JP2022037575A - Image evaluation predicting device and method - Google Patents

Abstract

To enable the sensibility evaluation of image data to be calculated efficiently.SOLUTION: An image evaluation predicting device 1 predicts the sensibility evaluation of image data. The image evaluation predicting device comprises a learning unit 11, and a prediction unit 12 for predicting the sensibility evaluation of image data on the basis of the result of learning by the learning unit. The learning unit acquires image data D11 and content data that includes text data D12 corresponding to the image data, sensually analyzes text data in the content data and calculates sensibility evaluation, classifies the image data in the content data, and generates a prediction model, on the basis of the classification result of image data and the sensibility evaluation of text data, which outputs the sensibility evaluation that corresponds to the classification result of inputted image data. The prediction unit inputs relevant image data to the prediction model and thereby calculates the sensibility evaluation of the relevant image data.SELECTED DRAWING: Figure 1

Description

The present invention relates to an image evaluation prediction device and a method.

A technique for evaluating the impression that a design image gives to an appeal target person (Patent Document 1) and a technique for calculating the degree of attention of an image (Patent Document 2) are known.

Japanese Unexamined Patent Publication No. 2017-091310 Japanese Unexamined Patent Publication No. 2014-149713

The prior art described in Patent Document 1 evaluates the impression that a design image gives to an appeal target person from the paper quality and color usage of the printing paper. The prior art described in Patent Document 2 only calculates the degree of attention of an image based on the number of texts referring to the image, does not predict the emotions of the viewer of the image, and is also improved in terms of convenience. There is room.

The present invention has been made in view of the above problems, and an object of the present invention is to provide an image evaluation prediction device and a method capable of calculating an emotional evaluation of image data more efficiently.

In order to solve the above problem, the image evaluation prediction device according to one viewpoint of the present invention is an image evaluation prediction device that predicts the emotional evaluation of image data, and is an image based on the learning unit and the learning result by the learning unit. It is equipped with a prediction unit that predicts the emotional evaluation of the data, and the learning unit acquires content data including image data and text data corresponding to the image data, and emotionally analyzes the text data in the content data to make it emotional. Prediction that calculates the evaluation, classifies the image data in the content data, and outputs the emotional evaluation corresponding to the input classification result of the image data based on the classification result of the image data and the emotional evaluation of the text data. The model is generated, and the prediction unit calculates the emotional evaluation of the target image data by inputting the target image data into the prediction model.

According to the present invention, it is possible to calculate an emotional evaluation corresponding to a classification result of any input target image data.

The functional block diagram of the image evaluation prediction apparatus which concerns on 1st Example. A hardware configuration diagram of a computer used as an image evaluation predictor. A flowchart showing the learning process. A flowchart showing the prediction process. A table showing the results of text processing. A table showing the processed data. An example of a model that predicts evaluation. The functional block diagram of the image evaluation prediction apparatus which concerns on 2nd Embodiment. Output example of image evaluation prediction device. The functional block diagram of the image evaluation prediction apparatus which concerns on 3rd Example. An explanatory diagram showing how the target image data is processed and evaluated. Explanatory drawing which shows the outline of the image evaluation prediction apparatus which concerns on 4th Embodiment. The functional block diagram of the image evaluation prediction apparatus which concerns on 5th Embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The image evaluation prediction device according to the present embodiment calculates the emotional evaluation by emotionally analyzing the text data associated with the image data, while classifying the image data, and is related to the classification result of the image data and the image data. By associating with the emotional evaluation of the text data, a prediction model for calculating the emotional evaluation of the input image data is generated. By inputting the target image data into this prediction model, it is possible to predict the emotions generated by the viewer of the target image data.

The emotional evaluations here are, for example, "fun", "joy", "bright", "refreshing", "pleasure", "sad", "spicy", "anger", "surprise", and "discomfort". , "Dislike" is a type of emotion caused by the sensibility of the viewer (reader or user) who sees the image data.

The image evaluation prediction device of the present embodiment can predict, for example, what kind of emotion the image data to be announced on SNS (Social Networking Service), blog, net news, etc. will make the viewer feel. .. Further, by using the image evaluation prediction device, for example, it is possible to objectively assume in advance an emotional evaluation such as a company logo mark or a design of a new product.

The image evaluation prediction device of the present embodiment can also select a layer of viewers (hereinafter, also referred to as a user layer or a reader layer; may also be referred to as a viewer) of the image data to be the target of the emotional evaluation. .. This is because even if the same image data is used, the impression (sensitivity evaluation) received by different user groups who view it will be different. Therefore, the image evaluation prediction device of the present embodiment defines a plurality of user groups from the creator attributes (user attributes) such as gender, age, place of residence, language used, hobbies, tastes, careers, occupations, etc., and is the same. It is also possible to predict how the target image data of the above will obtain an emotional evaluation in each user group.

The image evaluation prediction device of the present embodiment can also change the emotional evaluation by processing the target image data. That is, the image evaluation prediction device processes the target image data so as to approach the designated Kansei evaluation, and inputs the processed target image data to the prediction model to obtain a new Kansei evaluation.

The image evaluation prediction device can automatically process the target image data by, for example, changing the resolution, color, and the like of the target image data. Alternatively, the image evaluation prediction device may provide the designer or the creator of the content data with advice information on the image processing policy. The image evaluation prediction device can output advice information such as "women in their twenties and thirties tend to feel" healing "feelings in images containing a lot of pastel colors." The designer or the creator of the content data who has confirmed the advice information may manually process the target image data using image processing software.

The image evaluation prediction device of the present embodiment uses content data in which the image data and the text data corresponding to the image data exist in the same post as an analysis target. As a result, it can be assumed that the emotional evaluation calculated from the text data of a certain post is related to the image data in the same post. The same post includes, for example, one post on a microblog with a limited number of characters, one article on net news or a blog, and the like. The image data of this embodiment is mainly still image data.

The image evaluation prediction device of the present embodiment can be configured as, for example, an on-premises or cloud service. Further, the image evaluation prediction device of the present embodiment may be provided with some functions in the server and some other functions in the user terminal.

The first embodiment will be described with reference to FIGS. 1 to 6. FIG. 1 shows a functional configuration of the image evaluation prediction device 1 of the present embodiment.

The image evaluation prediction device 1 is connected to at least one information source 2 and at least one user terminal 3 via a communication network CN. The communication network connecting the information source 2 and the image evaluation prediction device 1 and the communication network connecting the user terminal 3 and the image evaluation prediction device 1 may be common or different.

The image evaluation prediction device 1 can be configured by using a general-purpose computer 100 as described later in FIG. At least a part of the image evaluation prediction device 1 may be configured from a dedicated device. Further, one or a plurality of image evaluation prediction devices 1 may be constructed by linking a plurality of general-purpose computers 100.

The information source 2 is, for example, a server that provides an SNS service or the like. The user terminal 3 is, for example, a mobile information terminal such as a notebook personal computer or a tablet terminal, or a mobile phone (including a so-called smartphone).

The configuration of the image evaluation prediction device 1 will be described. The image evaluation prediction device 1 is, for example, a content data storage unit 10, a learning unit 11, a prediction unit 12, an evaluation prediction image storage unit 13, and a provision unit 14.
To prepare for.

The content data storage unit 10 has a function of storing content data acquired from the information source 2. The content data storage unit 10 receives and stores content data from the information source 2 periodically or irregularly. Content data stored in a storage medium such as an optical disk, a flash memory, or a hard disk can also be stored in the content data storage unit 10. Content data may be acquired from an information source 2 such as a mobile information terminal or a mobile phone and stored in the content data storage unit 10.

The learning unit 11 has a function of generating an evaluation prediction model that outputs an emotional evaluation corresponding to the classification result of the image data. The learning unit 11 includes, for example, an image processing unit 111, a text processing unit 112, a processed data storage unit 113, an evaluation prediction model generation unit 114, and an evaluation prediction model storage unit 115.

The prediction unit 12 has a function of predicting an emotional evaluation of the evaluation prediction image by inputting the evaluation prediction image acquired from the evaluation prediction image storage unit 13 into the evaluation prediction model. The prediction unit 112 includes, for example, an image processing unit 121 and an evaluation prediction unit 122.

The evaluation prediction image is image data to be predicted for emotional evaluation, and corresponds to "target image data". Hereinafter, the evaluation prediction image may be referred to as a target image. In addition, image data may be abbreviated as an image.

The configuration of the learning unit 11 will be described. The image processing unit 111 has a function of converting the image data D11 in the content data into multidimensional numerical data. The image processing unit 111 can also detect an object included in the image data D11 and calculate the feature amount of the image data D11, for example.

The text processing unit 112 has a function of performing emotion analysis on the text data in the content data. The text processing unit 112 also performs preprocessing (morphological analysis, dependency analysis, etc.) for emotion analysis.

The processed data storage unit 113 stores the processing result of the image processing unit 111 and the processing result of the text processing unit 112 in association with each other.

The evaluation prediction model generation unit 114 classifies the image data D11 based on the data stored in the processed data storage unit 113. The classification method is not particularly limited. For example, it may be classified based on the pixel value of RGB, or it may be classified based on the feature points of the image. The evaluation prediction model generation unit 114 reaggregates the classified data and generates an evaluation prediction model using the reaggregated data.

The evaluation prediction model storage unit 115 stores the evaluation prediction model generated by the evaluation prediction model generation unit 114.

The configuration of the prediction unit 12 will be described. The prediction unit 12 inputs the image data D21 (evaluation prediction image data) read from the evaluation prediction image storage unit 13 to the image processing unit 121 to perform a predetermined process. The processing content of the image processing unit 121 is the same as the processing content of the image processing unit 111 of the learning unit 11. The image data processed by the image processing unit 121 is input to the evaluation prediction unit 122.

The prediction result data D22, which is the evaluation result of the evaluation prediction unit 122, is associated with the evaluation prediction image data D21 and stored in the evaluation prediction image storage unit 13. Hereinafter, the prediction result data D22 may be abbreviated as the prediction result D22.

The providing unit 14 has, for example, a function of exchanging information with the user terminal 3. The providing unit 14 receives the evaluation prediction image data from the user terminal 3 and stores it in the evaluation prediction image storage unit 13. When the processing of the evaluation prediction unit 122 is completed, the providing unit 14 reads the prediction result D22 from the evaluation prediction image storage unit 13 and transmits it to the user terminal 3 of the evaluation request source.

FIG. 2 is a hardware configuration diagram when the image evaluation prediction device 1 is configured by using the computer 100. The computer 100 includes, for example, a processor 101, a main storage device 102, an auxiliary storage device 103, an input device 104, an output device 105, and a communication device 106.

The auxiliary storage device 103 stores a computer program for realizing the learning unit 11, the prediction unit 12, the evaluation prediction image storage unit 13, and the providing unit 14 described in FIG. 1 on the computer 100. The processor 101 reads a computer program from the auxiliary storage device 103 into the main storage device 102 and executes it, thereby causing the computer 100 to function as the image evaluation prediction device 1.

The communication device 106 communicates with the information source 2 and the user terminal 3 via the communication network CN. The communication device 106 may communicate with the information source 2 and the user terminal 3 peer-to-peer by using infrared communication, short-range wireless communication, or the like. Alternatively, the image evaluation prediction device 1 can collect information from the information source 2 or receive evaluation prediction image data via the storage medium MM.

FIG. 3 is a flowchart showing the processing of the learning unit 11. The learning unit 11 acquires content data from the content data storage unit 10 (S11). The learning unit 11 processes the image data D11 included in the content data based on the feature amount (S12), and calculates the emotional evaluation of the text data included in the content data based on the feature amount. (S13). Step S12 and step S13 may be processed in parallel.

The learning unit 11 stores the image data processed in step S12 and the emotional evaluation obtained in step S13 in the processed data storage unit 113 (S14). The learning unit 11 generates an evaluation prediction model from the data stored in the processed data storage unit 113, and stores the generated evaluation prediction model in the evaluation prediction model storage unit 115 (S15).

FIG. 4 is a flowchart showing the processing of the prediction unit 12. The prediction unit 12 acquires the target image data D21 from the evaluation prediction image storage unit 13 (S21), and performs image processing on the target image data D21 based on the feature amount (S22). The prediction unit 12 calculates the emotional evaluation corresponding to the target image data D21 from the processed image data and the evaluation prediction model (S23), and outputs the calculation result (S24).

FIG. 5 shows the processing result T1 of the text processing unit 112. As a result of the text processing, T1 associates, for example, the text data with the emotional evaluation calculated by emotionally analyzing the text data.

FIG. 6 shows the processed data T2 stored in the processed data storage unit 113. The processed data T2 includes, for example, a text data column C21, an image data column C22, and an emotional evaluation column C23. A plurality of image data may be associated with one text data.

FIG. 7 is an example of the evaluation prediction model T3. The image processing unit 111 performs a classification process for grouping similar images. For each category C21, a text data column C22 belonging to the category, an image data column C23 belonging to the category, and a sensitivity evaluation column C24 are associated with each other.

According to this embodiment configured in this way, a prediction model is automatically generated by associating the classification result of the image data with the emotional evaluation of the text data related to the image data without conducting a questionnaire survey or the like. can do.

Therefore, by using the image evaluation prediction device 1 of this embodiment, it is possible to predict the emotions caused by the image data to be posted on the SNS, blog, net news, etc., or to design a company logo mark or a new product. It is possible to objectively estimate the reputation in the market such as.

The second embodiment will be described with reference to FIGS. 8 and 9. In each of the following examples including this embodiment, the differences from the first embodiment will be mainly described. In this embodiment, a user layer of image data is defined, and what kind of emotional evaluation the image data receives in each user layer is predicted.

FIG. 8 is a functional configuration diagram of the image evaluation prediction device 1A according to the present embodiment. When the image evaluation prediction device 1A collects the content data from the information source 2, the information D13 of the creator of the content data is also collected and stored in the content data storage unit 10.

As the user information D13, for example, a profile sentence described in a content data announcement medium such as an SNS or a blog can be used. The user information estimation unit 116 can also estimate at least a part of the user information D13 by analyzing the content data posted by the creator in the past.

The user information D13 includes, for example, at least one attribute such as gender, age, place of residence, hobbies, tastes, occupation, and mother tongue of the user (creator of content data, reader of content data). The user information D13 is associated with the processing result of the image processing unit 111 and the processing result of the text processing unit 112, and is stored in the processed data storage unit 113.

FIG. 9 is an explanatory diagram showing how the emotional evaluation D22 of the same target image data D21 is predicted for each user layer obtained by classifying from a plurality of user information. The image evaluation prediction device 1A outputs an emotional evaluation of the selected user attribute (user layer). The target user layer may be selected in advance, may be automatically selected according to the type of target image data, or all user layers may be selected.

As shown in the lower part of FIG. 9, the image evaluation prediction device 1A is used for each user group consisting of a plurality of user attributes, such as "female in her thirties", "male who likes anime", and "teen in Japan". , The emotional evaluations D22 (1) to D22 (3) of the target image data are output.

This embodiment configured in this way also has the same effect as that of the first embodiment. Further, according to this embodiment, since the reaction when the selected user group sees the target image data can be predicted, it is convenient for companies, local governments, individuals, etc. who use the image evaluation prediction device 1A. Is improved.

The third embodiment will be described with reference to FIGS. 10 and 11. The image evaluation prediction device 1B of the present embodiment processes the target image data so that the emotional evaluation by the selected user group approaches the desired emotional evaluation.

FIG. 10 is a functional configuration diagram of the image evaluation prediction device 1B of this embodiment. The configuration of FIG. 10 differs from the configuration of FIG. 8 in that the prediction unit 12 includes a result processing unit 123 and an image processing unit 124.

The result processing unit 123 detects a difference between the result of the emotional evaluation and the designated emotional evaluation D3 (see FIG. 11), and calculates an image processing method for reducing the difference. The image processing unit 124 processes the target image data D21 based on the instruction of the result processing unit 123. The processed target image data D21 is stored in the evaluation prediction image storage unit 13, and is input to the image processing unit 121 again, and the evaluation prediction unit 122 calculates the emotional evaluation. This cycle is repeatedly executed until the difference between the emotional evaluation of the target image data D21 and the designated emotional evaluation D3 is within a predetermined value, or the user of the image evaluation prediction device 1B instructs to stop. You may. Image processing may be performed little by little, and the user's judgment may be sought each time.

FIG. 11 shows a state in which the target image data D21 is processed and evaluated so that the emotional evaluation of the target image data D21 approaches the designated emotional evaluation D3. In the example of FIG. 11, the desired condition of "giving a surprise to a woman in her thirties" is specified. The desired conditions include an emotional evaluation in the target user group. The desired condition may be input as a natural sentence, a switch for selecting a user group, and a switch for selecting an index of emotional evaluation and its value (for example, surprise: 0.5, joy: 0.3, etc.). It may be input from a screen including.

For example, the result processing unit 123 changes the color (RGB value) of the target image data D21 so that the emotional evaluation approaches the emotional evaluation under the specified desired conditions. As is known in color engineering and color psychology, the impression that color gives to people is great. Therefore, the user layer (reader layer) in which the target image data D21 is selected by changing the color of the target image data D21 as a whole or changing the color of a part of the area in the target image data D21. ) The emotional evaluation given to) changes.

In the prediction result D22, the emotional evaluation predicted to occur in the user group is expressed by a plurality of emotional indexes and their values. The emotional index is a type of main emotion such as "surprise", "joy", "anger", and "sadness". By confirming the prediction result D22, the user of the image evaluation prediction device 1B can determine whether or not the target emotional evaluation is approaching. This judgment can also be automated using deep learning or the like.

This embodiment configured in this way also has the same effects as those of the first embodiment and the second embodiment. Further, according to the present embodiment, the target image data can be processed so as to approach the emotional evaluation specified for the selected user group, which is convenient for the user of the image evaluation prediction device 1B. Is improved.

The fourth embodiment will be described with reference to FIG. The image evaluation prediction device 1C of the present embodiment corrects the emotional evaluation of the target image data D21 by the emotional evaluation of the text data D23 used in association with the target image data D21.

FIG. 12 is a schematic diagram of the image evaluation prediction device 1C. The image evaluation prediction device 1C includes a learning unit 11, a prediction unit 12, and a correction unit 15, which are not shown. This embodiment can be combined with any of the first to third embodiments. Therefore, the image evaluation prediction device 1C has a function of collecting user information and defining a user layer, or an emotional evaluation of the target image data D21 so as to approach the emotional evaluation specified for the selected user layer. Can be provided with either or both of the functions for processing the target image data D21.

The text data D23 associated with the target image data D21 is, for example, an image (product, service, person, animal or plant, etc.) that is the subject of the target image data D21, such as a catch phrase, an advertising message, a product name, or a service name. ) Is the text data associated with.

The names of products and services may affect the emotional evaluation calculated from the image data. Therefore, the image evaluation prediction device 1C calculates the emotional evaluation of the text data D23 associated with the target image data D21 by the text processing unit 112. The correction unit 15 of the image evaluation prediction device 1C corrects the emotional evaluation of the target image data D21 by using the emotional evaluation calculated from the text data D23. The image evaluation prediction device 1C outputs the corrected prediction result D22. As shown in FIG. 12, it is also possible to output the corrected Kansei evaluation for each of the selected user groups.

This embodiment configured in this way also has the same effects as those of the first to third embodiments. Further, according to this embodiment, since the emotional evaluation of the target image data D21 can be corrected by using the emotional evaluation of the text data D23 associated with the target image data D21, the image data D21 and the text data D23 can be corrected. It is possible to investigate how the emotional evaluation changes depending on the combination of the above, and it is possible to support the selection of the optimum combination.

The fifth embodiment will be described with reference to FIG. This embodiment can be combined with any of the first to fourth embodiments.

The image evaluation prediction device 1D of this embodiment is provided with a prediction unit 12 in the user terminal 3D. That is, the evaluation prediction model generated by the learning unit 11 in the image evaluation prediction device 1D is transmitted to the user terminal 3D and stored. The user acquires the target image data D21 using either the camera 31 or the web browser 32 of the user terminal 3D, and stores the acquired target image data D21 in the evaluation prediction image storage unit 13.

The prediction unit 12 of the user terminal 3D calculates an emotional evaluation of the target image data D21 read from the evaluation prediction image storage unit 13, and outputs the calculated emotional evaluation via the user interface device 33. The user interface device 33 is, for example, a display device. A device in which a touch panel and a display are integrated may be used. The display may be configured as a headset used for virtual reality and the like. The image evaluation prediction devices 1, 1A, 1B, 1C, and 1D described in each embodiment can also provide the prediction result D22 in the virtual reality space.

This embodiment configured in this way also has the same effect as that of each of the above-mentioned embodiments. Further, according to the present embodiment, since the prediction unit 12 is provided in the user terminal 3D, the user (user of the present invention) can take a picture of an object of concern during a walk or shopping, for example. You can check the emotional evaluation. As a result, it is possible to support marketing activities and the like, and the usability for the user is improved.

The present invention is not limited to the above-described embodiment. Those skilled in the art can make various additions and changes within the scope of the present invention. In the above-described embodiment, the configuration is not limited to the configuration example shown in the attached drawings. The configuration and processing method of the embodiment can be appropriately changed within the range of achieving the object of the present invention.

Further, each component of the present invention can be arbitrarily selected, and the invention including the selected configuration is also included in the present invention. Further, the configurations described in the claims can be combined in addition to the combinations specified in the claims.

1,1A-1D: Image evaluation prediction device, 2: Information source, 3,3D: User terminal, 10: Content data storage unit, 11: Learning unit, 12: Prediction unit, 13: Evaluation prediction image storage unit, 14: Providing unit, 15: correction unit, 31: camera, 32: web browser, 33: user interface device, 111: image processing unit, 112: text processing unit, 113: processed data storage unit, 114: evaluation prediction model generation unit. , 115: Evaluation prediction model storage unit, 116: User information estimation unit, 121: Image processing unit, 122: Evaluation prediction unit, 123: Result processing unit, 124: Image processing unit

Claims (8)

An image evaluation prediction device that predicts the emotional evaluation of image data.
It includes a learning unit and a prediction unit that predicts the emotional evaluation of image data based on the learning results of the learning unit.
The learning unit
Acquire content data including image data and text data corresponding to the image data,
Sentiment analysis is performed on the text data in the content data to calculate the emotional evaluation.
The image data in the content data is classified and
Based on the classification result of the image data and the emotional evaluation of the text data, a prediction model that outputs the emotional evaluation corresponding to the input classification result of the image data is generated.
The prediction unit
By inputting the target image data into the prediction model, the emotional evaluation of the target image data is calculated.
Image evaluation prediction device. The learning unit further acquires the information of the creator of the content data, emotionally analyzes the text data for each user attribute generated from the information of the creator, and calculates the emotional evaluation.
The prediction unit calculates an emotional evaluation of the target image data for the selected user attribute.
The image evaluation prediction device according to claim 1. The prediction unit further includes an image processing unit that processes the target image data so as to approach a designated Kansei evaluation.
The image evaluation prediction device according to claim 2. The predetermined text data associated with the target image data is emotionally analyzed to calculate a correction emotional evaluation, and the prediction unit calculates the emotional evaluation calculated from the target image data from the predetermined text data. Correct using the calculated sentimental evaluation for correction,
The image evaluation prediction device according to claim 3. The content data is content data in which the image data or the link information indicating the image data and the text data corresponding to the image data exist in the same post.
The image evaluation prediction device according to claim 1. An image evaluation prediction device that predicts the emotional evaluation of image data.
Input based on the emotional evaluation calculated by emotionally analyzing the text data in the content data including the image data and the text data corresponding to the image data, and the classification result of the image data in the content data. A prediction model that outputs an emotional evaluation corresponding to the classification result of the image data
An evaluation prediction unit that calculates the emotional evaluation of the target image data by inputting the target image data into the prediction model.
Image evaluation prediction device including. It is an image evaluation prediction method that predicts the emotional evaluation of image data using a computer.
Acquire content data including image data and text data corresponding to the image data,
Sentiment analysis is performed on the text data in the content data to calculate the emotional evaluation.
The image data in the content data is classified and
Based on the classification result of the image data and the emotional evaluation of the text data, a prediction model that outputs the emotional evaluation corresponding to the classification result of the input image data is generated.
By inputting the target image data into the prediction model, the emotional evaluation of the target image data is calculated.
Image evaluation prediction method. A computer program that makes a computer function as an image evaluation predictor that predicts the emotional evaluation of image data.
A function to acquire image data and content data including text data corresponding to the image data, and
A function to calculate emotional evaluation by sentiment analysis of text data in the content data, and
The function to classify the image data in the content data and
Based on the classification result of the image data and the emotional evaluation of the text data, a function to generate a prediction model that outputs an emotional evaluation corresponding to the classification result of the input image data, and
A function to calculate the emotional evaluation of the target image data by inputting the target image data into the prediction model, and
A computer program that makes a computer realize.

Images