it to the display device MN.

The display device MN outputs (displays) each estimation result screen transmitted from the server S1 (St5).

As described above, the advertising effectiveness estimation system 100 according to the first embodiment can estimate (estimate) the advertising effectiveness of the advertising medium even if the advertising medium is not located within the detection range of the camera. Therefore, if at least a part of the second detection area of the advertising medium is included within the detection range of the already installed camera (that is, the first detection area), the advertising effectiveness estimation system 100 is able to add a new camera to the camera. The advertising effectiveness of advertising media can be estimated (estimated) without installing cameras or changing the camera installation position. Furthermore, the advertising effectiveness estimation system 100 can estimate (estimate) the advertising effectiveness of advertising media corresponding to each of the plurality of second detection areas included within the detection range of the camera (that is, the first detection area). Therefore, it is possible to estimate (estimate) the advertising effectiveness of multiple advertising media using one camera.

Next, with reference to FIG. 7, a procedure for creating an expanded estimation model of the advertising effectiveness estimation system 100 will be described. FIG. 7 is a flowchart illustrating an example of an expanded estimation model creation procedure of the advertising effectiveness estimation system 100 according to the first embodiment. Note that the expanded estimation model creation procedure shown in FIG. 7 is executed in the process of step St3 shown in FIG. 6.

The control unit 21 of the server S1 acquires the actual number of viewers actually measured visually by the worker and the detected number of viewers detected by each of the cameras C1, ... (St31). .

The expanded estimation model creation unit 22 sets the number of people detected by the person detection process as an explanatory variable and the number of people actually measured visually by the worker as an objective variable, and performs multiple regression analysis (that is, machine learning). (St32). It goes without saying that the method for creating the expanded estimation model is not limited to this. For example, the coefficients may be determined using heuristics rather than machine learning.

The expanded estimation model creation unit 22 executes machine learning and creates one or more expanded estimation models (for example, a model 22A for calculating the number of people who can be exposed to advertisements, a model 22B for calculating the number of people viewing advertisements, a model 22C for calculating the number of people watching advertisements, and a model for calculating the number of people watching advertisements). A time calculation model 22D, etc.) is created (St33).

Next, the estimation result screen SC1 will be explained with reference to FIG. FIG. 8 is a diagram illustrating an example of an estimation result screen SC1 that compares the measured value and the estimated value of advertising effectiveness.

The estimation result screen SC1 is an example of the expanded estimation result, and shows the average number of people who can be contacted by advertisements per day (in other words, an estimated value) estimated (estimated) using the number of people who can be contacted by advertisements calculation model 22A, This is a bar graph for comparing the average number of people who can be exposed to advertisements per day (in other words, the actual measurement value) as measured by visual confirmation by workers, and the vertical axis shows the average number of people who can be exposed to advertisements per day. and the day of the week is shown on the horizontal axis. Thereby, the advertising effectiveness estimation system 100 can visualize the difference between the actual value and the estimated value to the user.

Next, the person attribute-based estimation result screen SC2 will be described with reference to FIG. FIG. 9 is a diagram illustrating an example of the advertising effectiveness estimation result screen SC2 by person attribute.

The estimation result screen SC2 by person attribute is an example of the expanded estimation result, and shows the average number of people who can be exposed to advertisements per day (in other words, the estimated value) estimated (estimated) using the number of people who can be exposed to advertisements calculation model 22A. ), based on the person's age information, which is the attribute information of the person, and the average number of advertisement viewers estimated (estimated) using a bar graph classified by person attribute and time of day and the number of advertisement viewers calculation model 22B. It is shown as a line graph showing the visibility rate of the advertising medium obtained by dividing by the number of people who can contact the advertisement. Note that the person attribute-specific estimation result screen SC2 shown in FIG. 9 is an expanded estimation result in which persons whose attribute information is "male" are classified by age group and time zone.

In the person attribute-specific estimation result screen SC2, the vertical axis indicates the average number of people who can contact the advertisement per day and the visibility rate of the advertising medium, and the horizontal axis indicates the time period. Thereby, the advertising effectiveness estimation system 100 can visualize to the user the expanded estimation results (advertising effectiveness) by time period and by person attribute.

Furthermore, the person attribute-based estimation result screen SC2 includes display target selection fields SL1, SL2, and SL3 that can accept designation operations regarding the expanded estimation results displayed on the person attribute-based estimation result screen SC2.

The display target selection field SL1 accepts an operation for specifying an advertising medium to be displayed. In the example shown in FIG. 9, the display target selection field SL1 specifies advertising media installed at "B station".

The display target selection field SL2 accepts an operation for specifying a weekday or a holiday as the detection date and time of the detection information used for the expanded estimation. In the example shown in FIG. 9, "weekdays" is specified in the display target selection column SL2.

The display target selection field SL3 accepts an operation for specifying the day of the week or the date and time as the detection date and time of the detection information used in the expanded estimation. In the example shown in FIG. 9, "all" (that is, Monday to Sunday) is specified in the display target selection field SL3. Specifically, if you check only Monday, only Monday's data will be displayed. Note that the definition of holidays can be set arbitrarily (for example, holidays include or do not include holidays, etc.).

Next, with reference to FIG. 10, the station/day-of-week estimation result screen SC3 will be described. FIG. 10 is a diagram illustrating an example of the advertising effect estimation result screen SC3 by station and day of the week.

The estimation result screen SC3 by station and day of the week is an example of the expanded estimation result, and shows the average number of people who can be exposed to advertisements per day (in other words, estimated) using the calculation model 22A for calculating the number of people who can be exposed to advertisements. The visibility rate of the advertising medium, which is obtained by dividing the average number of advertisement viewers estimated (estimated) using the bar graph classified by day of the week and the number of advertisement viewers calculation model 22B, by the average number of people who can contact the advertisement. This is shown in the line graph shown below. In addition, the station/day-of-week estimation result screen SC3 shown in FIG. Display side by side. This allows the user to compare at a glance the enlarged estimation results ST1 and ST2 of two advertising media installed at different locations.

In the estimation result screen SC3 by station and day of the week, the vertical axis shows the average number of people who can contact the advertisement per day and the visibility rate of advertising media, and the horizontal axis shows the time of day. Thereby, the advertising effectiveness estimation system 100 can visualize to the user the expanded estimation results (advertising effectiveness) by time period and by person attribute.

As described above, the server S1 according to Embodiment 1 (an example of an advertising effectiveness estimation server) operates in a first detection area AR11 that includes at least a portion of detection areas AR10 and AR20 (an example of a viewing area) where advertising media can be viewed. , AR21 (an example of an imaging area), the communication unit 20 (of the acquisition unit) acquires the captured video captured by at least one camera C1, . . . An expanded estimation model is created to estimate the advertising effectiveness of advertising media AD11, AD21 in detection areas AR10, AR20 based on the first detection information and the second detection information regarding people in detection areas AR10, AR20. An expanded estimation model creation unit 22 (an example of a model creation unit) that estimates and outputs advertising effectiveness using the expanded estimation model and the first detection information, and a visibility effect measurement unit 23 (an example of an advertising effectiveness estimation unit) ) and. Note that the one or more computers referred to here include, for example, at least the server S1.

As a result, the server S1 according to the first embodiment can detect the captured images captured by the cameras C1, . . . and the captured images (that is, within the detection range of the cameras C1, . Using an expanded estimation model created using the detection information detected from the detection information detected from the detection range of the camera C1,... It is possible to estimate the advertising effectiveness of advertising media for people who pass through the following range. In other words, the server S1 detects the detection range of the cameras C1, ... even if the detection range (the first detection areas AR11, AR21) does not include the entire range of the detection areas AR10, AR20 where advertising media can be viewed. The advertising effectiveness of advertisements outside the detection range can be estimated with higher accuracy.

Further, as described above, the first detection information and the second detection information in the first embodiment include the number of detected people. The communication unit 20 acquires the first detection information. The expanded estimation model creation unit 22 uses the first detection information and the second detection information to generate an expanded estimation model for estimating the advertising effectiveness of advertising media. As a result, the server S1 according to the first embodiment uses the correct number of people appearing in the captured video and the detected number of people appearing in the captured video detected by the person detection unit 12 to detect the number of people detected by the person detection unit 12. It is possible to generate an expanded estimation model that has learned the detection error of . Therefore, the server S1 can estimate the advertising effect of the advertisement outside the detection range of the camera with higher accuracy.

Further, as described above, the visibility effect measurement unit 23 of the server S1 according to the first embodiment uses the expanded estimation model and the first detection information to obtain an estimated value (an example of the first advertising effect, for example, The estimated value of the number of people who can contact the advertisement shown in FIG. 8 is estimated and output. Thereby, the server S1 according to the first embodiment can output the estimated advertising effect (for example, the estimated number of people who can contact the advertisement) to the user.

Further, as described above, the first detection information acquired from the server S1 according to the first embodiment includes position information of a person. The expanded estimation model creation unit 22 generates different advertising effects (for example, the number of people who can contact the advertisement, the advertisement Create an expanded estimation model to estimate the number of viewers, number of people watching advertisements, etc. The visual recognition effect measurement unit 23 estimates and outputs the advertising effect using the expanded estimation model and the first detection information. Thereby, the server S1 according to the first embodiment can estimate each of the plurality of advertising effects with higher accuracy using the expanded estimation model.

Further, as described above, the first detection information acquired from the server S1 according to the first embodiment includes the position information of the person and the face angle information of the person. The enlarged estimation model creation unit 22 extracts the captured video, the position information of the person relative to the advertising medium (e.g., the position of the person relative to the advertising medium, the distance between the advertising surface of the advertising medium and the person), and the face angle information of the person (e.g. , a person's face angle, line of sight angle, etc.), an expanded estimation model is created to estimate different advertising effects (for example, the number of people viewing the advertisement or the number of people gazing at the advertisement). The visibility effect measuring unit 23 estimates and outputs the advertising effect (for example, the number of people viewing the advertisement or the number of people watching the advertisement) using the expanded estimation model and the first detection information. Thereby, the server S1 according to the first embodiment can estimate each of the plurality of advertising effects with higher precision using the expanded estimation model. Note that the face angle may be the line of sight angle or both.

Further, as described above, the detection information acquired by the server S1 according to the first embodiment includes the position information of the person and the flow line information of the person in the first detection area. The expanded estimation model creation unit 22 uses a plurality of models to estimate different advertising effects (for example, the number of advertisement viewers or the number of advertisement viewers) based on the distance between the advertising medium and the person and the flow line information of the person. Create an expanded estimation model. The visibility effect measuring unit 23 estimates and outputs a plurality of advertising effects using a plurality of enlarged estimation models and detection information. Thereby, the server S1 according to the first embodiment can estimate each of the plurality of advertising effects with higher accuracy using the expanded estimation model.

Further, as described above, the first detection information acquired by the server S1 according to the first embodiment includes position information of a person. The expanded estimation model creation unit 22 measures the length of time a person stayed in the viewing area for each detected person based on the location information of the person, and uses the measured length of stay to calculate advertising effectiveness based on the length of stay. Create an expanded estimation model for estimating (for example, gaze time, etc.). The visibility effect measurement unit 23 estimates and outputs the advertising effect using the expanded estimation model and the length of stay. Thereby, the server S1 according to the first embodiment can estimate the gaze time and the like as an example of advertising effectiveness with higher precision using the expanded estimation model.

In addition, as described above, the server S1 according to the first embodiment uses the environmental information of the installation location where the advertising medium is installed and the expanded estimation model that is set for each installation location and measures advertising effectiveness based on the environmental information. The device further includes a memory 25 (an example of a storage unit) that stores . The visibility effect measurement unit 23 estimates and outputs the advertising effect using an expanded estimation model corresponding to the installation location of the advertising medium. Thereby, the server S1 according to the first embodiment can estimate the advertising effect with higher accuracy by using the expanded estimation model based on the environmental information of the installation location where the advertising medium is installed.

Further, as described above, when the visibility effect measuring unit 23 in the server S1 according to the first embodiment determines that the expanded estimation model corresponding to the installation location of the advertising medium is not stored in the memory 25, the environmental information of the installation location Another installation location having environmental information similar to that of the other installation location is selected, and an expanded estimation model corresponding to the other selected installation location is used to estimate and output the advertising effect. As a result, the server S1 according to the first embodiment applies the environmental correction coefficient used for estimating the advertising effectiveness of advertising media installed in similar installation locations to the expanded estimation model for estimating the advertising effectiveness of advertising media. By doing so, advertising effectiveness can be estimated with higher accuracy.

Further, as described above, the first detection information acquired by the server S1 according to the first embodiment includes attribute information of a person. The visibility effect measurement unit 23 estimates the advertising effect using the expanded estimation model and the first detection information, and outputs the estimated advertising effect for each person's attribute information (for example, gender, age group, etc.) . Thereby, the server S1 according to the first embodiment can estimate the advertising effect for each attribute information using the expanded estimation model. That is, the user can more effectively select advertisement content to be posted on each advertising medium based on the estimated advertising effectiveness.

Further, as described above, the first detection information acquired by the server S1 according to the first embodiment includes detection time information at which a person was detected. The visibility effect measuring unit 23 estimates the advertising effectiveness using the expanded estimation model and the first detection information, and outputs the estimated advertising effectiveness for each time period based on the detection time information. Thereby, the server S1 according to the first embodiment can estimate the advertising effectiveness for each time period using the expanded estimation model. That is, the user can more effectively select advertisement content to be posted on each advertising medium based on the estimated advertising effectiveness.

Although various embodiments have been described above with reference to the accompanying drawings, the present disclosure is not limited to such examples. It is clear that those skilled in the art can come up with various changes, modifications, substitutions, additions, deletions, and equivalents within the scope of the claims, and It is understood that it falls within the technical scope of the present disclosure. Further, each of the constituent elements in the various embodiments described above may be arbitrarily combined without departing from the spirit of the invention.

The present disclosure is useful as a presentation of an advertising effectiveness estimation server, an advertising effectiveness estimation method, and an advertising effectiveness estimation program that more accurately estimate the advertising effectiveness of advertisements outside the detection range of a camera.

10, 20 Communication section 11, 21 Control section 12 Person detection section 13A Face detection section 13B Person detection section 14, 25 Memory 15 Display section 16 Input section 22 Expanded estimation model creation section 23 Visibility effect measurement section 24 Expanded estimation section 100 Advertising effect Estimation system AD11, AD21 Advertising medium AR11, AR21 First detection area AR12, AR22 Second detection area C1, C11, C21 Camera MN Display device NW Network

Claims (13)

Obtaining a captured image captured by at least one camera that captures an image capturing area including at least a part of a viewing area where advertising media can be viewed, and first detection information regarding a person detected from the captured image. Department and
a model creation unit that creates an expanded estimation model that estimates an advertising effect of the advertising medium in the viewing area based on the first detection information and second detection information regarding the person in the viewing area;
an advertising effectiveness estimation unit that estimates and outputs the advertising effectiveness using the expanded estimation model and the first detection information;
Advertising effectiveness estimation server. The first detection information and the second detection information include the number of people detected,
The acquisition unit includes:
obtaining the first detection information;
The model creation section includes:
generating the expanded estimation model for estimating the advertising effectiveness of the advertising medium using the first detection information and the second detection information;
The advertising effectiveness estimation server according to claim 1. The advertising effectiveness estimation department is
estimating and outputting a first advertising effect using the expanded estimation model and the first detection information;
The advertising effectiveness estimation server according to claim 2. The first detection information includes position information of the person,
The model creation section includes:
creating an expanded estimation model that estimates different advertising effects based on the distance between the advertising medium and the person;
The advertising effectiveness estimation department is
estimating and outputting the advertising effect using the expanded estimation model and the first detection information;
The advertising effectiveness estimation server according to claim 1. The first detection information includes position information of the person and face angle information of the person,
The model creation section includes:
creating an expanded estimation model for estimating different advertising effects based on position information of the person with respect to the advertising medium and face angle information of the person;
The advertising effectiveness estimation department is
estimating and outputting the advertising effect using the expanded estimation model and the first detection information;
The advertising effectiveness estimation server according to claim 1. The first detection information includes position information of the person and flow line information of the person in the imaging area,
The model creation section includes:
creating an expanded estimation model that estimates different advertising effects based on the distance between the advertising medium and the person and the flow line information of the person;
The advertising effectiveness estimation department is
estimating and outputting the advertising effect using the expanded estimation model and the detected information;
The advertising effectiveness estimation server according to claim 1. The first detection information includes position information of the person,
The model creation section includes:
Measuring the length of time the person stayed in the viewing area based on the location information of the person,
Using the measured stay time, create an expanded estimation model that estimates advertising effectiveness based on the stay time,
The advertising effectiveness estimation department is
estimating and outputting the advertising effect using the expanded estimation model and the stay time;
The advertising effectiveness estimation server according to claim 1. further comprising a storage unit that stores environmental information of the installation location where the advertising medium is installed, and the expanded estimation model that is set for each installation location and measures the advertising effectiveness based on the environmental information,
The advertising effectiveness estimation department is
estimating and outputting the advertising effect using the expanded estimation model corresponding to the installation location of the advertising medium;
The advertising effectiveness estimation server according to claim 1. The advertising effectiveness estimation department is
If it is determined that the expanded estimation model corresponding to the installation location of the advertising medium is not stored in the storage unit, selecting another installation location having environmental information similar to the environmental information of the installation location; estimating and outputting the advertising effect using the expanded estimation model corresponding to the other selected installation location;
The advertising effect estimation server according to claim 8. The first detection information includes attribute information of the person,
The advertising effectiveness estimation department is
estimating the advertising effect using the expanded estimation model and the first detection information;
outputting the estimated advertising effect for each attribute information of the person;
The advertising effectiveness estimation server according to claim 1. The first detection information includes detection time information when the person was detected,
The advertising effectiveness estimation department is
estimating the advertising effect using the expanded estimation model and the first detection information;
outputting the estimated advertising effectiveness by time zone based on the detection time information;
The advertising effectiveness estimation server according to claim 1. An advertising effectiveness estimation method executed by one or more computers capable of communicating with at least one camera that captures an imaging area that includes at least a portion of a viewing area where advertising media can be viewed, the method comprising:
acquiring a captured video captured by the camera and first detection information regarding a person detected from the captured video;
creating an expanded estimation model for estimating the advertising effectiveness of the advertising medium in the viewing area based on the first detection information and second detection information regarding the person in the viewing area;
estimating and outputting the advertising effect using the created expanded estimation model and the first detection information;
Advertising effectiveness estimation method. An advertising effectiveness estimation program executed by one or more computers capable of communicating with at least one camera that images an imaging area including at least a part of a viewing area where advertising media can be viewed,
acquiring a captured video captured by the camera and first detection information regarding a person detected from the captured video;
creating an expanded estimation model for estimating the advertising effectiveness of the advertising medium in the viewing area based on the first detection information and second detection information regarding the person in the viewing area;
estimating and outputting the advertising effect using the created expanded estimation model and the first detection information;
Advertising effectiveness estimation program.

Images