JP7100513B2 - Video screening equipment, computer programs and video screening methods - Google Patents

Description

The present invention relates to a video examination device , a computer program, and a video examination method.

In various public facilities such as station premises, station platforms, buildings such as trains, buses, and buildings, commercial facilities such as stores and movie theaters, and airports, an advertisement distribution system that sends information such as advertisements to display devices connected to the network. Is used (see Patent Document 1).

In an advertisement distribution service using such an advertisement distribution system, for example, an advertising agency performs operations such as confirmation of empty slots, application for advertisements, and submission of advertisement materials such as images from advertisers. In addition, the media company (administrator) that manages the advertisement distribution manages the application of the advertisement, examines whether the advertisement material contains inappropriate expressions, and manages the schedule of the advertisement distribution.

Japanese Unexamined Patent Publication No. 2001-338216

However, since the reviewer manually checks whether the advertising material contains inappropriate expressions, the reviewer is burdened and the review takes a long time. There is a problem.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide a video screening device , a computer program, and a video screening method capable of shortening the screening time of video for advertisement.

The video examination device according to the embodiment of the present invention is a video examination device that examines an image for advertisement, and is included in a video acquisition unit that acquires an image for advertisement and an image acquired by the image acquisition unit. A pass / fail information acquisition unit that acquires pass / fail information on whether or not at least one of images, characters, or sounds passes a predetermined examination condition, and a learning processing unit that learns a learning model using the video and pass / fail information as teacher data. Be prepared.

The video examination device according to the embodiment of the present invention is a video examination device that examines a video for advertisement, and has a video acquisition unit that acquires the video for advertisement and at least images, characters, or sounds contained in the video. One is provided with a learning model learned based on pass / fail information of whether or not a predetermined examination condition is passed, and the learning model is whether or not the image acquired by the image acquisition unit passes the examination condition. Output pass / fail information.

The information processing device according to the embodiment of the present invention is an information processing device provided with a display screen, and pass / fail information on whether or not the video submitted for each of a plurality of advertising agencies passes a predetermined examination condition. It is equipped with a display processing unit that displays a list of.

The computer program according to the embodiment of the present invention has a process of acquiring an image for advertisement on a computer and whether or not at least one of images, characters or sounds contained in the acquired image passes a predetermined examination condition. The process of acquiring the pass / fail information and the process of learning the learning model using the video and the pass / fail information as teacher data are executed.

The computer program according to the embodiment of the present invention has a process of acquiring a video for advertisement on a computer, and whether or not at least one of images, characters, or sounds contained in the video passes a predetermined examination condition. Using the learning model learned based on the information, a process of outputting pass / fail information as to whether or not the acquired video passes the examination condition is executed.

The video examination method according to the embodiment of the present invention is a video examination method for examining a video for advertisement, in which a video for advertisement is acquired and at least one of images, characters, or sounds included in the acquired video. The pass / fail information as to whether or not one passes a predetermined examination condition is acquired, and the learning model is trained using the video and the pass / fail information as teacher data.

The video screening method according to the embodiment of the present invention is a video screening method for screening a video for advertisement, in which a video for advertisement is acquired and at least one of an image, a character, or an audio contained in the video is predetermined. Using the learning model learned based on the pass / fail information of whether or not the examination condition is passed, the pass / fail information of whether or not the acquired video passes the examination condition is output.

According to the present invention, it is possible to shorten the examination time of the video for advertisement.

It is a schematic diagram which shows an example of the structure of the image examination system of this embodiment. It is a block diagram which shows an example of the structure of a media company terminal apparatus. It is a block diagram which shows an example of the structure of a video examination apparatus. It is a schematic diagram which shows the 1st example of the structure of a learning model. It is a schematic diagram which shows the process performed in a convolution layer. It is a schematic diagram which shows the process performed in a pooling layer. It is a schematic diagram which shows the 2nd example of the structure of a learning model. It is a schematic diagram which shows an example of the upload screen displayed on the display panel of the media company terminal apparatus. It is a schematic diagram which shows an example of the examination result list screen displayed on the display panel of the media company terminal apparatus. It is a schematic diagram which shows an example of the examination result detail screen displayed on the display panel of the media company terminal apparatus. It is a schematic diagram which shows the other example of the examination result list screen displayed on the display panel of the media company terminal apparatus. It is explanatory drawing which shows an example of a video format. It is a flowchart which shows an example of the processing procedure of the processing part in a learning mode. It is a flowchart which shows an example of the processing procedure of the processing part in the examination mode. It is a flowchart which shows an example of the processing procedure of the control part of the media company terminal apparatus.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic diagram showing an example of the configuration of the video examination system of the present embodiment. The video screening system examines client terminal devices 10, ... 10, media company terminal devices 20 and 20 as information processing devices provided in the media company, and video for advertising. The device 50 is connected via a network 1 such as the Internet. The video examination device 50 can provide, for example, an automatic examination of video for advertisement to the media company terminal device 20 as a cloud service.

In the client terminal device 10 of the advertising agency, for example, operations such as confirmation of an empty frame of an advertisement, application for an advertisement, and submission of an advertisement material such as a video from an advertiser are performed. In addition, the media company is a management company that manages the distribution of advertisements, and the media company terminal device 20 of the media company manages advertisement applications, examines whether the advertisement material contains inappropriate expressions, and advertises. Business such as delivery schedule management is performed.

For the video for advertisement that has passed the examination, the distribution person in charge of the media company notifies the advertising agency of the passing of the examination through the terminal device 20 of the media company, and at the same time, for example, in the station yard, the station platform, the train, etc. It is distributed to display devices connected to the network in buildings such as buses and buildings, commercial facilities such as stores and movie theaters, and various public facilities such as airports.

FIG. 2 is a block diagram showing an example of the configuration of the media company terminal device 20. The media company terminal device 20 includes a control unit 21, a communication unit 22, a storage unit 23, a display panel 24, a display processing unit 25, and an operation unit 26 that control the entire device. The media company terminal device 20 can be configured by, for example, a desktop computer, a notebook personal computer, a tablet, or the like. The control unit 21 can be composed of a CPU, a ROM, a RAM, and the like.

The communication unit 22 has a function of communicating with the client terminal device 10 and the video examination device 50 via the network 1, and can transmit and receive required information.

The storage unit 23 is composed of a hard disk, a flash memory, or the like, and includes information about an advertising agency including an advertiser, video data submitted by the advertising agency, and data related to processing performed by the media company terminal device 20 (for example,). Data of the processing result), data received from the video examination device 50, and the like can be stored.

The display panel 24 can be composed of a liquid crystal panel, an organic EL (Electro Luminescence) display, or the like.

The display processing unit 25 controls the display of the required information on the display panel 24. The display processing unit 25 can display, for example, pass / fail information on whether or not the video submitted for each advertising agency passes a predetermined examination condition. In this case, the video examination device 50 automatically performs a determination process of whether or not the predetermined examination condition is passed. The details of the display processing of the display processing unit 25 will be described later.

The operation unit 26 is composed of, for example, a hardware keyboard, a mouse, or the like, and can operate an icon or the like displayed on the display panel 24, input characters, or the like. The operation unit 26 may be configured with a touch panel.

FIG. 3 is a block diagram showing an example of the configuration of the video examination device 50. The video examination device 50 includes a control unit 51, a communication unit 52, a memory 53, a format determination unit 54, an examination DB 55, a format conversion unit 56, and a processing unit 57 that control the entire device. The processing unit 57 includes a learning model 58, a learning processing unit 59, a teacher data generation unit 60, an examination result output unit 61, and a language processing unit 62. The video examination device 50 can be composed of one or a plurality of servers. The control unit 51 can be composed of a CPU, ROM, RAM, and the like.

The communication unit 52 has a function of communicating with the client terminal device 10 and the media company terminal device 20 via the network 1, and can transmit and receive required information. Specifically, the communication unit 52 can receive the video (video data) for advertisement transmitted by the media company terminal device 20. Further, the communication unit 52 can transmit the pass / fail information of the examination of the video for advertisement to the terminal device 20 of the media company.

The memory 53 can store the video received by the communication unit 52 in frame units. The video includes images, characters, sounds, and the like. When the frame rate is, for example, 30 fps, the video contains 30 frames (images) per second.

The format determination unit 54 determines whether or not the format of the video received by the communication unit 52 matches a predetermined format, and outputs the determination result. The predetermined format is a predetermined format for distributing the video. The details of the format will be described later.

If the format of the video received by the communication unit 52 does not match the predetermined format, the format conversion unit 56 can convert the video to a predetermined format. The format determination unit 54 and the format conversion unit 56 can also be provided on a moving image conversion server (not shown) other than the video examination device 50.

The examination DB 55 can store information necessary for examination, and is, for example, a database of information such as video examination standards for each media company in a format that can be accessed by the processing unit 57. Further, the examination DB 55 stores dictionary data necessary for natural language processing.

Next, the processing unit 57 will be described. The processing unit 57 includes, for example, a CPU (for example, a multi-processor equipped with a plurality of processor cores), a GPU (Graphics Processing Units), a DSP (Digital Signal Processors), an FPGA (Field-Programmable Gate Arrays), or the like. It can be configured by combining hardware. It can also be combined with a quantum processor.

The language processing unit 62 has a voice recognition processing function, and can convert the voice included in the video into text data (character string). Further, the language processing unit 62 has an OCR (Optical Character Recognition / Reader) function, and can convert characters included in the video into text data. Since voice or characters are usually output or displayed over a plurality of frames, the text data conversion process by the language processing unit 62 is performed not in frame units but in a plurality of frames. The language processing unit 62 can also read a document such as a manuscript.

The language processing unit 62 has a morphological analysis function, and uses dictionary data to extract a word, which is the smallest meaningful unit, from the converted text data. Further, the language processing unit 62 has a word embedding processing function, and can convert the extracted word into a vector having a predetermined number of dimensions (a vector composed of a predetermined number of vector components).

The learning model 58 can use a multi-layered neural network (deep learning), for example, a convolutional neural network can be used, but other machine learning may be used. By learning by the learning processing unit 59, the learning model 58 can automatically determine whether or not the video for advertisement passes a predetermined examination condition, and can output pass / fail information. In the present specification, the learning model 58 includes a pre-learning model, a learning model, and a trained model. The learning model 58 includes, for example, a learning model for examining an image included in a video and a learning model for examining text extracted from the video. Hereinafter, the details of the learning model 58 will be described.

FIG. 4 is a schematic diagram showing a first example of the configuration of the learning model 58. The learning model 58 shown in FIG. 4 is a convolutional neural network for images, in which an input layer 581, a convolutional layer 582, a pooling layer 583, a convolutional layer 584, a pooling layer 585, a fully connected layer 586, and an output layer 587 are included. They are connected in order. The number of convolutional layers, pooling layers, and fully connected layers is for convenience, and is not limited to the number shown in FIG. Also, for convenience, the activation function layer is omitted. An image of an image is input to the input layer 581 in frame units. The learning model 58 can determine whether or not the image passes a predetermined examination condition in a frame unit or a plurality of frame units. That is, the output layer 587 outputs the pass / fail information of the image in frame units.

The output layer 587 may be composed of two output nodes and output the probability of passing a predetermined examination condition and the probability of failing (reliability), or at least one of passing or failing reliability. Is divided into a plurality of divisions (for example, reliability = 100%, 95%, 90%, 85%, 80%, ..., 0%), output nodes are provided for each division, and the probability of each division is set. It may be output.

Further, the output layer 587 may be composed of a plurality of output nodes, each node may determine the reason why the image does not pass a predetermined examination condition, and the probability of each reason may be output. The reason can be, for example, "exposed skin", "sexual expression", "violent expression", etc., but is not limited to these.

FIG. 5 is a schematic diagram showing the processing performed in the convolution layer. The input / output data of the convolution layer is also called a feature map, and the input data of the convolution layer is also called an input feature map, and the output data of the convolution layer is also called an output feature map. The input feature map of the first-stage convolution layer is an input frame-by-frame image. The process performed in the convolution layer (also referred to as "convolution operation") is a filter operation by a convolution filter (also referred to as "filter").

As shown in FIG. 5, the input feature map is 8 × 8 pixels. Further, the size of the filter is set to 3 × 3 pixels. In the convolution operation, the filter window is slid at regular intervals to the input feature map, the filter element is multiplied by the corresponding element of the input feature map, the sum is calculated, and the calculated sum is output as the output feature map. Store in the corresponding pixel of. In the example of FIG. 5, the calculation result of the area S corresponding to the filter F1 of the input feature map is stored in the pixel S1 of the output feature map. Further, the calculation result of the area S corresponding to the filter F2 of the input feature map is stored in the pixel S2 of the output feature map. Similarly, the calculation result of the area S corresponding to the filter F3 of the input feature map is stored in the pixel S3 of the output feature map. By moving the filter pixel by pixel and performing the same operation, the output feature map becomes 6 × 6 pixels in size. Here, by using the three filters F1, F2, and F3, three output feature maps can be obtained.

In the training of the training model 58, as parameters related to the filter, for example, the value of the element of the filter, the number of filters (3 in the example of FIG. 5), the size of the filter (3 × 3 in the example of FIG. 5), and the filter. The movement width of (also referred to as “slide”, 1 pixel in the example of FIG. 5), padding that fills the periphery (edge area) of the input feature map with 0, and the like are optimized. The convolutional layer allows the spatial features of the image to be extracted.

FIG. 6 is a schematic diagram showing a process performed in the pooling layer. The pooling layer performs a process of reducing the size of the two-dimensional feature map output from the convolution layer. Specifically, the process of consolidating the local area of the image into one element is performed. For example, as shown in FIG. 6, in a feature map (output feature map) of 6 × 6 pixels, a 2 × 2 local area (window W) is aggregated into “4” which is the maximum value of each element. ing. Since the slide of the window W is equal to the size of the window W and slides by 2 pixels in the example of FIG. 6, the feature map of 6 × 6 pixels is reduced to 3 × 3 pixels. With the pooling layer, for example, even if the feature portion is slightly deformed or displaced in the image, the feature portion can be extracted by absorbing the difference due to the deformation or displacement.

FIG. 7 is a schematic diagram showing a second example of the configuration of the learning model 58. The learning model 58 can include both the configuration shown in FIG. 4 and the configuration shown in FIG. 7. The learning model 58 shown in FIG. 7 is a convolutional neural network for text, in which an input layer 1581, a convolutional layer 1582, a pooling layer 1583, a fully connected layer 1584, and an output layer 1585 are connected in this order. The number of convolutional layers, pooling layers, and fully connected layers is for convenience, and is not limited to the number shown in FIG. 7. A string of words extracted from the text data is input to the input layer 1581 by the language processing unit 62. In the example of FIG. 7, six words (words 1 to 6) are extracted from a series of voices or character strings and input to the input layer 1581. Each word 1 to 6 is embedded and represented by a vector having a predetermined number of dimensions (for example, k). The number of words is not limited to 6.

In the convolution layer 1582, for example, the convolution operation is performed using six filters F11 to F16. The filters F11 and F12 have a size of 4 × k, and by sliding one by one on the input feature map, a feature map composed of three elements (size of 3 × 1) is output. Further, the filters F13 and F14 have a size of 3 × k, and by sliding one by one on the input feature map, a feature map composed of four elements (size of 4 × 1) is output. do. Similarly, the filters F15 and F16 have a size of 2 × k, and by sliding one by one on the input feature map, a feature map composed of five elements (size of 4 × 1) can be obtained. Output.

The pooling layer 1583 extracts the maximum value element from each element of the output feature map.

The output layer 1585 may be composed of two output nodes and output the probability of passing a predetermined examination condition and the probability of failing (reliability), or at least one of passing or failing reliability. Is divided into a plurality of divisions (for example, reliability = 100%, 95%, 90%, 85%, 80%, ..., 0%), output nodes are provided for each division, and the probability of each division is set. It may be output. The learning model 58 for text is not limited to the configuration shown in FIG. 7.

Further, the output layer 1585 may be composed of a plurality of output nodes, each node may determine the reason why the voice or the character does not pass the predetermined examination condition, and the probability of each reason may be output. The reason can be, for example, the existence of a sentence or voice meaning expressions such as "world's best", "industry's first", and "universal", but is not limited thereto.

A dictionary database that stores keywords such as "world's best", "industry's first", and "universal" is prepared in advance together with the learning model 58 as shown in FIG. 7, and a dictionary is included in the input text data. It is also possible to determine whether or not a predetermined examination condition is passed by determining whether or not a word matching a keyword in the database is included. That is, the dictionary database in which the keywords are stored and the learning model 58 may be combined.

Further, instead of the learning model 58 as shown in FIG. 7, for example, a recurrent neural network (RNN) can be used as the learning model 58 for text. In the recurrent neural network, time series information of a plurality of words can be considered by using the intermediate layer of the previous time together with the input layer of the next time for learning.

The teacher data generation unit 60 has a function as a pass / fail information acquisition unit, and acquires pass / fail information as to whether or not at least one of the images, characters, or sounds included in the acquired video passes a predetermined examination condition. , Generate teacher data. The predetermined examination condition can be, for example, a condition as to whether or not the image, character, or sound included in the video is an appropriate expression in terms of advertising expression. For example, if inappropriate expressions are included, the examination conditions will not be passed (failed). The pass / fail information includes pass / fail judgment results. Further, the pass / fail information includes data corresponding to each output node of the output layer.

The learning processing unit 59 trains the learning model 58 using the teacher data (the acquired video and the pass / fail information for the video). This makes it possible to optimize the value of the filter element, the number of filters, the size of the filter, the movement width of the filter, and the like.

When a video for advertisement is screened by a learning model 58 trained using teacher data, it is possible to obtain a pass / fail judgment as to whether or not the video passes a predetermined screening condition with a probability. For example, if the probability of pass judgment is equal to or higher than a predetermined threshold value, it is possible to omit the manual examination work because the advertisement video does not contain inappropriate expressions, and the examination time of the advertisement video can be reduced. It can be shortened. In addition, if the probability of failure determination is equal to or higher than a predetermined threshold value, it is possible to concentrate on the video for confirmation work, assuming that the video for advertisement contains inappropriate expressions, and the video for advertisement can be performed. The examination time can be shortened.

The learning processing unit 59 can train the learning model 58 for each media company by using the teacher data for each media company (the video acquired for each media company and the pass / fail information for the video).

The media company is a management company that manages the distribution of advertisements, and it is necessary to examine the video for advertisement submitted from various advertising agencies (advertisers). On the other hand, the screening criteria for whether or not the image, sound, or text contained in the video for advertising is an appropriate expression may differ from media company to media company. For example, the screening criteria for a certain media company are strict. , The screening criteria of another media company may be relatively loose.

Since the learning processing unit 59 trains the learning model 58 using the teacher data of each media company, the algorithms and parameters (for example, parameters related to the filter) included in the learning model 58 are specialized for each media company. The learning model 58 optimized by can be obtained. As a result, even if there are variations or differences in the screening criteria of the media companies, the learning model 58 suitable for each media company can be used, and the accuracy of the screening of the video for advertisement can be improved. In particular, it is possible to provide an optimal learning model 58 to a media company that distributes advertisements to all public places including mobile bodies such as transportation stations.

Further, the learning processing unit 59 can train the learning model 58 so as to output the reliability of the determination result when the learning model 58 determines that the acquired video does not pass the examination condition.

As a result, when the video for advertisement is examined using the learning model 58, when the reliability (probability) of the rejection determination is low, it is unlikely that the video for advertisement contains inappropriate expressions. As a result, manual screening work can be omitted, and the screening time for advertising video can be shortened. In addition, when the reliability of the rejection judgment is high, it is highly possible that the video for advertisement contains inappropriate expressions, and the confirmation work can be concentrated on the video for advertisement. The examination time can be shortened.

Further, the learning processing unit 59 can train the learning model 58 so that the learning model 58 outputs identification information for identifying a frame that does not pass the examination condition among the plurality of frames constituting the acquired video. ..

If the length and frame rate of the video are preset and the frame that does not pass the examination conditions (may be multiple consecutive frames) is known, for example, an inappropriate expression for the frame several seconds after the start of the video. It is easy to know if there is any.

Further, the learning processing unit 59 trains the learning model 58 so that the learning model 58 outputs image information or character information in the frames that do not pass the examination condition among the plurality of frames constituting the acquired video. Can be done.

For example, when an inappropriate expression is drawn on an image in a video, a thumbnail (image information) of the image can be output. Further, when the voice or character in the video contains an inappropriate expression, the text (character information) of the voice or character can be output. This makes it possible to grasp inappropriate expressions in the video.

Next, a case where the video for advertisement is automatically examined by using the learning model 58 learned by the learning processing unit 59 will be described.

FIG. 8 is a schematic diagram showing an example of the upload screen 201 displayed on the display panel 24 of the media company terminal device 20. The media company terminal device 20 can upload (transmit) the video submitted from a plurality of advertising agencies to the video review device 50. On the upload screen 201, an ID, an agency name, an advertiser name, an advertisement content, a submitted material (video), and an icon 202 for executing upload are displayed. The reviewer of the media company selects the submitted material to be automatically reviewed by the video review device 50 from the list of submitted materials submitted by the advertising agency and sends it to the video review device 50. Can be done.

The processing unit 57 (specifically, the learning model 58) of the video examination device 50 outputs pass / fail information as to whether or not the acquired video passes the examination conditions. Specifically, the learning model 58 can output the probability of pass / fail determination as to whether or not the acquired video passes a predetermined examination condition. The video examination device 50 transmits pass / fail information to the media company terminal device 20 via the communication unit 52.

As a result, the examiner of the media company can omit the manual examination work if the probability of the pass judgment is equal to or higher than a predetermined threshold value, assuming that the image does not contain inappropriate expressions. The examination time can be shortened. In addition, if the probability of failure judgment is equal to or higher than a predetermined threshold value, it is possible to concentrate on the video and perform confirmation work on the assumption that the video contains inappropriate expressions, shortening the video examination time. can do.

Further, when the video is acquired for each of a plurality of advertising agencies, the processing unit 57 outputs a list of pass / fail information as to whether or not the video acquired for each of the plurality of advertising agencies passes the examination condition. The video examination device 50 transmits a list of pass / fail information to the media company terminal device 20 via the communication unit 52.

This makes it possible to distinguish between videos that require manual confirmation work and videos that do not require manual confirmation work among multiple videos acquired by multiple advertising agencies, shortening the screening time for advertising videos. can do.

Further, when the processing unit 57 determines that the acquired video does not pass the examination condition, the processing unit 57 can output the reliability of the determination result.

FIG. 9 is a schematic diagram showing an example of the examination result list screen 210 displayed on the display panel 24 of the media company terminal device 20. For example, on the examination result list screen 210, the ID, the agency name, the submitted material (video), the examination result, the reliability, the icon 211 for displaying the examination result detail screen, and the examination completion notification are displayed on the advertising agency. An icon 212 or the like for transmission to the client terminal device 10 is displayed.

That is, the display processing unit 25 has pass / fail information on whether or not the video submitted for each of the plurality of advertising agencies passes the predetermined examination conditions (in the example of FIG. 9, the examination result is OK and its reliability. And the examination result is NG and its reliability) can be displayed. This makes it possible to distinguish between videos that require manual confirmation work and videos that do not require manual confirmation work among multiple videos acquired by multiple advertising agencies, shortening the screening time for advertising videos. can do.

Further, when it is determined that the video does not pass the examination condition, the display processing unit 25 may display the reliability of the determination result for each of a plurality of advertising agencies. As a result, the media company's reviewer omits the manual review work because if the reliability (probability) of the rejection judgment is low, it is unlikely that the video contains inappropriate expressions. It is possible to shorten the examination time of the video. In addition, if the reliability of the rejection judgment is high, it is highly possible that the video contains inappropriate expressions, and the confirmation work can be concentrated on the video, which reduces the examination time of the video. It can be shortened.

In the example of FIG. 9, if the person in charge of examination judges that the examination result of the submitted material (D0001) whose ID is M1 is OK and the reliability of the examination result is 90%, the examination is completed. By operating the notification icon 212, the examination completion notification is transmitted to the client terminal device 10 of the advertising agency AAA.

That is, the icon 212 has a function as a notification unit, and when it is determined that the acquired video passes the screening conditions, the media company terminal device 20 completes the screening for the advertising agency that submitted the video ( You may pass the examination). Here, the person in charge of examination can appropriately determine to what extent the reliability of the examination result OK is to notify the advertising agency of the completion of the examination.

Further, a predetermined threshold value (for example, 90%) is set in advance, and when the learning model 58 outputs an examination pass whose pass determination probability is equal to or higher than the predetermined threshold value, the examination completion (examination pass) notification is automatically performed. May be transmitted to the client terminal device 10.

With the above configuration, it is possible to notify the advertising agency of the passing of the examination without manually confirming the content of the video, so that the work efficiency of the person in charge of examination can be improved.

Next, detailed information on the examination results will be described.

FIG. 10 is a schematic diagram showing an example of the examination result detail screen 220 displayed on the display panel 24 of the media company terminal device 20. In the example of FIG. 10, the case where the submitted material is D0005, the examination result is NG, and the reliability of the examination result is 90% is illustrated. That is, in the examination result detail screen 220 of FIG. 10, in the example of FIG. 9, the icon 211 for displaying the examination result of the submitted material D0005 of EEE by the advertising agency with the ID of M4 is operated. Is displayed by.

When the processing unit 57 of the video screening device 50 determines that the submitted video (submitted material: D0005) does not pass the screening conditions, which part of the video has an inappropriate advertising expression is determined. , Can be displayed in the image area 221 and the voice / character area 222. In the image area 221, the frames from the beginning to the end of the video are displayed in order, and the frames (which may be a plurality of frames) 223 and 224 having improper drawing are displayed in an identifiable manner. The display mode of the examination result differs between the frame 223 and the frame 224 according to the reliability of NG (for convenience, the patterns are different in the figure). For example, a frame whose examination result has a high reliability of NG is displayed darkly, or a color that stands out (for example, red) and a frame whose examination result has a low reliability of NG is displayed lightly, or a color that is slightly conspicuous. (For example, yellow).

The examination result output unit 61 has a function as a second output unit, and when it is determined that the acquired video does not pass the examination conditions, identification information, image information, or character information is determined according to the reliability of the determination result. Can be output in different display modes. For example, the higher the reliability, the more the identification information, the image information, or the character information can be output in a display mode that is easy to identify. As a result, the priority of the manual confirmation work can be known, and the work efficiency can be improved.

Further, when the person in charge of examination performs a predetermined operation (for example, click, double-click, etc.) on the frame 223, the thumbnail (image information) of the frame 223 can be displayed. Further, the display mode of the thumbnail may be different depending on the reliability of the examination result. For example, reliable thumbnails may be framed and highlighted, or may be given a prominent color. As a result, the examiner can actually confirm the inappropriate drawing on the image. Although the thumbnail is not shown, it may be displayed on a pop-up screen near the frame, or a thumbnail area may be provided in advance in a predetermined area in the examination result detail screen 220.

When the examiner moves the slide 227 to the position of the frame 223, the position of the frame 223 (for example, the elapsed time from the start time of the video: 2 minutes and 40 seconds in the example of the figure) is displayed on the reason explanation screen 229. The reliability of the examination result NG regarding the frame 223 (90% in the example of the figure), the reason why the examination result is NG (exposed skin in the example of the figure), and the like are displayed. When the examiner moves the slide 227 to the position of the frame 224, the reason explanation screen 229 displays the same information about the frame 224.

In the audio / character area 222, frames from the beginning to the end of the video are displayed in order, and frames with inappropriate audio or characters (which may be a plurality of frames) 225 and 226 are identifiablely displayed. The display mode of the examination result differs between the frame 225 and the frame 226 according to the reliability of NG (for convenience, the patterns are different in the figure). For example, a frame whose examination result has a high reliability of NG is displayed darkly, or a color that stands out (for example, red) and a frame whose examination result has a low reliability of NG is displayed lightly, or a color that is slightly conspicuous. (For example, yellow).

When the examiner moves the slide 228 to the position of the frame 226, the position of the frame 226 (for example, the elapsed time from the start time of the image: 3 minutes and 20 seconds in the example of the figure) is displayed on the reason explanation screen 230. The reliability of the examination result NG regarding the frame 226 (90% in the example of the figure), the reason why the examination result is NG (in the example of the figure, the expression of "the best in the world"), etc. are displayed. When the examiner moves the slide 228 to the position of the frame 225, the reason explanation screen 230 displays the same information about the frame 225. Further, the display mode of the character information such as "the best in the world" may be different depending on the reliability of the examination result. For example, highly reliable character information may be highlighted with a frame or given a prominent color.

If the reliability of the examination result NG for each frame is different, the average reliability of each frame may be used as the reliability of the examination result NG of the entire submitted material, and the reliability of each frame is the highest. The one with a low degree may be regarded as the reliability of the examination result NG of the entire submitted material.

As described above, the processing unit 57 can output the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired video. As shown in FIG. 10, the identification information may be colored or patterned on the frame, the frame of the frame may be highlighted, or a mark may be attached to the upper part or the lower part of the frame. Since the length and frame rate of the video are set in advance, if you know the frame that does not pass the examination conditions (may be multiple consecutive frames), for example, it is inappropriate to express the frame several seconds after the start of the video. It is easy to know if there is any.

The operation unit 26 of the media company terminal device 20 has a function as a reception unit, and the operation unit 26 receives a predetermined operation. The predetermined operation is, for example, when a list of pass / fail information on whether or not the submitted video passes the predetermined examination conditions is displayed, in order to confirm the pass / fail content in more detail in the list. It can be an operation, and as illustrated in FIG. 9, it can be an operation for the icon 211 for displaying the detail screen.

When the display processing unit 25 accepts a predetermined operation, the display processing unit 25 can display identification information for identifying a frame that does not pass the examination condition among a plurality of frames constituting the video. Since the length and frame rate of the video are set in advance, if you know the frame that does not pass the examination conditions (may be multiple consecutive frames), for example, it is inappropriate to express the frame several seconds after the start of the video. It is easy to know if there is any. This makes it possible to easily confirm frames containing inappropriate expressions that do not pass the examination.

Further, the processing unit 57 can output the image information in the frame that does not pass the examination condition. For example, if an image in a video has an inappropriate expression, a thumbnail of the image can be output. This makes it possible to grasp inappropriate expressions in the video.

Further, the processing unit 57 can output the character information in the frame that does not pass the examination condition. For example, if the audio or text in the video contains inappropriate expressions, the text of the audio or text can be output. This makes it possible to grasp inappropriate expressions in the video.

Further, the processing unit 57 can output information indicating the reason why the frame determined not to pass the examination condition does not pass the examination condition. The reason for not passing the examination condition can be, for example, if the image contains an inappropriate drawing in terms of advertising expression, it can be a wording indicating what kind of drawing it is, and it can be in voice or text. If an inappropriate word is included in, it can be the word.

The display processing unit 25 can display image information (thumbnails) in a frame that does not pass the examination conditions. Further, the display processing unit 25 can display the character information in the frame that does not pass the examination condition. In addition, the display processing unit 25 can display information indicating the reason why the frame does not pass the examination condition. As a result, the person in charge of examination can immediately understand the reason for not passing the examination conditions, so that, for example, it is not necessary to examine different examination criteria for each media company, the examination work can be simplified, and the examination time can be shortened.

FIG. 11 is a schematic diagram showing another example of the examination result list screen 240 displayed on the display panel 24 of the media company terminal device 20. The difference of the examination result list screen 210 illustrated in FIG. 9 is that the columns for format determination and format conversion are displayed. The format determination unit 54 determines whether or not the video for advertisement is in a predetermined format.

FIG. 12 is an explanatory diagram showing an example of a check target for determining the format of a video. The video format check target (format check item) can be divided into a moving image parameter, an audio parameter, and a still image parameter. Video parameters include, for example, file size, video width and height, video length, bit rate, frame rate, and necessity of conversion from VBR (Variable Bit Rate) to CBR (Constant Bit Rate). Check, and if these parameters are not in the specified format, convert to the specified format. For audio parameters, for example, check the sampling frequency, the necessity of conversion from stereo to monaural, the bit rate and the quantization bit that determines the sound quality, and if these parameters are not in the specified format, Convert to a given format. Further, as the parameters for the still image, for example, the file size, the width and height of the still image, and the necessity of conversion from CMYK to RGB are checked, and if these parameters are not in a predetermined format, they are predetermined. Convert to the format of. The format for distributing the video is defined as a predetermined format, but the format of the video produced by the advertiser may not always match the format for distribution.

When it is determined that the format of the acquired video does not match the predetermined format, the format conversion unit 56 converts the format of the acquired video into the predetermined format. Specifically, the format conversion can be performed by operating the format conversion icon of the submitted material (D0002 in the example of FIG. 11) whose format determination is NG.

The processing unit 57 has a function as a first output unit, and outputs a list of determination results as to whether or not the video acquired for each of the plurality of advertising agencies is in a predetermined format. As a result, it is possible to distinguish between the video that requires the format correction and the video that does not require the format correction among the plurality of videos acquired for each of the plurality of advertising agencies, and it is possible to improve the work efficiency.

In the example of FIG. 9 or FIG. 11, the learning model 58 of the video examination device 50 outputs that the reliability of the examination result NG is 80% with respect to the submitted material D0002 whose ID is M2. However, when the person in charge of examination confirms the details of the submitted material D0002, it may be determined that the inappropriate advertising expression is not included in light of the examination criteria of the media company. Similarly, even though the learning model 58 of the video examination device 50 outputs that the reliability of the examination result OK is 80% for the submitted material D0004 whose ID is M4, the examiner is in charge of the examination. When the details of the submitted material D0004 are confirmed, it may be determined that an inappropriate advertising expression is included in light of the examination criteria of the media company. In such a case, the learning model 58 can be retrained.

That is, the learning processing unit 59 can relearn the learning model 58 using the acquired video and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model 58 as teacher data. For example, if the video determined by the learning model 58 to pass the examination contains inappropriate expressions, or if the video determined by the learning model 58 to fail the examination does not contain inappropriate expressions, the learning model The pass / fail information output by 58 can be corrected so that the learning model 58 can be trained again. Thereby, the determination accuracy of the learning model 58 can be further improved.

FIG. 13 is a flowchart showing an example of the processing procedure of the processing unit 57 in the learning mode. The learning mode is a mode in which the learning model 58 is trained by the learning processing unit 59. Hereinafter, the subject of processing will be described as the processing unit 57 for convenience. The processing unit 57 sets the learning model 58 for the media company (S11). The processing unit 57 acquires the video data for training (S12), acquires the pass / fail information of the video data for training (S13), and generates the teacher data (S14).

The processing unit 57 learns and updates the learning model (S15), and determines the presence or absence of data (training data) from another media company (S16). If there is data from another media company (YES in S16), the processing unit 57 continues the processing after step S11, and if there is no data from another media company (NO in S16), the processing unit 57 ends the processing.

With the above configuration, the learning model optimized for each media company can be stored, and even if the screening criteria differ for each media company, the pass / fail of the screening conditions can be determined using the learning model suitable for the media company. Judgment can be made.

Although not shown, test data (video data) is prepared separately from training data (teacher data), test data is input to the trained learning model 58, and the learning model 58 is evaluated. be able to.

FIG. 14 is a flowchart showing an example of the processing procedure of the processing unit 57 in the examination mode. The examination mode is a mode for determining whether or not the video for advertisement passes a predetermined examination condition. The processing unit 57 acquires the video data for each advertising agency (S21) and determines the format of the video data (S22).

The processing unit 57 inputs the video data into the trained learning model 58 (S23), and stores the pass / fail information of the video data (S24). If the video data does not pass the examination conditions, the pass / fail information includes the rejected frame, the image information in the frame, the character information, and the like. The processing unit 57 outputs the examination result of the video data for each advertising agency (S25), and transmits the examination result and the format determination result to the media company terminal device 20 (S26).

The processing unit 57 determines whether or not to relearn the learning model 58 (S27), and when relearning (YES in S27), the processing unit 57 relearns the learning model 58 using the video data and pass / fail information as teacher data. (S28), the process is terminated. When no re-learning is performed (NO in S27), the processing unit 57 ends the processing.

FIG. 15 is a flowchart showing an example of the processing procedure of the control unit 21 of the media company terminal device 20. The control unit 21 receives the pass / fail information of the video data for each of the plurality of advertising agencies from the video screening device 50 (S31), and displays the screening result list screen (S32). The control unit 21 determines whether or not the examination completion notification icon is operated (S33), and when the examination completion notification icon is operated (YES in S33), the control unit 21 transmits the examination completion notification to the client terminal device 10 (S34). , The process of step S35 described later is performed.

If there is no operation of the examination completion notification icon (NO in S33), the control unit 21 determines whether or not the detail screen icon is operated (S35), and if there is an operation of the detail screen icon (YES in S35), the examination is performed. The result details screen is displayed (S36). The control unit 21 determines whether or not there is a frame selection operation of the examination result NG (S37).

When there is a frame selection operation of the examination result NG (YES in S37), the control unit 21 displays the reason explanation screen (S38) and performs the process of step S39 described later. When there is no operation of the detail screen icon (NO in S35) or when there is no frame selection operation of the examination result NG (NO in S37), the control unit 21 determines whether or not to end the process (S39). ).

If the process is not terminated (NO in S39), the control unit 21 continues the process after step S33, and if the process is terminated (YES in S39), the process is terminated.

The control unit 51, the processing unit 57, the format determination unit 54, and the format conversion unit 56 of the present embodiment can also be realized by using a general-purpose computer including a CPU (processor), a GPU, a RAM (memory), and the like. That is, as shown in FIGS. 13 and 14, a computer program that defines the procedure for each process is loaded into a RAM (memory) provided in the computer, and the computer program is executed by the CPU (processor) on the computer. The control unit 51, the processing unit 57, the format determination unit 54, and the format conversion unit 56 can be realized. The computer program may be recorded and distributed on a recording medium. The learning model 58 trained by the video examination device 50 and the computer program based on the learning model 58 may be distributed and installed in the media company terminal device 20 via the network 1.

In the above-described embodiment, when the video examination device 50 only examines the video and does not learn the learning model 58, the learning processing unit 59 and the teacher data generation unit 60 may not be provided. Further, the format determination unit 54 and the format conversion unit 56 are not essential configurations and may not be provided.

The video examination device according to the present embodiment is a video examination device that examines an image for advertisement, and is a video acquisition unit that acquires an image for advertisement and an image included in the image acquired by the image acquisition unit. It is provided with a pass / fail information acquisition unit that acquires pass / fail information on whether or not at least one of characters or audio passes a predetermined examination condition, and a learning processing unit that learns a learning model using the video and pass / fail information as teacher data.

The computer program according to the present embodiment has a process of acquiring video for advertisement on a computer, and whether or not at least one of images, characters, or sounds contained in the acquired video passes a predetermined examination condition. A process of acquiring information and a process of learning a learning model using the video and pass / fail information as teacher data are executed.

The video examination method according to the present embodiment is a video examination method for examining a video for advertisement, in which at least one of an image, a character, or an audio contained in the acquired video is predetermined by acquiring the video for advertisement. The pass / fail information as to whether or not the examination condition is passed is acquired, and the learning model is trained using the video and the pass / fail information as teacher data.

The video acquisition unit acquires video for advertising. The video is, for example, a moving image, and is composed of a plurality of frames (also referred to as frame images).

The pass / fail information acquisition unit acquires pass / fail information as to whether or not at least one of the images, characters, or sounds included in the acquired video passes the predetermined examination conditions. The predetermined examination condition can be, for example, a condition as to whether or not the image, character, or sound included in the video is an appropriate expression in terms of advertising expression. For example, if inappropriate expressions are included, the examination conditions will not be passed (failed). Pass / fail information includes pass and fail.

The learning processing unit trains the learning model using the acquired video and pass / fail information as teacher data. As the learning model, a multi-layered neural network (deep learning) can be used, for example, a convolutional neural network can be used, but other machine learning may be used.

When a video for advertisement is screened by a trained learning model trained using teacher data, it is possible to obtain a pass / fail judgment as to whether or not the video passes a predetermined screening condition. For example, if the probability of pass judgment is equal to or higher than a predetermined threshold value, it is possible to omit the manual examination work because the advertisement video does not contain inappropriate expressions, and the examination time of the advertisement video can be reduced. It can be shortened. In addition, if the probability of failure determination is equal to or higher than a predetermined threshold value, it is possible to concentrate on the video for confirmation work, assuming that the video for advertisement contains inappropriate expressions, and the video for advertisement can be performed. The examination time can be shortened.

In the video examination device according to the present embodiment, the video acquisition unit acquires the video for advertisement for each media company, and the pass / fail information acquisition unit acquires the pass / fail information for each media company, and the learning. The processing unit trains the learning model for each media company.

The video acquisition department acquires video for advertising for each media company. The media company is a management company that manages the distribution of advertisements, and examines videos for advertisements submitted by various advertising agencies (advertisers).

The pass / fail information acquisition department acquires pass / fail information for each media company. That is, the pass / fail information acquisition unit acquires pass / fail information as to whether or not at least one of the images, characters, or sounds contained in the video for advertisement acquired by each media company passes the predetermined examination conditions. .. The screening criteria for inappropriate expressions may differ from media company to media company. For example, the screening criteria of one media company may be strict, but the screening criteria of another media company may be relatively loose. be.

The learning processing unit trains the learning model using the teacher data of each media company. It is possible to obtain a learning model in which the algorithms and parameters included in the learning model are optimized for each media company. As a result, even if there are variations or differences in the screening criteria of the media companies, it is possible to use a learning model suitable for each media company, and it is possible to improve the accuracy of the screening of the video for advertisement.

In the video examination device according to the present embodiment, the learning model outputs the reliability of the determination result when it is determined that the acquired image does not pass the examination condition.

The learning model outputs the reliability of the judgment result when it is judged that the acquired video does not pass the examination condition. For example, if the reliability (probability) of the rejection judgment is low, it is unlikely that the video for advertisement contains inappropriate expressions, and the manual examination work can be omitted. It is possible to shorten the examination time of the video. In addition, if the reliability of the rejection judgment is high, it is highly possible that the video for advertisement contains inappropriate expressions, and the confirmation work can be concentrated on the video for advertising. It is possible to shorten the examination time of the video.

In the video examination device according to the present embodiment, the learning model outputs identification information for identifying a frame that does not pass the examination condition among a plurality of frames constituting the acquired video.

The learning model outputs identification information that identifies a frame that does not pass the examination condition among a plurality of frames constituting the acquired video. If the length and frame rate of the video are preset and the frame that does not pass the examination conditions (may be multiple consecutive frames) is known, for example, an inappropriate expression for the frame several seconds after the start of the video. It is easy to know if there is any.

In the video examination device according to the present embodiment, the learning model outputs image information or character information in a frame that does not pass the examination condition among a plurality of frames constituting the acquired image.

The learning model outputs image information or character information in a plurality of frames constituting the acquired video that do not pass the examination conditions. For example, if an image in a video has an inappropriate expression, a thumbnail of the image can be output. Further, if the voice or character in the video contains an inappropriate expression, the text of the voice or character can be output. This makes it possible to grasp inappropriate expressions in the video.

The video examination device according to the present embodiment is a video examination device that examines a video for advertisement, and includes a video acquisition unit that acquires the video for advertisement and at least one of an image, a character, or an audio contained in the video. The learning model includes a learning model learned based on pass / fail information on whether or not a predetermined examination condition is passed, and the learning model includes pass / fail information on whether or not the video acquired by the video acquisition unit passes the examination condition. Is output.

The computer program according to the present embodiment has a process of acquiring a video for advertisement on a computer and pass / fail information on whether or not at least one of images, characters, or sounds contained in the video passes a predetermined examination condition. Using the learning model learned based on the above, a process of outputting pass / fail information as to whether or not the acquired video passes the examination condition is executed.

The video examination method according to the present embodiment is a video examination method for examining a video for advertisement, and at least one of an image, a character, or an audio contained in the video is a predetermined examination condition for acquiring the video for advertisement. Using the learning model learned based on the pass / fail information on whether or not the image passes, the pass / fail information on whether or not the acquired video passes the examination conditions is output.

The video acquisition unit acquires video for advertising. The video is, for example, a moving image, and is composed of a plurality of frames (also referred to as frame images).

The learning model is learned based on pass / fail information as to whether or not at least one of the images, characters, or sounds contained in the video passes the predetermined examination conditions. The predetermined examination condition can be, for example, a condition as to whether or not the image, character, or sound included in the video is an appropriate expression in terms of advertising expression. For example, if inappropriate expressions are included, the examination conditions will not be passed (failed). Pass / fail information includes pass and fail.

The learning model outputs pass / fail information as to whether or not the video acquired by the video acquisition unit passes the examination conditions. The learning model can obtain a pass / fail judgment as to whether or not the video passes a predetermined examination condition with a probability. For example, if the probability of pass judgment is equal to or higher than a predetermined threshold value, it is possible to omit the manual examination work because the advertisement video does not contain inappropriate expressions, and the examination time of the advertisement video can be reduced. It can be shortened. In addition, if the probability of failure determination is equal to or higher than a predetermined threshold value, it is possible to concentrate on the video for confirmation work, assuming that the video for advertisement contains inappropriate expressions, and the video for advertisement can be performed. The examination time can be shortened.

In the video examination device according to the present embodiment, the video acquisition unit acquires video for advertisement for each of a plurality of advertising agencies, and in the learning model, the video acquired for each of the plurality of advertising agencies is said. Outputs a list of pass / fail information as to whether or not the examination conditions are passed.

The video acquisition unit acquires video for advertising for each of a plurality of advertising agencies.

The learning model outputs a list of pass / fail information as to whether or not the video acquired for each of a plurality of advertising agencies passes the examination conditions. This makes it possible to distinguish between videos that require manual confirmation work and videos that do not require manual confirmation work among multiple videos acquired by multiple advertising agencies, shortening the screening time for advertising videos. can do.

The video examination device according to the present embodiment has a determination unit for determining whether or not the video for advertisement is in a predetermined format, and whether or not the video acquired for each of a plurality of advertising agencies is in the predetermined format. It is provided with a first output unit that outputs a list of determination results.

The determination unit determines whether or not the video for advertisement is in a predetermined format. The format is file information for constituting video (video and still image), for example, file name (file format), compression method, video bit rate, frame rate, resolution, audio, audio bit rate, frequency, and so on. Includes information such as channel, video length, etc. The format for distributing the video is defined as a predetermined format, but the format of the video produced by the advertiser may not always match the format for distribution.

The first output unit outputs a list of determination results as to whether or not the video acquired for each of the plurality of advertising agencies is in a predetermined format. As a result, it is possible to distinguish between the video that requires the format correction and the video that does not require the format correction among the plurality of videos acquired for each of the plurality of advertising agencies, and it is possible to improve the work efficiency.

In the video examination device according to the present embodiment, the learning model outputs the reliability of the determination result when it is determined that the acquired image does not pass the examination conditions.

The learning model outputs the reliability of the judgment result when it is judged that the acquired video does not pass the examination condition. For example, if the reliability (probability) of the rejection judgment is low, it is unlikely that the video for advertisement contains inappropriate expressions, and the manual examination work can be omitted. It is possible to shorten the examination time of the video. In addition, if the reliability of the rejection judgment is high, it is highly possible that the video for advertisement contains inappropriate expressions, and the confirmation work can be concentrated on the video for advertising. It is possible to shorten the examination time of the video.

In the video examination device according to the present embodiment, the learning model outputs identification information for identifying a frame that does not pass the examination condition among a plurality of frames constituting the acquired video.

The learning model outputs identification information that identifies a frame that does not pass the examination condition among a plurality of frames constituting the acquired video. Since the length and frame rate of the video are set in advance, if you know the frame that does not pass the examination conditions (may be multiple consecutive frames), for example, it is inappropriate to express the frame several seconds after the start of the video. It is easy to know if there is any.

In the video examination device according to the present embodiment, the learning model outputs image information in a frame that does not pass the examination conditions.

The learning model outputs the image information in the frame that does not pass the examination condition. For example, if an image in a video has an inappropriate expression, a thumbnail of the image can be output. This makes it possible to grasp inappropriate expressions in the video.

In the video examination device according to the present embodiment, the learning model outputs character information in a frame that does not pass the examination conditions.

The learning model outputs the character information in the frame that does not pass the examination condition. For example, if the audio or text in the video contains inappropriate expressions, the text of the audio or text can be output. This makes it possible to grasp inappropriate expressions in the video.

When it is determined that the acquired video does not pass the examination condition, the image examination device according to the present embodiment displays the identification information, the image information, or the character information in different display modes depending on the reliability of the determination result. A second output unit for outputting is provided.

When it is determined that the acquired video does not pass the examination condition, the second output unit outputs the identification information, the image information, or the character information in different display modes according to the reliability of the determination result. For example, the higher the reliability, the more the identification information, the image information, or the character information can be output in a display mode that is easy to identify. As a result, the priority of the manual confirmation work can be known, and the work efficiency can be improved.

The video examination device according to the present embodiment includes a learning processing unit that relearns the learning model using the video acquired by the video acquisition unit and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data. ..

The learning processing unit relearns the learning model using the video acquired by the video acquisition unit and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data. For example, if the video that the learning model determines to pass the examination contains inappropriate expressions, or if the video that the learning model determines to fail the examination does not contain inappropriate expressions, the learning model outputs. It is possible to correct the pass / fail information and train the learning model again. This makes it possible to further improve the determination accuracy of the learning model.

The information processing device according to the present embodiment is an information processing device provided with a display screen, and is a list of pass / fail information on whether or not the video submitted for each of a plurality of advertising agencies passes a predetermined examination condition. It is provided with a display processing unit for displaying.

The display processing unit displays a list of pass / fail information as to whether or not the video submitted for each of a plurality of advertising agencies passes a predetermined examination condition. This makes it possible to distinguish between videos that require manual confirmation work and videos that do not require manual confirmation work among multiple videos acquired by multiple advertising agencies, shortening the screening time for advertising videos. can do.

In the information processing apparatus according to the present embodiment, when it is determined that the video does not pass the examination condition, the display processing unit displays the reliability of the determination result for each of the plurality of advertising agencies.

When it is determined that the video does not pass the examination condition, the display processing unit displays the reliability of the determination result for each of a plurality of advertising agencies. For example, if the reliability (probability) of the rejection judgment is low, it is unlikely that the video for advertisement contains inappropriate expressions, and the manual examination work can be omitted. It is possible to shorten the examination time of the video. In addition, if the reliability of the rejection judgment is high, it is highly possible that the video for advertisement contains inappropriate expressions, and the confirmation work can be concentrated on the video for advertising. It is possible to shorten the examination time of the video.

The information processing apparatus according to the present embodiment includes a reception unit that accepts a predetermined operation, and when the display processing unit accepts the operation, the information processing unit does not pass the examination condition among the plurality of frames constituting the video. Display identification information that identifies the frame.

The reception unit accepts a predetermined operation. The predetermined operation is, for example, when a list of pass / fail information on whether or not the submitted video passes the predetermined examination conditions is displayed, in order to confirm the pass / fail content in more detail in the list. It can be an operation.

When the operation is accepted, the display processing unit displays identification information for identifying a frame that does not pass the examination condition among a plurality of frames constituting the video. Since the length and frame rate of the video are set in advance, if you know the frame that does not pass the examination conditions (may be multiple consecutive frames), for example, it is inappropriate to express the frame several seconds after the start of the video. It is easy to know if there is any. This makes it possible to easily confirm frames containing inappropriate expressions that do not pass the examination.

In the information processing apparatus according to the present embodiment, the display processing unit displays information indicating the reason why the frame identified by the identification information does not pass the examination condition.

The display processing unit displays information indicating the reason why the frame identified by the identification information does not pass the examination condition. The reason can be, for example, when the image contains an inappropriate drawing in terms of advertising expression, the wording can indicate what kind of drawing it is, and the inappropriate word in the voice or characters. If is included, it can be the word.

The information processing apparatus according to the present embodiment includes a notification unit that notifies the advertising agency that submitted the video of the screening pass when it is determined that the acquired video passes the screening conditions.

If it is determined that the acquired video passes the screening conditions, the notification unit notifies the advertising agency that submitted the video of the screening pass. For example, if the learning model outputs a screening pass with a probability of passing judgment equal to or higher than a predetermined threshold, it is possible to notify the advertising agency of the screening pass without manually checking the content of the video. Efficiency can be improved.

1 Network 10 Client terminal device 20 Media company terminal device 21 Control unit 22 Communication unit 23 Storage unit 24 Display panel 25 Display processing unit 26 Operation unit 50 Video examination device 51 Control unit 52 Communication unit 53 Memory 54 Format judgment unit 55 Examination DB
56 Format conversion unit 57 Processing unit 58 Learning model 59 Learning processing unit 60 Teacher data generation unit 61 Examination result output unit 62 Language processing unit

Claims (22)

It is a video screening device that screens video for advertising.
The video acquisition department that acquires video for advertising, and
A pass / fail information acquisition unit that acquires pass / fail information on whether or not at least one of images, characters, or sounds included in the video acquired by the video acquisition unit passes a predetermined examination condition.
With a learning processing unit that trains a learning model using the video and pass / fail information as teacher data,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said With an output unit that outputs image information or character information in a frame that does not pass the examination conditions in different display modes
A video screening device equipped with. It is a video screening device that screens video for advertising.
The video acquisition department that acquires video for advertising, and
A pass / fail information acquisition unit that acquires pass / fail information on whether or not at least one of the images, characters, or sounds included in the video acquired by the video acquisition unit passes a predetermined examination condition.
With a learning processing unit that trains a learning model using the video and pass / fail information as teacher data
Equipped with,
The learning processing unit
The learning model is retrained using the video acquired by the video acquisition unit and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.
Video review device. The video acquisition unit
Acquire video for advertising for each media company,
The pass / fail information acquisition unit
Obtain the pass / fail information for each media company,
The learning processing unit
The video examination device according to claim 1 or 2 , wherein the learning model is trained for each media company. The learning model is
The video examination device according to claim 2, wherein when it is determined that the acquired video does not pass the examination conditions, the reliability of the determination result is output. The learning model is
The video examination device according to claim 2 or 4 , which outputs identification information for identifying a frame that does not pass the examination conditions among a plurality of frames constituting the acquired video. The learning model is
The video examination device according to claim 2, claim 4 or claim 5 , which outputs image information or character information in a frame that does not pass the examination conditions among a plurality of frames constituting the acquired image. .. It is a video screening device that screens video for advertising.
The video acquisition department that acquires video for advertising, and
With a learning model learned based on pass / fail information on whether at least one of the images, characters, or sounds contained in the video passes the prescribed examination conditions.
Equipped with
The learning model is
Outputs pass / fail information as to whether or not the video acquired by the video acquisition unit passes the examination conditions.death,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said Further provided with an output unit that outputs image information or character information in a frame that does not pass the examination conditions in different display modes.
Video review device. It is a video screening device that screens video for advertising.
The video acquisition department that acquires video for advertising, and
With a learning model learned based on pass / fail information on whether at least one of the images, characters, or sounds contained in the video passes the prescribed examination conditions.
Equipped with
The learning model is
Outputs pass / fail information as to whether or not the video acquired by the video acquisition unit passes the examination conditions.death,
Further provided is a learning processing unit for re-learning the learning model using the video acquired by the video acquisition unit and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.
Video review device. The video acquisition unit
Get the video for the advertisement of each of multiple advertising agencies,
The learning model is
The video examination device according to claim 7 or 8 , which outputs a list of pass / fail information as to whether or not the video acquired for each of the plurality of advertising agencies passes the examination conditions. A determination unit that determines whether or not the video for advertisement is in a predetermined format, and
The video according to any one of claims 7 to 9, further comprising a list output unit that outputs a list of determination results of whether or not the video acquired for each of a plurality of advertising agencies is in the predetermined format. Examination device. The learning model is
The video examination device according to claim 8 , wherein when it is determined that the acquired video does not pass the examination conditions, the reliability of the determination result is output. The learning model is
The video examination device according to claim 8 or 11 , which outputs identification information for identifying a frame that does not pass the examination conditions among a plurality of frames constituting the acquired video. The learning model is
The video examination device according to claim 12 , which outputs image information in a frame that does not pass the examination conditions. The learning model is
The video examination device according to claim 12 or 13 , which outputs character information in a frame that does not pass the examination conditions. On the computer
The process of acquiring video for advertising and
The process of acquiring pass / fail information on whether or not at least one of the images, characters, or sounds contained in the acquired video passes the predetermined examination conditions.
With the process of learning the learning model using the video and pass / fail information as teacher data.,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said Processing to output image information or text information in a frame that does not pass the examination conditions in different display modes
A computer program that runs a computer program. On the computer
The process of acquiring video for advertising and
The process of acquiring pass / fail information on whether or not at least one of the images, characters, or sounds contained in the acquired video passes the predetermined examination conditions.
With the process of learning the learning model using the video and pass / fail information as teacher data.,
Processing to relearn the learning model using the acquired video and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.When
A computer program that runs. On the computer
The process of acquiring video for advertisement and
Whether or not the acquired video passes the screening conditions using the learning model learned based on the pass / fail information on whether or not at least one of the images, characters, or sounds contained in the video passes the predetermined screening conditions. And the process of outputting pass / fail information,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said Processing to output image information or text information in a frame that does not pass the examination conditions in different display modes
A computer program that runs. On the computer
The process of acquiring video for advertising and
Whether or not the acquired video passes the screening conditions using the learning model learned based on the pass / fail information on whether or not at least one of the images, characters, or sounds contained in the video passes the predetermined screening conditions. And the process of outputting pass / fail information,
A process of retraining the learning model using the acquired video and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.
A computer program that runs a computer program. It is a video screening method that screens video for advertising.
Video for advertisingThe video acquisition departmentAcquired,
Pass / fail information on whether or not at least one of the images, characters, or sounds contained in the acquired video passes the prescribed examination conditions.The pass / fail information acquisition departmentAcquired,
A learning model using the above video and pass / fail information as teacher dataThe learning processing departmentLet me learn,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said The output unit outputs the image information or character information in the frame that does not pass the examination conditions in different display modes.
Video screening method. It is a video screening method that screens video for advertising.
Video for advertisingThe video acquisition departmentAcquired,
Pass / fail information on whether or not at least one of the images, characters, or sounds contained in the acquired video passes the prescribed examination conditions.The pass / fail information acquisition departmentAcquired,
A learning model using the above video and pass / fail information as teacher dataThe learning processing departmentLet me learn,
The learning processing unit relearns the learning model using the acquired video and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.
Video screening method. It is a video screening method that screens video for advertising.
Video for advertisingThe video acquisition departmentAcquired,
Whether or not the acquired video passes the examination conditions using a learning model learned based on the pass / fail information of whether or not at least one of the images, characters, or sounds included in the video passes the predetermined examination conditions. Pass / fail informationThe learning modeloutputdeath,
When it is determined that the acquired video does not pass the examination condition, the identification information for identifying the frame that does not pass the examination condition among the plurality of frames constituting the acquired image according to the reliability of the determination result, the said The output unit outputs the image information or character information in the frame that does not pass the examination conditions in different display modes.
Video screening method. It is a video screening method that screens video for advertising.
Video for advertisingThe video acquisition departmentAcquired,
Whether or not the acquired video passes the examination conditions using a learning model learned based on the pass / fail information of whether or not at least one of the images, characters, or sounds included in the video passes the predetermined examination conditions. Pass / fail informationThe learning modeloutputdeath,
The learning processing unit relearns the learning model using the acquired video and the modified pass / fail information obtained by modifying the pass / fail information output by the learning model as teacher data.
Video screening method.

Images