Estimation device, estimation method, and program

Abstract

The present invention provides an estimation device, estimation method, and program capable of estimating a user's evaluation of an image being evaluated. [Solution] One aspect of the estimation device of the present disclosure comprises: a first acquisition unit that acquires a first image related to a predetermined user; a second acquisition unit that acquires a second image to be evaluated; and an estimation unit that outputs an evaluation of the second image by the predetermined user, estimated based on the first and second images.

Description

Technical Field

[0001] The present disclosure relates to an estimation device, an estimation method, and a program.

Background Art

[0002] Conventionally, techniques for evaluating an object of interest of a user or generating a profile of the user using an image group associated with the user have been known (see, for example, Patent Documents 1 and 2).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the conventional technology as described above, even if the object of interest or profile of the user can be evaluated, it is impossible to estimate what evaluation the user makes on the image to be evaluated.

[0005] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide an estimation device, an estimation method, and a program capable of estimating a user's evaluation of an image to be evaluated.

Means for Solving the Problems

[0006] One aspect of the estimation device of the present disclosure includes a first acquisition unit that acquires a first image related to a predetermined user, a second acquisition unit that acquires a second image to be evaluated, and an estimation unit that outputs an evaluation of the predetermined user for the second image estimated based on the first image and the second image.

[0007] One aspect of the estimation method of this disclosure includes a first acquisition step in which a first acquisition unit acquires a first image related to a predetermined user; a second acquisition step in which a second acquisition unit acquires a second image to be evaluated; and an estimation step in which an estimation unit outputs an evaluation of the second image by the predetermined user, estimated based on the first and second images.

[0008] One aspect of the program of this disclosure causes a computer to perform a first acquisition step of acquiring a first image related to a predetermined user, a second acquisition step of acquiring a second image to be evaluated, and an estimation step of outputting an evaluation of the second image by the predetermined user, estimated based on the first and second images. [Effects of the Invention]

[0009] According to this disclosure, it is possible to provide an estimation device, estimation method, and program capable of estimating a user's evaluation of an image being evaluated. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is a block diagram showing an example of the configuration of the estimation system of this embodiment. [Figure 2] Figure 2 is a block diagram showing an example of the hardware configuration of the estimation device and information processing device of this embodiment. [Figure 3] Figure 3 is a block diagram showing an example of the functional configuration of the estimation device of this embodiment. [Figure 4] Figure 4 is a configuration diagram showing an example of the second trained model of this embodiment. [Figure 5] Figure 5 shows an example of the estimation results of this embodiment. [Figure 6] Figure 6 is a flowchart showing an example of the learning process performed in the estimation device of this embodiment. [Figure 7] Figure 7 is a flowchart showing an example of the estimation process performed by the estimation device of this embodiment. [Modes for carrying out the invention]

[0011] Hereinafter, embodiments of the present disclosure (hereinafter simply referred to as "these embodiments") will be described in detail with reference to the drawings. However, this disclosure is not limited to the following embodiments. Furthermore, the following embodiments and modifications can be combined as appropriate.

[0012] In the following, we will explain this disclosure using the example of estimating the appeal of a product package to a predetermined target group.

[0013] First, the configuration of the estimation system in this embodiment will be described.

[0014] Figure 1 is a block diagram showing an example of the configuration of the estimation system 1 of this embodiment. As shown in Figure 1, the estimation system 1 comprises an estimation device 10, an information processing device 20, and a platform 30. The estimation device 10, the information processing device 20, and the platform 30 are connected via a network 2. The network 2 can be implemented, for example, by the Internet. The network 2 may be a wired network, a wireless network, or a mixture of wired and wireless networks.

[0015] The estimation device 10 estimates user evaluations of images to be evaluated, and can be implemented, for example, by a server computer. In this embodiment, the estimation device 10 uses the platform 30 to identify users belonging to a predetermined target group, and generates a trained model by machine learning on images related to the identified users. The estimation device 10 inputs images of product packaging, which are the images to be evaluated, into the trained model to estimate the appeal of the product packaging to the predetermined target group.

[0016] The information processing device 20 is a terminal device for users used to utilize the estimation device 10, and examples thereof include a PC (Personal Computer), a tablet terminal, or a smartphone, etc.

[0017] The platform 30 is a web-based platform that provides web services, and examples thereof include at least any one of, for example, an SNS (Social Networking Service), a blog service, an online review service, a word-of-mouth site, and an EC (Electronic Commerce) site, etc., but are not limited thereto. In the present embodiment, the platform 30 may be any one as long as users can register an account and post comments and images.

[0018] FIG. 2 is a block diagram showing an example of the hardware configuration of the estimation device 10 and the information processing device 20 of the present embodiment.

[0019] First, the hardware configuration of the estimation device 10 will be described. As shown in FIG. 2, the estimation device 10 includes a control device 11, a main storage device 12, an auxiliary storage device 13, a communication device 14, and various buses 19. The control device 11, the main storage device 12, the auxiliary storage device 13, and the communication device 14 are connected via the various buses 19. Thus, the estimation device 10 of the present embodiment has a general hardware configuration using a normal computer.

[0020] The control device 11 controls the overall operation of the estimation device 10. Examples of the control device 11 include at least any one of, for example, a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), etc., but are not limited thereto. The CPU and the GPU may be one or more, and may be a single core or a multi-core.

[0021] Examples of main memory 12 include, but are not limited to, ROM (Read Only Memory) and RAM (Random Access Memory). ROM stores various programs, such as programs for controlling the estimation device 10 and programs for estimating user evaluations of images under evaluation. RAM is used as a workspace for the control device 11 to perform various controls based on the programs stored in ROM.

[0022] The auxiliary storage device 13 stores the various programs and data mentioned above. The programs mentioned above only need to be stored in at least one of the main storage device 12 and the auxiliary storage device 13. Examples of the auxiliary storage device 13 include, but are not limited to, at least one of existing storage devices capable of magnetic, electrical, or optical storage, such as an HDD (Hard Disk Drive), SSD (Solid State Drive), and DVD (Digital Versatile Disc). The auxiliary storage device 13 may be built into the estimation device 10 or externally connected to the estimation device 10 via an interface such as USB (Universal Serial Bus). Furthermore, the auxiliary storage device 13 may be a NAS (Network Attached Storage) connected via a network such as a LAN (Local Area Network) or WAN (Wide Area Network).

[0023] The communication device 14 is used to communicate with the information processing device 20 and the platform 30 via the network 2. Examples of the communication device 14 include, but are not limited to, a communication device for a wired LAN or a wireless communication device for a wireless LAN.

[0024] In addition to the above configuration, the estimation device 10 may further include hardwired circuits such as ICs (Integrated Circuits), ASICs (Application Specific Integrated Circuits), and FPGAs (Field-Programmable Gate Arrays) that are specific to the estimation device 10.

[0025] Next, the hardware configuration of the information processing device 20 will be described. As shown in Figure 2, the information processing device 20 comprises a control device 21, a main memory 22, an auxiliary memory 23, a communication device 24, an input device 25, a display device 26, and various buses 29. The control device 21, the main memory 22, the auxiliary memory 23, the communication device 24, the input device 25, and the display device 26 are connected via the various buses 29.

[0026] The control device 21 controls the overall operation of the information processing device 20. The implementation method of the control device 21 is the same as that of the control device 11.

[0027] The implementation method for the main memory 22 is the same as that for the main memory 12, so a detailed explanation will be omitted. The ROM of the main memory 22 stores various programs, including programs for controlling the information processing device 20.

[0028] The auxiliary storage device 23 stores the various programs and data described above. Note that the various programs described above only need to be stored in at least one of the main storage device 22 and the auxiliary storage device 23. The implementation method for the auxiliary storage device 23 is the same as that for the auxiliary storage device 13, so a detailed explanation is omitted.

[0029] The communication device 24 is used to communicate with the estimation device 10 via the network 2. Since the implementation method for the communication device 24 is the same as that for the communication device 14, a detailed explanation is omitted.

[0030] The input device 25 is used for various inputs, selections, and specifications, and serves as a user interface with the user. Examples of input devices 25 include, but are not limited to, keyboards, mice, and touch panels. The input device 25 may be built into the information processing device 20 or connected externally to the information processing device 20 via an interface such as USB.

[0031] The display device 26 displays various screens, such as a screen showing the user's evaluation of the image to be evaluated, which has been estimated by the estimation device 10, and serves as a user interface between the device and the user. Examples of displays for the display device 26 include, but are not limited to, liquid crystal displays, organic electro-luminescence (OLED) displays, and touch panel displays. The display device 26 may be an internal display built into the information processing device 20, or an external display connected to the information processing device 20 via a display interface such as HDMI®.

[0032] Figure 3 is a block diagram showing an example of the functional configuration of the estimation device 10 of this embodiment. As shown in Figure 3, the estimation device 10 includes an acquisition unit 101, an analysis unit 103, a first acquisition unit 105, a learning unit 107, a model storage unit 109, a second acquisition unit 111, and an estimation unit 113. The acquisition unit 101, analysis unit 103, first acquisition unit 105, learning unit 107, second acquisition unit 111, and estimation unit 113 can be realized, for example, by the control device 11, main memory 12, and communication device 14 described in Figure 2. For example, the control device 11 reads a program for estimating the user's evaluation of the image to be evaluated, which is stored in the main memory 12 (ROM) or auxiliary memory 13, and expands it into the main memory 12 (RAM). The control device 11 realizes each of the above-mentioned functional units by executing various processes according to the expanded program. Here, we have explained the case where each of the above-mentioned functional units is implemented as software, but at least a part of each of the above-mentioned functional units may be implemented as hardware. In this case, the functional unit to be implemented as hardware can be implemented, for example, by the hardwired circuit described above. Alternatively, any of the above-mentioned functional units may be implemented through the cooperation of software and hardware.

[0033] The model memory unit 109 can be implemented, for example, by the auxiliary storage device 13 described in Figure 2.

[0034] The collection unit 101 collects (scrapes) multiple accounts that are publicly available on a web platform. Specifically, the collection unit 101 receives an instruction from the information processing device 20 to generate a trained model (second trained model) for estimating user ratings of images to be evaluated. Upon receiving this instruction, the collection unit 101 begins collecting accounts from the platform 30. For example, the collection unit 101 collects 1000 accounts from any social networking service (SNS).

[0035] The collection unit 101 may collect accounts from a single platform or from multiple platforms. Furthermore, the collection unit 101 may randomly collect accounts from platforms or collect accounts that meet predetermined account collection conditions. Examples of accounts that meet predetermined account collection conditions include accounts that have reacted to competing products of the product that uses the image being evaluated on its packaging. Reactions include, but are not limited to, following the competing product's account, giving positive ratings ("likes") to posts about the competing product, or posting about the competing product. The predetermined account collection conditions may be set in advance or specified by the information processing device 20.

[0036] When the collection unit 101 collects multiple accounts from the platform 30, it collects the content of posts for each collected account. For example, the collection unit 101 collects posts that meet predetermined post content collection conditions. Posts that meet the predetermined post content collection conditions include, but are not limited to, posts included in a specified period, posts containing specified keywords, or posts of a specified number. The predetermined post content collection conditions may be predetermined or specified by the information processing device 20. For example, the collection unit 101 collects 200 posts from the most recent posts for each account.

[0037] The analysis unit 103 analyzes the content of posts from multiple accounts collected by the collection unit 101 (by performing psychographic analysis) to identify accounts of users belonging to a predetermined target group and selects the users of those accounts as predetermined users. Specifically, the analysis unit 103 features the content of posts from each account collected by the collection unit 101 and compares it with features associated with the predetermined target group to identify accounts of users belonging to the predetermined target group.

[0038] For example, the analysis unit 103 analyzes the content of posts from multiple accounts using a first pre-trained model. The first pre-trained model is a machine learning model that performs natural language processing, and examples include, but is not limited to, pre-trained models such as Word2Vec, which converts words into feature vectors.

[0039] Specifically, the analysis unit 103 inputs the collected post content into the first trained model for each account collected by the collection unit 101, vectorizes the words contained in the post content, and calculates the average of the vectors. For each account, the analysis unit 103 compares the average of the calculated vectors with the vector values, which are features associated with a predetermined target layer, to determine the similarity. The analysis unit 103 identifies accounts whose calculated similarity exceeds a threshold as accounts of users belonging to a predetermined target layer. Note that the accounts of users belonging to a predetermined target layer may be singular or multitude.

[0040] The designated target group may be a commonly used target group, or it may be a target group independently defined by the manufacturer of the product that uses the image being evaluated on its packaging. Commonly used target groups include, for example, target groups used in television ratings such as the F1 demographic, or target groups defined by birth year, such as millennials or Gen Z. Independently defined target groups include, for example, the active group, which targets active users, or the smart group, which targets users with efficient and smart lifestyles.

[0041] A vector value, which is a feature associated with a given target group, could be, for example, the vector value of a keyword frequently used by users belonging to that target group. For example, if the keyword frequently used by the aforementioned smart group is "minimal," then the vector value of the word "minimal" could be used. Note that the vector value associated with a given target group may be the vector value of a single keyword or the vector value of multiple keywords.

[0042] This analysis unit 103 processes data to efficiently identify accounts of users who possess attributes such as being of a specific age group, a specific gender, or having interests in specific matters (users belonging to a predetermined target group).

[0043] The first acquisition unit 105 acquires a first image associated with a predetermined user. Specifically, the first acquisition unit 105 acquires an image associated with a predetermined user's account that is publicly available on a web platform as the first image. More specifically, on platform 30, the first acquisition unit 105 acquires multiple images as the first image for each predetermined user's account identified by the analysis unit 103, which is posted within that account. For example, the first acquisition unit 105 acquires 200 images as the first image from an arbitrary SNS, which is posted within a predetermined user's account. Thus, since the first image is an image posted within a predetermined user's account, it includes at least one image that the predetermined user is interested in or concerned with.

[0044] The first acquisition unit 105 acquires images that satisfy predetermined image acquisition conditions. Images that satisfy predetermined image acquisition conditions include, for example, images included in a specified period or a specified number of images, but are not limited to these. The predetermined image acquisition conditions may be predetermined or specified by the information processing device 20. For example, the first acquisition unit 105 acquires 200 images from the latest images posted within a specified user's account.

[0045] The learning unit 107 trains a second pre-trained model using a plurality of first images acquired by the first acquisition unit 105. The second pre-trained model is a pre-trained model for estimating the evaluation of a given user of a second image to be evaluated. More specifically, the second pre-trained model is a pre-trained model for estimating the appeal of a second image to a given user, and more specifically, a pre-trained model for estimating the appeal of a second image to a given target group to which the given user belongs. The second image to be evaluated may, for example, be an image applied to a product or the packaging of such a product, but is not limited to this.

[0046] The learning unit 107 adds multiple first images acquired by the first acquisition unit 105 to the dataset and trains the second pre-trained model using the dataset. The learning unit 107 may also newly train the second pre-trained model using the dataset, or perform additional training (fine tuning). In this implementation, the learning unit 107 uses the dataset to further train an existing CNN (Convolutional Neural Network) model, such as ResNet (Residual Neural Networks), but is not limited to this.

[0047] The learning unit 107 further trains the second pre-trained model using, for example, multiple first images acquired by the first acquisition unit 105 as ground truth images. The learning unit 107 may also further train the second pre-trained model using images whose vectors are the exact opposite of those of the multiple first images acquired by the first acquisition unit 105 as incorrect truth images. Images whose vectors are the exact opposite of those of the first images can be acquired by the first acquisition unit 105 from, for example, the platform 30.

[0048] The learning unit 107 may perform preprocessing, such as noise reduction or normalization, on the multiple first images acquired by the first acquisition unit 105, and add the multiple preprocessed first images to the dataset. Alternatively, the learning unit 107 may train the second trained model by excluding the first image from the multiple first images acquired by the first acquisition unit 105 that contains information that allows for product identification. In this case, it is possible to suppress the influence of non-design elements of the image, such as text information, on the estimation of appeal. In this case, the learning unit 107 can simply exclude the first image containing information that allows for product identification from the multiple first images acquired by the first acquisition unit 105 and add it to the dataset.

[0049] The model memory unit 109 stores the second trained model learned by the learning unit 107.

[0050] Figure 4 is a diagram showing an example of a second pre-trained model of this embodiment. As shown in Figure 4, the second pre-trained model is composed of an input layer 51, a contrast extraction layer 52, edge extraction layers 53-1 and 53-2, higher-order feature extraction layers 54-1 to 54-S (where S is a natural number), and recognition cell layers 55-1 and 55-2, etc. The second pre-trained model extracts various parts of the input image and features from the image, and as a final output, it outputs the appeal of the input image to a predetermined target layer.

[0051] The second acquisition unit 111 acquires the second image to be evaluated. As mentioned above, the second image to be evaluated is, for example, an image applied to the product or the packaging of the product. Specifically, the second acquisition unit 111 acquires the second image to be evaluated from the information processing device 20.

[0052] The estimation unit 113 outputs a predetermined user's evaluation of the second image, estimated based on the first and second images. Specifically, the estimation unit 113 estimates a predetermined user's evaluation of the second image based on the features of the first image and the features of the second image. More precisely, the estimation unit 113 estimates a predetermined user's evaluation of the second image based on the similarity between the first and second images. The similarity is determined based on the features of the first image and the features of the second image.

[0053] In this embodiment, the estimation unit 113 inputs the second image acquired by the second acquisition unit 111 into the second trained model stored in the model storage unit 109 to estimate a predetermined user's evaluation of the second image, and outputs it to, for example, the information processing device 20. The predetermined user's evaluation of the second image may be whether they view the second image favorably, whether they feel aversion towards the second image, or any other evaluation. Specifically, the estimation unit 113 estimates the appeal of the second image to the predetermined user, and more specifically, estimates the appeal of the second image to a predetermined target group to which the predetermined user belongs. The estimation based on similarity using the features of the first image and the features of the second image described above is performed within the second trained model. This makes it possible to estimate the appeal of an image applied to a product or its packaging to a predetermined target group.

[0054] The evaluation of the second image by a given user (the appeal of the second image) may be expressed as a numerical score such as likelihood, or as a multi-level evaluation using a heatmap or similar method.

[0055] Figure 5 shows an example of the estimation results of this embodiment, illustrating the appeal of the second image to a predetermined target group using a heatmap. The heatmap 71 shown in Figure 5 is created by multiplying the image features extracted by each of the higher-order feature extraction layers 54-1 to 54-S of the second trained model by weights and summing them up, then superimposing the results onto the second image. In this case, the second image is the packaging of a beverage can.

[0056] Image features 64-1 of higher-order feature extraction layer 54-1 are multiplied by the weight W1 of higher-order feature extraction layer 54-1. Image features 64-2 of higher-order feature extraction layer 54-2 are multiplied by the weight W2 of higher-order feature extraction layer 54-2. Image features 64-S of higher-order feature extraction layer 54-S are multiplied by the weight W of higher-order feature extraction layer 54-S. S It is multiplied by .

[0057] Furthermore, when the features of each image are combined, the lower right portion receives the most attention. As shown in heatmap 61, this indicates that the design of the lower right portion of the beverage can packaging is highly appealing to the target audience.

[0058] Next, the operation of the estimation system of this embodiment will be described.

[0059] Figure 6 is a flowchart showing an example of the learning process performed in the estimation device 10 of this embodiment.

[0060] First, the collection unit 101 collects N (N≧2) accounts from the platform 30 (step S101).

[0061] Next, the data collection unit 101 initializes the variable i to 0 (step S103).

[0062] Next, the collection unit 101 collects M (M≧2) posts from account i (step S105).

[0063] Next, the analysis unit 103 vectorizes the words contained in the M posts collected by the collection unit 101 and calculates the average of the vectors (step S107).

[0064] Next, the analysis unit 103 compares the average value of the obtained vector with the vector value, which is a feature associated with a predetermined target layer, to determine the similarity, and determines whether the obtained similarity is above a threshold (step S109).

[0065] If the similarity is above a threshold (Yes in step S109), the first acquisition unit 105 acquires L first images from account i's posts, and the learning unit 107 adds L first images to the dataset (step S111).

[0066] On the other hand, if the similarity is below the threshold (No in step S109), the process in step S111 is not performed.

[0067] Next, the collection unit 101 increments the variable i (step S113) and determines whether the variable i is greater than or equal to N (step S115).

[0068] If variable i is less than N (No in step S115), return to step S105 and collect M posts from the following account i.

[0069] On the other hand, if the variable i is greater than or equal to N (Yes in step S115), the acquisition of the first image for training is terminated, and the training unit 107 trains the second trained model using the dataset (step S117).

[0070] Figure 7 is a flowchart showing an example of the estimation process performed by the estimation device 10 of this embodiment.

[0071] First, the second acquisition unit 111 acquires the second image to be evaluated (step S201).

[0072] Next, the estimation unit 113 inputs the second image acquired by the second acquisition unit 111 into the second trained model (step S203).

[0073] Next, the estimation unit 113 outputs the estimated appeal of the second image to a given user (step S205).

[0074] As described above, according to this embodiment, it is possible to easily estimate the evaluation of a predetermined user for the image being evaluated.

[0075] In particular, in this embodiment, data is collected from a web-based platform, and the collected data is processed using natural language to identify user accounts belonging to a predetermined target group, and training images are obtained from those accounts. Therefore, according to this embodiment, a trained model for estimating the appeal of an image to be evaluated for a predetermined target group can be easily trained.

[0076] Furthermore, in this embodiment, the trained model is trained using the method described above to estimate the appeal of the image to be evaluated to a predetermined target group. Therefore, according to this embodiment, it becomes possible to evaluate the package's appeal objectively and quantitatively, which was difficult with conventional subjective evaluation methods such as questionnaire evaluations.

[0077] Furthermore, in this embodiment, the latest data can be collected from a web-based platform, allowing for real-time capture of changes in consumer preferences and trends. Therefore, according to this embodiment, it becomes possible to quickly adjust package designs in response to market trends.

[0078] Furthermore, in this embodiment, psychographic analysis can be used to segment the target audience and individually estimate the appeal to each segment. Therefore, according to this embodiment, it becomes possible to formulate a more precise marketing strategy.

[0079] Furthermore, this embodiment enables significant cost and time savings compared to conventional methods such as questionnaire evaluation. Additionally, since the limitations on sample size are eliminated, more reliable results can be obtained.

[0080] Furthermore, according to this embodiment, it is possible to estimate the appeal of not only the company's own product packaging but also that of competitors' product packaging, making it possible to objectively grasp the position of the company's own products in the market.

[0081] Furthermore, this embodiment enables effective package design improvements by identifying elements with low appeal and clarifying areas for improvement.

[0082] Furthermore, according to this embodiment, since data can be collected from multiple platforms on the Web, differences between platforms and differences in appeal on each platform can also be understood.

[0083] Furthermore, according to this embodiment, the second trained model is continuously updated through additional training, so the prediction accuracy improves over time, and more reliable estimation results can be obtained.

[0084] Furthermore, according to this embodiment, by utilizing the system not only for evaluating the packaging of existing products but also from the packaging design stage of new products, it becomes possible to select a highly appealing design before market launch.

[0085] (Variation 1) In the above embodiment, the analysis unit 103 may further analyze additional information to identify user accounts belonging to a predetermined target group. The additional information includes at least one of the following: information other than the posted content associated with the account, and information obtained by analyzing the information associated with the account.

[0086] Other information associated with an account besides the content of posts may include, but is not limited to, the number of followers, the language used, keywords, hashtags, emoji usage patterns, and references to specific brands or products.

[0087] Information obtained by analyzing information associated with an account may include, but is not limited to, the frequency of posts, the sentiment of the posts, and the time of day the user is active.

[0088] The additional information includes at least the information that is used for feature analysis. Examples of information that is used for feature analysis include, but is not limited to, at least one of the following: posting frequency, number of followers, sentiment of the posted content, language used, and activity time.

[0089] The additional information includes at least information that is analyzed without being feature-enhanced. Examples of information that is analyzed without being feature-enhanced include, but are not limited to, at least one of the following: keywords, hashtags, emoji usage patterns, frequency of image postings, and references to specific brands or products.

[0090] (program) The programs executed by the estimation device 10 in the above embodiments and each of the above modifications are provided as installable or executable files stored on a computer-readable storage medium such as a CD-ROM, CD-R, memory card, DVD, or flexible disk (FD).

[0091] Furthermore, the programs executed by the estimation device 10 in the above embodiments and each of the above modifications may be stored on a computer connected to a network such as the Internet and provided by allowing download via the network. Alternatively, the programs executed by the estimation device 10 in the above embodiments and each of the above modifications may be provided or distributed via a network such as the Internet. Furthermore, the programs executed by the estimation device 10 in the above embodiments and each of the above modifications may be pre-installed in ROM or the like and provided.

[0092] The programs executed in the estimation device 10 of the above embodiments and each of the above modifications are configured as modules for realizing the above-described parts on a computer. In actual hardware, for example, the CPU reads the learning program from the HDD into RAM and executes it, thereby realizing the above-described parts on the computer.

[0093] The above embodiments and their respective modifications are merely examples of how this disclosure may be implemented, and they do not restrict the technical scope of this disclosure. Therefore, this disclosure can be implemented in various ways without departing from its essence or its main features. For example, the above embodiments and their respective modifications may be combined as appropriate on a component basis. Also, for example, some components may be removed from the total components in the above embodiments and their respective modifications.

[0094] This disclosure includes the following aspects:

[0095] (1) A first acquisition unit that acquires a first image related to a predetermined user, A second acquisition unit that acquires the second image to be evaluated, An estimation unit that outputs a predetermined user evaluation of the second image, estimated based on the first image and the second image, An estimation device equipped with the following features.

[0096] (2) The first acquisition unit acquires an image associated with the predetermined user account that is publicly available on a Web platform as the first image. The estimation device described in (1) above.

[0097] (3) The first image includes at least one of the images of interest to the predetermined user, The estimation device described in (1) or (2) above.

[0098] (4) A collection unit that collects multiple accounts that are publicly available on a web platform, The system includes an analysis unit that analyzes the content of posts from multiple collected accounts to identify accounts of users belonging to a predetermined target group, and selects the users of those accounts as the predetermined users. The estimation device described in (1) above.

[0099] (5) The analysis unit identifies accounts of users belonging to the predetermined target group by characterizing the content of posts for each account and comparing it with the features associated with the predetermined target group. The estimation device described in (4) above.

[0100] (6) The analysis unit analyzes the content of posts from the multiple accounts using the first trained model. The estimation device described in (4) or (5) above.

[0101] (7) The analysis unit further analyzes additional information, which includes at least one of the information other than the posted content associated with the account and the information obtained by analyzing the information associated with the account, to identify the accounts of users belonging to the predetermined target group. The estimation device described in (4) or (5) above.

[0102] (8) The additional information includes at least information that is feature-quantified and analyzed, The estimation device described in (7) above.

[0103] (9) The additional information includes at least information that is analyzed without being feature-enhanced. The estimation device described in (7) above.

[0104] (10) The evaluation of the second image by the predetermined user is the appeal of the second image to the predetermined user, The estimation device described in (1) above.

[0105] (11) The estimation unit estimates the predetermined user's evaluation of the second image based on the features of the first image and the features of the second image. The estimation device described in (1) above.

[0106] (12) The estimation unit estimates the evaluation of the second image by a predetermined user based on the similarity between the first image and the second image. The estimation device described in (1) above.

[0107] (13) The similarity is determined based on the features of the first image and the features of the second image. The estimation device described in (12) above.

[0108] (14) The first acquisition unit acquires multiple first images, The estimation unit inputs the second image into a second trained model that has been trained from a plurality of first images to estimate the predetermined user's evaluation of the second image. An estimation device as described in (1) and any one of (10) to (13) above.

[0109] (15) Further comprising a learning unit that trains a second trained model using multiple first images, The evaluation of the second image by the predetermined user is the appeal of the second image to the predetermined user. The second image above is an image applied to the product or the packaging of the product. The learning unit trains the second trained model by excluding the first image from among the multiple first images that contains information that can identify the product. The estimation device described in (14) above.

[0110] (16) A first acquisition step in which the first acquisition unit acquires a first image related to a predetermined user, The second acquisition unit performs a second acquisition step in which it acquires the second image to be evaluated, Estimation step: The estimation unit outputs the predetermined user evaluation of the second image, which it has estimated based on the first image and the second image. An estimation method that includes [this].

[0111] (17) A first acquisition step of acquiring a first image related to a predetermined user, The second acquisition step involves obtaining the second image to be evaluated, An estimation step that outputs the predetermined user's evaluation of the second image, estimated based on the first image and the second image; A program that causes a computer to execute something. [Explanation of Symbols]

[0112] 1 Estimation System 2 Network 10 Estimation device 20 Information Processing Devices 30 platforms 101 Collection Department 103 Analysis Department 105 First acquisition part 107 Learning Department 109 Model Memory Unit 111 Second Acquisition Department 113 Estimation Department

Claims

1. A first acquisition unit that acquires a first image related to a predetermined user, A second acquisition unit acquires a second image to be evaluated, An estimation unit that outputs a predetermined user's evaluation of the second image, estimated based on the first image and the second image, An estimation device equipped with the following features.

2. The first acquisition unit acquires an image associated with a predetermined user account that is publicly available on a Web platform as the first image. The estimation device according to claim 1.

3. The first image includes at least one of the images of interest to the predetermined user, The estimation device according to claim 1 or 2.

4. A collection unit that collects multiple accounts that are publicly available on a web platform, The system includes an analysis unit that analyzes the content of posts from multiple collected accounts to identify accounts of users belonging to a predetermined target group, and selects the users of those accounts as the predetermined users. The estimation device according to claim 1.

5. The analysis unit identifies user accounts belonging to the predetermined target group by characterizing the content of posts for each account and comparing it with the features associated with the predetermined target group. The estimation device according to claim 4.

6. The analysis unit analyzes the content of posts from the multiple accounts using the first trained model. The estimation device according to claim 4 or 5.

7. The analysis unit further analyzes additional information, which includes at least one of the information other than the posted content associated with the account and the information obtained by analyzing the information associated with the account, to identify the accounts of users belonging to the predetermined target group. The estimation device according to claim 4 or 5.

8. The aforementioned additional information includes at least information that is feature-enhanced and analyzed. The estimation device according to claim 7.

9. The aforementioned additional information includes at least information that is analyzed without being feature-enhanced. The estimation device according to claim 7.

10. The evaluation of the second image by the predetermined user is the appeal of the second image to the predetermined user. The estimation device according to claim 1.

11. The estimation unit estimates the predetermined user's evaluation of the second image based on the features of the first image and the features of the second image. The estimation device according to claim 1.

12. The estimation unit estimates the evaluation of the second image by a predetermined user based on the similarity between the first image and the second image. The estimation device according to claim 1.

13. The similarity is determined based on the features of the first image and the features of the second image. The estimation device according to claim 12.

14. The first acquisition unit acquires multiple first images, The estimation unit inputs the second image into a second trained model that has been learned from a plurality of first images to estimate the evaluation of the second image by a predetermined user. The estimation device according to claim 1 and any one of claims 10 to 13.

15. The system further includes a learning unit that trains a second pre-trained model using multiple first images, The evaluation of the second image by the predetermined user is the appeal of the second image to the predetermined user. The second image is an image applied to the product or the packaging of the product. The learning unit trains the second trained model by excluding the first image from among the plurality of first images that contains information that can identify the product. The estimation device according to claim 14.

16. The first acquisition unit performs a first acquisition step of acquiring a first image related to a predetermined user, The second acquisition unit performs a second acquisition step in which it acquires the second image to be evaluated, Estimation step: The estimation unit outputs the predetermined user evaluation of the second image, which it has estimated based on the first image and the second image. An estimation method that includes [this].

17. A first acquisition step involves acquiring a first image related to a specified user, The second acquisition step involves obtaining the second image to be evaluated, An estimation step that outputs the predetermined user's evaluation of the second image, which is estimated based on the first image and the second image; A program that causes a computer to execute something.

Images