Television broadcast intelligent advertisement monitoring system

Abstract

The application discloses a television broadcast intelligent advertisement monitoring system and belongs to the technical field of broadcast television monitoring. The system comprises a video analysis module for obtaining change data of a television broadcast video, generating an advertisement possibility evaluation index, a video splitting module for generating an advertisement video according to the advertisement possibility evaluation index, a sample production demand analysis module for obtaining product characteristic values and video lengths of the advertisement video, generating an advertisement sample generation demand value, a production demand judgment module for judging whether sample production is needed for the advertisement video, a violation analysis module for generating an advertisement violation degree evaluation index, and a violation judgment module for judging whether a violation phenomenon occurs in the advertisement video according to the advertisement violation degree evaluation index. The application realizes accurate advertisement splitting through multi-modal feature fusion, dynamically updates a dynamic sample library, and combines an audio, length and text multi-dimensional violation detection model to significantly improve efficiency and reduce artificial dependence in the whole process.

Description

Technical Field

[0001] This invention belongs to the field of broadcast television monitoring technology, and in particular relates to an intelligent advertising monitoring system for television broadcasts. Background Technology

[0002] With intensifying market competition and the booming development of the advertising industry, advertising has become a crucial means of brand communication. Radio and television advertising, as one of the main forms of traditional media, still holds an important position in many industries, especially among certain audience groups. However, with the continuous improvement of advertising regulations in various countries and the increased oversight of the advertising industry by regulatory agencies, the compliance of advertising content has become a key issue in advertising placement. For example, in some countries and regions, advertising needs to comply with specific time and content restrictions, and adhere to specific ethical and social norms. Radio and television advertising testing can effectively help companies ensure the compliance of their advertising content and avoid legal risks arising from illegal advertising.

[0003] Traditional monitoring of broadcast television advertisements mainly relies on manually creating sample databases and reviewing each advertisement individually. However, this method is inefficient, costly, and susceptible to subjective bias. In particular, it lacks the ability to automatically separate different versions of advertisements for the same product played consecutively, resulting in a high rate of missed detections. This makes it difficult to meet the needs of real-time monitoring. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a smart advertising monitoring system for television broadcasting, which solves the aforementioned problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a smart advertising monitoring system for television broadcasting, comprising:

[0006] The video analytics module is used to acquire change data of television broadcast videos and generate an advertising probability assessment index; the change data includes audio intensity change values, scene switching frequency, and subtitle change values.

[0007] The video splitting module is used to split TV broadcast videos based on the advertising probability assessment index and generate advertising videos;

[0008] The sample production requirement analysis module is used to obtain the product feature values ​​and video duration of the advertising video, and generate advertising sample generation requirement values; among them, product features refer to the feature values ​​of products that continuously appear in the advertising video.

[0009] The requirement judgment module is used to generate requirement values ​​based on the advertising sample and determine whether the advertising video needs to be sampled.

[0010] The violation analysis module is used to obtain the audio intensity, playback duration and text content of the advertising video if sample production is required, to establish a violation analysis model and generate an advertising violation assessment index.

[0011] The violation judgment module is used to assess the index based on the degree of ad violation and determine whether the ad video has violated any rules.

[0012] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0013] Further technical solutions: The specific methods for generating the advertising likelihood assessment index include:

[0014] Set a monitoring period, divide the monitoring period into several monitoring time periods, and obtain the audio intensity change value, scene switching frequency and subtitle change value of the TV broadcast video within the monitoring time period; where the audio intensity change value refers to the difference between the audio intensity of the current monitoring time period and the audio intensity of the adjacent monitoring time period; the subtitle change value refers to the difference between the subtitle position of the current monitoring time period and the subtitle position of the adjacent monitoring time period.

[0015] Generate an audio intensity change assessment value based on the audio intensity change value;

[0016] Generate a scene switching frequency evaluation value based on the scene switching frequency;

[0017] Generate a subtitle change assessment value based on the subtitle change value;

[0018] Through the formula:

[0019] Q poss =B TV *α+P TV *β+L TV *γ;

[0020] Ad generation probability assessment index Q poss ;

[0021] In the formula, B TV This represents the evaluation value of audio intensity change, P. TV This represents the evaluation value for scene switching frequency, L. TV This represents the evaluation value of the subtitle changes. α, β, and γ are all weighted proportions, and α+β+γ=1.

[0022] Further technical solution: The specific method for obtaining the subtitle change value is as follows:

[0023] A two-dimensional coordinate system is established with the length of the television broadcast video frame as the X-axis and the width as the Y-axis. The positions of the subtitles in the television broadcast video are then substituted into the two-dimensional coordinate system to generate the subtitle coordinates.

[0024] Obtain the coordinates of all subtitles in a television broadcast video;

[0025] Through the formula:

[0026]

[0027] Generate subtitle offset value ΔL;

[0028] In the formula, X1 represents the X-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, X2 represents the X-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, Y1 represents the Y-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, and Y2 represents the Y-axis coordinate value of the adjacent subtitle coordinate in the timing sequence.

[0029] The subtitle change value is generated based on the subtitle offset value ΔL; where the subtitle change value refers to the average value of all subtitle offset values ​​ΔL.

[0030] Further technical solution: The method for generating the audio intensity change evaluation value is as follows:

[0031] Based on the audio intensity change value and the audio intensity change threshold, an audio intensity change difference is generated; whereby the audio intensity change difference refers to the difference between the audio intensity change value and the audio intensity change threshold.

[0032] An audio intensity change evaluation value is generated based on the audio intensity change difference and the audio intensity change threshold; where the audio intensity change evaluation value refers to the ratio between the audio intensity change difference and the audio intensity change threshold.

[0033] The specific method for generating the scene switching frequency evaluation value is as follows:

[0034] A switching frequency difference is generated based on the scene switching frequency and the scene switching frequency threshold; where the switching frequency difference refers to the difference between the scene switching frequency and the scene switching frequency threshold.

[0035] A scene switching frequency evaluation value is generated based on the switching frequency difference and the scene switching frequency threshold; where the scene switching frequency evaluation value refers to the ratio between the switching frequency difference and the scene switching frequency threshold.

[0036] The specific method for generating the subtitle change evaluation value is as follows:

[0037] A subtitle change deviation value is generated based on the subtitle change value and the subtitle change threshold; the subtitle change deviation value refers to the difference between the subtitle change value and the subtitle change threshold.

[0038] A subtitle change evaluation value is generated based on the subtitle change deviation value and the subtitle change threshold; where the subtitle change evaluation value refers to the ratio between the subtitle change deviation value and the subtitle change threshold.

[0039] Further technical solutions: The specific methods for splitting television broadcast video include:

[0040] The advertising probability assessment index is compared with the advertising probability assessment index threshold to determine whether there is any abnormality in the status of the television broadcast video.

[0041] When the advertising probability assessment index is greater than the advertising probability assessment index threshold, the state of the television broadcast video is determined to be abnormal.

[0042] Identify the time points when all TV broadcast videos show abnormal status and generate advertising videos accordingly.

[0043] The method for cropping the video area is as follows:

[0044] All time points when the status of a television broadcast video becomes abnormal are sequentially marked in chronological order; the marking method is to mark the time point when the status of the television broadcast video becomes abnormal as C. i , i = 1, 2, 3, ..., m, where m represents the total number of time points when the status of all television broadcast videos becomes abnormal;

[0045] All abnormal times in the status of television broadcast videos are divided into odd-numbered times and even-numbered times.

[0046] The video content corresponding to adjacent odd-numbered time nodes and even-numbered time nodes is marked as an advertisement video, and the advertisement video starts from an odd-numbered time node.

[0047] Further technical solution: The method for generating the advertising sample demand value is as follows:

[0048] Obtain product characteristic values ​​and video duration for advertising time;

[0049] Based on the product feature values, a product feature matching evaluation value is generated; whereby the product feature matching evaluation value refers to the ratio between the product feature value and the most recent product feature value in the advertising sample video;

[0050] Through the formula:

[0051]

[0052] The generated video duration matches the evaluation value T. ad ;

[0053] In the formula, T iThis indicates the video duration of the advertisement video, T1, T2, T... x These are the most recent video durations of all the advertising sample videos, where x represents the number of advertising sample videos corresponding to the most recent product feature value, and a1, a2, a... x All are proportionality constants, and a1 + a2 + ... + a x =1;

[0054] Based on the product feature matching assessment value and the video duration matching assessment value, an advertising sample generation demand value is generated; whereby the advertising sample generation demand value refers to the weighted sum of the product feature matching assessment value and the video duration matching assessment value.

[0055] Further technical solution: The specific method for generating the advertising violation assessment index is as follows:

[0056] Obtain the audio intensity, playback time, and text content of the ad video;

[0057] Generate an audio intensity violation rating based on the audio intensity of the advertisement video;

[0058] Generate a playback duration violation evaluation value based on the playback duration of the advertisement video;

[0059] Generate a text content violation evaluation value based on the text content of the advertisement video;

[0060] Based on the violation evaluation values ​​of audio intensity, playback time, and text content, a violation analysis model is established to generate an evaluation index for the degree of advertising violation.

[0061] Further technical solution: The specific method for generating the audio intensity violation evaluation value is as follows:

[0062] An audio intensity warning deviation value is generated based on the audio intensity of the advertisement video and the audio intensity warning value; whereby the audio intensity warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value.

[0063] An audio intensity violation evaluation value is generated based on the audio intensity warning deviation value and the audio intensity warning value; the audio intensity violation evaluation value refers to the ratio between the audio intensity warning deviation value and the audio intensity warning value.

[0064] The specific method for generating the playback duration violation evaluation value is as follows:

[0065] Based on the playback duration of the advertisement video and the playback duration warning value, a playback duration warning deviation value is generated; whereby the playback duration warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value.

[0066] A playback duration violation evaluation value is generated based on the playback duration warning deviation value and the playback duration warning value; the playback duration violation evaluation value refers to the ratio between the playback duration warning deviation value and the playback duration warning value.

[0067] Further technical solution: The specific method for generating the text content violation evaluation value is as follows:

[0068] The text content of advertising videos is categorized into non-violation text content and non-violation text content.

[0069] The system obtains the number of non-compliant text contents and generates a text content violation evaluation value. The text content violation evaluation value refers to the ratio between the number of non-compliant texts and the total number of text contents in the advertising video.

[0070] Further technical solution: The expression of the violation analysis model is specifically as follows:

[0071] K = (1 + W) b )*(1+W h )*(1+W m );

[0072] In the expression, K represents the index for assessing the degree of advertising violation, and W... b This represents the audio intensity violation rating, W. h This represents the violation score for playback time, W. m This represents the evaluation value for violations of text content.

[0073] This invention provides a smart advertising monitoring system for television broadcasting, which has the following advantages compared with the prior art:

[0074] This invention achieves accurate ad segmentation through multimodal feature fusion, updates the dynamic sample library in real time, and combines audio, duration, and text multi-dimensional violation detection models to significantly improve efficiency and reduce reliance on manual labor through full-process automation. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the structure of the intelligent advertising monitoring system for television broadcasting provided in an embodiment of the present invention. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0077] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0078] Please see Figure 1 The present invention provides a smart advertising monitoring system for television broadcasting, comprising:

[0079] The video analytics module is used to acquire change data of television broadcast videos and generate an advertising probability assessment index; the change data includes audio intensity change values, scene switching frequency, and subtitle change values.

[0080] The video splitting module is used to split TV broadcast videos based on the advertising probability assessment index and generate advertising videos;

[0081] The sample production requirement analysis module is used to obtain the product feature values ​​and video duration of the advertising video, and generate advertising sample generation requirement values; among them, product features refer to the feature values ​​of products that continuously appear in the advertising video.

[0082] It should be explained that the product feature value is determined by identifying the features of all objects appearing in the advertising video, and the feature value of the object with the highest frequency of appearance is marked as the product feature value.

[0083] The requirement judgment module is used to generate requirement values ​​based on the advertising sample and determine whether the advertising video needs to be sampled.

[0084] The violation analysis module is used to obtain the audio intensity, playback duration and text content of the advertising video if sample production is required, to establish a violation analysis model and generate an advertising violation assessment index.

[0085] It should be noted that the text content includes the text content in the advertisement video and the text content converted from the audio.

[0086] The violation judgment module is used to assess the index based on the degree of ad violation and determine whether the ad video has violated any rules.

[0087] In a preferred embodiment of the present invention, the method for generating the advertising likelihood assessment index specifically includes:

[0088] Set a monitoring period, divide the monitoring period into several monitoring time periods, and obtain the audio intensity change value, scene switching frequency and subtitle change value of the TV broadcast video within the monitoring time period; where the audio intensity change value refers to the difference between the audio intensity of the current monitoring time period and the audio intensity of the adjacent monitoring time period; the subtitle change value refers to the difference between the subtitle position of the current monitoring time period and the subtitle position of the adjacent monitoring time period.

[0089] It should be explained that adjacent monitoring periods refer to the earlier monitoring periods in the time sequence; for example, if the current monitoring period is the nth moment of a television broadcast video, then the adjacent monitoring period is the nxth moment, where n and x are both positive numbers;

[0090] Generate an audio intensity change assessment value based on the audio intensity change value;

[0091] Generate a scene switching frequency evaluation value based on the scene switching frequency;

[0092] It should be explained that the scene switching frequency refers to the frequency of scene switching in television broadcast video; specifically, the scene switching frequency is the ratio between the number of scene switching times in television broadcast video during the monitoring period and the duration of the monitoring period.

[0093] Generate a subtitle change assessment value based on the subtitle change value;

[0094] An advertising probability assessment index is generated based on the audio intensity change assessment value, the scene switching frequency assessment value, and the subtitle change assessment value.

[0095] For example, through the formula:

[0096] Q poss =B TV *α+P TV *β+L TV *γ;

[0097] Ad generation probability assessment index Q poss ;

[0098] In the formula, B TV This represents the evaluation value of audio intensity change, P. TV This represents the evaluation value for scene switching frequency, L. TV This represents the evaluation value of the subtitle changes, where α, β, and γ are all weighted proportions, and α+β+γ=1;

[0099] It should be noted that the values ​​of α, β, and γ are determined by those skilled in the art, and the methods for determining these values ​​include, but are not limited to, the analytic hierarchy process (AHP).

[0100] In a preferred embodiment of the present invention, the method for obtaining the subtitle change value is as follows:

[0101] A two-dimensional coordinate system is established with the length of the television broadcast video frame as the X-axis and the width as the Y-axis. The positions of the subtitles in the television broadcast video are then substituted into the two-dimensional coordinate system to generate the subtitle coordinates.

[0102] Obtain the coordinates of all subtitles in a television broadcast video;

[0103] Through the formula:

[0104]

[0105] Generate subtitle offset value ΔL;

[0106] In the formula, X1 represents the X-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, X2 represents the X-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, Y1 represents the Y-axis coordinate value of the adjacent subtitle coordinate in the timing sequence, and Y2 represents the Y-axis coordinate value of the adjacent subtitle coordinate in the timing sequence.

[0107] The subtitle change value is generated based on the subtitle offset value ΔL; where the subtitle change value refers to the average value of all subtitle offset values ​​ΔL.

[0108] In a preferred embodiment of the present invention, the method for generating the audio intensity change evaluation value is as follows:

[0109] Based on the audio intensity change value and the audio intensity change threshold, an audio intensity change difference is generated; whereby the audio intensity change difference refers to the difference between the audio intensity change value and the audio intensity change threshold.

[0110] It should be noted that the audio intensity change threshold refers to the maximum value of audio intensity change in a television broadcast; furthermore, audio intensity refers to the sound intensity of a television broadcast video.

[0111] An audio intensity change evaluation value is generated based on the audio intensity change difference and the audio intensity change threshold; where the audio intensity change evaluation value refers to the ratio between the audio intensity change difference and the audio intensity change threshold.

[0112] In a preferred embodiment of the present invention, the method for generating the scene switching frequency evaluation value is as follows:

[0113] A switching frequency difference is generated based on the scene switching frequency and the scene switching frequency threshold; where the switching frequency difference refers to the difference between the scene switching frequency and the scene switching frequency threshold.

[0114] It should be noted that the threshold value for scene switching frequency is set by those skilled in the art.

[0115] A scene switching frequency evaluation value is generated based on the switching frequency difference and the scene switching frequency threshold; where the scene switching frequency evaluation value refers to the ratio between the switching frequency difference and the scene switching frequency threshold.

[0116] In a preferred embodiment of the present invention, the method for generating the subtitle change evaluation value is as follows:

[0117] A subtitle change deviation value is generated based on the subtitle change value and the subtitle change threshold; the subtitle change deviation value refers to the difference between the subtitle change value and the subtitle change threshold.

[0118] A subtitle change evaluation value is generated based on the subtitle change deviation value and the subtitle change threshold; the subtitle change evaluation value refers to the ratio between the subtitle change deviation value and the subtitle change threshold.

[0119] It should be noted that the subtitle change threshold is a set value, which is set by those skilled in the art.

[0120] As a preferred embodiment of the present invention, the method of splitting television broadcast video specifically includes:

[0121] The advertising probability assessment index is compared with the advertising probability assessment index threshold to determine whether there is any abnormality in the status of the television broadcast video.

[0122] When the advertising probability assessment index is less than or equal to the advertising probability assessment index threshold, the state of the television broadcast video is determined to be normal; at this time, the smaller the advertising probability assessment index, the more normal the state of the television broadcast video.

[0123] When the advertising probability assessment index is greater than the advertising probability assessment index threshold, the state of the television broadcast video is determined to be abnormal; at this time, the higher the advertising probability assessment index, the more abnormal the state of the television broadcast video.

[0124] Identify the time points when all TV broadcast videos show abnormal status and generate advertising videos accordingly.

[0125] The method for cropping the video area is as follows:

[0126] All time points when the status of a television broadcast video becomes abnormal are sequentially marked in chronological order; the marking method is to mark the time point when the status of the television broadcast video becomes abnormal as C. i , i = 1, 2, 3, ..., m, where m represents the total number of time points when the status of all television broadcast videos becomes abnormal;

[0127] All abnormal times in the status of television broadcast videos are divided into odd-numbered times and even-numbered times.

[0128] The video content corresponding to adjacent odd-numbered and even-numbered time nodes is marked as an advertisement video, and the starting point of the advertisement video is an odd-numbered time node; for example, if the starting point of the advertisement video is C1, then the ending point of the advertisement video is C2.

[0129] In a preferred embodiment of the present invention, the method for generating the advertising sample demand value is as follows:

[0130] Obtain product characteristic values ​​and video duration for advertising time;

[0131] Based on the product feature values, a product feature matching evaluation value is generated; whereby the product feature matching evaluation value refers to the ratio between the product feature value and the most recent product feature value in the advertising sample video;

[0132] It should be explained that the most recent product feature value of the advertising sample video refers to the product feature value with the smallest difference among all the product feature values ​​of the advertising sample videos; for example, if the product feature value is set as the product edge perimeter, then the most recent product feature value of the advertising sample video is the product edge perimeter with the smallest difference among all the product edge perimeters of the advertising sample videos (i.e., the product feature value in the advertising video).

[0133] Based on the video length, a video length matching evaluation value is generated; where the video length matching evaluation value refers to the ratio between the video length and the length of the most recent video of the advertising sample video;

[0134] It should be noted that the most recent video duration of the advertising sample video refers to the duration of the advertising sample video corresponding to the most recent product feature value.

[0135] In this embodiment, if there are multiple advertising sample videos corresponding to the most recent product feature value of the advertising sample video, the ratio processing is performed on each video, and the processing results are weighted to generate a video duration matching evaluation value.

[0136] For example, through the formula:

[0137]

[0138] The generated video duration matches the evaluation value T. ad ;

[0139] In the formula, T i This indicates the video duration of the advertisement video, T1, T2, T... x These are the most recent video durations of all the advertising sample videos, where x represents the number of advertising sample videos corresponding to the most recent product feature value, and a1, a2, a... x All are proportionality constants, and a1 + a2 + ... + a x =1;

[0140] It needs to be explained that a1, a2, and a x The values ​​are all

[0141] Based on the product feature matching assessment value and the video duration matching assessment value, an advertising sample generation demand value is generated; whereby the advertising sample generation demand value refers to the weighted sum of the product feature matching assessment value and the video duration matching assessment value.

[0142] In a preferred embodiment of the present invention, the method for determining whether the advertising video needs to be sampled is as follows:

[0143] Compare the advertising sample generation demand value with the advertising sample generation demand threshold;

[0144] It should be noted that the threshold for generating advertising samples is a set value, which is determined by those skilled in the art.

[0145] When the advertising sample generation demand value is less than the advertising sample generation demand threshold, it is determined that the advertising video needs to produce a sample; at this time, the smaller the advertising sample generation demand value, the higher the demand for advertising video to produce a sample.

[0146] When the advertising sample generation demand value is greater than or equal to the advertising sample generation demand threshold, it is determined that the advertising video does not need to produce samples. In this case, the larger the advertising sample generation demand value, the lower the demand for advertising video to produce samples.

[0147] In a preferred embodiment of the present invention, the method for generating the advertising violation assessment index is as follows:

[0148] Obtain the audio intensity, playback time, and text content of the ad video;

[0149] Generate an audio intensity violation rating based on the audio intensity of the advertisement video;

[0150] Generate a playback duration violation evaluation value based on the playback duration of the advertisement video;

[0151] Generate a text content violation evaluation value based on the text content of the advertisement video;

[0152] Based on the violation evaluation values ​​of audio intensity, playback time, and text content, a violation analysis model is established to generate an evaluation index for the degree of advertising violation.

[0153] In a preferred embodiment of the present invention, the method for generating the audio intensity violation evaluation value is as follows:

[0154] An audio intensity warning deviation value is generated based on the audio intensity of the advertisement video and the audio intensity warning value; whereby the audio intensity warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value.

[0155] It should be noted that the audio intensity warning value refers to the maximum audio intensity of a standard advertising video;

[0156] An audio intensity violation evaluation value is generated based on the audio intensity warning deviation value and the audio intensity warning value; the audio intensity violation evaluation value refers to the ratio between the audio intensity warning deviation value and the audio intensity warning value.

[0157] In a preferred embodiment of the present invention, the method for generating the playback duration violation evaluation value is as follows:

[0158] Based on the playback duration of the advertisement video and the playback duration warning value, a playback duration warning deviation value is generated; whereby the playback duration warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value.

[0159] It should be noted that the playback duration warning value refers to the maximum playback duration of a standard advertising video;

[0160] A playback duration violation evaluation value is generated based on the playback duration warning deviation value and the playback duration warning value; the playback duration violation evaluation value refers to the ratio between the playback duration warning deviation value and the playback duration warning value.

[0161] In a preferred embodiment of the present invention, the method for generating the text content violation evaluation value is as follows:

[0162] The text content of advertising videos is categorized into non-violation text content and non-violation text content.

[0163] The system obtains the number of non-compliant text contents and generates a text content violation evaluation value. The text content violation evaluation value refers to the ratio between the number of non-compliant texts and the total number of text contents in the advertising video.

[0164] In a preferred embodiment of the present invention, the expression of the violation analysis model is specifically as follows:

[0165] K = (1 + W) b )*(1+W h )*(1+W m );

[0166] In the expression, K represents the index for assessing the degree of advertising violation, and W... b This represents the audio intensity violation rating, W. h This represents the violation score for playback time, W. m This represents the evaluation value for violations of text content.

[0167] In a preferred embodiment of the present invention, the method for determining whether an advertising video violates regulations is as follows:

[0168] Compare the advertising violation assessment index with the advertising violation assessment index threshold;

[0169] It should be noted that the threshold for assessing the degree of advertising violation is a set value, which is determined by relevant personnel in this field.

[0170] When the ad violation assessment index is less than or equal to the ad violation assessment index threshold, the ad video is determined not to have violated any rules. In this case, the smaller the ad violation assessment index, the lower the likelihood that the ad video has violated any rules.

[0171] When the advertising violation assessment index is greater than the advertising violation assessment index threshold, the advertising video is judged to have violated regulations. At this time, the higher the advertising violation assessment index, the higher the probability that the advertising video has violated regulations.

[0172] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart advertising monitoring system for television broadcasting, characterized in that, The system includes: The video analytics module is used to acquire change data of television broadcast videos and generate an advertising probability assessment index; the change data includes audio intensity change values, scene switching frequency, and subtitle change values. The video splitting module is used to split TV broadcast videos based on the advertising probability assessment index and generate advertising videos; The sample production requirement analysis module is used to obtain the product feature values ​​and video duration of the advertising video, and generate advertising sample generation requirement values; among them, product features refer to the feature values ​​of products that continuously appear in the advertising video. The requirement judgment module is used to generate requirement values ​​based on the advertising sample and determine whether the advertising video needs to be sampled. The violation analysis module is used to obtain the audio intensity, playback duration and text content of the advertising video if sample production is required, to establish a violation analysis model and generate an advertising violation assessment index. The violation judgment module is used to evaluate the index based on the degree of ad violation and determine whether the ad video has violated any rules. The method for generating the required values ​​for the advertising sample is as follows: Obtain the product feature values ​​and video duration of the advertising video; Based on the product feature values, a product feature matching evaluation value is generated; whereby the product feature matching evaluation value refers to the ratio between the product feature value and the most recent product feature value in the advertising sample video; Through the formula: ； The generated video duration matches the evaluation value. ; In the formula, This indicates the video length of the advertisement video. , , , where represents the most recent video duration of all ad sample videos, and x represents the number of ad sample videos corresponding to the most recent product feature value. , , All are proportionality coefficients, and + +....+ =1; The most recent product feature value of the advertising sample video refers to the product feature value with the smallest difference between the product feature values ​​of all advertising sample videos and the product feature value of the advertising video; The most recent video duration of the advertising sample video refers to the duration of the advertising sample video corresponding to the most recent product feature value of the advertising sample video. Based on the product feature matching evaluation value and the video duration matching evaluation value, an ad sample generation demand value is generated; whereby the ad sample generation demand value refers to the weighted sum of the product feature matching evaluation value and the video duration matching evaluation value. The specific methods for splitting television broadcast video include: The advertising probability assessment index is compared with the advertising probability assessment index threshold to determine whether there is any abnormality in the status of the television broadcast video. When the advertising probability assessment index is greater than the advertising probability assessment index threshold, the state of the television broadcast video is determined to be abnormal. Identify the time points when all TV broadcast videos show abnormal status and generate advertising videos accordingly. The method for cropping the video area is as follows: All time points when the status of a television broadcast video becomes abnormal are sequentially marked in chronological order; wherein, the marking method is to mark the time points when the status of a television broadcast video becomes abnormal as Ci, i=1,2,3,....m, where m represents the total number of time points when the status of all television broadcast videos becomes abnormal; All abnormal times in the status of television broadcast videos are divided into odd-numbered times and even-numbered times. The video content corresponding to adjacent odd-numbered time nodes and even-numbered time nodes is marked as an advertisement video, and the advertisement video starts from an odd-numbered time node.

2. The intelligent advertising monitoring system for television broadcasting according to claim 1, characterized in that, The specific methods for generating the advertising likelihood assessment index include: Set a monitoring period, divide the monitoring period into several monitoring time periods, and obtain the audio intensity change value, scene switching frequency and subtitle change value of the TV broadcast video within the monitoring time period; where the audio intensity change value refers to the difference between the audio intensity of the current monitoring time period and the audio intensity of the adjacent monitoring time period; the subtitle change value refers to the difference between the subtitle position of the current monitoring time period and the subtitle position of the adjacent monitoring time period. Generate an audio intensity change assessment value based on the audio intensity change value; Generate a scene switching frequency evaluation value based on the scene switching frequency; Generate a subtitle change assessment value based on the subtitle change value; Through the formula: ； Ad generation probability assessment index ; In the formula, This represents the evaluation value of audio intensity change. This represents the evaluation value for the frequency of scene switching. This represents the evaluation value of the subtitle changes. α, β, and γ are all weighted proportions, and α+β+γ=1.

3. The intelligent advertising monitoring system for television broadcasting according to claim 2, characterized in that, The specific method for obtaining the subtitle change value is as follows: A two-dimensional coordinate system is established with the length of the television broadcast video frame as the X-axis and the width as the Y-axis. The positions of the subtitles in the television broadcast video are then substituted into the two-dimensional coordinate system to generate the subtitle coordinates. Obtain the coordinates of all subtitles in a television broadcast video; Through the formula: ; Generate subtitle offset value ; In the formula, This represents the X-axis coordinate value of the preceding subtitle in the time sequence. This represents the X-axis coordinate value of the subtitle that is sequentially adjacent to the next subtitle. This represents the Y-axis coordinate value of the preceding subtitle in the time sequence. This represents the Y-axis coordinate value of the subtitle that is sequentially adjacent to the next subtitle; Based on subtitle offset value Generate subtitle change values; where subtitle change values ​​refer to all subtitle offset values. The average value.

4. The intelligent advertising monitoring system for television broadcasting according to claim 2, characterized in that, The specific method for generating the audio intensity change evaluation value is as follows: Based on the audio intensity change value and the audio intensity change threshold, an audio intensity change difference is generated; whereby the audio intensity change difference refers to the difference between the audio intensity change value and the audio intensity change threshold. An audio intensity change evaluation value is generated based on the audio intensity change difference and the audio intensity change threshold; where the audio intensity change evaluation value refers to the ratio between the audio intensity change difference and the audio intensity change threshold. The specific method for generating the scene switching frequency evaluation value is as follows: A switching frequency difference is generated based on the scene switching frequency and the scene switching frequency threshold; where the switching frequency difference refers to the difference between the scene switching frequency and the scene switching frequency threshold. A scene switching frequency evaluation value is generated based on the switching frequency difference and the scene switching frequency threshold; where the scene switching frequency evaluation value refers to the ratio between the switching frequency difference and the scene switching frequency threshold. The specific method for generating the subtitle change evaluation value is as follows: A subtitle change deviation value is generated based on the subtitle change value and the subtitle change threshold; the subtitle change deviation value refers to the difference between the subtitle change value and the subtitle change threshold. A subtitle change evaluation value is generated based on the subtitle change deviation value and the subtitle change threshold; where the subtitle change evaluation value refers to the ratio between the subtitle change deviation value and the subtitle change threshold.

5. The intelligent advertising monitoring system for television broadcasting according to claim 1, characterized in that, The specific method for generating the advertising violation assessment index is as follows: Obtain the audio intensity, playback time, and text content of the ad video; Generate an audio intensity violation rating based on the audio intensity of the advertisement video; Generate a violation score based on the playback duration of the advertisement video; Generate a text content violation evaluation value based on the text content of the advertisement video; Based on the violation evaluation values ​​of audio intensity, playback time, and text content, a violation analysis model is established to generate an evaluation index for the degree of advertising violation.

6. The intelligent advertising monitoring system for television broadcasting according to claim 5, characterized in that, The specific method for generating the audio intensity violation evaluation value is as follows: An audio intensity warning deviation value is generated based on the audio intensity of the advertisement video and the audio intensity warning value; whereby the audio intensity warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value. An audio intensity violation evaluation value is generated based on the audio intensity warning deviation value and the audio intensity warning value; the audio intensity violation evaluation value refers to the ratio between the audio intensity warning deviation value and the audio intensity warning value. The specific method for generating the playback duration violation evaluation value is as follows: Based on the playback duration of the advertisement video and the playback duration warning value, a playback duration warning deviation value is generated; whereby the playback duration warning deviation value refers to the difference between the audio intensity of the advertisement video and the audio intensity warning value. A playback duration violation evaluation value is generated based on the playback duration warning deviation value and the playback duration warning value; the playback duration violation evaluation value refers to the ratio between the playback duration warning deviation value and the playback duration warning value.

7. The intelligent advertising monitoring system for television broadcasting according to claim 5, characterized in that, The specific method for generating the text content violation evaluation value is as follows: The text content of advertising videos is categorized into non-violation text content and non-violation text content. The system obtains the number of non-compliant text contents and generates a text content violation evaluation value. The text content violation evaluation value refers to the ratio between the number of non-compliant texts and the total number of text contents in the advertising video.

8. The intelligent advertising monitoring system for television broadcasting according to claim 5, characterized in that, The specific expression of the violation analysis model is as follows: ； In the expression, K represents the index for assessing the degree of advertising violation. This represents the audio intensity violation rating. This indicates the violation score based on playback time. This represents the evaluation value for violations of text content.

Images