JP7247784B2 - Interest estimation system - Google Patents

Description

The present disclosure relates to interest estimation systems.

A system is known that detects whether or not a person is seated on a seat in order to grasp the seating situation in a facility where content is provided (see Patent Document 1).

JP-A-2003-61779

The system described above has a function of assisting the user in securing a seat, but does not have a function of estimating the interest of the seated user in the content after the start of provision of the content.

One aspect of the present disclosure aims to provide an interest estimation system capable of estimating a user's interest in content.

One aspect of the present disclosure is a detection device (2) that detects the posture of a user of content, acquires the direction of the user's line of sight from the posture detected by the detection device (2), , and a processing device (4) for estimating the level of user's interest in content.

According to such a configuration, it is possible to estimate the user's level of interest in the content (that is, the degree of gaze) based on the movement of the user's line of sight. Based on this result, it is possible to improve the user's interest by changing the contents of the content, the method of providing the content, and the like. In addition, since the user's line of sight is used for estimation, the level of interest can be estimated without using language information (that is, user's utterance content) that is affected by analysis accuracy.

In one aspect of the present disclosure, the processing device (4) may estimate the user's level of interest in the content based on the frequency with which the line-of-sight direction changes. According to such a configuration, it is possible to determine the degree of attention of the user to the content with higher accuracy than in the case of determining only the maintenance time of the line-of-sight direction.

In one aspect of the present disclosure, the processing device (4) estimates the level of interest of the user in the content based on the time from the start of the visual guidance by the content until the direction of the visual line coincides with the direction of the visual guidance. may According to such a configuration, it is possible to easily and accurately determine the level of interest in content that includes visual guidance.

In one aspect of the present disclosure, the processing device (4) acquires the posture transition amount from the posture detected by the detection device (2), and determines whether the user's interest in the content is high based on the posture transition amount after the gaze guidance is started. can be estimated. According to such a configuration, it is possible to determine the degree of attention to the content including the line-of-sight guidance with higher accuracy than when determining only the line-of-sight direction.

In one aspect of the present disclosure, the detection device (2) includes a seat (21) placed in a space where content is provided, and at least one acceleration sensor (22A, 22B, 22C) placed on the seat (21). and may have The processing device (4) may estimate the posture transition amount of the user seated on the seat (21) based on the output of at least one acceleration sensor (22A, 22B, 22C). According to such a configuration, it is possible to detect the transition of the user's posture more accurately than detection using image processing such as a camera. In addition, it is possible to reduce the influence of environment such as brightness and noise of the space where the content is provided on the detection accuracy.

It should be noted that the symbols in each parenthesis above are examples showing the correspondence with the specific configurations and the like described in the embodiments described later, and the present disclosure is limited to the specific configurations and the like shown in the symbols in the parentheses. not something.

FIG. 1 is a schematic configuration diagram showing an interest estimation system according to an embodiment. FIG. 2 is a schematic diagram for explaining the correspondence between the posture of the user on the seat of FIG. 1 and the output of the acceleration sensor. FIG. 3A is a graph showing an example of changes in line-of-sight direction, and FIG. 3B is a graph showing an example of changes in interest scores. 4A is a flow diagram schematically showing estimation processing executed by the estimation unit in FIG. 1, and FIG. 4B is a flow diagram schematically showing interest score calculation processing. FIG. 5 is a schematic diagram of a moving object on which the interest estimation system of FIG. 1 is installed.

Embodiments to which the present disclosure is applied will be described below with reference to the drawings.
[1. First Embodiment]
[1-1. composition]
The interest estimation system 1 shown in FIG. 1 is installed, for example, in facilities such as stadiums, movie theaters, theaters, concert halls, and live houses, or mobile objects such as automobiles, railroad vehicles, ships, aircraft, and spaceships.

The interest estimation system 1 includes a detection device 2 that detects the posture of the content user and a processing device 4 that estimates the interest from the detected posture of the user.

<Detection device>
The detection device 2 includes a seat 21 , a first cushion acceleration sensor 22 A, a second cushion acceleration sensor 22 B, a back acceleration sensor 22 C, and a camera 23 .

The seat 21 and the camera 23 are arranged in a space where contents are provided in a facility or a mobile body. The “contents” include viewing events such as sports, board games, and video games, viewing events such as plays and musical performances, live-action or virtual images, music, scenery (that is, sightseeing), and the like.

In particular, the interest estimation system 1 can be effectively applied to interactive content whose content changes according to user behavior. Examples of such interactive content include a virtual reality (VR) space or an augmented reality (AR) space and a virtual person character placed in the virtual reality space or the augmented reality space. Content includes:

(sheet)
The seat 21 has a seat cushion 21A and a seat back 21B. The seat cushion 21A is a part for supporting the user's buttocks and the like. The seat back 21B is a part for supporting the user's back.

A first cushion acceleration sensor 22A and a second cushion acceleration sensor 22B, which will be described later, are arranged on the seat cushion 21A. A back acceleration sensor 22C, which will be described later, is arranged on the seat back 21B.

(Acceleration sensor)
The first cushion acceleration sensor 22A, the second cushion acceleration sensor 22B, and the back acceleration sensor 22C are three-axis acceleration sensors and are configured to output three-dimensional acceleration data.

The first cushion acceleration sensor 22A, the second cushion acceleration sensor 22B, and the back acceleration sensor 22C have a function of detecting triaxial angular velocities (that is, roll angular velocity, pitch angular velocity, and yaw angular velocity) as necessary. may have.

However, if the processing device 4, which will be described later, can estimate interest without using angular velocity as input data, from the viewpoint of reducing the cost and power consumption of the sensor, each acceleration sensor can detect only acceleration. It is preferable to have

The first cushion acceleration sensor 22A and the second cushion acceleration sensor 22B are embedded in the seat cushion 21A so as to be spaced apart from each other in the width direction of the seat cushion 21A, that is, side by side in the left-right direction.

Specifically, the first cushion acceleration sensor 22A is arranged on the right side of the center of the seat cushion 21A in the width direction. The second cushion acceleration sensor 22B is arranged on the left side of the center of the seat cushion 21A in the width direction.

The first cushion acceleration sensor 22A and the second cushion acceleration sensor 22B are arranged so as to overlap the user's hip point (that is, the outermost part of the femur). Moreover, the distance in the horizontal direction between the first cushion acceleration sensor 22A and the second cushion acceleration sensor 22B is, for example, 100 mm or more and 150 mm or less.

The back acceleration sensor 22C is embedded inside the seat back 21B. The back acceleration sensor 22C is arranged at the center of the seat back 21B in the width direction. The height of the back acceleration sensor 22C is, for example, the hip point plus a height of 150 mm or more and 400 mm or less.

(camera)
The camera 23 is a three-dimensional camera that captures the posture of the user seated on the seat 21 . The camera 23 acquires the orientation of the user's body as an image and outputs the image to the processing device 4 . Note that the detection device 2 may have a plurality of cameras 23 .

<Processing device>
The processing device 4 is a device for estimating the user's interest in the current content provided to the user sitting on the seat 21 .

The processing device 4 has an estimation unit 41 , a storage unit 42 and a control unit 43 . The processing device 4 is composed of, for example, a microcomputer including a microprocessor, storage media such as RAM and ROM, and an input/output unit. The processing device 4 may be attached to or built into the seat 21 or may be arranged apart from the seat 21 .

(Estimation part)
The estimation unit 41 acquires information for estimation from the posture detected by the detection device 2 . The information for estimation includes the direction of the user's line of sight and the amount of posture transition.

The line-of-sight direction used for estimation is acquired as, for example, the direction of the user's head. The direction of the user's head is estimated from the image acquired by the camera 23, the output of each acceleration sensor, or a combination thereof.

For example, the direction of the head is the tilt of the user's upper body with respect to the vertical direction, the tilt of the line segment connecting the user's shoulders with respect to the reference axis, and the vertical line segment connecting the user's neck and nose. It is determined by a combination of the tilt with respect to the direction. That is, the line-of-sight direction is obtained as a code representing a combination of these tilts.

Each of the three tilts mentioned above can be detected by image analysis of the camera 23 . Also, from the output of each acceleration sensor, it is possible to estimate the three tilts using a learning model constructed by machine learning.

The learning model is based on input data based on the outputs of the first cushion acceleration sensor 22A, the second cushion acceleration sensor 22B, and the back acceleration sensor 22C, and information related to the user's posture or the user's posture transition. It is constructed by machine learning using teacher data (that is, label data). The learning model is stored in the storage unit 42, which will be described later.

A learning model is a classifier (that is, a classification model) constructed by supervised machine learning, and is composed of, for example, a multilayer neural network. Examples of multilayer neural networks include CNN (Convolution Neural Network), DNN (Deep Neural Network), LSTM (Long Short-Term Memory), and the like.

Note that the learning model is not limited to a multilayer neural network, and a model other than a neural network may be used. For example, a learning model may be constructed using an algorithm such as SVC (class classification by support vector machine), random forest, or the like.

In the machine learning of the learning model stored in the storage unit 42, the output of each acceleration sensor (that is, three-dimensional acceleration or acceleration data obtained by adding three-dimensional angular velocity to three-dimensional acceleration) is used as input data. Also, as teacher data, a certain number of posture labels indicating a user's posture pattern or a certain number of transition labels indicating a user's posture transition pattern are used.

In the learning step of generating a learning model, a large amount of labeled data is analyzed by a machine learning device. Labeled data is acceleration data with corresponding posture labels or transition labels. A machine learning device learns feature quantities for classifying acceleration data into a plurality of labels from a large amount of labeled data, and constructs a learning model.

A learning model is constructed using a machine learning device (not shown). A learning model constructed by the machine learning device is output to the storage unit 42 . Note that the machine learning device may be incorporated in the processing device 4 .

As shown in FIG. 2 , for example, when the user's posture transitions from a forward leaning posture P1 to an upright posture P2, the estimating unit 41 converts each acceleration at the time of transition to a learning model in which posture labels are teacher data. The sensor output O is input, and the posture label "upright" is added to the input data. As a result, the user's posture of "upright" is estimated.

Alternatively, the estimating unit 41 inputs the sensor output O to the learning model using the posture transition label as teacher data, and assigns the posture transition label "raising the body backward" to the input data. As a result, the user's posture transition of "raising the body backward" is estimated.

The posture labels given to the input data by the estimation unit 41 predefine multiple posture patterns to be estimated. The posture label of this embodiment is, for example, a code indicating the direction of the line of sight.

The posture transition amount used for estimation together with the direction of the line of sight is obtained as a feature amount calculated from the output of each acceleration sensor. This feature amount is calculated based on calculations such as integration and averaging of the acquired output. Table 1 shows an example of a combination of the content of the feature amount calculated by the estimation unit 41, the input data used for calculating the feature amount, and the method of calculating the feature amount.

The "seat back acceleration" and "seat back angular velocity" in Table 1 are outputs of the back acceleration sensor 22C, and the "seat cushion acceleration" and "seat cushion angular velocity" are the outputs of the first cushion acceleration sensor. 22A and the output of the second cushion acceleration sensor 22B.

The estimating unit 41 estimates the content based on the direction of the user's line of sight and the posture transition amount obtained based on the detection result of the detection device 2 (that is, the image obtained by the camera 23 and the output of each acceleration sensor). Estimate the level of user interest.

Specifically, the estimation unit 41 performs a first estimation based on the gaze maintenance time, a second estimation based on the frequency of changes in the gaze, a third estimation based on joint gaze, and a fourth estimation based on the magnitude of the reaction. .

In the first estimation, the estimation unit 41 estimates the user's level of interest in the content based on the gaze direction maintenance time. Specifically, the user's interest score is added when the gaze direction maintenance time is equal to or greater than a predetermined first threshold.

Note that "maintaining the direction of the line of sight" means that the "code indicating the direction of the line of sight" acquired by the estimation unit 41 does not change, and the direction code of the line of sight does not change (that is, the same Minor changes in line of sight (included in the direction code) are included in "maintenance of direction of line of sight". The first threshold may be set according to the details of the content, and may change according to the progress of the content. The first threshold is, for example, 1.5 seconds.

In the second estimation, the estimation unit 41 estimates the user's level of interest in the content based on the frequency of changes in the line-of-sight direction (that is, the line-of-sight deviation rate). Specifically, the user's interest score is added when the number of changes in the line-of-sight direction within a certain period of time is equal to or less than a predetermined second threshold.

As in the first estimation, "change in the direction of the line of sight" means that the "code indicating the direction of the line of sight" changes. The second threshold may be set according to the details of the content and the first threshold, and may change according to the progress of the content. The second threshold is, for example, 2 times/5 seconds.

In the third estimation, the estimation unit 41 determines whether the user's interest in the content is high based on the time from the start of the visual guidance by the content until the direction of the visual line matches the direction of the visual guidance (that is, transition to joint gaze). to estimate the

Specifically, the user's interest score is added when the direction of the line-of-sight guidance of the content matches the direction of the user's line of sight while the joint gaze scene is provided by the content.

FIG. 3A shows an example of a change in the direction of the user's line of sight. FIG. 3A shows changes in the direction of the line of sight of the user when the direction of the line of sight is classified into "right", "straight", and "left". The dashed line indicates the direction of the user's line of sight acquired by the estimation unit 41, and the dashed line indicates the direction in which the content guides the user's line of sight (that is, the direction the user wants the user to look at). A solid line is a portion where the direction of the user's line of sight and the direction of line of sight guidance match.

In period T1 in FIG. 3A, the direction of the user's line of sight is not fixed. In other words, the user turns his or her head left and right to look for the content itself or the direction the content is guiding.

Note that the estimating unit 41 evaluates such reciprocation of the line-of-sight direction as a state in which the user has begun to take an interest in the content, and adds the interest score of the user when the number of reciprocations exceeds a certain value. good too.

In the period T2 following the period T1, the direction of the line of sight is maintained for a certain period of time, so the interest score is added by the first estimation. In addition, since the direction of the line of sight also changes in accordance with the change in the guiding direction of the content, the interest score is added by the third estimation.

In the period T3 following the period T2, the direction of the line of sight is maintained for a certain period of time, and the line of sight deviates only once, so the interest score is added by the first estimation and the second estimation.
In the period T4 following the period T3, the direction of the line of sight changes and the guidance direction of the content is not followed, so the interest score is not added by the first estimation and the third estimation.

A fourth estimate is performed in conjunction with the third estimate. In the fourth estimation, the degree of user's interest in content is estimated based on the posture transition amount after the start of visual guidance by the content.

Specifically, the estimation unit 41 calculates at least one feature value shown in Table 1 above, and adds the user's interest score each time the feature value exceeds the threshold value. Also, the estimation unit 41 may add an interest score when all of the plurality of feature amounts exceed the threshold.

Furthermore, the estimating unit 41, when the time from the start of the visual guidance by the content until at least one feature value reaches a constant value is equal to or less than a predetermined threshold value (that is, when the transition speed is equal to or greater than a constant value) , the user's interest score may be added.

Note that the feature values shown in Table 1 are examples. Therefore, the estimation unit 41 does not need to calculate all the feature values shown in Table 1. Also, the estimation unit 41 may perform the fourth estimation using feature amounts not shown in Table 1 (however, limited to those calculated based on the output of the acceleration sensor).

For example, the estimation unit 41 may calculate the user's interest score separately for each user's posture (that is, line-of-sight direction code). For example, when it is estimated that the user is highly interested in the content when the user is looking to the right, the interest score for the "right direction" is added. Similarly, when it is estimated that the user is highly interested in the content when the user is looking leftward, the "leftward" interest score is added.

Interest scores change over time, as shown in FIG. 3B. In other words, the first to fourth estimations are performed at regular time intervals, and the estimation unit 41 outputs the sum of the interest scores resulting from these estimations at regular time intervals.

It should be noted that the magnitude of the interest score added by the first estimation to the fourth estimation may be different. In addition, the magnitude of the added interest score in each estimation may be changed according to the details of the content.

The estimation unit 41 estimates the degree of interest (that is, the degree of interest or the degree of immersion) in the content supplied at the time of calculating the interest score, based on the magnitude of the interest score obtained in this way. That is, if the interest score is high, it is estimated that the interest in the content is high, and if the interest score is low, it is estimated that the interest in the content is low.

In addition, the estimation unit 41 can determine which posture has the highest interest score by comparing the interest scores for each posture, and can estimate the direction in which the user showed the greatest interest when the content was provided.

The estimation unit 41 outputs the calculated interest score at each time to the storage unit 42 and the control unit 43 as a result of estimating the interest in the content.

(storage unit)
The storage unit 42 records, as data, the estimation result of the user's interest in the content (that is, the interest score) by the estimation unit 41 . The storage unit 42 also stores a learning model for estimating the posture of the user.

The data stored in the storage unit 42 is output by writing to a recording medium, transmission to a server via a communication circuit, printing on paper, or the like. The output data can be used as reference information for providing more suitable services to users, big data for analyzing trends of interests of unspecified number of users, and the like.

The data stored in the storage unit 42 is also referred to by the control unit 43 . By directly referring to the data in the storage unit 42, the control unit 43 controls the details of the content based on the past interest estimation results, as will be described later.

(control part)
The control unit 43 controls the details of the content based on the estimation result of the user's interest in the content (that is, the interest score) by the estimation unit 41 .

For example, if the current content has a low interest score, the control unit 43 starts providing different content instead of the current content. The control unit 43 may also add a new event or change the story (for example, change the character's personality) to make the user interested in the current content. Furthermore, the control unit 43 may provide the user with a stimulus (for example, sound, smell, wind, vibration, etc.) recognized by perception other than sight. Therefore, the controller 43 may have hardware for outputting such a stimulus.

Further, when the interest score of the current content is high, the control unit 43 adds an event, changes the story, provides stimulation, etc. so as to promote the user's interest and improve the degree of immersion in the content. conduct.

Further, the control unit 43 may combine the past interest score stored in the storage unit 42 with the current interest score to determine the user's preference, and control the details of the currently provided content.

<Processing of estimation unit>
An example of the estimation process performed by the estimation unit 41 will be described below with reference to the flowchart of FIG. 4A.

In this process, the estimation unit 41 first receives the posture of the user detected by the detection device 2, and acquires information for estimation (that is, the direction of the line of sight and the amount of posture transition) used for estimating the interest (step S10). ). Next, the estimation unit 41 executes the interest score calculation process of FIG. 4B based on the information for estimation (step S20).

In the interest score calculation process, the estimation unit 41 determines whether or not the user's interest is high according to the first estimation (step S210). If the interest is high (S210: YES), the estimation unit 41 adds a certain number of points to the interest score (step S220). On the other hand, when the interest is low (S210: NO), the estimation unit 41 does not add the interest score.

After the first estimation, the estimation unit 41 determines whether the user is highly interested by the second estimation (step S230), determines whether the user is highly interested by the third estimation (step S250), Then, it is determined whether or not the user's interest is high based on the fourth estimation (step S270).

In each estimation, if the interest is high (S230, S250, S270: YES), the estimation unit 41 adds a certain number of points to the interest score (steps S240, S260, S280). On the other hand, if the interest is low (S230, S250, S270: NO), the estimation unit 41 does not add the interest score. Note that the first to fourth estimations may be performed in series as shown in FIG. 4B, or may be performed in parallel.

After calculating the interest score, the estimation unit 41 determines whether the content has ended (step S30). If the content has ended (S30: YES), the estimation unit 41 ends the process. On the other hand, if the content has not ended (that is, the content continues) (S30: NO), the estimation unit 41 repeats the processing from step S10.

Note that the estimating unit 41 has a step of instructing the control unit 43 to perform control such as content switching, event addition, story change, stimulation provision, etc. according to the magnitude of the interest score. good too.

<Moving body>
An example will be described below in which the interest estimation system 1 is installed in the moving body 10 shown in FIG.

The moving object 10 has an internal space 10A in which the seat 21 of the interest estimation system 1 is arranged, and is a device capable of self-propelled on the ground using power such as electric power. However, the moving body 10 having the internal space 10A is not limited to one that travels on the ground.

The internal space 10A of the moving body 10 constitutes a guest room in which the user U who uses the interest estimation system 1 stays. 10 A of internal spaces of the mobile body 10 are movable spaces (namely, mobility space) which are not fixed to one point.

The moving body 10 has an automatic driving function and can automatically travel along a predetermined route. The mobile body 10 has at least one window that separates the interior space 10A from the exterior of the mobile body 10, and the user U boarding the interior space 10A of the mobile body 10 can see the outside of the mobile body 10 through the window. You can see the scenery of However, the moving body 10 does not necessarily have to have an automatic driving function.

In the interior space 10A of the moving body 10, in addition to a seat 21, a camera 23, and a character seat 11 on which a character V (to be described later) is seated, there are not shown desks, lighting, air conditioning equipment, displays, and operation control equipment (for example, equipment such as input devices such as touch panels) are arranged.

The processing device 4 of the interest estimation system 1 is arranged inside the mobile object 10 . Also, the content is displayed on the display unit 12 . In this embodiment, the display unit 12 is configured by a wearable device worn by the user U. As shown in FIG.

As a wearable device that configures the display unit 12, for example, a head-mounted display that is worn on the head of the user U can be used. The head-mounted display has a display that covers the eyes of the user U, and a sensor for head tracking such as a gyro sensor and an acceleration sensor.

By using a head-mounted display as the display unit 12, the position and orientation of the user U can be measured, and the degree of immersion of the user U in the augmented reality space can be enhanced. However, the wearable device used as the display unit 12 is not limited to the head mounted display.

Moreover, the display unit 12 is not limited to a wearable device. For example, a flat display arranged in the internal space 10A may be used as the display unit 12. FIG. Further, the window forming the internal space 10A may be provided with a display function and used as the display unit 12. FIG.

The processing device 4 forms an augmented reality space based on the real space based on the image of the internal space 10A captured by an imaging unit (not shown). In addition, the processing device 4 forms in real time a provided image including the augmented reality space and the character V arranged in the augmented reality space, based on pre-stored information on the character V and a program for moving the character V. .

The provided image formed by the processing device 4 is transmitted to the display unit 12 in real time. For example, a change in scenery seen from the moving body 10, a change in lighting in the internal space 10A, and the like are reflected in the instantly provided image. As a result, the user U can recognize these changes in the augmented reality space in real time via the display unit 12 .

In the content displayed on the display unit 12, the character V guides the user U's line of sight through conversation, gestures, and the like. The processing device 4 estimates the user U's interest in the character V by using the guiding direction of the line of sight of the character V and the direction of the user's line of sight.

For example, if the user U has little interest, the processing device 4 provides content in which the character V acts as a guide. On the other hand, if the user U is highly interested, the processing device 4 provides content in which the character V behaves like a lover.

In addition, the processing device 4 selects a scene in which the direction of the user U's line of sight is maintained (that is, the position where the moving body 10 was running, the scenery that was visible to the user U, etc.). Infer the subject of interest.

[1-2. effect]
According to the embodiment detailed above, the following effects are obtained.
(1a) Based on the movement of the user's line of sight, the user's level of interest in the content (that is, the degree of attention) can be estimated. Based on this result, it is possible to improve the user's interest by changing the contents of the content, the method of providing the content, and the like. In addition, since the user's line of sight is used for estimation, the level of interest can be estimated without using language information (that is, user's utterance content) that is affected by analysis accuracy.

(1b) Using the second estimation using the frequency of changes in the line-of-sight direction, the degree of attention of the user to the content can be determined with higher accuracy than when only the first estimation based on the line-of-sight direction maintenance time is used. can judge.

(1c) It is possible to easily and accurately determine the level of interest in the content that includes the line-of-sight guidance by the third estimation using the match time between the line-of-sight direction and the guidance direction of the content.

(1d) By the fourth estimation using the posture transition amount, it is possible to determine the degree of attention to the content including the gaze guidance with higher accuracy than when determining only the third estimation based on the direction of the gaze. .

(1e) By estimating the posture transition amount by the acceleration sensors 22A, 22B, and 22C arranged on the seat 21, it is possible to detect the posture transition of the user more accurately than detection using image processing such as a camera. can. In addition, it is possible to reduce the influence of environment such as brightness and noise of the space where the content is provided on the detection accuracy.

[2. Other embodiments]
Although the embodiments of the present disclosure have been described above, it is needless to say that the present disclosure is not limited to the above embodiments and can take various forms.

(2a) In the interest estimation system 1 of the above embodiment, the processing device 4 does not necessarily have to perform the second estimation, the third estimation, and the fourth estimation. The processing device 4 may perform only the first estimate, only the first and second estimates, or only the first and third estimates.

(2b) In the interest estimation system 1 of the above embodiment, the detection device 2 may have detection equipment other than the acceleration sensors 22A, 22B, 22C and the camera 23 . That is, the detection device 2 does not necessarily have the seat 21, the acceleration sensors 22A, 22B, and 22C, and the camera 23 as long as the posture of the user can be detected.

(2c) In the interest estimation system 1 of the above embodiment, the processing device 4 does not necessarily have to control the details of the content based on the estimation result of the user's interest in the content. Also, the processing device 4 does not necessarily have to record the estimation result of the user's interest in the content as data.

(2d) In the interest estimation system 1 of the above embodiment, the number of acceleration sensors may be one, two, or four or more. For example, only the two cushion acceleration sensors 22A and 22B may be arranged on the seat 21 .

(2e) The function of one component in the above embodiments may be distributed as multiple components, or the functions of multiple components may be integrated into one component. Also, part of the configuration of the above embodiment may be omitted. Also, at least a part of the configuration of the above embodiment may be added, replaced, etc. with respect to the configuration of the other above embodiment. It should be noted that all aspects included in the technical idea specified by the wording in the claims are embodiments of the present disclosure.

DESCRIPTION OF SYMBOLS 1... Interest estimation system, 2... Detection apparatus, 4... Processing apparatus, 10... Moving body,
10A... Internal space, 11... Character sheet, 12... Display section, 21... Sheet,
21A... seat cushion, 21B... seat back,
22A... Acceleration sensor for first cushion, 22B... Acceleration sensor for second cushion,
22C...back acceleration sensor, 23...camera, 41...estimating section, 42...storage section,
43... Control part.

Claims (3)

a detection device that detects the posture of a content user;
a processing device that acquires the direction of the user's line of sight from the posture detected by the detection device and estimates the level of interest of the user in the content based on the maintenance time of the direction of the line of sight;
with
The processing device estimates that the user's interest in the content increases as the frequency of change in the line-of-sight direction decreases , and the line-of-sight direction changes from the start of the line-of-sight guidance by the content. An interest estimation system for estimating the level of interest of the user in the content based on the direction of visual guidance and the time it takes to match . An interest estimation system according to claim 1 ,
The processing device acquires a posture transition amount from the posture detected by the detection device, and estimates a level of interest of the user in the content based on the posture transition amount after the gaze guidance is started. estimation system. The interest estimation system according to claim 2 ,
The detection device is
a sheet arranged in a space where the content is provided;
at least one acceleration sensor located on the seat;
has
The interest estimation system, wherein the processing device estimates the posture transition amount of the user seated on the seat based on the output of the at least one acceleration sensor.

Images