Examination device, examination method, and program

Abstract

To enable appropriate examination in each examination range even in the presence of examination ranges of a plurality of different examination types in an examination target.SOLUTION: An examination device provided herein is configured to detect examination type switching points in examination target data based at least on either of content information and attribute information of the examination target data, detect examination ranges based on the switching points in the examination target data, and preform examination processing on each examination range.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] Commercial footage broadcast by broadcasting stations and advertisements placed on web pages are subject to various inspections, such as whether the text contained in commercial footage or advertising images is appropriate.

[0003] A technology for conducting such an inspection (screening) of advertisements is disclosed in Patent Document 1. Patent Document 1 discloses a technology for extracting a plurality of video components and determining the suitability of each of the plurality of video components as an advertisement based on a learning model. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2023-124762 Summary of the Invention [Problem to be solved by the invention]

[0005] Even when the subject of the examination had multiple examination scopes for different types of examination, it was required to conduct examinations for each examination scope appropriately.

[0006] An object of the present disclosure is to provide an examination device, an examination method, and a program that solve the above-mentioned problems. [Means for solving the problem]

[0007] An examination device according to one aspect of the present disclosure includes a switching detection means for detecting a switching point in the examination type in the examination target data based on at least one of content information and attribute information of the examination target data, an examination range detection means for detecting an examination range based on the switching point in the examination target data, and an examination processing means for performing examination processing for each examination range.

[0008] In one embodiment of the present disclosure, an examination device has an examination method that detects a change in examination type in the data to be examined based on at least one of content information and attribute information of the data to be examined, detects an examination range based on the change in the data to be examined, and performs examination processing for each examination range.

[0009] A program according to one aspect of the present disclosure causes a computer to function as a switching detection means for detecting switching points in the test type in the test data based on at least one of the content information and attribute information of the test data, a test range detection means for detecting the test range based on the switching points in the test data, and a test processing means for performing test processing for each test range. [Effects of the Invention]

[0010] According to the above aspect, even if the test subject has test ranges of multiple different test types, it is possible to appropriately test each test range. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 2 is a functional block diagram of an examination device according to the present disclosure. [Figure 2] FIG. 2 is a hardware configuration diagram of an examination device according to the present disclosure. [Figure 3] FIG. 10 is a first diagram illustrating an example of processing by a switching detection unit according to the present disclosure. [Figure 4] FIG. 10 is a second diagram illustrating an example of processing by the switching detection unit according to the present disclosure. [Figure 5]A diagram showing the processing flow of the examination device according to the present disclosure. [Figure 6] A diagram showing other functional blocks of the examination device according to the present disclosure. [Figure 7] A diagram showing another processing flow of the examination device according to the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0012] The examination device, examination method, and program of the present disclosure will be described below with reference to the drawings.

[0013] FIG. 1 is a functional block diagram of the examination device. The inspection device 1 is a computer that performs inspection processing based on the data to be inspected and outputs the inspection results. The computer of the inspection device 1 executes an inspection program. As a result, the inspection device 1 performs the functions of a control unit 11, an acquisition unit 12, a content analysis unit 13, a meta information generation unit 14, a switch detection unit 15, an inspection range detection unit 16, and an inspection processing unit 17. Note that inspection refers to determining, for example, whether the information contained in the data to be inspected complies with legal regulations, whether it complies with the self-imposed restrictions of industry associations, and whether it complies with the self-imposed restrictions of each business that handles the data to be inspected. This determines whether the integrity of the data to be inspected is guaranteed.

[0014] The control unit 11 controls each functional unit. The acquisition unit 12 acquires the examination target data and attribute information linked to the examination target data. The content analysis unit 13 acquires predetermined content information contained in the data to be examined. The meta information generating unit 14 generates meta information that can identify the type of examination content indicated by the examination target data using the content information and attribute information. The change detection unit 15 detects a change point of the examination type in the examination target data based on at least one of the content information and attribute information of the examination target data. The examination range detection unit 16 detects the examination range based on the change point of the examination type in the examination target data. The examination processing unit 17 performs examination processing for each examination range.

[0015] Figure 2 is a hardware configuration diagram of the examination device. 2, the examination device 1 is a computer equipped with various hardware components such as a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a database 104, and a communication module 105. The examination device 1 may be configured as a single computer, or may be configured as a single examination device consisting of multiple computers connected to each other for communication.

[0016] FIG. 3 is a first diagram illustrating an example of processing by the switching detection unit. When the data to be examined is an advertising video, the transition detection unit 15 of the examination device 1 detects the playback time of a scene transition in the advertising video. For example, when a scene (31) with a forest background switches to a scene (32) of a specific beverage, which is the advertised product, the transition detection unit 15 detects the playback time of the scene transition. The playback time indicates the time elapsed from the start of the advertising video. The transition detection unit 15 may be a processing unit that detects the point in time when the examination type changes in the data to be examined based on at least one of the content information and attribute information of the data to be examined. In other words, the transition detection unit 15 is a processing unit that detects the point in time when the examination type changes in the advertising video, which is the data to be examined. In the example of Figure 3, the transition detection unit 15 detects the time when a scene of a forest, which is not subject to examination and does not particularly display the product name, switches to a scene that displays the product name of the beverage, which is the subject of examination.

[0017] FIG. 4 is a second diagram illustrating an example of processing by the switching detection unit. When the data to be examined is a storyboard, the transition detection unit 15 of the examination device 1 detects the position (coordinates) within the image of the transition between frames in the storyboard. For example, the transition detection unit 15 determines that the data is not eligible for examination based on the first picture (41) and the second picture (42) and the corresponding first picture description (410) and second picture description (420). Also, the transition detection unit 15 determines that the data is eligible for examination based on the third picture (43) and the fourth picture (44) and the corresponding third picture description (430) and fourth picture description (440). In this case, the change detection unit 15 performs processing to detect the position (coordinates) in the image of the storyboard 40 at the boundary between the range including the second picture (42) and the description of the second picture (420) and the range including the third picture (43) and the description of the third picture (430) as the position indicating the change point. Note that in this case too, the change detection unit 15 detects the change point of the test type in the test target data based on at least one of the content information and attribute information of the test target data. In other words, the change detection unit 15 is a processing unit that detects the position (coordinates) in the image that is the change point of the test type in the storyboard, which is the test target data.

[0018] Here, when the data to be examined is a video, the change detection unit 15 may detect the time indicating the change point of the examination type in the video based on at least one of the content information and the attribute information.

[0019] Furthermore, when the data to be examined is a storyboard, the transition detection unit 15 may detect the coordinates between multiple image diagrams in the storyboard that indicate the transition positions of the examination types in the storyboard based on at least one of the content information and the attribute information.

[0020] The test data may be images other than storyboards. The transition detection unit 15 may detect coordinates between multiple images within an image that indicate the transition position of the test type in the image based on at least one of the content information and the attribute information. A storyboard may be an example of an image. Although the test data according to the present disclosure will be described as an advertising video or storyboard, the test data may be other data. For example, the test data may be data such as a script storyboard, a poster, a bulletin board, or text.

[0021] FIG. 5 is a diagram showing the processing flow of the examination device. Next, the processing flow of the examination device will be explained step by step. First, the examination device 1 receives the data to be examined (step S101). The examination device 1 may receive the data to be examined from another device, or may receive the data to be examined from an input interface based on user operation. The data to be examined may have attribute information attached to it. In this case, the examination device 1 receives the attribute information (step S102). The attribute information may be included in the data to be examined. The examination device 1 may not receive the attribute information. In that case, the examination device 1 may generate the attribute information from the information included in the data to be examined. The control unit 11 outputs the data to be examined and the attribute information to the acquisition unit 12. The acquisition unit 12 outputs the data to be examined to the content analysis unit 13. When the acquisition unit 12 acquires the attribute information, it outputs the attribute information to the meta information generation unit 14.

[0022] The content analysis unit 13 detects the type of data to be examined. The content analysis unit 13 may detect the type based on the extension of the data to be examined, or may detect the type of data to be examined based on the type input via an input interface based on user operation. The content analysis unit 13 may acquire attribute information and detect the type from the attribute information.

[0023] The content analysis unit 13 analyzes the content of the data to be examined. The content analysis unit 13 outputs content information, which is the result of analyzing the content of the data to be examined (step S103). The content analysis unit 13 may input the data to be examined into a neural network using a content analysis model generated by prior machine learning, acquire content information contained in the data to be examined, and output the content information. The content analysis model may be one of multiple content analysis models that can be identified depending on the type of data to be examined. The content analysis unit 13 may input attribute information from the acquisition unit 12, input the data to be examined and the attribute information into a neural network using a content analysis model identified depending on the data to be examined and the attribute information, acquire content information contained in the data to be examined, and output the content information. Note that, as described above, when the examination device 1 generates attribute information based on the data to be examined, the content analysis unit 13 may generate the attribute information.

[0024] The content information may be information related to the subject of examination that is identified based on the character information of the audio or text included in the subject of examination data. For example, if the subject of examination data is an advertisement video, the content information may include the name of the product advertised in the advertisement video, the name of the advertising company that placed the advertisement, the conversation of people appearing in the advertisement video, and other sentences that appear in the video.

[0025] For example, if the data to be examined is a storyboard, the content information may include the name of the product being advertised as displayed in the storyboard image, the name of the advertising agency placing the advertisement, the content of the conversations between people in the image description, and text displayed in other images or in the image description.

[0026] The attribute information also includes information other than content information related to the advertising video or storyboard shown by the data under review, such as the creator of the data under review (name of the creating company), the type of advertising target, and the genre of the product under review.

[0027] The content analysis unit 13 outputs the content information to the meta information generation unit 14. When the content analysis unit 13 generates attribute information, it outputs the attribute information to the meta information generation unit 14. The meta information generation unit 14 generates meta information using the content information and attribute information. Meta information is information that classifies detailed information about the examination target contained in the content information and attribute information. In other words, the content analysis unit 13 classifies the examination target data based on the meta information and attribute information (step S104).

[0028] The meta information generating unit 14 generates meta information related to the text information based on the text information included in the content information. For example, if the text information includes a product name, the meta information related to the text information may be the product genre. The meta information generating unit 14 generates meta information related to the logo information based on the logo information included in the content information. For example, if the logo information is a company logo, the meta information related to the logo information may be the company's industry. The meta information generating unit 14 generates meta information related to the personal information based on the personal information included in the content information. For example, if the personal information is a person's name, the meta information may be related to the personal information or an occupation (actor, entertainer, etc.) corresponding to the person's name. The meta information generating unit 14 generates meta information related to the object information based on the object information included in the content information. For example, if the object information is an elephant or a giraffe, the meta information related to the object information may be the animal, and if the object information is a product name, the meta information related to the object information may be the product genre. The meta information generating unit 14 generates meta information related to the famous place information based on the famous place information included in the content information. For example, if the famous place information is the address of Tokyo Tower, the meta information related to the famous place information may be information such as Tokyo, which is the name of the prefecture that includes the address of Tokyo Tower. The meta information generating unit 14 generates meta information related to a border line based on the border line included in the content information. For example, when an object to be examined included in an advertising video or storyboard is surrounded by a border line, information such as the type of border line (dotted or solid, a numerical value indicating the thickness, color, presence or absence of a border line) may be generated as meta information related to the border line. The meta information generating unit 14 generates meta information related to the attribute information based on the text included in the attribute information.

[0029] The content information may also include OCR (Optical Character Recognition) information obtained from video images, brightness, audio, subtitles, color, barcodes, two-dimensional codes, etc., and the meta information generation unit 14 may generate meta information by meta-converting this information.

[0030] The meta information generation unit 14 outputs the meta information as classification information to the switching detection unit 15 (step S105). As an example, it is assumed that, based on the analysis results and attribute information obtained by the content analysis unit 13, the information about the advertising video is that it is a "wine advertisement," the title is "N Wine," the production company is "A Company," and the producer is "Yamada Taro." In this case, the meta information generation unit 14 generates "CM," "alcohol," and "main information type for each frame in the video" as meta information and outputs them to the switching detection unit 15.

[0031] The transition detection unit 15 acquires meta information. Based on the meta information, the transition detection unit 15 identifies a transition detection model from one or more transition detection models. The transition detection unit 15 inputs the test target data and the meta information into a neural network that uses the transition detection model, and as a result, detects transition point information that indicates transition points in the test target data (step S106). The transition detection unit 15 outputs the transition point information to the test range detection unit 16 (step S107). As described above, the transition point information indicates the time within the advertising video and the position within the storyboard.

[0032] The test range detection unit 16 acquires switching point information. The test range detection unit 16 detects the test range in the data to be tested based on the switching points (step S108). For example, if the data to be tested is an advertising video, the test range detection unit 16 identifies the range of the video to be tested (from the start time to the end time) based on the time of the switching point. If there are multiple switching points, the test range is the range of the video separated by those times that includes the information of the test target. Furthermore, if the data to be tested is a storyboard, the test range detection unit 16 identifies the range within the image that includes the test target (such as a rectangular range or a circular range) based on the position of the switching point within the storyboard. If there are multiple positions (coordinates) that indicate the switching points, the test range is the range identified based on those positions that includes the information of the test target. The examination range detection unit 16 may detect the examination range using an examination range detection model identified based on the type of examination target data and meta information (or meta information of the examination range). If the examination target has multiple examination ranges, these examination ranges may be examination ranges of different examination types. By performing examination processing of different types for each examination range in this way, examination for each examination range can be carried out appropriately.

[0033] The examination range detection unit 16 links the examination target data, information on the specified examination range, the type of examination target data for which the examination range has been specified, and meta information, and records them in a database (steps S109, S110, and S111). The examination range detection unit 16 outputs the start of examination processing for this recorded information to the examination processing unit 17. This output may include an identifier that identifies the combination of the examination range information, the type of examination target data for which the examination range has been specified, and the meta information.

[0034] The inspection processing unit 17 reads the inspection target data, inspection range information, the type of inspection target data that identifies the inspection range, and meta information, which are recorded in the database linked to the identification information. The inspection processing unit 17 identifies one inspection judgment model from multiple inspection judgment models based on the type (step S112). The inspection processing unit 17 inputs data within the inspection range of the inspection target data into a neural network using the inspection judgment model, and obtains the inspection judgment result. This inspection judgment result includes information indicating NG in the inspection target data indicated by the inspection range. The information indicating NG includes information on characters that are prohibited from being displayed, characters that are prohibited from being broadcast, parts that need to be corrected depending on the advertising company and the details of the corrections, and parts that need to be corrected depending on the type of advertising target and the details of the corrections. The information indicating NG is information that has been determined to be a risky expression during inspection. The examination judgment result may include an identifier indicating the reason for determining the expression as risky in the examination, a specified identifier corresponding to the reason for determining the expression as risky, information regarding the range or position determined to be a risky expression (coordinates within the image and playback time), information specifying the examination range (coordinate range within the image and playback time), etc. A risky expression may be an example of an expression that is determined to be inappropriate or potentially inappropriate in the examination. The examination processing unit 17 outputs the examination judgment result to a specified output destination. The output destination may be a monitor, a memory unit, etc. If an examination range exists for one examination target data, the examination processing unit 17 performs examination processing in the same way for each examination range.

[0035] According to the above process, even when there are multiple test ranges for different test types in test target data such as video, storyboards, and images, the test device 1 can appropriately test each test range.

[0036] Furthermore, according to the above-described processing, the inspection device 1 can shorten the inspection time for each inspection range even when there are multiple inspection ranges of different inspection types in the inspection target data such as video, storyboards, and images, thereby reducing the burden on the inspector.

[0037] The content analysis model may be a machine learning model obtained by machine learning data on a large number of relationships between input and output, for example, when test subject data and attribute data are input and correct content information is output. The content analysis model may be a machine learning model generated by machine learning using the test device 1 or another device.

[0038] The above-mentioned switching detection model may be a machine learning model obtained by machine learning data on a large number of relationships between inputs and outputs, for example, where the test data, content information, and attribute information are input and the correct answer switching points are output. The switching detection model may be a machine learning model generated by machine learning using the test device 1 or another device.

[0039] The above-mentioned test range detection model may be a machine learning model obtained by machine learning data on a large number of input-output relationships, for example, where test target data, content information, attribute information, and test range transition points are input, and the correct test range is output. The test range detection model may be a machine learning model generated by machine learning using the test device 1 or another device. The above-mentioned test judgment model may be a machine learning model obtained by machine learning data on a large number of input-output relationships, for example, where test target data, content information, attribute information, and test range information are input, and the correct answer information including information that should be considered to be risky expression in the test of that test range (such as text, logos, colors, lines, and images), an identifier indicating the reason for determining the expression as risky, and a specified identifier corresponding to the reason for determining the expression as risky. The test judgment model may be a machine learning model generated by machine learning using the test device 1 or another device.

[0040] FIG. 6 is a diagram showing other functional blocks of the examination device. FIG. 7 is a diagram showing another processing flow of the examination device. As shown in FIG. 6, the examination device 1 has at least the functions of a switching detection means 61, an examination range detection means 62, and an examination processing means 63. The detection means 61 detects a change point of the examination type in the examination target data based on at least one of the content information and attribute information of the examination target data (step S701). The examination range detection means 62 detects the examination range based on the switching points in the examination target data (step S702). The examination processing means 63 performs examination processing for each examination range (step S703).

[0041] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

[0042] Some or all of the above-described embodiments can be described as, but are not limited to, the following supplementary notes.

[0043] (Appendix 1) A change detection means for detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; A test range detection means for detecting a test range based on the switching point in the test target data; A testing processing means for performing testing processing for each of the testing ranges; An examination device comprising:

[0044] (Appendix 2) The examination processing means performs examination processing for each examination range using examination rules specified based on the type of the examination range. 1. The testing device described in Appendix 1.

[0045] (Appendix 3) The data to be examined is a video; The change detection means detects a time indicating a change point of the examination type in the video based on at least one of the content information and the attribute information. 1. A testing device as described in Appendix 1 or Appendix 2.

[0046] (Appendix 4) The data to be examined is a storyboard, The change detection means detects coordinates between a plurality of image drawings in the storyboard that indicate a change position of the examination type in the storyboard based on at least one of the content information and the attribute information. 1. A testing device as described in Appendix 1 or Appendix 2.

[0047] (Appendix 5) The test data is an image, The change detection means detects coordinates between a plurality of images in the image that indicate a change position of the examination type in the image based on at least one of the content information and the attribute information. 1. A testing device as described in Appendix 1 or Appendix 2.

[0048] (Appendix 6) The switching detection means detects a switching point of the examination type in the examination target data using a detection model of a switching point according to the examination target data. An examination device according to any one of appendices 1 to 5.

[0049] (Appendix 7) The test range detection means detects the test range using a detection model of the test range. An examination device according to any one of appendices 1 to 6.

[0050] (Appendix 8) Detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; Detecting an examination range based on the switching point in the examination target data; Testing is carried out for each of the above test ranges Examination method.

[0051] (Appendix 9) Performing inspection processing for each inspection range using inspection rules identified based on the type of inspection range The examination method is as described in Appendix 8.

[0052] (Appendix 10) The data to be examined is a video; Detecting a time indicating a change in examination type in the video based on at least one of the content information and the attribute information. The examination method described in Appendix 8 or Appendix 9.

[0053] (Appendix 11) The data to be examined is a storyboard, Detecting coordinates between a plurality of image diagrams in the storyboard that indicate switching positions of examination types in the storyboard based on at least one of the content information and the attribute information. The examination method described in Appendix 8 or Appendix 9.

[0054] (Appendix 12) The test data is an image, Detecting coordinates between a plurality of images in the image that indicate switching positions of examination types in the image based on at least one of the content information and the attribute information. The examination method described in Appendix 8 or Appendix 9.

[0055] (Appendix 13) Detecting a change point of the examination type in the examination target data using a detection model of a change point according to the examination target data. The examination method described in any one of Appendix 8 to Appendix 12.

[0056] (Appendix 14) Detecting the test range using the test range detection model. The examination method described in any one of Appendix 8 to Appendix 13.

[0057] (Appendix 15) Computer, a change detection means for detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; a test range detection means for detecting a test range based on the switching point in the test target data; A test processing means for performing test processing for each test range; A program that functions as a

[0058] (Appendix 16) The examination processing means performs examination processing for each examination range using examination rules specified based on the type of the examination range. 15. The program described in Appendix 15.

[0059] (Appendix 17) The data to be examined is a video; The change detection means detects a time indicating a change point of the examination type in the video based on at least one of the content information and the attribute information. 17. The program according to claim 15 or 16.

[0060] (Appendix 18) The data to be examined is a storyboard, The change detection means detects coordinates between a plurality of image drawings in the storyboard that indicate a change position of the examination type in the storyboard based on at least one of the content information and the attribute information. 17. The program according to claim 15 or 16.

[0061] (Appendix 19) The test data is an image, The change detection means detects coordinates between a plurality of images in the image that indicate a change position of the examination type in the image based on at least one of the content information and the attribute information. 17. The program according to claim 15 or 16.

[0062] (Appendix 20) The switching detection means detects a switching point of the examination type in the examination target data using a detection model of a switching point according to the examination target data. 19. A program according to any one of claims 15 to 19.

[0063] (Appendix 21) The test range detection means detects the test range using a detection model of the test range. 21. A program according to any one of claims 15 to 20. [Explanation of symbols]

[0064] 1. Testing device 11 Control section 12... Acquisition part 13...Content analysis department 14. Meta information generation unit 15 Switching detection unit 16. Test range detection unit 17 Examination Department

Claims

1. A change detection means for detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; A test range detection means for detecting a test range based on the switching point in the test target data; A testing processing means for performing testing processing for each of the testing ranges; An examination device comprising:

2. The examination processing means performs examination processing for each examination range using examination rules specified based on the type of the examination range. The examination device according to claim 1.

3. The data to be examined is a video; The change detection means detects a time indicating a change point of the examination type in the video based on at least one of the content information and the attribute information. The examination device according to claim 2.

4. The data to be examined is a storyboard, The change detection means detects coordinates between a plurality of drawings in the storyboard that indicate a change position of the examination type in the storyboard based on at least one of the content information and the attribute information. The examination device according to claim 2.

5. The test data is an image, The change detection means detects coordinates specifying a range of a plurality of figures in the image that indicate a change position of the examination type in the image based on at least one of the content information and the attribute information. The examination device according to claim 2.

6. The switching detection means detects a switching point of the examination type in the examination target data using a detection model of a switching point according to the examination target data. The examination device according to claim 2.

7. The test range detection means detects the test range using a detection model of the test range. The examination device according to claim 2 .

8. Detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; Detecting an examination range based on the switching point in the examination target data; Testing is carried out for each of the above test ranges Examination method.

9. Computer, a change detection means for detecting a change point of the examination type in the examination target data based on at least one of content information and attribute information of the examination target data; a test range detection means for detecting a test range based on the switching point in the test target data; A test processing means for performing test processing for each test range; A program that functions as a

Images