JP7164231B1 - Casting device, method and program - Google Patents

Abstract

Kind Code: A1 To provide a technology capable of efficiently executing a document selection process in selecting a candidate (candidate model) that meets a request. A casting apparatus (200) includes a storage unit (220) that stores composite data including a plurality of face images photographed from a plurality of angles for each model, a selection unit (240) that selects a plurality of candidate models, and a selection unit (240) that selects a plurality of candidate models. The extracting unit 250 extracts one face image from the composite data stored in the storage unit 220 for each candidate model extracted. A display control unit 260 that arranges the face images of each candidate model side by side and displays a list on the display device so that the orientation of the body is directed toward the center and the height of the eyes of each candidate model is substantially constant. Prepare. [Selection drawing] Fig. 3

Description

The present invention relates to a casting device, method and program.

Advertisers may use models, influencers, celebrities, etc. ( (hereinafter collectively referred to as models), and promotes sales through advertisements, publication in magazines, appearances on television, etc.

When contracting with a model, the advertiser (hereafter, the client is collectively referred to as the advertiser, including cases where the client is indirectly requested by a production company, advertising agency, etc.) is responsible for casting work such as arranging models and managing portrait rights. Ask the casting agency to select a candidate model. Upon receiving such a request, the casting agency selects multiple models (for example, 200 models) that match the request from among the many models belonging to each partner model agency, and selects each selected model. After providing the model composite to the advertiser, hold an audition with the advertiser. Here, a "composite" is a group of photographs in which a model's appearance and expression can be confirmed from various angles, and is used as an appealing material during an audition.

Currently, advertisers need to carefully examine a large number of model composites before deciding on a model. It is difficult to compare models uniformly.

Therefore, advertisers go through a process of inviting many selected models, conducting face-to-face auditions, and deciding which model to contract, but the selection process leading up to the audition takes a lot of time and money. have to spend. In addition, even at the office to which the model belongs, it is necessary to spend a lot of time on the process of document selection, etc. until the model participates in one audition, and the model decision process is inefficient. problem was pointed out.

The present invention has been made in view of the circumstances described above, and an object of the present invention is to provide a technique capable of efficiently narrowing down candidates (candidate models) that meet a request.

A casting device according to an aspect of the present disclosure is a casting device that assists an advertiser in determining a model, and is a storage unit that stores composite data including a plurality of face images shot from a plurality of angles for each model. a selection unit that selects a plurality of candidate models; an extraction unit that extracts one face image from the composite data stored in the storage unit for each of the selected candidate models; and each extracted candidate Based on the facial images of the models, the facial images of the candidate models are arranged such that the face or body of each candidate model is directed substantially toward the center and the eye height of each candidate model is substantially constant. and a display control unit for horizontally arranging and displaying a list on a display device.

According to the predetermined aspect of the present invention, it is possible to efficiently narrow down the candidate models that match the request.

It is a figure showing an example of composition of a model decision support system concerning a 1st embodiment. It is a figure which shows an example of a hardware configuration of an advertiser terminal and a casting device. It is a block diagram which shows the functional structure of a casting apparatus. It is the figure which illustrated the registration screen of composite data. FIG. 11 is a diagram illustrating a candidate model proposal screen using an eye-catching image ICP; It is a block diagram showing a functional configuration of a casting device according to a second embodiment. FIG. 4 is an explanatory diagram of analysis operation by an analysis unit; It is the figure which illustrated the learning process of the learning model in the classification part. It is the figure which illustrated the registration content of composite DB.

Preferred embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that, in each figure, the same reference numerals have the same or similar configurations.

A. First Embodiment <System Configuration>
FIG. 1 is a diagram showing an example of the configuration of a model determination support system 1000 according to the first embodiment.
This system 1000 converts a group of face photographs (face images) (hereinafter referred to as composite data), which are used as appealing materials for each model, into electronic data and uses them to create candidate models that meet the request of an advertiser who is a client. can be efficiently narrowed down.

The model determination support system 1000 includes an advertiser terminal 100 operated by an advertiser and a casting device 200 that supports model determination by the advertiser. The advertiser terminal 100 and the casting device 200 can communicate with each other via the network N.

One or more portions of network N may be wired or wireless networks. Network N can be, by way of example and not limitation, an Ad Hoc Network, an Intranet, an Extranet, a Virtual Private Network (VPN), a Local Area Network (LAN), a wireless Wireless LAN (WLAN), Wide Area Network (WAN), Wireless WAN (WWAN), Metropolitan Area Network (MAN), Part of the Internet, Public Switched Telephone Network (Public Part of Switched Telephone Network (PSTN), mobile phone networks, ISDNs (Integrated Service Digital Networks), wireless LANs, LTE (Long Term Evolution), CDMA (Code Division Multiple Access), Bluetooth (registered trademark), satellite communication, etc., or a combination of two or more thereof.

The advertiser terminal 100 is a terminal operated by an advertiser (for example, a person in charge of a company that promotes products or services), and is composed of a personal computer, a handheld computer device, a smart phone, a mobile phone, a wearable terminal, or the like. . The advertiser appropriately operates the advertiser terminal 100 to request the casting device 200 to cast a model.

The casting apparatus 200 is composed of, for example, a server computer, and is managed by an administrator of a company that conducts model casting work (hereinafter referred to as a casting administrator). The casting device 200 uses composite data of each model stored in an on-premises type or cloud type database to select a model according to a request from an advertiser (details will be described later). It should be noted that the advertiser terminal 100 and the casting device 200 do not necessarily have to be one computer, and may be composed of a plurality of computers distributed over the network N and storages.

<Hardware configuration>
FIG. 2 is a diagram showing an example of the hardware configuration of the advertiser terminal 100 and the casting device 200. As shown in FIG. The advertiser terminal 100 and the casting device 200 each include a CPU (Central Processing Unit) 1, a storage device 2 such as a ROM (Read Only Memory) or a RAM (Random Access Memory), and a communication IF (Interface) for performing wired or wireless communication. 3. It has an input device 4 for receiving input operations and an output device 5 for outputting information. The input device 4 is, for example, a keyboard, mouse, touch panel, microphone, or the like. The output device 5 is, for example, a display (screen) and/or a speaker.

The functional configuration and functional overview of the casting apparatus 200 will be described below with reference to FIGS. 3 to 5. FIG.

<Functional configuration and function overview>
FIG. 3 is a block diagram showing the functional configuration of the casting device 200. As shown in FIG.

Casting device 200 includes a reception unit 210 , a storage unit 220 , a registration unit 230 , a selection unit 240 , an extraction unit 250 and a display control unit 260 .
The reception unit 210 , the registration unit 230 , the selection unit 240 , the extraction unit 250 , and the display control unit 260 are implemented by the CPU 1 of the casting device 200 executing various programs stored in the storage device 2 . Note that the program can be stored in a storage medium such as a USB memory or a CD-ROM.

The accepting unit 210 accepts a model casting request and the like in accordance with the operation of the advertiser terminal 100 by the advertiser, and also accepts an input operation and the like by the casting manager.

The storage unit 220 is implemented by the storage device 2, and stores programs for implementing the model casting function, various data, and the like. The storage unit 220 also includes a composite database (DB) 221 that stores composite data of each model.

The composite data includes a group of photographs of faces prepared for each model as appealing materials (for example, about seven photographs in PDF or JPEG format for one model). Composite data is provided by the model office to which each model belongs. When the casting manager receives new model composite data from a model agency or the like, it registers it in the composite DB 221 .

FIG. 4 is a diagram exemplifying a composite data registration screen W1.
The registration screen W1 has an entry field EF1 for entering basic information of each model (hereinafter referred to as model basic information) such as model name, pronunciation, gender, date of birth, affiliated office, race, etc., as well as composite data CMD. to the composite DB 221, and a registration button B2 for registering the eye-catching image ICP.

The eye-catching image ICP is one of a plurality of facial photographs forming the composite data CMD, and is an eye-catching facial photograph used when comparing with other models. In this embodiment, the casting manager selects an eye-catching image ICP and edits the eye-catching image ICP (for example, trimming, color tone correction, etc.) by manually operating a mouse or the like.

FIG. 5 is a diagram exemplifying a candidate model proposal screen W2 using an eye-catching image ICP.
On the proposal screen W2, eye-catching images ICP of a plurality of candidate models are displayed side by side as a list. The proposal screen W2 is displayed on the display panels of the casting device 200 and the advertiser terminal 100, or the like. In this embodiment, as shown in FIG. 5, eye-catching images ICP of a plurality of candidate models are arranged in a horizontal direction and displayed as a list, so that the plurality of candidate models can be compared and examined more efficiently than the conventional document screening. It is possible. Further, each candidate model displayed on the proposal screen W2 is designed such that the face or body is directed toward the center and the eye level is substantially constant, for example, as follows. there is It should be noted that these measures are only examples, and other methods may be used so that the face or body of each candidate model faces the center and the height of the eyes is kept substantially constant.

<Example of trimming of eye-catching image ICP>
Trimming is performed so that the image size of the eye-catching image ICP of each candidate model is substantially constant and the proportion of the face of each candidate model in the eye-catching image ICP is substantially constant (see FIG. 5).

<Example of device for display (vertical direction) of eye-catching image ICP>
・Arrange the eye and jaw lines of each candidate model so that they are almost constant (see Fig. 5).

<Example of device for display (horizontal direction) of eye-catching image ICP>
・Place each candidate model whose face or body is facing the front in the center, place each candidate model facing right on the right side as seen from the judging side, and place each candidate model facing left from the judging side. It is placed on the left side as seen (see FIG. 5).

By implementing the above-described measures, the advertiser can compare and consider these candidate models more efficiently than the conventional document screening.

Returning to FIG. 3, the registration unit 230 registers the composite data CMD and the like in the composite DB 221 based on the operation of the casting manager. When the casting manager receives new composite data CMD from, for example, a model agency or the like, he or she follows the registration screen W1 shown in FIG. perform an operation. As a result, in the composite DB 221, composite data CMD including model basic information, an eye-catching image ICP, etc. are registered for each model.

The selection unit 240 selects a plurality of candidate models according to candidate model selection conditions input by a casting manager. For example, when the casting manager inputs that a female model in her thirties should be selected (that is, inputs the selection conditions for candidate models), the selection unit 240 accesses the composite DB 221 and collects the composite data of each model. See CMD. Then, the selection unit 240 selects a plurality of candidate models that match the input candidate model selection conditions by referring to the model basic information included in each composite data CMD, and notifies the extraction unit 250 of them.

When the selection unit 240 notifies the extraction unit 250 of the selected candidate models, the extraction unit 250 refers to the composite DB 221 , extracts the eye-catching image ICP of each selected candidate model, and sends it to the display control unit 260 .

When the display control unit 260 receives the eye-catching image ICP of each candidate model extracted by the extraction unit 250, the display control unit 260 adjusts the orientation of the face or body of each candidate model so that it faces the center and the height of the eyes is approximately A proposal screen W2 (see FIG. 5) is generated in which the eye-catching images ICP of each candidate model are horizontally arranged so as to be uniform, and this is displayed as a list on the display panels of the casting device 200 and the advertiser terminal 100. FIG.

The advertiser confirms the proposal screen W2 displayed on the advertiser terminal 100, and compares and considers these candidate models. Then, for example, several candidate models are selected from among the considered candidate models, and an actual audition is held to finally determine the model that matches the image of the company's product. Thus, according to this embodiment, it is possible to efficiently narrow down the candidate models that meet the request of the advertiser who is the client.

In the above embodiment, an actual audition is held to finalize the model that matches the image of the company's product. may decide.
Also, in the above embodiment, the female model was explained as an example, but the present invention can also be applied to male models, child models, and the like. Furthermore, it can be applied to models other than humans, such as dogs and cats, and any other animal model.

Further, in the above-described embodiment, the case where the casting manager manually edits the eye-catching image ICP (for example, trimming, color tone correction, etc.) has been described. May be edited. Specifically, the casting device 200 is provided with a processing unit that edits and processes the eye-catching image ICP. The processing unit is equipped with well-known machine learning functions such as neural networks and deep learning. The processing unit uses a machine learning function to perform trimming, color tone correction, and the like on the above-described eye-catching image ICP. For example, in the case of trimming, the processing unit uses a machine learning function so that the image size of the eye-catching image ICP of each candidate model is substantially constant, and each candidate occupying the eye-catching image ICP Crop the model's face so that the proportions are almost constant. With such a configuration, it is possible to greatly reduce the workload of the casting manager.

B. Second Embodiment In the first embodiment, a casting manager manually inputs selection conditions for candidate models according to a theme desired by an advertiser, and selects candidate models. On the other hand, in a second embodiment described below, a case will be described in which the casting device 200 automatically selects candidate models according to the theme desired by the advertiser.

FIG. 6 is a block diagram showing the functional configuration of a casting device 200 according to the second embodiment. The casting device 200 is the same as the casting device 200 shown in FIG. Therefore, the same reference numerals are given to the corresponding parts, and detailed explanations thereof are omitted.

The analysis unit 270 has an image recognition tool (for example, an image recognition API using machine learning), analyzes at least one facial photograph included in the composite data CMD of each model, and classifies the analysis results. Output to the unit 280 .

FIG. 7 is an explanatory diagram of the analysis operation by the analysis unit 270. As shown in FIG.
As shown in FIG. 7, the analysis unit 270 analyzes the facial photograph of the model to determine the feature points of the model's face, i. , position of eyebrows, etc.) as well as the model's facial expression (joy, sadness, anger, surprise, etc.). The analysis unit 270 outputs analysis results including the acquired analysis data of facial parts (hereinafter referred to as facial parts data) and facial expression analysis data (hereinafter referred to as facial expression data) to the classification unit 280 .

The classification unit 280 has a learning model LM, and classifies the image type of each model by using the analysis results supplied from the analysis unit 270 as well as the casting history information and learning data accumulated so far. "Image type" is, for example, "beautiful" (image of a well-proportioned and beautiful face), "cute" (a cute image that makes you think of an idol), "elegant type" (an image that makes you think of a graceful lady, etc.). ), “cool” (image that suggests dignified coolness), etc., and expresses the impression that the user receives from the model. Here, four image types are exemplified, but the number of image types to be classified can be appropriately set/changed according to the design or the like.

In addition, "casting history information" includes face parts data and facial expression data (i.e., past model information on casting results). The storage unit 220 stores a certain number of past casting history information (for example, several thousands).

As described below, the classification unit 280 uses the data of each model reclassified by a person with a lot of casting experience (for example, a casting manager; below, assumes a casting manager) after classification as training data. By learning, a learning model LM with high prediction accuracy is constructed.
FIG. 8 is a diagram illustrating the learning process of the learning model in the classification section 280. As shown in FIG.
For example, assume that two "beautiful" candidate models (model A and model B) have been listed in response to a request from an advertiser. Finally, when model A is selected as a "beautiful" model by the casting manager and model B is not selected as a "beautiful" model, the classification unit 280 classifies model A's facial part data and It learns that facial expression data is appropriate as a model of "beautiful". On the other hand, the classification unit 280 learns that the facial parts data and expression data of model B are not appropriate as a model of "beautiful". In this way, the classification unit 280 uses the facial parts data and facial expression data of each model reclassified by a person with extensive casting experience as learning data (teacher data), thereby increasing the classification accuracy of the image type of each model. is increasing.

When the classification unit 280 receives the facial parts data and the expression data of the new model from the analysis unit 270, it inputs it to the learning model LM, and classifies the image type of this model (for example, “cute”). , the degree of the image type (for example, "○ ×%"; hereinafter, image degree) is obtained. Then, the classification unit 280 includes the image classification data representing the image type classification result (including the degree of image, etc.) together with the face part data and display data of the model in the composite data CMD of the model, and stores them in the composite DB 221. (see Figure 9).

Upon receiving a request for the type of model that the advertiser wishes to use (for example, “elegant”) from the advertiser terminal 100 , the reception unit 210 outputs this to the selection unit 240 .
Upon receiving the advertiser's request, the selection unit 240 selects a plurality of applicable candidate models by referring to the composite DB 221 in response to the advertiser's request. For example, when an advertiser requests an "elegant" model, the selection unit 240 selects a plurality of "elegant" candidate models by referring to the image classification data included in the composite data CMD. and notifies the extracting unit 250.

The extraction unit 250 extracts the eye-catching images ICP of the plurality of notified candidate models from the composite DB 221 and sends them to the display control unit 260 .

When the display control unit 260 receives the eye-catching image ICP of each candidate model extracted by the extraction unit 250, the display control unit 260 adjusts the orientation of the face or body of each candidate model to face the center, as in the first embodiment described above. Also, a proposal screen W2 (see FIG. 5) in which the eye-catching images ICP of each candidate model are arranged horizontally so that the height of the eyes is substantially constant is displayed on the display panels of the casting device 200 and the advertiser terminal 100. List.

According to such a configuration, the casting device 200 automatically selects candidate models according to the theme desired by the advertiser. Since the casting device 200 learns the learning model LM using the data of each model reclassified by a person with a lot of casting experience as learning data, it is possible to increase the accuracy of selecting the candidate model.

C. Modifications The present invention is not limited to the above-described embodiments, and can be implemented in various other forms without departing from the gist of the present invention. Therefore, the above-described embodiment is merely an example in all respects, and should not be construed as limiting.

In the above embodiment, the case where one eye-catching image ICP is registered for each model has been described, but a plurality of eye-catching images ICP may be registered. Instead of selecting the eye-catching image ICP from among the photographs forming the composite data CMD, another photograph not included in the composite data CMD may be selected as the eye-catching image. Furthermore, in the above embodiment, image types such as "beautiful", "cute", "elegant", and "cool" were exemplified. Can be set/changed. For example, in the case of a male model, image types such as "beautiful", "cute", "cool", and "unique" may be set.

Further, in this specification, the term "unit" does not simply mean a physical configuration, but also includes the case where the processing executed by the "unit" is implemented by software. Also, even if the processing executed by one “unit” or device is realized by two or more physical configurations or devices, the processing executed by two or more “units” or devices can be implemented by one physical means or device. may be realized by

DESCRIPTION OF SYMBOLS 1000... Model decision support system 100... Advertiser terminal 200... Casting apparatus 210... Reception part 220... Storage part 221... Composite DB 230... Registration part 240... Selection part 250... Extraction part 260... Display control unit 270 Analysis unit 280 Classification unit LM Learning model CMD Composite data ICP Featured image W1 Registration screen W2 Proposal screen.

Claims (8)

A casting device that assists a client in determining a model,
a storage unit for storing composite data including a plurality of face images photographed from a plurality of angles for each model;
a selection unit that selects a plurality of candidate models;
an extraction unit for extracting one face image from the composite data stored in the storage unit for each selected candidate model;
Based on the extracted face image of each candidate model, the above-described a display control unit for horizontally arranging face images of each candidate model and displaying a list on a display device;
A casting device comprising a an analysis unit that analyzes the face image of each model and outputs analysis results;
a classifying unit that classifies the model type of each model based on the analysis results of each model and information on casting results of past models;
further comprising a reception unit that receives input of the model type,
The selection unit
2. The casting apparatus according to claim 1, wherein models classified into the input model type are selected as said candidate models. 3. The casting apparatus according to claim 2, wherein the analysis unit analyzes the face image of each model to output analysis results including facial parts data and facial expression data. The composite data includes model basic information,
4. The selection unit according to any one of claims 1 to 3, wherein the selection unit refers to the model basic information included in the composite data of each model to select the candidate models that match selection conditions for input candidate models. 3. Casting equipment according to paragraph. The plurality of face images include an eye-catching image for comparison,
The extraction unit extracts the eye-catching image from the composite data for each selected model,
Based on the extracted eye-catching image of each candidate model, the display control unit is arranged so that the orientation of the face or body of each candidate model is directed substantially toward the center and the height of the eyes of each candidate model is substantially adjusted. 5. The casting device according to any one of claims 1 to 4, wherein the eye-catching images of the candidate models are arranged side by side and listed on the display device so as to be constant. further comprising a processing unit for trimming the eye-catching image,
6. The casting device according to claim 5 , wherein said extraction unit extracts said trimmed eye-catching image from composite data for each selected model candidate. A method for assisting a client in determining a model, performed by a computer having a storage unit that stores composite data including multiple facial images taken from multiple angles for each model, the method comprising:
a selection step of selecting a plurality of candidate models;
an extracting step of extracting one facial image from the composite data stored in the storage unit for each selected candidate model;
Based on the extracted face image of each candidate model, the above-described and a display control step of displaying the facial images of each candidate model side by side in a list on a display device. A program for assisting a client in determining a model, comprising a computer having a storage unit for storing composite data including multiple facial images taken from multiple angles for each model,
a selection step of selecting a plurality of candidate models;
an extracting step of extracting one facial image from the composite data stored in the storage unit for each selected candidate model;
Based on the extracted face image of each candidate model, so that the orientation of the face or body of each candidate model is directed substantially toward the center and the height of the eyes of each candidate model is substantially constant, and a display control step of arranging the face images of the candidate models side by side and displaying the list on a display device.

Images