WO2022097457A1

WO2022097457A1 - Design evaluating device, design evaluating system, design evaluating method, design evaluating program, and learning device

Info

Publication number: WO2022097457A1
Application number: PCT/JP2021/038376
Authority: WO
Inventors: 真一郎斉藤; 暁井上; 勝一浦谷; 健二寺田; 伸彦高嶋; 麻由香加羽澤
Original assignee: コニカミノルタ株式会社
Priority date: 2020-11-06
Filing date: 2021-10-18
Publication date: 2022-05-12
Also published as: JPWO2022097457A1

Abstract

[Problem] To provide a design evaluating device, a design evaluating system, a design evaluating method, a design evaluating program, and a learning device with which a user's kansei (subjective impression) with respect to an image of an object being evaluated can be visualized. [Solution] This design evaluating device includes: an image acquiring unit; an attraction degree estimating unit for estimating a degree of attraction of an image on the basis of characteristics of the image; a display control unit; an expression providing unit for providing to the user a plurality of expressions relating to the design of the object being evaluated; a line of sight information acquiring unit for acquiring information relating to the line of sight of the user with respect to an image being displayed on the display device; an attraction degree measuring unit for measuring the degree of attraction of the image on the basis of the information relating to the line of sight; a difference calculating unit for calculating the difference between the estimated degree of attraction estimated by the attraction degree estimating unit and the measured degree of attraction measured by the attraction degree measuring unit; a kansei information acquiring unit for acquiring kansei information indicating a relationship between the expressions and the user's kansei relating to the design; and a kansei extracting unit for using the kansei information to extract a correlation between the difference and the plurality of expressions.

Description

Design evaluation device, design evaluation system, design evaluation method, design evaluation program, and learning device

The present invention relates to a design evaluation device, a design evaluation system, a design evaluation method, a design evaluation program, and a learning device.

The design of product packaging and advertisements has become an important factor in attracting consumer attention in the market. Traditionally, sellers have implemented image strategies aimed at attracting the attention of consumers and leading to the purchase of products. However, in general, the evaluation of design is greatly influenced by the subjectivity of individual consumers, and it is not always possible to obtain similar evaluations among a plurality of individuals.

It is generally said that it is difficult to understand consumers' purchasing decisions, thoughts and motives. Studies have shown that in fact, about 90-95% of the thinking process occurs in the subconscious as consumers decide their thoughts and actions (eg, Kahneman, 2013; Zaitman, 2003). Against this background, it is important for product marketing practitioners who handle the brand of their own products and for the brand itself, how to establish the subconscious mind that the brand is looking for in consumers. It is also important to actually understand the degree of fixation.

In addition, when selling or promoting products through various media such as magazines, televisions, and the Internet in addition to stores, the design of product packages and the design of advertisements are major factors that influence the sales of products. In recent years, there have been an increasing number of cases in which consumers purchase products through media such as televisions and the Internet, relying on the designs of product packages and advertisements. As a sales strategy, it is conceivable to design the product to attract the attention of many consumers, but there is a problem that the evaluation varies among individuals in the preliminary survey by the subjects.

On the other hand, in Patent Document 1 below, an image of an object to be evaluated and an item for evaluating the sensitivity of the image are presented to the subject, and the subject is made to evaluate the image, and the subject evaluates the image. While doing so, a technique for measuring the gaze transition of a subject is disclosed. The purpose of this technique is to calculate a unified sensibility evaluation for the image by adding the subject's line-of-sight information to the variation in the sensibility evaluation between the subjects.

In addition, a technique for calculating the correlation between the subject's line-of-sight position on the image of the evaluation object and the prominence of the evaluation object (which indicates the ease of visual attention of humans and is also called "salience") is known. ing. For example, in Patent Document 2 below, a line-of-sight tracking device 3 for measuring a subject's line of sight, a physiological index measuring device 4 for measuring a subject's physiological index, and a subject's psychological state are estimated from the subject's physiological index. A line-of-sight analysis system 1 including at least a line-of-sight analysis device 2 developed so as to be able to analyze the line-of-sight of the subject in association with the psychological state of the subject is disclosed. In the region where the concentration of the subject's line-of-sight position is equal to or higher than the concentration threshold and the degree of saliency (hereinafter referred to as "severity") is equal to or higher than the saliency threshold, the frequency of the line of sight is high. In addition, since it is an area with a high degree of prominence, it is considered to be an area of interest that may affect the impression of the subject.

However, if the objects in the image are highly prominent, the user may look at the objects in the image even if they are not interested, which may affect the estimation of the degree of interest that indicates the degree of the user's interest in the objects. There is. In this regard, a technique for estimating the user's interest in an object is known so that the longer the line-of-sight residence time for the object is, the higher the interest is, and the smaller the salency is, the higher the interest is. (For example, see Patent Document 3 below).

Japanese Patent No. 3772514 Japanese Patent No. 6201520 Japanese Patent No. 5602155

However, the technique of Patent Document 1 aims to reduce the variation in evaluation among individuals with respect to the image of the evaluation target. Therefore, when a seller or the like formulates a sales strategy in consideration of the subjectivity of an individual consumer, there is a problem that it is difficult to understand what kind of individual sensibility works on the image of the evaluation target.

In addition, the line-of-sight position of the subject in the image of the evaluation object and the prominence of the evaluation object are not always highly correlated. Therefore, even with the technique of Patent Document 2, there is a problem that it is not possible to know what kind of individual sensibility works on the image of the evaluation object. Further, the technique of Patent Document 3 aims to reduce an error in estimating the degree of interest, and what kind of individual sensibilities work on the image of the evaluation object also by the technique of Patent Document 3. I don't know.

The present invention has been made in view of the above-mentioned problems. Therefore, an object of the present invention is to provide a design evaluation device, a design evaluation system, a design evaluation method, a design evaluation program, and a learning device capable of visualizing a user's feelings regarding an image of an evaluation object.

The above problem of the present invention is solved by the following means.

(1) An image acquisition unit that acquires at least one image of an evaluation target, an attraction degree estimation unit that estimates the degree of attraction of the image based on the characteristics of the image, and a display for displaying the image on a display device. A control unit, an expression providing unit that provides a plurality of expressions related to the design of the evaluation object to the user, a line-of-sight information acquisition unit that acquires information about the user's line of sight with respect to the image displayed on the display device, and the above. Difference between the attractiveness measuring unit that measures the attractiveness of the image based on the information about the line of sight, the estimated attractiveness estimated by the attractiveness estimation unit, and the measured attractiveness measured by the attractiveness measuring unit. The difference calculation unit for calculating the difference, the sensitivity information acquisition unit for acquiring the sensitivity information indicating the relationship between the expression and the user's sensitivity regarding the design, and the difference and the plurality of expressions using the sensitivity information. Sensitivity extraction unit that extracts relevance and design evaluation device.

(2) The difference calculation unit sets a plurality of regions that do not overlap each other with respect to the image, calculates the difference in each region, and the sensitivity extraction unit calculates the difference with the difference in each region. The design evaluation device according to (1) above, which extracts the relationship with the plurality of expressions.

(3) Further, the attraction degree conversion value calculation unit for calculating the attraction degree conversion value based on the measurement attraction degree and the sensitivity information in each of the regions is further provided, and the sensitivity information acquisition unit has the above-mentioned design. The user's response to the expression is received, the sensitivity information is acquired based on the response, and the sensitivity extraction unit extracts the relationship based on the correlation between the difference and the attraction conversion value. The design evaluation device according to (2) above.

(4) Further having an attribute information acquisition unit for acquiring information regarding the user's attributes, the sensitivity extraction unit relates the difference to the plurality of expressions and information regarding the user's attributes in each of the regions. The design evaluation device according to (3) above, which extracts sex.

(5) The design evaluation device according to (2) above, wherein the sensitivity extraction unit extracts the relationship by a statistical method using the plurality of expressions as explanatory variables and the difference as an objective variable.

(6) Further has an attribute information acquisition unit for acquiring information related to the user's attributes.
The design according to (2) above, wherein the sensitivity extraction unit extracts the relationship by a statistical method using the information about the plurality of expressions and the attributes of the user as explanatory variables and the difference as the objective variable. Evaluation device.

(7) Further having an attribute information acquisition unit for acquiring information on the user's attributes, the sensitivity extraction unit uses the sensitivity information and information on the user's attributes as explanatory variables, and machine learning using the difference as an objective variable. The design evaluation device according to (2) above, which extracts the relationship using the machine learning model obtained in the above.

(8) The design evaluation device according to any one of (1) to (7) above, wherein the sensitivity extraction unit extracts the relationship with a plurality of the users.

(9) The Kansei information acquisition unit acquires the user's evaluation regarding the design, and calculates the Kansei information based on the user's evaluation, in any one of the above (1) to (8). Described design evaluation device.

(10) The Kansei information acquisition unit acquires the response time required for the user to complete the response to the expression, and calculates the Kansei information based on the response time (1). )-(8). The design evaluation device according to any one of (8).

(11) The line-of-sight information acquisition unit continuously provides information on the line of sight from the time the expression is provided to the user by the expression providing unit until the user's response to the expression is received by the sensitivity information acquisition unit. The design evaluation device according to any one of (1) to (10) above.

(12) The design evaluation device according to any one of (1) to (11) above, wherein the display control unit simultaneously displays the image and the expression on the display device.

(13) The design evaluation device according to any one of (1) to (12) above, wherein the expression includes at least words.

(14) The image acquisition unit acquires a plurality of images of the evaluation target, and the attraction degree estimation unit estimates the attraction degree of each image based on the characteristics of each image, and displays the display. The control unit displays each of the images on the display device, and the attraction degree measuring unit measures the degree of attraction of each of the images based on the information regarding the line of sight. The design evaluation device according to any one of the above.

(15) In the above (14), the image acquisition unit acquires an image of the evaluation object to which decorative printing is applied and an image of the evaluation object to which the decorative printing is not applied. Described design evaluation device.

(16) The display device, a line-of-sight measuring device that measures the user's line of sight with respect to the image displayed on the display device, and outputs information about the line of sight to the line-of-sight information acquisition unit, and the above (1) to (1). A design evaluation system comprising the design evaluation device according to any one of 15).

(17) A step (a) of acquiring at least one image of an evaluation target, a step (b) of estimating the degree of attraction of each image based on the characteristics of the image, and the image being displayed on a display device. A step (c) for displaying, a step (d) for providing the user with a plurality of expressions relating to the design of the evaluation object, and a step for acquiring information regarding the user's line of sight with respect to the image displayed on the display device (d). e), the step (f) for measuring the degree of attraction of the image based on the information about the line of sight, the estimated degree of attraction estimated in the step (b), and the measurement measured in the step (f). The step (g) for calculating the difference from the degree of attraction, the step (h) for acquiring the sensitivity information indicating the relationship between the expression and the user's sensitivity regarding the design, and the difference with the difference using the sensitivity information. A design evaluation method comprising the step (i) of extracting the relationship with the plurality of expressions.

(18) In the step (g), a plurality of regions that do not overlap each other are set for the image, the difference is calculated in each region, and in the step (i), the region is described. The design evaluation method according to (17) above, which extracts the relationship between the difference and the plurality of expressions.

(19) Prior to the step (i), there is further a step (j) for acquiring information regarding the user's attributes, and in the step (i), the sensitivity information and the information regarding the user's attributes are explanatory variables. The design evaluation method according to (17) or (18) above, wherein the difference is used as an objective variable to extract the relationship.

(20) A design evaluation program for causing a computer to execute the process included in the design evaluation method according to any one of (17) to (19) above.

(21) The sensitivity information and the attraction based on the storage unit that stores the learning data including the sensitivity information as the explanatory variable and the attraction degree difference as the objective variable, and the learning data stored in the storage unit. A learning device having a model generation unit that generates a machine learning model by machine learning the relationship with the degree difference.

According to the present invention, with respect to the image of the evaluation object, the degree of attraction based on the user's sensibilities is calculated, and the relationship between the degree of attraction and a plurality of expressions related to the image of the evaluation object is extracted. You can visualize the user's feelings about the image of.

It is a schematic diagram which looked at the design evaluation system which concerns on 1st Embodiment from above. It is a block diagram exemplifying the schematic hardware configuration of the design evaluation system shown in FIG. 1. It is a functional block diagram illustrating the main function of the control device shown in FIG. It is a figure which exemplifies the sensitivity question item about the design proposal of a food package. It is a flowchart which illustrates the design evaluation method of the control device which concerns on 1st Embodiment. It is a flowchart following FIG. 5A. It is a schematic diagram which illustrates the design proposal of the evaluation object. It is a schematic diagram which illustrates the plurality of areas 1 to 4 set for the image of a design proposal. It is a figure which illustrates the saliency map of the image of a design proposal. It is a figure which illustrates the estimated degree of attraction in the area 1 to 4 of the image of a design proposal. It is a schematic diagram which illustrates the case where the image of a design proposal is displayed in a standard state. It is a schematic diagram which exemplifies the distribution of the line-of-sight position of a user who is looking at an image of a design proposal. It is a figure which illustrates the measurement attraction degree in the area 1 to 4 of the image of a design proposal. It is a figure which exemplifies the difference of the degree of attraction in the area 1 to 4 of the image of a design proposal. It is a schematic diagram exemplifying the display of the image of the design proposal and the display of the sensitivity question item (“innovative”). It is a schematic diagram exemplifying the display of the image of the design proposal and the display of the sensitivity question item (“delicious”). It is a schematic diagram exemplifying the display of the image of the design proposal and the display of the sensitivity question item (“like”). It is a figure which illustrates the measurement attraction degree in the area 1 to 4 of the image of a design proposal for each sensitivity question item. It is a figure which exemplifies the score input about the design proposal by the user for each sensitivity question item. It is a figure which exemplifies the attractiveness conversion value in the area 1 to 4 of the image of a design proposal for each sensitivity question item. It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“innovative”). It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“delicious”). It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“like”). It is a figure which exemplifies each variable in the case of extracting the relationship between the difference in attractiveness degree and a plurality of Kansei question items by using multiple regression analysis. It is a figure which illustrates the regression coefficient β ₂ to β ₃ calculated by the multiple regression analysis in FIG. It is a schematic block diagram illustrating a learning device that generates a machine learning model based on the difference in attractiveness and the sensitivity information (and the attribute of the user). It is a schematic diagram which looked at the design evaluation system which concerns on 2nd Embodiment from above. It is a flowchart which illustrates the design evaluation method of the control device which concerns on 2nd Embodiment. It is a flowchart following FIG. 23A. It is a schematic diagram which illustrates the design proposal 1 of the evaluation object. It is a schematic diagram which illustrates the area setting of the image of the design proposal 1. It is a schematic diagram which illustrates the area setting of the image of the design proposal 2. It is a schematic diagram which illustrates the saliency map of the image of the design proposal 1. It is a schematic diagram which illustrates the saliency map of the image of the design proposal 2. It is a figure which illustrates the estimated degree of attraction in the area 1 to 4 of the image of the design proposal 1. FIG. It is a figure which illustrates the estimated degree of attraction in the area 1 to 4 of the image of the design proposal 2. It is a schematic diagram which illustrates the case where the images of

design proposals

1 and 2 are displayed in a standard state. It is a schematic diagram which illustrates the distribution of the line-of-sight position of the user who is looking at the image of

design proposals

1 and 2. It is a schematic diagram which illustrates the measurement attractiveness degree in the area 1 to 4 of the image of the

design proposals

1 and 2. It is a schematic diagram which illustrates the difference of attractiveness degree in the area 1 to 4 of the image of the

design proposals

1 and 2. It is a schematic diagram exemplifying the display of the image of the

design proposals

1 and 2 and the sensitivity question item (“innovative”). It is a schematic diagram which exemplifies the display of the image of the

design proposals

1 and 2 and the sensitivity question item (“delicious”). It is a schematic diagram exemplifying the display of the image of the

design proposals

1 and 2 and the sensitivity question item (“like”). It is a schematic diagram which exemplifies the distribution of the line-of-sight position of a user when the sensitivity question item (“innovative”) is displayed. It is a schematic diagram exemplifying the distribution of the line-of-sight position of the user when the sensitivity question item (“delicious”) is displayed. It is a schematic diagram exemplifying the distribution of the line-of-sight position of the user when the sensitivity question item (“like”) is displayed. It is a schematic diagram which exemplifies the measurement attraction degree for a sensitivity question item (“innovative”). It is a schematic diagram which exemplifies the degree of attraction of measurement for a sensitivity question item (“delicious”). It is a schematic diagram which exemplifies the degree of attraction of measurement for a sensitivity question item (“like”). It is a figure which exemplifies the calculation result of the reaction intensity at the time of selecting a design proposal for each sensitivity question item. It is a graph exemplifying the degree of attraction conversion value in the area 1 to 4 of the image of the

design proposals

1 and 2 for each sensitivity question item. It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“innovative”). It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“delicious”). It is a graph exemplifying the correlation between the attractiveness difference and the attractiveness conversion value for the sensitivity question item (“like”).

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same elements are designated by the same reference numerals, and duplicate description will be omitted. In addition, the dimensional ratios in the drawings are exaggerated for convenience of explanation and may differ from the actual ratios.

(First Embodiment)
FIG. 1 is a schematic view of the design evaluation system 100 according to the first embodiment as viewed from above, and FIG. 2 is a block diagram illustrating a schematic hardware configuration of the design evaluation system 100 shown in FIG. .. Further, FIG. 3 is a functional block diagram illustrating the main functions of the control device 300 shown in FIG. Further, FIG. 4 is a diagram illustrating the sensitivity question items regarding the design proposal of the food package.

The design evaluation system 100 of the present embodiment presents an image 30 of the evaluation object and words 40 relating to the design of the evaluation object, and with respect to the image 30, an attractiveness (visual attention) based on the user 10's sensibilities. Ease) and the relationship between the plurality of words 40 are extracted. As a result, among these words 40, the words 40 that are in harmony with the sensibility of the user 10 are determined with respect to the design of the evaluation object. The evaluation object may be, for example, a product designed by a designer, a product package, an advertisement, or the like.

As shown in FIGS. 1 and 2, the design evaluation system 100 includes a line-of-sight measuring device 200, a control device 300, a display device 400, an input device 500, an audio output device 600, and a communication device 700. These configurations are communicably connected to each other.

<Gaze measuring device 200>
The line-of-sight measuring device 200 continuously measures the line-of-sight 20 of the user 10 and transmits information about the line-of-sight to the control device 300. As will be described later, the control device 300 functions as a design evaluation device, causes the display 410 of the display device 400 to display the image 30 of the evaluation target, and presents the image 30 to the user 10. The information regarding the line of sight is, for example, the position (hereinafter referred to as “line-of-sight position”) 50 that the user 10 is looking at on the image 30 of the evaluation object, the line-of-sight direction of the left and right eyes of the user 10, or the left and right eyes of the user 10. It can be the position of the pupil of.

The line-of-sight measuring device 200 can be attached to the head of the user 10, for example. A known eye tracking technique can be adopted for the line-of-sight measurement by the line-of-sight measuring device 200. The eye tracking technique may be a non-contact technique such as a corneal reflex method or a scleral reflex method. Further, a contact-type technique such as a search coil method or an eye potential method, or a line-of-sight measurement method other than these may be used.

When measuring the line of sight using the corneal reflex method, the line-of-sight measuring device 200 irradiates each of the left and right eyes of the user 10 with weak near-infrared light, and images the left and right eyes of the user 10 with a camera, for example. .. Then, the line-of-sight measuring device 200 calculates the center position of the reflected light of the near-infrared light on the corneal surface and the center position of the pupil of the eye for each of the left and right eyes of the user 10 from the image of the camera, and based on these. The line of sight 20 of the user 10 is detected. As a result, the line of sight 20 of the user 10 is accurately measured while the user 10 is gazing at the image of the evaluation object. The line-of-sight measuring device 200 calculates the line-of-sight position of the user 10 in the image 30 of the evaluation object based on the information on the line-of-sight direction of the user 10 in the three-dimensional space, the center positions of the pupils of the left and right eyes of the user 10, and the like. .. As information on the line of sight, the line-of-sight directions of the left and right eyes of the user 10, the center positions of the pupils of the left and right eyes of the user 10, and the like are transmitted from the line-of-sight measuring device 200 to the control device 300, and the control device 300 relates to the line of sight. It may be configured to calculate the line-of-sight position 50 based on the information.

<Control device 300>
The control device 300 functions as a design evaluation device and controls the design evaluation system 100 in an integrated manner. The control device 300 may be a personal computer, a smartphone, a PDA (Personal Digital Assistant) in which a design evaluation program described later is installed, a main body of a tablet terminal (control device), or the like. As shown in FIG. 2, the control device 300 includes a CPU (Central Processing Unit) 310, a RAM (Random Access Memory) 320, a ROM (Read Only Memory) 330, an auxiliary storage unit 340, and the like.

The CPU 310 executes an OS (Operating System) and a design evaluation program deployed in the RAM 320, and controls the operation of the line-of-sight measuring device 200, the display device 400, the input device 500, the voice output device 600, and the communication device 700. The design evaluation program is stored in the ROM 330 or the auxiliary storage unit 340 in advance. Further, the RAM 320 stores data or the like temporarily generated by the processing of the CPU 310. The ROM 330 stores a program executed by the CPU 310, data used for executing the program, parameters, and the like.

The auxiliary storage unit 340 has, for example, an HDD (Hard Disk Drive), an SSD (Solid State Drive), or the like, and can store a plurality of images of an evaluation target. In these images, the evaluation target is imaged by a camera, a scanner, or the like, and for example, JPEG (Joint Photographic Experts Group) format, TIFF (tag image file format) format, PNG (Portable Network image storage), etc. Has been done. Further, the auxiliary storage unit 340 stores a plurality of words presented to the user 10. Details of these terms will be described later. In addition, the auxiliary storage unit 340 also stores information regarding the attributes of the user 10 when performing a monitor test described later.

<Display device 400>
The display device 400 includes a display 410, and is used to present an image of the evaluation object and words related to the design proposal of the evaluation object to the user 10. Further, after the user 10 evaluates the design proposal of the evaluation target object, the graph of the correlation between the attractiveness difference and the attractiveness conversion value (for example, FIGS. 18A to 18C), which will be described later, and the attractiveness difference The extraction result (for example, FIG. 20) of the relationship with the plurality of sensitivity question items is displayed on the display 410.

<Input device 500>
The input device 500 includes, for example, a keyboard, a mouse, and the like, and is used for the user 10 to perform various instructions (inputs) such as character input by the keyboard, the mouse, and various settings. In the present embodiment, it is used for inputting information (described later) regarding the attributes of the user 10 in the monitor test, input when the user 10 evaluates the design proposal of the evaluation object, and the like. Further, the input device 500 has a timekeeping ability, and after the image of the evaluation object and the words related to the design proposal of the evaluation object are displayed on the display of the display device 400, the user 10 uses the evaluation object. The time taken to complete the evaluation of the design proposal (hereinafter referred to as "response time") is measured.

<Audio output device 600>
The voice output device 600 has a speaker, and can provide the user 10 with words related to the design proposal of the evaluation object by voice.

<Communication device 700>
The communication device 700 has, for example, a communication circuit such as a network interface card (NIC: Network Interface Card), and transmits data to and from an external device through a communication network (not shown).

As shown in FIG. 3, in the control device 300, the CPU 310 executes the design evaluation program, so that the image acquisition unit 351 and the attribute information acquisition unit 352, the attraction degree estimation unit 353, the attraction degree measurement unit 354, and the difference calculation unit 355 are executed. , Display control unit 356, expression providing unit 357, line-of-sight information acquisition unit 358, sensitivity information acquisition unit 359, attraction conversion value calculation unit 360, sensitivity extraction unit 361, and sensitivity output unit 362.

The image acquisition unit 351 acquires at least one image of the evaluation target object stored in the auxiliary storage unit 340, for example. Further, the image acquisition unit 351 can also acquire an image of the evaluation target object from an external server or the like via the communication device 700.

The attribute information acquisition unit 352 acquires the attributes and personas of the user 10 (hereinafter referred to as "information regarding the attributes of the user 10"). The attributes of the user 10 can be classified into, for example, demographic attributes, psychological attributes, and behavioral attributes.

Demographic attributes include, for example, attributes such as age (age), gender, and place of residence (place of residence, place of work). Psychological attributes include, for example, attributes such as values and propensity to consume. The behavioral attributes include, for example, attributes such as product purchase history and action range.

For example, the attribute information acquisition unit 352 displays a questionnaire or the like including a question on the display device 400, and prompts the user 10 to answer the questionnaire. The questionnaire may include, for example, questions such as the age (age) of the user 10 as a demographic attribute, gender, and the consumption behavior of the user 10 as a psychological attribute. Questions about consumer behavior include, for example, whether they often buy things that others buy or branded things.

In addition, by asking the user 10 what he / she recently purchased, the method often used for purchasing the product, etc. in a questionnaire, and obtaining the user 10's web browsing history, etc., ethology You can get the target attribute.

Furthermore, it is also possible to acquire attributes by analyzing the behavior (behavior) of the user 10 based on the image information or the like collected in advance. For example, it is possible to estimate the age from the image of the face of the user 10 or to judge the personality or the like from the behavior of the user 10 or the movement of the line of sight. For example, if the camera in the store captures the behavior of the user 10 in front of the display shelf and the user 10 spends a lot of time looking at the description of the product package before purchasing, he or she thinks carefully or logically. Can be determined to be. Further, by using the questionnaire and the analysis of the behavior of the user 10 together, it is possible to improve the accuracy of determining (acquiring) the attribute.

The persona is, for example, a fictitious consumer image of purchasing products and services related to occupations, job titles, annual income, hobbies, values, lifestyles, etc. Information about the attributes of the user 10 is stored in, for example, the auxiliary storage unit 340.

The attraction degree estimation unit 353 estimates (simulates) the degree of prominence of the image of the evaluation target. In the present embodiment, the ease of visual attention of a human being due to the characteristics of an image is referred to as "significance", and the numerical value representing the degree of remarkableness is referred to as "significance". For example, the part where the human retina reacts strongly when the human looks at the evaluation object has a high degree of prominence. The attractiveness estimation unit 353 generates a saliency map showing the distribution of saliency with respect to the entire image of the evaluation target. When the saliency map is displayed in color, the portion having high saliency is displayed in a warm color such as red, and the portion having low saliency is displayed in a cool color. Blue corresponds to the ground level with the lowest saliency.

In this embodiment, known techniques (for example, C. Koch, L. Itti and E. Niebur, “A model saturation-based visual attention for rapid scene analysis”, IEEE Transaction on Pattern Analysis and Machine Intelligence, Vol. 20 ( Based on the technique described in 1998)), a salience map of the image of the evaluation object is generated. Further, a method for generating a saliency map is also described in, for example, Japanese Patent Application Laid-Open No. 2019-119164, and since it is a known technique, detailed description thereof will be omitted, but the outline thereof is as follows.

The attraction degree estimation unit 353 creates a first image group in which a plurality of images obtained by sampling the image of the evaluation object a plurality of times every other pixel and enlarging them to the same size as the image of the evaluation object are created. A second image group is created by extracting the edge features of each image included in one image group. Further, the attraction degree estimation unit 353 creates a third image group that extracts the characteristics of the brightness of each image included in the first image group, and R (red), G (red), G (red), G (red) of each image included in the first image group. A fourth image group is created by extracting the characteristics of the hues of green), B (blue), and Y (yellow). Then, the attraction degree estimation unit 353 calculates the degree of saliency for each unit region based on at least one of the second image group, the third image group, or the fourth image group. Further, the attraction degree estimation unit 353 has a fifth image group from which the differences between the R (red) and G (green) images included in the fourth image group are extracted, and B (blue) included in the fourth image group. A sixth image group obtained by extracting the difference between the Y (yellow) images is created. Then, the attraction degree estimation unit 353 creates a saliency map by normalizing and adding up the second image group, the third image group, the fifth image group, and the sixth image group.

Further, the attraction degree estimation unit 353 calculates the estimated attraction degree for each area of a plurality of areas (hereinafter, also referred to as "areas") set in the image of the evaluation target object based on the degree of prominence. The method of calculating the estimated attractiveness will be described later.

The attractiveness measuring unit 354 measures the attractiveness of the image of the evaluation target. On the image of the evaluation object, the line-of-sight position of the user is concentrated on a part that stands out due to the characteristics of the image, such as the color and shape of the object, that is, a part that has a high degree of prominence and a part that the user is interested in. Therefore, it is considered that the attractiveness of these parts (or areas) is enhanced. The attraction degree measuring unit 354 calculates the attraction degree (referred to as “measurement attraction degree”) in each area based on the information about the line of sight, that is, the result of the line-of-sight measurement, and stores it in the RAM 320. The details of the method of calculating the degree of attraction for measurement will be described later.

The difference calculation unit 355 calculates the difference between the measurement attraction degree calculated by the attraction degree measurement unit 354 and the estimated attraction degree estimated by the attraction degree estimation unit 353. More specifically, the difference calculation unit 355 calculates the difference between the measured attractiveness and the estimated attractiveness (hereinafter, also referred to as “attracting difference”) for each area and stores it in the RAM 320.

The display control unit 356 controls the display of the image of the evaluation target on the display 410 of the display device 400. The display control unit 356 also controls the display of words provided by the expression providing unit 357. The display control unit 356 displays, for example, an image and words of the evaluation target on the display 410 at the same time.

The expression providing unit 357 provides the user 10 with words regarding the design proposal of the evaluation target object. The words can be appropriately selected according to the type, field, content, etc. of the design proposal of the evaluation target in consideration of the emotional matters that the designer or product maker of the evaluation target wants to appeal to the consumer. Therefore, this word can be a word corresponding to the user 10's sensibility regarding the design of the evaluation object. For example, when the evaluation target is a food package, words such as "delicious", which is a word related to taste, and "like", which is a word related to the taste of the user 10, can be selected. In addition, the words may be configured to be preset by the performer of the monitor test or the like when the monitor test is performed by the subject.

As shown in FIG. 4, the expression providing unit 357 can provide the user 10 with a plurality of words regarding the design proposal of the evaluation target object. These words are set as "sensitivity question items" in the monitor test. FIG. 4 exemplifies the three words “innovative”, “delicious”, and “like” regarding the design proposal of a food package.

The word presented to the user 10 may be a word, a clause, or a sentence including a question, an adjective, a feature of the evaluation object, and the like. Further, instead of displaying the words on the display 410 of the display device 400, or in addition to displaying the words on the display 410, the words and sounds may be provided to the user 10 by voice by the speaker of the voice output device 600. Further, the user 10 may be provided with at least one of tactile, olfactory, and taste information in place of or in addition to visual or auditory information such as words. Hereinafter, in the present specification, such visual, auditory, tactile, olfactory, or taste information is referred to as "expression". Expressions can be sensitive question items. Hereinafter, the case where the expression is a word will be mainly described, but the case where the expression is something other than a word, for example, a smell or a taste, is the same as the case of a word.

The line-of-sight information acquisition unit 358 acquires information on the line-of-sight of the user 10 with respect to the image of the evaluation object displayed on the display device 400. The line-of-sight information acquisition unit 358 acquires, for example, the line-of-sight position of the user 10 on the image of the evaluation target as information on the line-of-sight from the line-of-sight measuring device 200.

The Kansei information acquisition unit 359 calculates the Kansei information of the user 10 regarding the design of the evaluation target object. In the present embodiment, the evaluation of the design proposal is performed, for example, by the user 10 answering the score of the design proposal for each sensitivity question item. The user 10 inputs the score of the design proposal by using the keyboard and the mouse of the input device 500. The Kansei information acquisition unit 359 receives the evaluation of the user 10 for each Kansei question item, calculates the Kansei information indicating the relationship between each Kansei question item and the user 10's Kansei regarding the design proposal based on the evaluation, and RAM320. Save to.

In the line-of-sight information acquisition unit 358, words are provided by the expression providing unit 357, and the words are continuously displayed on the display 410 of the display device 400 until the user 10's evaluation of the words is completed. Get information about the line of sight. That is, the presentation of words by the expression providing unit 357, the acceptance of the evaluation of the user 10 by the sensitivity information acquisition unit 359, and the measurement of the line of sight 20 by the line of sight measuring device 200 can be executed in parallel.

The attraction degree conversion value calculation unit 360 calculates the attraction degree conversion value based on the measured attraction degree and the sensitivity information in each area, and stores it in the RAM 320. The details of the method of calculating the attractiveness conversion value will be described later.

The Kansei extraction unit 361 extracts the relationship between the attractiveness difference and the plurality of Kansei question items (plural expressions) 41 to 43. The sensitivity extraction unit 361 extracts, for example, the relationship between the attractiveness difference and a plurality of words based on the correlation between the attractiveness difference and the attractiveness conversion value. In addition, the sensitivity extraction unit 361 can also extract the above-mentioned relationship by a statistical method or a machine learning method described later.

The sensitivity output unit 362 outputs a graph of the correlation between the attraction degree difference and the attraction degree conversion value, the extraction result of the relationship between the attraction degree difference and the plurality of sensitivity question items 41 to 43, and the like on the display 410.

<Design evaluation method>
5A and 5B are flowcharts illustrating the design evaluation method of the control device 300 according to the first embodiment, and FIG. 6 is a schematic diagram illustrating a design proposal of the evaluation object. The processing of the flowcharts of FIGS. 5A and 5B is realized by the CPU 310 executing the design evaluation program. Further, FIG. 7 is a schematic diagram illustrating a plurality of areas 1 to 4 set for the image of the design proposal, FIG. 8 is a diagram illustrating a salency map of the image of the design proposal, and FIG. 9 is a diagram. It is a figure which exemplifies the estimated degree of attraction in the area 1 to 4 of a design proposal. Further, FIG. 10 is a schematic diagram illustrating a case where the image of the design proposal is displayed in a standard state, and FIG. 11 is a schematic diagram illustrating the distribution of the line-of-sight position of the user 10 who is viewing the image of the design proposal. .. 12 and 13 are diagrams illustrating the measurement attraction degree and the attraction degree difference in areas 1 to 4 of the image of the design proposal, respectively, and FIGS. 14A to 14C show the image of the design proposal and the display of each sensitivity question item. It is a schematic diagram which illustrates. FIG. 15 is a diagram illustrating the degree of attraction for measurement in areas 1 to 4 of the image of the design proposal for each sensitivity question item, and FIG. 16 illustrates the score input for the design proposal by the user 10 for each sensitivity question item. FIG. 17 is a diagram illustrating the attractiveness conversion value in the areas 1 to 4 of the image of the design proposal for each sensitivity question item. 18A to 18C are graphs illustrating the correlation between the attractiveness difference and the attractiveness conversion value for each sensitivity question item.

In the following, for example, a case of evaluating a package design of a certain food will be illustrated, and a design evaluation method of the control device 300 will be described. In the present embodiment, it is assumed that after the design proposal of the package is completed, the user 10 acts as a subject and conducts a monitor test for evaluating the design proposal.

In the monitor test, the design evaluation system 100 displays an image of the design proposal and words related to the design on the display 410, and with respect to the image of the design proposal, the degree of attraction based on the user 10's sensibilities and the relationship with a plurality of words. To extract. More specifically, it is as follows.

First, an image of the evaluation target is acquired (step S101). The image acquisition unit 311 acquires, for example, an image of a package of a certain food among a plurality of images stored in the auxiliary storage unit 340 as an image of the evaluation target (see FIG. 6). Hereinafter, the image of the design proposal of this evaluation object is referred to as a "design proposal image". The design proposal of FIG. 6 is, for example, a novel design proposal that combines an “hourglass” and a “food ingredient”.

Next, the information regarding the attribute of the user 10 is acquired (step S102). The attribute information acquisition unit 352 reads, for example, information about the attributes of the user 10 from the auxiliary storage unit 340. For example, in this embodiment, it is assumed that the user 10 is a woman in her twenties and works for a company in the city center. In this case, the information regarding the attributes of the user 10 includes information such as gender: female, age: 20s, location area: Tokyo, and the like.

Next, the degree of attraction is estimated for each area of the image of the evaluation target (step S103). As shown in FIG. 7, the attraction degree estimation unit 353 identifies an object (character, figure, etc.) in the design proposal image 30 by using, for example, a known image recognition technique, and designs based on the identification result. A plurality of areas (for example, areas 1 to 4) divided so as not to overlap each other are set for the draft image 30. Area 1 is an area including the character string "hourglass of salt 糀" at the right end of the design proposal image 30, and area 2 is an area including an illustration of "hourglass" at the center of the design proposal image 30. Further, the area 3 is an area including the character string "0000OO / CORPORATION" at the lower part of the design proposal image 30, and the area 4 is a mariage of "oysters and lemons" at the left end of the design proposal image 30. It is an area containing the character string.

Subsequently, as shown in FIG. 8, the attractiveness estimation unit 353 generates a saliency map for the design proposal image 30. In the example of the figure, in the "hourglass of salt 糀" in area 1, the letters "糀" (A1) and "total" (A2), the central part (A3) of the illustration of "hourglass" in area 2, and the area There is a highly prominent part on the character 3 (A4 to A6). In addition, the outline of the illustration of the "hourglass" in area 2 (A7) and the text (A8) of the "mariage of oysters and lemons" in area 4 have a medium degree of prominence. In the case of color display, the portions A1 to A6 having high saliency may be displayed in colors such as red and yellow, and the portions A7 and A8 having moderate saliency may be displayed in colors such as green.

The part of the design proposal image 30 with a high degree of prominence is considered to be highly attractive because it attracts the attention of the user 10. The attraction degree estimation unit 353 calculates the estimated attraction degree in each area. In the present embodiment, the estimated degree of attraction in each area represents the degree of ease of visual attention of the user 10 due to the characteristics of the design proposal image 30, for example, the degree of prominence in each area for all areas. It can be defined as a percentage of the sum of saliency. Therefore, the higher the degree of saliency, the higher the estimated degree of attraction in each area, and the lower the degree of saliency, the lower the estimated degree of attraction in each area. The total estimated attraction is 1.0 for all areas.

FIG. 9 illustrates the estimated degree of attraction in areas 1 to 4. In the example of the figure, the estimated attractiveness in area 2 is 0.52, which is higher than the estimated attractiveness in

other areas

1, 3 and 4. Therefore, it is considered that the area 2 is more likely to attract the attention of the user 10 than the

other areas

1, 3 and 4. That is, the design proposal image 30 originally has an attractiveness of more than 50% in the area 2 due to the prominence of the area 2. Therefore, it is considered that the area 2 of the design proposal image 30 attracts the attention of not only the user 10 but also other users.

Next, the image of the evaluation target is displayed (step S104). As shown in FIG. 10, the display control unit 356 controls the design proposal image 30 to be displayed on the display 410 of the display device 400. Hereinafter, in the present embodiment, displaying only the design proposal image 30 without accompanying the sensitivity question item is referred to as "display in the standard state". The display in the standard state is performed for a predetermined time (for example, 10 seconds).

Next, the information regarding the line of sight of the user 10 with respect to the image of the evaluation object displayed on the display device 400 is acquired (step S105). The line-of-sight measuring device 200 measures the line of sight of the user 10 with respect to the design proposal image 30 displayed on the display device 400, and transmits information about the line of sight to the line-of-sight information acquisition unit 358. Before the start of the monitor test, the line-of-sight measuring device 200 is calibrated according to the eyes of the user 10. In the calibration, the geometric features of the user 10's eyes are acquired in order to accurately calculate the user 10's line-of-sight position.

The line-of-sight measuring device 200 continuously measures the line-of-sight of the user 10 over the predetermined time, and during this period, the display control unit 356 displays the locus of the line-of-sight position of the user 10 on the display 410, that is, the distribution of the line-of-sight position at the predetermined time. Is displayed. As a result, the person performing the monitor test or the like can easily grasp the change over time in the line-of-sight position of the user 10 on the display surface of the display 410. The data of the distribution of the line-of-sight position at a predetermined time is stored in the RAM 320.

When the line-of-sight position of the user 10 stays at the same position in the design proposal image 30 or its surroundings for a long time, it is estimated that the user 10 is gazing at the position or its surroundings, and the user 10 is gazing at the position or its surroundings. It is considered that you are interested in the contents of the part.

As shown in FIG. 11, the distribution of the line-of-sight position can be exemplified by a heat map. For example, in the case of a color heat map, the part where the line of sight is concentrated is painted in a warm color system (the part where the line of sight is dark in the figure), and the part where the line of sight is weakly concentrated is painted in a cool color system. For example, in the lower part of the illustration of the "hourglass", there is a concentrated zone (dark part, the same applies hereinafter) B1 with a strong line of sight, and in the upper part, there is a concentrated zone B2 (a light colored part, the same applies hereinafter).

Next, the degree of attraction is measured for each area of the image of the evaluation target (step S106). The attraction degree measuring unit 354 measures the attraction degree for each area of the design proposal image 30 based on the information regarding the line of sight. In the present specification, the measurement attraction is defined as, for example, the ratio of the residence time of the line-of-sight position in each area to the total residence time of each area. Here, the total residence time of each area is equal to the above-mentioned predetermined time. Therefore, the longer the line-of-sight position stays in the same area, the higher the degree of attraction for measurement in that area, and the shorter the position, the lower the degree of attraction for measurement. The total degree of attraction for measurement is 1.0 for all areas.

FIG. 12 illustrates the degree of attraction for measurement in areas 1 to 4. In the example of the figure, the measurement attraction degree in the area 2 is 0.82, which is higher than the measurement attraction degree in the other areas 1, the area 3, and the area 4. Therefore, it is considered that the area 2 attracts the attention of the user 10 more than the

other areas

1, 3 and 4. In the example of the figure, the line-of-sight position of the user 10 is in the area 2 at a time exceeding 80% of the predetermined time.

Next, the difference between the measured attractiveness and the estimated attractiveness is calculated for each area (step S107). The difference calculation unit 355 calculates the difference between the measured attractiveness and the estimated attractiveness (attracting difference) for each area. FIG. 13 illustrates the difference in attractiveness in areas 1 to 4. As described above, it is considered that the measured attractiveness includes the attractiveness (that is, the estimated attractiveness) due to the prominence of the design proposal image 30. Therefore, by subtracting the estimated attractiveness from the measured attractiveness, the attractiveness based on the sensibility of the user 10 can be calculated.

When the attractiveness difference is approximately 0 (zero) in each area of the design proposal image 30 (or where the attractiveness difference is approximately 0 in the design proposal image 30), the attraction due to the prominence of the design proposal image 30 Only the degree is included, and the degree of attraction based on the user 10's sensibilities is considered to be approximately 0 or extremely small. Further, when the attractiveness difference is a positive value (or the place where the attractiveness difference is a positive value in the design proposal image 30), the attractiveness based on the user's sensibilities is included in addition to the attractiveness caused by the prominence. It is thought that it has been done. It is considered that this is because the sensibility of the user 10 is stimulated by the above-mentioned portion of the design proposal image 30, and the line-of-sight position of the user 10 is moved to the above-mentioned portion. On the other hand, when the attractiveness difference is a negative value (or where the attractiveness difference is a negative value in the design proposal image 30), or when the attractiveness difference is a positive value but small, the attractiveness due to the remarkableness is caused. It is probable that the actual degree of attraction was almost the same as or lower than that, even though it was estimated.

As described above, the attractiveness difference corresponds to the attractiveness based on the sensibility of the user 10, and the higher the attractiveness difference, the greater the contribution of the user 10's sensibility to the measured attractiveness.

Next, as shown in FIG. 5B, a plurality of sensitivity question items 41 to 43 regarding the design proposal of the evaluation target are provided to the user 10 (step S108). The expression providing unit 357 selects three words, for example, "innovative", "delicious", and "like" as the sensitivity question items.

As shown in FIGS. 14A to 14C, for example, while the design proposal image 30 is displayed, the three sensitivity question items 41 to 43 are displayed one by one in order, and the user 10 displays these sensitivity question items 41. It is assumed that the design proposal is evaluated with respect to ~ 43. In this case, the display control unit 356 controls, for example, to display these three sensitivity question items 41 to 43 on the display 410 one by one in order at predetermined time intervals. Here, the predetermined time interval is a time during which the user 10 can sufficiently evaluate the design proposal for each sensitivity question item, and can be set to, for example, 10 seconds in accordance with the "time limit" described later.

When displaying each sensitivity question item on the display 410, the display control unit 356 asks the user 10 for each sensitivity question item to be displayed, for example, "Enter a score with the keyboard (1 to 10 points). ) ”, A message requesting the evaluation of the design proposal is also displayed.

Also, the monitor test is a test related to latent consciousness. Therefore, the user 10 evaluates the design proposal for each sensitivity question item in the shortest possible time, has three sensitivity question items, and switches to the next sensitivity question item at 10-second intervals, etc. in the monitor test. Get an explanation before the start. The display control unit 356 may be configured to include the time limit in the message and display it on the display 410 in order to confirm the time limit to the user 10. Further, the sensitivity question items are not limited to three, and may be two or four or more.

Next, the information regarding the line of sight of the user 10 with respect to the image of the evaluation object displayed on the display device 400 is acquired (step S109). The line-of-sight measuring device 200 measures the line of sight of the user 10 with respect to the design proposal image 30 displayed on the display device 400, and transmits information about the line of sight to the line-of-sight information acquisition unit 358.

User 10 evaluates the design proposal according to his / her own sensibility for each sensibility question item. Specifically, the user 10 inputs the score of the design proposal by using the keyboard and the mouse of the input device 500.

Further, the input of the score can be set so as to be performed within a predetermined time limit (for example, 10 seconds). For example, if the answer is not given within the time limit, the input of the score for the displayed question sensitivity item is completed, and if there is an undisplayed sensitivity question item, the answer is moved to the next sensitivity question item. Can be controlled to do so. This is to obtain the evaluation of the design proposal made according to the sensibility of the user 10 and to not put an undue burden on the user 10 (subject). It should be noted that the present invention is not limited to such a case, and if even one of the three sensitivity question items is not answered within the time limit, the monitor test may be forcibly terminated.

In this embodiment, a time limit is set for the answer so that the user 10 can intuitively evaluate the design proposal based on his / her own sensibility without evaluating the design proposal based on knowledge or logical thinking. Sensitive question items for which the answer of the user 10 exceeds the time limit are treated as no answer or invalid answer.

If there is no answer or an invalid answer, it is not possible to obtain specific information about the user 10's sensibility, but it is also considered that the user 10 could not evaluate the design proposal within the time limit. Be done. Therefore, even if the user 10 cannot evaluate the design proposal within the time limit and cannot obtain a valid answer, the information regarding the line of sight of the user 10 is saved and the user 10 cannot evaluate the design proposal. It can be used as information about the line of sight.

Although it is possible to redo the test for sensitive question items that were unanswered or invalid answers, the measurement results obtained by redoing the test can be used for the purpose of supplementing the measurement results of valid answers.

In addition, the Kansei information acquisition unit 359 has a high possibility that the answer of the user 10 is a latent (according to Kansei) answer or a rational judgment based on the length of the answer time for each Kansei question item. Can be determined. That is, it is presumed that when the response time is short, the user 10 makes a potential judgment, and when the response time is long, the user 10 makes a rational judgment.

The temporal threshold for dividing potential / rational is not particularly limited, and can be appropriately set according to the complexity of the design of the evaluation object. For example, if the design is not complicated as in the design proposal of FIG. 6, the threshold value can be set to 4 to 5 seconds. On the other hand, when the design is complicated or the product description is included in the design, the threshold value is set to 9 to 10 seconds after extending the time limit in consideration of the time for the user 10 to read the description. It may be set to a degree.

Next, the Kansei information acquisition unit 359 determines whether or not the evaluation of the design proposal by the user 10 has been completed for each Kansei question item (step S110). As a result of the determination, if the evaluation is not completed (step S110: NO), the process returns to the process of S109.

On the other hand, when the evaluation of the design proposal is completed (step S110: YES), the measurement attraction degree in each area is calculated for each sensitivity question item (step S111). As described above, the degree of attraction for measurement is calculated as the ratio of the residence time of the line-of-sight position in each area to the total residence time of each area, and the total residence time of each area here is the sensitivity question. Equal to the response time for the item. Therefore, the total residence time in each area may differ depending on the sensitivity question items 41 to 43. FIG. 15 illustrates the degree of attraction for measurement in areas 1 to 4 for each sensitivity question item. For example, the degree of attraction for measurement when "innovative" is presented is 0.04, 0.83, 0.10, and 0.03, respectively, in areas 1 to 4. The total of all areas of measurement attraction is 1.0.

Next, the sensitivity information of the user 10 regarding the design of the evaluation target is calculated (step S112). The Kansei information acquisition unit 359 acquires the score input by the user 10 when presenting each Kansei question item with respect to the design proposal. FIG. 16 illustrates the points input by the user 10 regarding the design proposal when each sensitivity question item is presented. In the example shown in the figure, the scores for presenting "innovative", "delicious", and "like" are 6, 3, and 8, respectively.

The Kansei information acquisition unit 359 receives the above points as the evaluation of the user 10 for each Kansei question item, and calculates the Kansei information indicating the relationship between each Kansei question item and the user 10's sensitivity regarding the design proposal based on the evaluation. do.

For example, the Kansei information acquisition unit 359 calculates Kansei information for each Kansei question item so as to be proportional to the evaluation of the user 10 for each Kansei question item. For example, as illustrated in FIG. 16, when the user 10's evaluation of the design proposal is 6 points, 3 points, and 8 points for "innovative", "delicious", and "like", respectively. Set the ratio of the sensitivity information of each sensitivity question item to 6: 3: 8.

Next, the attractiveness conversion value is calculated for each area (step S113). The Kansei information acquisition unit 359 calculates an attractiveness conversion value for each area based on the measured attractiveness and the Kansei information in each area for each Kansei question item. In the present embodiment, the attractiveness conversion value is represented by the product of the measured attractiveness and the sensitivity information in each area. In the example of FIG. 15, when the sensitivity question item is "innovative", the measurement attraction degree of area 1 is 0.04 (the part shown in gray in FIG. 15), and the sensitivity information is 6 (in gray in FIG. 16). The part shown). Therefore, the attractiveness conversion value is 0.04 × 6 = 0.24 ≈0.2 (the part shown in gray in FIG. 17). It is calculated in the same way for other areas. That is, the attractiveness conversion value is a value obtained by weighting the measured attractiveness of each area with the sensitivity information corresponding to each sensitivity question item. FIG. 17 illustrates the attractiveness conversion value of each area for each sensitivity question item.

Next, the relationship between the difference in attractiveness and the plurality of sensitivity question items 41 to 43 is extracted (step S114). As shown in FIGS. 18A to 18C, the sensitivity output unit 362 outputs, for example, a graph showing the correlation between the attractiveness difference and the attractiveness conversion value to the display 410. As a result, the sensibilities of the user 10 for the plurality of sensibility question items 41 to 43 are visualized with respect to the design proposal image 30.

In the example shown in FIG. 18A, when the sensitivity question item is "innovative", the attraction degree difference is a negative value in

areas

1, 3 and 4, and the attraction degree difference is a positive value at 0.3 in area 2. ing. The same applies to the sensibility question items of FIGS. 18B and 18C for "delicious" and "like". Therefore, in the examples shown in FIGS. 18A to 18C, the sensitivity extraction unit 361 has a relatively attractive degree difference of 0.3 in the area 2 for any of "innovative", "delicious" and "like". It is a large positive value, and in

areas

1, 3 and 4, the relevance that the attractiveness difference is a negative value is extracted. Further, the sensibility extraction unit 361 determines that the degree of attraction based on the sensibility of the user 10 in the area 2 is high and the degree of attraction based on the sensibility of the user 10 in the

areas

1, 3 and 4 is low based on this relationship.

Further, in FIG. 17, when the sensitivity question items 41 to 43 are "innovative", "delicious", and "like", the attractiveness conversion values of the area 2 are 5.0, 1.0, and respectively. It is 7.4. The attractiveness conversion values of "innovative" and "like" are higher than the attractiveness conversion values of "delicious". That is, when the sensitivity question items are "innovative" and "like", when the attractiveness difference is large, the attractiveness conversion value is also large, and the attractiveness difference and the attractiveness are compared with the case where "it looks delicious". High correlation with converted value.

The attractiveness difference reflects the attractiveness based on the sensibility of the user 10, but does not take into account the influence of the sensibility question item on the user 10. On the other hand, since the attraction degree conversion value is the product of the measurement attraction degree and the sensitivity information, the influence of the sensitivity question item on the user 10 is added to the sensitivity information. Therefore, in a graph of Cartesian coordinates with the attraction degree difference on the horizontal axis and the attraction degree conversion value on the vertical axis, the higher the correlation between the attraction degree difference and the attraction degree conversion value, the more the influence of the sensitivity question item on the user 10. It is judged to be strong.

In this way, the sensitivity extraction unit 361 extracts the relationship between the attraction degree difference and the plurality of sensitivity question items based on the attraction degree difference and the attraction degree conversion value. This makes it possible to determine which sensitivity question item causes the difference in the degree of attraction to be large.

In addition, the Kansei extraction unit 361 extracts the relationship between the attractiveness difference and a plurality of Kansei question items and user attributes in each area. As described above, the attributes of the user 10 who is the subject are set to be in their twenties, have a female gender, and have a place of residence in a company in the city center. It is considered that there is a possibility that the area 2 attracts the eyes of the user 10 as compared with other areas due to such an attribute (background) of the user 10. Further, due to the attributes of the user 10, it is considered that the impression of "innovative" and "like" is good in the area 2, while the appeal of "delicious" is weak.

As described above, in the flowcharts shown in FIGS. 5A and 5B, one design proposal image 30 of the evaluation target is acquired, and the estimated attraction degree is calculated based on the characteristics of the design proposal image 30. Further, the design proposal image 30 is displayed on the display device 400, and a plurality of sensitivity question items 41 to 43 regarding the design of the evaluation object are provided to the user 10. Subsequently, information regarding the line of sight of the user 10 with respect to the design proposal image 30 displayed on the display device 400 is acquired, and the measurement attraction degree of the image is calculated based on the information. Then, the attractiveness difference, which is the difference between the estimated attractiveness and the measured attractiveness, is calculated, and the relationship between the attractiveness difference and the plurality of Kansei question items is extracted.

Before performing the processing of the flowcharts shown in FIGS. 5A and 5B, a process of displaying a dummy image (not shown) and a dummy sensitivity question item on the display 410 and having the subject practice the contents of the monitor test is included. It may be configured as follows. As a result, the subject can face the actual monitor test while being accustomed to the contents of the monitor test. It should be noted that the measurement results of the practice using the dummy image are not added to the measurement results of the actual monitor test.

Further, in the above example, the case where the Kansei information acquisition unit 359 acquires the evaluation (score) of the user 10 regarding the design proposal and calculates the Kansei information based on the evaluation of the user 10 has been described. The calculation method is not limited to such cases. The Kansei information acquisition unit 359 may be configured to acquire the answer time of the user 10 for the Kansei question item and calculate the Kansei information based on the answer time. Sensitivity information can be calculated, for example, as the reciprocal of the response time.

Further, the camera mounted on the design evaluation system 100 captures the facial expression and posture of the user 10, and the captured image is subjected to image processing such as pattern recognition to perform the sensitivity information of the user 10 for the sensitivity question item. Can also be estimated. As a result, it is possible to acquire Kansei information while suppressing the burden on the user 10 depending on the answer.

In addition, in collecting data for a large number of subjects, it is considered effective to conduct a monitor test in an internet environment using the Internet or the like.

(Modification 1)
FIG. 19 is a diagram illustrating each variable when the relationship between the attractiveness difference and the plurality of Kansei question items 41 to 43 is extracted using the multiple regression analysis, and FIG. 20 is a diagram illustrating the multiple regression analysis in FIG. It is a figure which illustrates the regression coefficient β ₂ to β ₃ calculated by.

In FIGS. 18A to 18C, the case where the correlation between the attractiveness difference and the attractiveness conversion value is displayed in a graph for the plurality of Kansei question items 41 to 43 is illustrated, but this embodiment is limited to such a case. Not done. For example, a statistical method (eg, simple regression analysis or multiple regression analysis) can be adopted to extract the relationship between the attractiveness difference and at least one sensitivity question item (and the attribute of the user 10).

In the statistical method, the sensitivity question item (and the attribute of the user 10) is used as the explanatory variable, and the attractiveness difference is used as the objective variable, and the correlation between the explanatory variable and the objective variable is analyzed by statistical processing.

For example, a case where multiple regression analysis is performed will be described. As shown in FIG. 19, when the number of subjects is n, the attractiveness conversion value for each sensitivity question item is X _ki , ε is an error term, the regression coefficient is β _k , and the attractiveness difference in each area is Y _i . The multiple regression equation is expressed by Y _i = β ₁ + β ₂ X _2i + β ₃ X _3i + ... + β _k X _ki + ε _i (i = 1, 2, 3, ..., N). Here, β ₁ , β ₂ , β ₃ , ..., β _k represent the influence of each explanatory variable, and the influence on the attractiveness difference can be quantitatively expressed by the magnitude of this regression coefficient.

This multiple regression equation includes k unknown regression coefficients, but these regression coefficients can be calculated by using the least squares method. Further, the error term is expressed by ε _i = Y _i − (β ₁ + β ₂ X _2i + β ₃ X _3i +… + β _k X _ki ), and the sum of squares S = Σε _i ² is calculated as the minimum ε _i . .. In order to minimize S, prepare the number of equations with the first partial derivative of each explanatory variable set to zero.

For example, for β ₁ , S = Σ {Y _i- (β ₁ + β ₂ X _2i + β ₃ X _3i + ... + β _k X _ki )} ² by modifying the above equation.
= Σ {Y _i ² -2Y _i (β ₁ + β ₂ X _2i + β ₃ X _3i +… + β _k X _ki ) + (β ₁ + β ₂ X _2i + β ₃ X _3i +… + β _k X _ki ) ² }
= Σ {Y _i ² -2Y _i (β ₁ + β ₂ X _2i + β ₃ X _3i +… + β _k X _ki ) + β ₁ ² + 2β ₁ (β ₂ X _2i + β ₃ X _3i +… + β _k X _ki ) + ( β ₂ X _2i + β ₃ X _3i + ... + β _k X _ki ) ² }.

The first partial derivative for β ₁ is δS / δβ ₁ = Σ {-2Y _i + 2β ₁ + 2 (β ₂ X2 _i + ... + β _k X _ki )} = -2Σ {Y _i -β _1- (β ₂ X) _2i + ... + β _k X _ki )}, and this first partial derivative is zero.

Similarly, the first partial differential equation for β ₂ is δS / δβ ₂ = 2Σ {X ₂₂ β2-X ₂ (Y-β ₁ -β ₃ X ₃ -...- β _k X _ki )}, and this first-order partial differential. Let the differential formula be zero. As a result, a system of equations of order k with k variables is obtained, and β ₁ , β ₂ , ..., Β _k can be uniquely obtained. Therefore, as shown in FIG. 20, by solving these simultaneous equations, it is possible to derive regression coefficients β ₂ to β ₄ indicating the degree of influence of the sensitivity question items.

(Modification 2)
A method of analyzing the relationship between an explanatory variable and an objective variable by machine learning, with the sensitivity question item (and the attribute of user 10) as the explanatory variable and the attractiveness difference as the objective variable (hereinafter referred to as "machine learning method"). Can also be adopted.

In the machine learning method, the sensitivity extraction unit 361 has a machine learning model in which supervised machine learning is performed using the sensitivity information (and information about the attributes of the user 10) as an explanatory variable and the attractiveness difference as an objective variable. A machine learning model is used to extract the relationship between the attractiveness difference and multiple sensitivity question items.

(Learning device)
FIG. 21 is a schematic block diagram illustrating a learning device 370 that generates a machine learning model based on the difference in attractiveness and the sensitivity information (and information on the attributes of the user 10). The control device 300 functions as a learning device 370 by the CPU 310 executing a learning program.

The learning device 370 has a learning data storage unit 371 and a model generation unit 372. The learning data storage unit 371 stores learning data (teacher data) including n sets of explanatory variables and objective variables acquired by performing a monitor test on n subjects. The learning data may include, for example, sensitivity information as an explanatory variable and attractiveness difference as an objective variable. Further, information about the attribute of the user 10 may be further included as an explanatory variable.

The model generation unit 372 generates a machine learning model that machine-learns the relationship between the attractiveness difference and the sensitivity information (and the attribute of the user 10) based on the learning data stored in the learning data storage unit 371. The generated machine learning model is transmitted to the Kansei extraction unit 361 of the control device 300, and can be used in extracting the relationship between the attractiveness difference and the plurality of Kansei question items 41 to 43.

By conducting a monitor test with a sufficiently large number of subjects and accumulating learning data, it is possible to improve the estimation accuracy of the attractiveness difference conversion value for a specific Kansei question item. As a result, it is possible to simulate the emotional effect of the persona on the design proposal without going through a monitor test by the subject.

In addition, when conditions such as persona and sensibility information are specified for a certain purchase target group using the learning device 370, the specified conditions are selected from the design database in which a large number of designs are stored in advance. It is possible to realize a system that extracts multiple design proposals that are close to each other. The designer in charge of the design proposal of the product package can utilize these multiple designs as a reference when creating a new design proposal or changing the design proposal. Furthermore, it is conceivable to create multiple patterns on the system side to change the design proposal and present it to the designer.

The control device 300 of the present embodiment described above has the following effects.

The control device 300 of the present embodiment calculates the degree of attraction based on the sensibility of the user 10 with respect to the design proposal image 30, and determines the relationship between the degree of attraction and the plurality of sensibility question items 41 to 43 regarding the design proposal image 30. Extract. Therefore, the user 10's sensibility regarding the design proposal image 30 can be visualized.

In addition, the designer who devised the design of the product package can determine the direction to be improved with respect to the sensitivity target required by the brand owner of the product based on the above-mentioned relationship extracted by the control device 300.

Also, with the persona that is the purchase target group in mind, a preferable design proposal is decided at the planning stage of the new product from multiple product package design proposals, but the above-mentioned extracted relationships are designed by the product brand owner. It can also be useful when deciding on a plan. For example, in product package design, where is the place where the degree of attraction based on human sensibilities is high on the design proposal image for a certain sensibility question item, and how the degree of attraction changes in different sensibility question items. It will be easier to compare. Therefore, when using it for product marketing, if the sensitive points that you want to appeal to the purchasing target group are clear at the stage of preparing multiple design proposals, which one should be selected from multiple design proposals? It can be used as a material for judging whether it is good or not.

In addition, since it is possible to grasp the magnitude of the influence that each sensitivity question item has on the degree of attraction based on the human sensitivity according to the persona set in the product sales strategy, the design proposal is selected in consideration of the persona of the purchasing target group. You can get an index when you do.

(Second embodiment)
In the first embodiment, a case has been described in which the relationship between the attractiveness difference (that is, the attractiveness based on the user's sensibility) and the plurality of sensibility question items is extracted with respect to the image of one evaluation object. In the second embodiment, a case where the relationship between the attractiveness difference and the plurality of Kansei question items is extracted with respect to the images of the plurality of evaluation objects will be described. In the following, in order to avoid duplication of description, detailed description of the same configuration as that of the first embodiment will be omitted.

22 is a schematic view of the design evaluation system 100 according to the second embodiment as viewed from above, and FIGS. 23A and 23B are flowcharts illustrating a design evaluation method of the control device 300 according to the second embodiment. .. The processing of the flowcharts of FIGS. 23A and 23B is realized by the CPU 310 executing the design evaluation program. Further, FIG. 24 is a schematic diagram illustrating the design proposal 1, FIGS. 25A and 25B are schematic views illustrating the area setting of the images of the

design proposals

1 and 2, respectively, and FIGS. 26A and 26B are design proposals, respectively. It is a schematic diagram which illustrates the saliency map of the image of 1 and 2. 27A and 27B are diagrams illustrating the estimated attractiveness in areas 1 to 4 of the images of the

design proposals

1 and 2, and FIG. 28 illustrates the case where the images of the

design proposals

1 and 2 are displayed in the standard state. It is a schematic diagram. Further, FIG. 29 is a schematic diagram illustrating the distribution of the line-of-sight position of the user who is viewing the images of the

design proposals

1 and 2, and FIGS. 30 and 31 are measurements in areas 1 to 4 of the images of the

design proposals

1 and 2. It is a schematic diagram which exemplifies each of the attractiveness degree and the attractiveness degree difference. Further, FIGS. 32A to 32C are schematic views illustrating the images of the

design proposals

1 and 2 and the display of each sensitivity question item, and FIGS. 33A to 33C are the line-of-sight positions of the user when each sensitivity question item is displayed. It is a schematic diagram which illustrates the distribution of. Further, FIGS. 34A to 34C are schematic views illustrating the degree of attraction for measurement for each sensitivity question item, and FIG. 35 is a diagram illustrating the calculation result of the reaction intensity when a design proposal is selected for each sensitivity question item. be. Further, FIG. 36 is a graph illustrating the attractiveness conversion values in the areas 1 to 4 of the images of the

design proposals

1 and 2 for each sensitivity question item, and FIGS. 37A to 37C show the attraction degree difference for each sensitivity question item. It is a graph exemplifying the correlation between and the attractiveness conversion value.

As shown in FIG. 22, in the present embodiment, the user 10 simultaneously uses the

images

31 and 32 of the two evaluation objects and the sensitivity question item (for example, “innovative”) 41 related to the design of the two evaluation objects. Present to. The user 10 selects a design proposal that seems to be suitable for the sensitivity question item from a plurality of design proposals of the evaluation target according to his / her own sensitivity.

First, as shown in FIG. 23A, an image of the evaluation target is acquired (step S201). In this embodiment, in addition to the design proposal (design proposal 2) used in the first embodiment, the design proposal (design proposal 1) shown in FIG. 24 is used. The design plan 1 is a design plan in which a concrete cooking example of "ingredients" is displayed by an illustration. The image acquisition unit 351 acquires the design proposal 1 image 31 and the design proposal 2 image 32 (hereinafter, also referred to as “

design proposal

1, 2

image

31, 32”).

Next, the information about the attribute of the user 10 is acquired (step S202). The attribute information acquisition unit 352 reads, for example, information about the attributes of the user 10 from the auxiliary storage unit 340.

Next, the degree of attraction is estimated for each area of the image of the evaluation target (step S203). As shown in FIGS. 25A and 25B, the attraction degree estimation unit 353 sets a plurality of areas (for example, areas 1 to 4) for each of the

design proposals

1 and 2

images

31 and 32. Further, as shown in FIGS. 26A and 26B, the attraction degree estimation unit 353 generates a saliency map for the

design proposals

1 and 2 and the

images

31 and 32, and stores the saliency map in the RAM 320. FIG. 27A exemplifies the estimated degree of attraction in areas 1 to 4 of the design proposal 1 image 31. Further, FIG. 27B illustrates the estimated degree of attraction in areas 1 to 4 of the design proposal 2 image 32.

Next, the image of the evaluation target is displayed (step S204). As shown in FIG. 28, in the standard state, the display control unit 356 causes the display 410 of the display device 400 to display the

design proposals

1 and 2

images

31 and 32 side by side at predetermined intervals.

Next, the information regarding the line of sight of the user 10 with respect to the image of the evaluation object displayed on the display device 400 is acquired (step S205). The line-of-sight measuring device 200 measures the line-of-sight of the user 10 with respect to the

design proposals

1 and 2

images

31 and 32 displayed on the display device 400, and transmits information about the line-of-sight to the line-of-sight information acquisition unit 358.

As shown in FIG. 29, the distribution of line-of-sight positions can be exemplified by a heat map. For example, in the central part of the illustration of the cooking example of the area 3 of the design proposal 1 image 31, there is a concentration zone C1 having a slightly strong line of sight. In the design proposal 2 image 32, there are strong concentration zones C2 to C4 on the characters of the area 1 and at the upper part and the lower part of the “hourglass” illustration of the area 2, respectively.

Next, the degree of attraction is measured for each area of the image of the evaluation target (step S206). The attraction degree measurement unit 354 calculates the measurement attraction degree for each area of the

design proposals

1 and 2

images

31 and 32 based on the information regarding the line of sight. FIG. 30 illustrates the degree of attraction for measurement in areas 1 to 4 of the

design proposals

1 and 2

images

31 and 32.

Next, the difference between the measured attractiveness and the estimated attractiveness is calculated for each area (step S207). The difference calculation unit 355 calculates the difference between the measured attraction degree and the estimated attraction degree (attraction degree difference) for each area of the

design proposals

1 and 2

images

31 and 32. FIG. 31 illustrates the difference in attractiveness in areas 1 to 4 of the

design proposals

1 and 2

images

31 and 32.

Next, a plurality of sensibility question items regarding the design proposal of the evaluation target are provided to the user (step S208). In the present embodiment, the control device 300 provides the user 10 with three sensitivity question items 41 to 43 of "innovative", "delicious", and "like" with respect to the design proposal 1 and the design proposal 2. Similar to the first embodiment, these Kansei question items are sequentially displayed one by one on the display 410 of the display device 400 at predetermined time intervals (for example, 10 seconds) (FIGS. 32A to 32C). See). In the example shown in FIG. 32A, the sensitivity question item "innovative" 41 is displayed at a position between the design proposal 1 image 31 and the design proposal 2 image 32.

When the display control unit 356 displays the sensitivity question items on the display 410, the display control unit 356 asks the user 10 about each sensitivity question item to be displayed, for example, "Design proposal 1 and design proposal matching the displayed words. A message asking you to select a design proposal is also displayed, such as "Please select from proposal 2". The selection of the proposed design can be set to be done within a predetermined time limit (eg, 10 seconds).

By displaying the sensitivity question item at the display position between the design proposal 1 image 31 and the design proposal 2 image 32, it is possible to prevent a psychological bias from being applied to any of the design proposals. , The position of the sensitivity question item is not limited to the intermediate display position.

Next, the information regarding the line of sight of the user 10 with respect to the image of the evaluation object displayed on the display device 400 is acquired (step S209). The line-of-sight measuring device 200 measures the line-of-sight of the user 10 with respect to the

design proposals

1 and 2

images

User 10 selects a design proposal according to his / her own sensibility for each sensibility question item. Specifically, the user 10 selects one of the design proposals 1 and the design proposal 2 by using the keyboard and the mouse of the input device 500 in order to determine the superiority or inferiority of these design proposals.

The input device 500 measures the response time from when each sensitivity question item is displayed until the user 10 selects the design proposal 1 or the design proposal 2 using the keyboard or the mouse. Further, the line-of-sight information acquisition unit 358 acquires information on the user's line of sight with respect to the

design proposals

1 and 2

images

31 and 32 displayed on the display device 400, and obtains the movement history of the user's line of sight within the response time of the user 10. measure. The movement history is displayed on the display 410 of the display device 400 as a distribution of line-of-sight positions (see FIGS. 33A to 33C).

For example, as shown in FIG. 33A, when the sensitivity question item is "innovative", the heat map of the design proposal 1 image 31 is relatively strong in the part where the upper character of the illustration of the cooking example is written. There is a concentrated zone C5 of the line of sight and a weak concentrated zone C6 in the lower part where the characters are written. On the other hand, in the heat map of the design proposal 2 image 32, there is a weak concentration zone C7 at the upper part of the illustration of the “hourglass” and a strong concentration zone C8 at the lower part.

Next, the Kansei information acquisition unit 359 determines whether or not the selection regarding the design proposal by the user 10 has been completed for each Kansei question item (step S210). As a result of the determination, if the selection is not completed (step S210: NO), the process returns to the process of S209.

On the other hand, when the selection is completed (step S210: YES), the degree of attraction is measured for each area (step S211). After the selection of the design proposal by the user 10 is completed, the attraction degree measuring unit 354 calculates the measurement invitation degree of the

design proposals

1, 2

images

31, 32 based on the information on the line of sight for each sensitivity question item. FIGS. 34A to 34C exemplify the degree of attraction of measurement in areas 1 to 4 for "innovative", "delicious", and "like", respectively.

Next, the sensitivity information of the user 10 regarding the design of the evaluation target is calculated (step S212). The Kansei information acquisition unit 359 calculates the Kansei information of the user 10 for each Kansei question item. In the present embodiment, for example, sensitivity information is calculated for each sensitivity question item so as to be proportional to the reciprocal of the response time (hereinafter, also referred to as “reaction intensity”). This is because the shorter the response time, the stronger the impression of the user 10's sensibility is by the sensibility question item, that is, it is considered that there is a strong connection between the sensibility question item and the sensibility of the user 10. In the present embodiment, for example, the sensitivity information is set to a value equal to the reaction intensity. As shown in FIG. 35, for example, when the response times for "innovative", "delicious", and "like" were 3.5 seconds, 8.5 seconds, and 6.9 seconds, respectively, the reaction intensity. Are about 2.9, about 1.2, and about 1.4.

Next, the attractiveness conversion value is calculated for each area (step S213). The Kansei information acquisition unit 359 calculates the Kansei degree conversion value for each Kansei question item based on the measured Kansei degree and the Kansei information in each area. In the present embodiment, the attractiveness conversion value is represented by the product of the measured attractiveness and the sensitivity information in each area. FIG. 36 illustrates the attractiveness conversion value in each area when each sensitivity question item is presented. For example, when the sensitivity question item is "innovative", the measurement attraction degree of the area 1 is 0.02 (the part shown in gray in FIG. 34A) with respect to the design proposal 1 image 31, and the sensitivity information (reaction time) is. Since it is 2.9 (the part shown in gray in FIG. 35), the attractiveness conversion value is 0.02 × 2.9 = 0.58 ≈ 0.06 (the part shown in gray in FIG. 36).

Next, the relationship between the difference in attractiveness and the plurality of Kansei question items 41 to 43 is extracted (step S214). As shown in FIGS. 37A to 37C, the sensitivity output unit 362 outputs, for example, a graph showing the correlation between the attractiveness difference and the attractiveness conversion value to the display 410. As a result, the sensibilities of the user 10 for the plurality of sensibility question items 41 to 43 are visualized with respect to the design proposal image 30.

In the example shown in FIG. 37A, when the sensitivity question item is "innovative", the difference in attractiveness between the area 3 of the design proposal 1 image 31 and the area 2 of the design proposal 2 image 32 is a positive value of 0.3 or more. It has become. The same applies to the sensibility question items of FIGS. 37B and 37C for "delicious" and "like". Therefore, in the examples shown in FIGS. 37A to 37C, the sensitivity extraction unit 361 has the area 3 of the design plan 1 image 31 and the design plan 2 for any of "innovative", "delicious" and "like". The relationship with the area 2 of the image 32 that the attractiveness difference takes a relatively large positive value of 0.3 or more is extracted. In addition, the sensitivity extraction unit 361 responds to any of "innovative", "delicious", and "like" in areas other than the area 3 of the design proposal 1 image 31 and the area 2 of the design proposal 2 image 32. The relevance that the attractiveness difference is approximately 0 or less is extracted.

Further, in the example shown in FIG. 36, the attractiveness conversion value of the area 2 of the design proposal 2 image 32 is 3.77, respectively, when the sensitivity question items are “innovative”, “delicious”, and “like”. , 0.12, and 2.23. The attractiveness conversion value is higher in the case of "innovative" and "like" than in the case of "delicious". That is, when the sensitivity question items are "innovative" and "like", when the attractiveness difference is large, the attractiveness conversion value is also large, and the attractiveness difference and the attractiveness are compared with the case where "it looks delicious". High correlation with converted value. In this way, the sensitivity extraction unit 361 extracts the relationship between the attraction degree difference and the plurality of sensitivity question items based on the attraction degree difference and the attraction degree conversion value. This makes it possible to determine which sensitivity question item causes the difference in the degree of attraction to be large.

In the above example, the case where two

design proposals

1 and 2 are used has been described, but the case where three or more design proposals are used can also be applied.

Further, in the above example, the case where the images of the evaluation objects have different designs from each other has been described. However, the present invention is not limited to such cases, and the images of the evaluation objects have the same design, one image is decoratively printed, and the other image is decoratively printed. It can be applied even if it does not exist. In this case, the image acquisition unit 351 acquires an image of the evaluation object to which decorative printing is applied and an image of the evaluation object to which decorative printing is not applied. In the present embodiment, the decorative printing is, for example, beat printing, varnishing printing, or the like.

In addition, when the power message of the selected design proposal is changed, the subject's emotional effect is investigated, and when decorative processing such as foil stamping printing or varnishing printing is added, the subject's emotional effect on the sense of quality is obtained. It becomes easy to check.

Further, in this embodiment as well, it goes without saying that the relationship between the attractiveness difference and the plurality of Kansei question items can be extracted by the statistical method or the machine learning method described in the first embodiment. By using such a method, it is possible to recommend a design proposal when specializing in a specific Kansei question item from a plurality of design proposals.

The present invention is not limited to the above-described embodiment, and can be variously modified within the scope of the claims.

For example, the design evaluation program may be provided by a computer-readable recording medium such as a USB memory, a flexible disk, or a CD-ROM, or may be provided online via a network such as the Internet. In this case, the program recorded on the computer-readable recording medium is usually transferred to a memory, storage, or the like and stored. Further, this design evaluation program may be provided as a single application software, or may be incorporated into the software of each device as a function of the server.

Further, in the embodiment, a part or all of the processing executed by the program can be replaced with hardware such as a circuit and executed.

This application is based on Japanese Patent Application No. 2020-185842 filed on November 6, 2020, the disclosure of which is incorporated as a whole by reference.

100 design evaluation system,
200 line-of-sight measuring device,
300 control unit,
310 CPU,
320 RAM,
330 ROM,
340 auxiliary storage,
351 Image acquisition department,
352 Attribute information acquisition department,
353 Invitation degree estimation department,
354 attraction measurement unit,
355 Difference calculation unit,
356 display control unit,
357 Expression Providing Department,
358 Line-of-sight information acquisition department,
359 Kansei Information Acquisition Department,
360 attraction conversion value calculation unit,
361 Sensitivity Extractor,
362 Sensitive output unit,
370 learning device,
371 Learning data storage unit,
372 model generator,
400 display device,
500 input device,
600 audio output device,
700 communication equipment.

Claims

An image acquisition unit that acquires at least one image of the evaluation target,
An attraction degree estimation unit that estimates the degree of attraction of the image based on the characteristics of the image,
A display control unit that displays the image on the display device,
An expression providing unit that provides users with a plurality of expressions related to the design of the evaluation object,
A line-of-sight information acquisition unit that acquires information about the user's line-of-sight with respect to the image displayed on the display device, and a line-of-sight information acquisition unit.
An attractiveness measuring unit that measures the attractiveness of the image based on the information about the line of sight, and an attractiveness measuring unit.
A difference calculation unit that calculates the difference between the estimated attraction degree estimated by the attraction degree estimation unit and the measurement attraction degree measured by the attraction degree measurement unit.
The Kansei information acquisition unit that acquires Kansei information indicating the relationship between the expression and the Kansei of the user regarding the design, and the Kansei information acquisition unit.
A design evaluation device having a sensitivity extraction unit that extracts a relationship between the difference and the plurality of expressions using the sensitivity information.
The difference calculation unit sets a plurality of regions that do not overlap each other with respect to the image, and calculates the difference in each region.
The design evaluation device according to claim 1, wherein the sensitivity extraction unit extracts the relationship between the difference and the plurality of expressions in each of the regions.
Further, it has an attraction degree conversion value calculation unit that calculates an attraction degree conversion value based on the measurement attraction degree and the sensitivity information in each of the regions.
The Kansei information acquisition unit receives the user's response to the expression regarding the design, and acquires the Kansei information based on the answer.
The design evaluation device according to claim 2, wherein the sensitivity extraction unit extracts the relationship based on the correlation between the difference and the attractiveness conversion value.
It also has an attribute information acquisition unit that acquires information about the user's attributes.
The design evaluation device according to claim 3, wherein the sensitivity extraction unit extracts the relationship between the difference and the information regarding the plurality of expressions and the user's attributes in each of the regions.
The design evaluation device according to claim 2, wherein the sensitivity extraction unit extracts the relationship by a statistical method using the plurality of expressions as explanatory variables and the difference as an objective variable.
It also has an attribute information acquisition unit that acquires information about the user's attributes.
The design evaluation according to claim 2, wherein the sensitivity extraction unit extracts the relationship by a statistical method using the information about the plurality of expressions and the attributes of the user as explanatory variables and the difference as the objective variable. Device.
It also has an attribute information acquisition unit that acquires information about the user's attributes.
The second aspect of the present invention, wherein the sensitivity extraction unit extracts the relationship by using a machine learning model in which the sensitivity information and the information related to the user's attributes are used as explanatory variables and machine learning is performed using the difference as an objective variable. Described design evaluation device.
The design evaluation device according to any one of claims 1 to 7, wherein the sensitivity extraction unit extracts the relevance of a plurality of the users.
The design evaluation device according to any one of claims 1 to 8, wherein the Kansei information acquisition unit acquires the user's evaluation regarding the design and calculates the Kansei information based on the user's evaluation.
The Kansei information acquisition unit acquires the response time required for the user to complete the response to the expression, and calculates the Kansei information based on the response time, claim 1 to 8. The design evaluation device according to any one of the items.
The line-of-sight information acquisition unit continuously acquires information about the line of sight from the time the expression is provided to the user by the expression providing unit until the user's response to the expression is received by the sensitivity information acquisition unit. , The design evaluation device according to any one of claims 1 to 10.
The design evaluation device according to any one of claims 1 to 11, wherein the display control unit simultaneously displays the image and the expression on the display device.
The design evaluation device according to any one of claims 1 to 12, wherein the expression includes at least words.
The image acquisition unit acquires a plurality of images of the evaluation target, and obtains a plurality of images.
The attraction degree estimation unit estimates the attraction degree of each image based on the characteristics of each image.
The display control unit causes the display device to display each of the images.
The design evaluation device according to any one of claims 1 to 13, wherein the attractiveness measuring unit measures the attractiveness of each of the images based on the information about the line of sight.
The design evaluation according to claim 14, wherein the image acquisition unit acquires an image of the evaluation object to which decorative printing is applied and an image of the evaluation object to which the decorative printing is not applied. Device.
With the display device
A line-of-sight measuring device that measures the user's line of sight with respect to the image displayed on the display device and outputs information about the line of sight to the line-of-sight information acquisition unit.
A design evaluation system comprising the design evaluation device according to any one of claims 1 to 15.
Step (a) to acquire at least one image of the evaluation target,
In step (b) of estimating the degree of attraction of each of the images based on the characteristics of the images,
In step (c) of displaying the image on the display device,
Step (d) of providing the user with a plurality of expressions relating to the design of the evaluation object, and
A step (e) of acquiring information on the user's line of sight with respect to the image displayed on the display device, and
The step (f) of measuring the degree of attraction of the image based on the information about the line of sight, and
A step (g) for calculating the difference between the estimated attractiveness estimated in the step (b) and the measured attractiveness measured in the step (f), and the step (g).
The step (h) of acquiring the sensibility information indicating the relationship between the expression and the sensibility of the user regarding the design, and
A design evaluation method comprising the step (i) of extracting the relationship between the difference and the plurality of expressions using the sensitivity information.
In the step (g), a plurality of regions that do not overlap each other are set for the image, and the difference is calculated in each region.
The design evaluation method according to claim 17, wherein in step (i), the relationship between the difference and the plurality of expressions is extracted in each of the regions.
Prior to the step (i), the step (j) for acquiring information regarding the attributes of the user is further included.
The design evaluation method according to claim 17 or 18, wherein in step (i), the sensitivity information and the information related to the user's attribute are used as explanatory variables, and the difference is used as an objective variable to extract the relationship.
A design evaluation program for causing a computer to execute the process included in the design evaluation method according to any one of claims 17 to 19.
A storage unit that stores sensibility information as an explanatory variable and learning data including an attraction difference as an objective variable.
A learning device having a model generation unit that generates a machine learning model by machine learning the relationship between the sensitivity information and the attraction degree difference based on the learning data stored in the storage unit.