A digital human high-quality video generation method and system based on data driving

Abstract

The application discloses a kind of based on data-driven digital person high-quality video generation method and system, it is related to data analysis technical field, including: obtaining historical digital person video, and current target video, extract the feature video in digital person video;Obtain the viewing user of viewing feature video, obtain target user according to the interactive behavior and viewing duration between viewing user and feature video;The behavior of target user is analyzed when watching feature video, and the audience user corresponding to target video is extracted;According to the adjustment record of the parameter value of the interactive parameter in the video changed by audience user, the early warning degree of target video is calculated, and intelligent prompt is carried out.The application obtains the audience user of digital person video by feature video and adjustment record, and the actual characteristics of audience user are intelligently analyzed, effectively generate high-quality, strong attractive video content, improve the viewing experience and satisfaction of user group watching video.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data analysis, and in particular to a digital human high-quality video generation method and system based on data driving. BACKGROUND

[0002] Digital human is a virtual character created through artificial intelligence technology, which integrates 3D modeling, speech synthesis and natural language processing and other cutting-edge technologies, and is widely used in short video creation field. As a new human-computer interaction carrier and digital life, it has high interaction ability and intelligent features, and brings unprecedented changes to the generation of oral video, and is gradually becoming a powerful tool in the hands of many creators.

[0003] At present, in the short video creation, the video generation depends on the preset template, lacks creativity and flexibility, and the digital human video is generated according to the fixed interaction parameters of the template preset. Although there is a simple adjustment of the interaction parameters by artificial means, it does not combine the characteristics of the audience group and the historical adjustment record for deep analysis, which may lead to the fact that the digital human video cannot adapt to the actual characteristics and needs of the audience group, affecting the viewing experience and satisfaction of the user group watching the video. SUMMARY

[0004] The purpose of the present application is to provide a digital human high-quality video generation method and system based on data driving to solve the problems existing in the prior art.

[0005] To achieve the above purpose, the present application provides the following technical scheme:

[0006] A digital human high-quality video generation method based on data driving, comprising the following steps:

[0007] Step S100: obtaining a digital human video generated by a digital human video template on a video creation platform, and a target video generated according to the video template at present; according to the type of the video template corresponding to the target video and the keywords in the script content of the target video, extracting a feature video from the digital human video;

[0008] Step S200: obtaining a viewing user watching the feature video on a video playing platform, extracting a segment of the feature video in which the digital human appears and performs oral broadcast, and obtaining a target user according to the viewing time of the viewing user and the interaction behavior with the feature video;

[0009] Step S300: analyzing the behavior of the target user pausing and rewinding when watching the feature video, and obtaining a keyword set of the target user in combination with the title of the video searched and played by the target user in history; according to the script content of the target video and the keyword set, extracting an audience user corresponding to the target video from the target user;

[0010] Step S400: According to the adjustment record of the parameter value of the interaction parameter in the characteristic video when the audience user watches the characteristic video, and the to-be-determined parameter value of the interaction parameter to be applied to the target video, the warning degree of the target video is calculated, and the target video and the to-be-determined parameter value are intelligently prompted according to the warning degree.

[0011] Further, step S100 includes:

[0012] Step S110: Obtain historical digital human videos generated by digital human video templates on a video creation platform, wherein each digital human video is a video that has been published on a video playing platform; obtain a target video V generated according to a video template Tem and a script content Tex;

[0013] Obtain the type corresponding to each video template, the type including the style, theme and applicable scene of the video template, and generate a corresponding type set; take the type set of the video template Tem as S Tem , obtain the video template WTem corresponding to a historical digital human video W, take the type set of the video template WTem as S WTem , according to the number N of the same types in the set S Tem and the set S WTem , obtain the first reliability X1=1-e -N of the digital human video W as a characteristic video, e is a natural constant;

[0014] Step S120: Extract all keywords in the script content Tex, remove the same keywords, and obtain a set S Tex ; extract all keywords in the script content WTex corresponding to the digital human video W, remove the same keywords, and obtain a set S WTex ; obtain a certain keyword KW1 in the set S Tex , calculate the similarity between each keyword in the set S WTex and the keyword KW1, take the maximum similarity as the target value of the keyword KW1, and further obtain the target value of each keyword in the set S Tex , and add and average to obtain the second reliability X2 of the digital human video W as a characteristic video;

[0015] In the present scheme, the characteristic video is a video that can select a suitable audience user for the target video, that is, a video similar to the target video, but since not every video is a characteristic video, it is necessary to select the characteristic video, which is obtained by analyzing the script content of the video template in the present step: 1, the first reliability of the digital human video W is determined through the type corresponding to the video template, since X1=1-e -NIn it, X1 increases as N increases, and when N > 0, the value of X1 ranges from 0 to 1. Then the value of the first reliability X1 ranges from 0 to 1. 2. Through the similarity of the corresponding keywords in the text content, if there is a greater similarity, then the text content of the digital human video W is more similar to the text content corresponding to the target video. Since obtaining the similarity between two keywords is a natural language processing technology with a value ranging from 0 to 1, the specific implementation process will not be elaborated here.

[0016] Pre-set the weights corresponding to the first reliability X1 and the second reliability X2 to obtain the total reliability X0; if X0 > K1 and |X1 - X2| < K2, where K1 is the first numerical threshold and K2 is the second numerical threshold, then the digital human video W is taken as the feature video.

[0017] Further, step S200 includes:

[0018] Step S210: Obtain a certain viewing user WU who views a certain feature video VID on the video playback platform, and set the initial behavior value of the viewing user WU to 0. If the viewing user WU likes the feature video VID, the behavior value is increased by h1. If the user comments, the behavior value is increased by h2. If the user collects, the behavior value is increased by h3, where h1 + h2 + h3 = 1, and then the final behavior value H is obtained.

[0019] Step S220: Extract the segments where the digital human appears on camera and gives an oral broadcast from the feature video VID, and obtain the total duration D0 of the segments. According to the viewing duration D of the viewing user WU for the videos within the segments WU , and the obtained behavior value H, the target degree G of the viewing user WU is obtained WU =H×D WU / D0. If the target degree G WU is greater than the preset degree threshold, then the viewing user WU is taken as the target user of the feature video VID, and thus all target users corresponding to each feature video are obtained.

[0020] Further, step S300 includes:

[0021] Step S310: Establish a keyword set S with an empty element corresponding to a certain target user TU TU ; Obtain the playback progress time when the target user TU executes the pause action while viewing the corresponding feature video FV, extract the statement text spoken by the digital human at the playback progress time, and input the keywords in the statement text into the set S TU ;

[0022] Obtaining the initial playing progress time T2 before the user TU performs the rollback action when watching the video FV, and the final playing progress time T1 after the rollback action, wherein the time T1 is before the time T2, extracting all the sentence texts said by the digital person within the time T1 to the time T2, and inputting the keywords in all the sentence texts into the set S TU ;

[0023] Obtaining the title of the video searched and played by the target user TU within C days before the time when the target user TU watches the feature video FV, and inputting the keywords in the title into the set S TU , and obtaining the final keyword set S TU ;

[0024] In the scheme, the keywords in the keyword set are words of interest to the user. When the user watches the digital person video and encounters the content said by the digital person that is of interest to the user, the user usually pauses to watch the subtitles or rolls back to deliberate. Then, the keywords of the corresponding sentence texts when pausing and rolling back can be input into the keyword set. The user usually searches the video title when searching, because such a title is also of interest to the user. Then, the keywords in the title can also be input into the keyword set, and the final keyword set is obtained.

[0025] Step S320: obtaining the set S Tex according to the script content Tex of the target video, obtaining a certain keyword KW2 in the set S Tex , calculating the similarity between each keyword in the set S TU and the keyword KW2, taking the maximum similarity as the target value of the keyword KW2, and obtaining the target value of each keyword in the set S Tex , and adding and averaging to obtain the correlation degree Y. If the correlation degree Y is greater than the correlation degree threshold, the target user TU is taken as the audience user of the target video.

[0026] Further, the step S400 includes: randomly extracting M adjustment records of the audience user changing the parameter value of the interactive parameter in the feature video, and obtaining the final parameter value of the interactive parameter after the change of each adjustment record, and obtaining the warning degree Z of the target video according to the pending parameter value of the target video , wherein e is a natural constant, P V is the pending parameter value of the target video, P m is the final parameter value corresponding to the mth adjustment record. If the warning degree Z is greater than the warning degree threshold, the target video and the pending parameter value are intelligently prompted.

[0027] The interaction parameters include speech speed, speech volume, video brightness, etc., and the adjustment records of the audience users are important basis for determining whether the pending parameter values of the target video are reasonable, for example, the audience users of the target video are usually middle-aged and old people, and most of the middle-aged and old people will adjust the speech volume to be larger in order to watch the video more appropriately, so when the pending parameter values of the target video deviate greatly from the parameter values corresponding to the actual adjustment records, that is, |P V -P m | is larger, the warning degree should also be larger.

[0028] A digital human high-quality video generation system based on data driving includes a feature video extraction module, a target user acquisition module, an audience user extraction module, and an interaction parameter analysis module.

[0029] The feature video extraction module is used to obtain digital human videos generated by digital human video templates on a video creation platform and a target video generated according to a video template at present, and extract feature videos in the digital human videos according to the type of the video template corresponding to the target video and the keywords in the script content of the target video.

[0030] The target user acquisition module is used to obtain a viewing user who watches a feature video on a video playing platform, extract a segment in which a digital human appears and performs oral broadcasting from the feature video, and obtain a target user according to the viewing duration of the viewing user and the interaction behavior with the feature video.

[0031] The audience user extraction module is used to analyze the behavior of the target user in pausing and rewinding when watching the feature video, and obtain a keyword set of the target user in combination with the title of the video searched and played by the target user in the past, and extract an audience user corresponding to the target video from the target user according to the script content of the target video and the keyword set.

[0032] The interaction parameter analysis module is used to calculate the warning degree of the target video according to the adjustment records of the audience user in changing the parameter values of the interaction parameters in the feature video and the pending parameter values of the interaction parameters to be applied to the target video, and intelligently prompt the target video and the pending parameter values according to the warning degree.

[0033] Further, the feature video extraction module includes a first reliability calculation unit and a feature video extraction unit.

[0034] The first reliability calculation unit is used to obtain digital human videos generated by digital human video templates on a video creation platform, obtain a target video generated according to a video template and a script content at present, obtain the type corresponding to each video template, generate a corresponding type set, and obtain the first reliability of the digital human video as a feature video.

[0035] The characteristic video extraction unit is configured to extract all keywords in the script content, establish a set, obtain a second reliability of the digital human video as a characteristic video, and obtain a total reliability according to the first reliability and the second reliability, and then obtain the characteristic video in the digital human video according to the total reliability.

[0036] Further, the audience user extraction module comprises a keyword set establishment unit and an audience user extraction unit.

[0037] The keyword set establishment unit is configured to establish a keyword set corresponding to the target user, analyze the pause and back behavior of the target user when watching the characteristic video, and combine the title of the video searched and played by the target user to obtain the final keyword set of the target user.

[0038] The audience user extraction unit is configured to extract the audience user corresponding to the target video from the target user according to the script content corresponding to the target video and the keyword set.

[0039] Compared with the prior art, the present application has the beneficial effects that: the present application provides a digital human high-quality video generation method and system based on data driving, comprising: obtaining historical digital human videos and the current target video, and extracting characteristic videos in the digital human videos; obtaining a watching user who watches the characteristic videos, and obtaining a target user according to the interactive behavior and watching time length between the watching user and the characteristic videos; analyzing the behavior of the target user when watching the characteristic videos, and extracting the audience user corresponding to the target video; calculating the warning degree of the target video according to the adjustment record of the audience user changing the parameter value of the interactive parameter in the video, and intelligently prompting. The present application obtains the audience user of the digital human video through the characteristic video and the adjustment record, intelligently analyzes the actual characteristics of the audience user, effectively generates high-quality and attractive video content, and improves the watching experience and satisfaction of the user group watching the video. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 It is a flowchart of the digital human high-quality video generation method based on data driving.

[0041] Figure 2 It is a structural diagram of the digital human high-quality video generation system based on data driving. DETAILED DESCRIPTION

[0042] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the protection scope of the present application.

[0043] Embodiment: As shown in the figure, the present application provides a data-driven digital human high-quality video generation method technical solution, including the following steps: Figure 1

[0044] Step S100: Obtain historical digital human videos generated through digital human video templates on a video creation platform, and a target video generated according to a video template at present; according to the type of the video template corresponding to the target video and the keywords in the script content of generating the target video, extract a feature video in the digital human video;

[0045] Step S110: Obtain historical digital human videos generated through digital human video templates on a video creation platform, wherein each digital human video is a video that has been published on a video playing platform; obtain a target video V generated according to a video template Tem and a script content Tex at present;

[0046] Obtain the type corresponding to each video template, the type including the style, theme and applicable scene of the video template, and generate a corresponding type set; take the type set of the video template Tem as S Tem , obtain the video template WTem corresponding to a historical digital human video W, take the type set of the video template WTem as S WTem , according to the number N of the same types in the set S Tem and the set S WTem , obtain the first reliability X1=1-e -N of the digital human video W as a feature video, e is a natural constant;

[0047] Step S120: Extract all keywords in the script content Tex, remove the same keywords, obtain a set S Tex , extract all keywords in the script content WTex corresponding to the digital human video W, remove the same keywords, obtain a set S WTex ; obtain a certain keyword KW1 in the set S Tex , calculate the similarity between each keyword in the set S WTex and the keyword KW1, take the maximum similarity as the target value of the keyword KW1, and then obtain the target value of each keyword in the set S Tex , and add and average as the second reliability X2 of the digital human video W as a feature video;​

[0048] In the present scheme, the characteristic video is a video that can select a suitable and audience user for the target video, that is, a video similar to the target video, but since not every video is a characteristic video, it is necessary to select a characteristic video, which is obtained in the present step by analyzing the script content of the video template: 1. Determine the first reliability of the digital human video W through the type corresponding to the video template. Since X1=1-e -N In the present scheme, the characteristic video is a video that can select a suitable and audience user for the target video, that is, a video similar to the target video, but since not every video is a characteristic video, it is necessary to select a characteristic video, which is obtained in the present step by analyzing the script content of the video template: 1. Determine the first reliability of the digital human video W through the type corresponding to the video template. Since X1=1-e

[0049] Pre-set the weight values corresponding to the first reliability X1 and the second reliability X2 to obtain the total reliability X0; if X0>K1 and |X1-X2|<K2 are satisfied, K1 is the first numerical threshold, and K2 is the second numerical threshold, then the digital human video W is taken as a characteristic video.

[0050] In the present embodiment, the weight values corresponding to the first reliability X1 and the second reliability X2 are respectively WX1 and WX2, and the total reliability X0=WX1×X1+WX2×X2 is obtained. Here, X0>K1 and |X1-X2|<K2 are set because the total reliability X0 represents the overall reliability value, which is an important factor for measuring whether the digital human video W can be a characteristic video. When the total reliability X0 is too small, it means that the digital human video W is unreliable and cannot be a characteristic video. |X1-X2| represents the absolute value of the difference between the two reliabilities. When one side is large and the other side is small, it means that the reliability is biased and unbalanced, which cannot be a characteristic video. Since the values of X1 and X2 are both 0 to 1, K1 is set to 0.6 and K2 is set to 0.3 in the present embodiment.

[0051] Step S200: Obtain a viewing user who watches a characteristic video on a video playback platform, extract a segment in which a digital human appears and performs a voice broadcast from the characteristic video, and obtain a target user according to the viewing duration of the viewing user and the interaction behavior with the characteristic video.

[0052] Step S210: Obtain a viewing user WU who watches a certain characteristic video VID on a video playback platform, set the initial behavior value of the viewing user WU to 0, if the viewing user WU likes the characteristic video VID, then the behavior value is increased by h1, if the viewing user WU comments, then the behavior value is increased by h2, if the viewing user WU collects, then the behavior value is increased by h3, h1+h2+h3=1, and then the final behavior value H is obtained.

[0053] Step S220: Extract the segment of the digital person appearing and performing oral broadcast from the feature video VID, and obtain the total duration D0 of the segment, and obtain the viewing duration D of the viewing user WU viewing the video in the segment WU , and obtain the behavior value H, and obtain the target degree G of the viewing user WU WU = H x D WU / D0, if the target degree G WU is greater than a preset degree threshold, the viewing user WU is regarded as a target user of the feature video VID, and then all target users corresponding to each feature video are obtained.

[0054] The target user obtained here is determined based on the interaction behavior and the viewing duration with the feature video. When a user likes, comments and collects a feature video, and the viewing duration is relatively long, it indicates that the content of the feature video meets the user, and then the user can be regarded as a target user of the feature video. However, the user is not necessarily an audience user of the target video, so the following analysis is needed to determine the audience user of the target video.

[0055] Step S300: Analyzing the behavior of pausing and rewinding of the target user when watching the feature video, and combining the title of the video searched and played by the target user to obtain a keyword set of the target user; extracting the audience user corresponding to the target video from the target user according to the script content corresponding to the target video and the keyword set;

[0056] Step S310: Establishing a keyword set S TU corresponding to a target user TU as empty; obtaining the playing progress time of the target user TU when watching the corresponding feature video FV, executing the pause action, extracting the sentence text spoken by the digital person at the playing progress time, and inputting the keywords in the sentence text into the set S TU ;

[0057] Obtaining the initial playing progress time T2 before the rewinding action and the final playing progress time T1 after the rewinding action of the user TU when watching the video FV, wherein the time T1 is before the time T2, extracting all sentence texts spoken by the digital person within the time T1 to the time T2, and inputting the keywords in all the sentence texts into the set S TU ;

[0058] Obtaining the title corresponding to the video searched and played by the target user TU within C days before the time when the target user TU watches the feature video FV as the starting point, and inputting the keywords in the title into the set S TU , and then obtaining the final keyword set S TU ;

[0059] In this solution, the keywords in the keyword set are words that users are interested in. When watching a digital human video, if the user encounters content that the digital human says that interests them, they will usually pause to watch the subtitles or rewind to consider. In this case, the keywords of the corresponding sentence text during the pause and rewind can be entered into the keyword set. When users conduct historical searches, they are usually attracted to the titles of the videos they are searching for. This is because such titles are also of interest to users. In this case, the keywords in the titles can also be entered into the keyword set to obtain the final keyword set.

[0060] Step S320: The set S is obtained based on the text content Tex corresponding to the target video. Tex , get the set S Tex A keyword KW2 in the set S TU The similarity between each keyword in and keyword KW2 is calculated, and the maximum similarity is used as the target value of keyword KW2, and then the set S is obtained. Tex The target value of each keyword in is added and the average value is obtained to obtain the correlation degree Y. If the correlation degree Y is greater than the correlation degree threshold, the target user TU is regarded as the audience user of the target video.

[0061] Step S400: Based on the adjustment records of the parameter values ​​of the interactive parameters in the video when the audience user watches the feature video, and the pending parameter values ​​of the interactive parameters currently to be applied to the target video, the warning level of the target video is calculated, and intelligent prompts are given to the target video and the pending parameter values ​​according to the warning level.

[0062] Step S400 includes: randomly extracting M adjustment records of the parameter values ​​of the interaction parameters in the video when the audience user watches the feature video, and obtaining the final parameter value after the interaction parameter of each adjustment record is changed, and obtaining the warning level of the target video according to the pending parameter value of the target video. , where e is a natural constant, P V is the undetermined parameter value of the target video, P m It is the final parameter value corresponding to the mth adjustment record. If the warning level Z is greater than the warning level threshold, an intelligent prompt is given for the target video and the pending parameter value.

[0063] Interaction parameters include speech speed, speech volume, and video brightness. For the target video, the adjustment records of the target users are an important basis for determining whether the target video's pending parameter values ​​are reasonable. For example, the target video's target users are usually middle-aged and elderly people, and most of them adjust the speech volume to a higher level in order to watch the video more appropriately. In this case, when the target video's pending parameter values ​​deviate significantly from the actual parameter values ​​corresponding to the adjustment records, that is, |P V -Pm | is larger, then the warning level should also be larger. Since the warning level Z ranges from 0 to 1, in this solution, the warning level threshold is 0.6. When the warning level Z is greater than 0.6, intelligent prompts are given for the target video and the undetermined parameter values.

[0064] The present invention also provides a data-driven digital human high-quality video generation system, such as Figure 2 Shown, including:

[0065] Feature video extraction module: used to obtain historical digital human videos generated using digital human video templates on the video creation platform, as well as target videos currently generated based on video templates; extract feature videos from digital human videos based on the type of video template corresponding to the target video and the keywords in the text content used to generate the target video;

[0066] Target user acquisition module: This module is used to acquire users who watch the featured video on the video playback platform. It extracts the clips of the digital human appearing and speaking from the featured video, and then determines the target user based on the viewing time and interaction behavior of the user with the featured video.

[0067] Audience user extraction module: This module analyzes the target user's pause and rewind behaviors when watching a featured video, and combines the titles of the videos that the target user has searched and played in the past to obtain the target user's keyword set. Based on the target video's corresponding copy and keyword set, the module extracts the audience user corresponding to the target video from the target user.

[0068] Interaction parameter analysis module: used to calculate the warning level of the target video based on the adjustment records of the parameter values ​​of the interaction parameters in the video when the audience users watch the feature video, as well as the pending parameter values ​​of the interaction parameters currently to be applied to the target video, and provide intelligent prompts for the target video and the pending parameter values ​​based on the warning level.

[0069] The feature video extraction module includes a first reliability calculation unit and a feature video extraction unit;

[0070] The first reliability calculation unit is used to obtain digital human videos generated in the past using digital human video templates on the video creation platform, and obtain the target video currently generated based on the video template and copy content; obtain the type corresponding to each video template, and generate a corresponding type set to obtain the first reliability of the digital human video as a feature video;

[0071] Feature video extraction unit: used to extract all keywords in the copy content, establish a set, and obtain the second reliability of the digital human video as a feature video; based on the first reliability and the second reliability, obtain the total reliability, and then obtain the feature video in the digital human video based on the total reliability.

[0072] The audience user extraction module comprises a keyword set establishing unit and an audience user extraction unit;

[0073] The keyword set establishing unit is configured to establish a keyword set corresponding to a target user and empty of elements, analyze the behaviors of pausing and rewinding of the target user when watching a characteristic video, and obtain a final keyword set of the target user in combination with the titles of the videos searched and played by the target user in the past;

[0074] The audience user extraction unit is configured to extract audience users corresponding to a target video from target users according to the script content corresponding to the target video and the keyword set.

[0075] It is apparent for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, but can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all aspects as illustrative and not restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and range of the equivalent elements of the claims are intended to be embraced in the present application. Any reference signs in the claims should not be considered as limiting the involved claims.

Claims

1. A data-driven digital human high-quality video generation method, characterized in that, The method comprises the following steps: Step S100: obtaining digital human videos generated by a digital human video template on a video creation platform in history, and a target video generated according to the video template at present; According to the type of the video template corresponding to the target video and the keywords in the script content of the target video, a feature video in the digital human video is extracted; Step S200: obtaining a viewing user who watches the feature video on a video playing platform, extracting a segment in which a digital human appears and performs oral broadcast from the feature video, and obtaining a target user according to the viewing duration of the viewing user and the interactive behavior with the feature video; Step S300: analyzing the behavior of the target user in pausing and rewinding when watching the feature video, and obtaining a keyword set of the target user in combination with the title of the video searched and played by the target user in history; extracting an audience user corresponding to the target video from the target user according to the script content of the target video and the keyword set; Step S400: calculating the warning degree of the target video according to the adjustment record of the parameter value of the interactive parameter changed by the audience user when watching the feature video, and the undetermined parameter value of the interactive parameter to be applied to the target video, and intelligently prompting the target video and the undetermined parameter value according to the warning degree.

2. The data-driven digital human high-quality video generation method of claim 1, wherein, Step S100 comprises: Step S110: obtaining digital human videos generated by a digital human video template on a video creation platform in history, wherein each digital human video is a video that has been published on a video playing platform; and obtaining a target video V generated according to a video template Tem and a script content Tex at present; Obtain the type corresponding to each video template, the type including the style, theme and applicable scene of the video template, and generate a corresponding type set; take the type set of the video template Tem as S Tem , obtain the video template WTem corresponding to a historical digital human video W, take the type set of the video template WTem as S WTem , obtain the first reliability X1 of the digital human video W as a feature video according to the number N of the same types in the set S Tem and the set S WTem , X1=1-e -N , e is a natural constant; Step S120: extract all keywords in the script content Tex, remove the same keywords, and obtain a set S Tex , extract all keywords in the script content WTex corresponding to the digital human video W, remove the same keywords, and obtain a set S WTex ; obtain a certain keyword KW1 in the set S Tex ; calculate the similarity between each keyword in the set S WTex and the keyword KW1, take the maximum similarity as the target value of the keyword KW1, and further obtain the target value of each keyword in the set S Tex , and add and average the target values to obtain the second reliability X2 of the digital human video W as a feature video. The total reliability X0 is obtained by pre-setting the weight values corresponding to the first reliability X1 and the second reliability X2; if X0>K1 and |X1-X2|<K2 are satisfied, K1 is a first numerical threshold, and K2 is a second numerical threshold, the digital human video W is taken as a feature video.

3. The data-driven digital human high-quality video generation method of claim 1, wherein, Step S200 comprises: Step S210: obtaining a viewing user WU who watches a certain feature video VID on a video playing platform, setting the initial behavior value of the viewing user WU as 0, if the viewing user WU likes the feature video VID, the behavior value is increased by h1, if the viewing user WU comments, the behavior value is increased by h2, if the viewing user WU collects, the behavior value is increased by h3, h1+h2+h3=1, and then the final behavior value H is obtained; Step S220: Extract the segment of the digital human appearing and performing oral broadcast from the feature video VID, and obtain the total duration D0 of the segment, and the viewing duration D of the user WU watching the video in the segment WU , and the obtained behavior value H, and the target degree G of the user WU WU =H×D WU / D0, if the target degree G WU is greater than the preset degree threshold, the user WU is regarded as the target user of the feature video VID, and then all target users corresponding to each feature video are obtained.

4. The data-driven digital human high-quality video generation method of claim 1, wherein, Step S300 comprises: Step S310: Establishing an empty keyword set S corresponding to an element of a target user TU TU ; obtaining the playing progress time when the target user TU performs the pause action when watching the corresponding feature video FV, extracting the sentence text spoken by the digital human at the playing progress time, and inputting the keywords in the sentence text into the set S TU ; obtaining an initial playing progress time T2 before a user TU performs a rollback action when watching a video FV, and a final playing progress time T1 after the rollback action, wherein the time T1 is before the time T2, extracting all sentence texts said by the digital human within the time T1 to the time T2, and inputting keywords in the all sentence texts into a set S TU ; A title corresponding to a video searched and played by the target user TU within C days before the time when the target user TU watches the feature video FV is obtained, and a keyword in the title is input into the set S TU In this way, the final keyword set S is obtained TU ; Step S320: obtaining a set S Tex according to the script content Tex corresponding to the target video Tex , obtaining a certain keyword KW2 in the set S TU , calculating the similarity between each keyword in the set S Tex and the keyword KW2, taking the maximum similarity as the target value of the keyword KW2, and then obtaining the target value of each keyword in the set S Tex , and adding and averaging to obtain the correlation degree Y, if the correlation degree Y is greater than the correlation degree threshold, the target user TU is taken as the audience user of the target video.

5. The data-driven digital human high-quality video generation method of claim 1, wherein, Step S400 comprises: randomly extracting M adjustment records of the audience user changing the parameter value of the interaction parameter in the characteristic video when watching the characteristic video, and obtaining the final parameter value of the interaction parameter after the adjustment of each adjustment record, and obtaining the warning degree of the target video according to the pending parameter value of the target video wherein e is a natural constant, P V is the pending parameter value of the target video, P m is the final parameter value corresponding to the mth adjustment record, and if the warning degree Z is greater than the warning degree threshold, the target video and the pending parameter value are intelligently prompted.

6. A digital human high-quality video generation system for performing a data-driven based digital human high-quality video generation method according to any one of claims 1-5, characterized in that, The system comprises a feature video extraction module, a target user acquisition module, an audience user extraction module and an interactive parameter analysis module; The feature video extraction module is used for obtaining digital human videos generated by a digital human video template on a video creation platform in history, and a target video generated according to the video template at present; and extracting a feature video in the digital human video according to the type of the video template corresponding to the target video and the keywords in the script content of the target video; The target user acquisition module is used for obtaining a viewing user who watches the feature video on a video playing platform, extracting a segment in which a digital human appears and performs oral broadcast from the feature video, and obtaining a target user according to the viewing duration of the viewing user and the interactive behavior with the feature video; The audience user extraction module is configured to analyze the pausing and rewinding behaviors of the target user when watching the characteristic video, and obtain a keyword set of the target user in combination with the titles of the videos searched and played by the target user in the past; and extract audience users corresponding to the target video from the target user according to the script content corresponding to the target video and the keyword set. The interaction parameter analysis module is configured to calculate a warning degree of the target video according to the adjustment records of the parameter values of the interaction parameters in the target video changed by the audience users when watching the characteristic video, and the to-be-determined parameter values of the interaction parameters to be applied to the target video, and intelligently prompt the target video and the to-be-determined parameter values according to the warning degree.

7. The digital human high-quality video generation system of claim 6, wherein, The characteristic video extraction module includes a first reliability calculation unit and a characteristic video extraction unit. The first reliability calculation unit is configured to obtain digital human videos generated by using digital human video templates on a video creation platform, and obtain a target video generated according to a video template and a script content; obtain the types corresponding to each video template, generate a corresponding type set, and obtain a first reliability of the digital human video as a characteristic video; The characteristic video extraction unit is configured to extract all keywords in the script content, establish a set, and obtain a second reliability of the digital human video as a characteristic video; obtain a total reliability according to the first reliability and the second reliability, and then obtain a characteristic video in the digital human video according to the total reliability.

8. The digital human high-quality video generation system of claim 6, wherein, The audience user extraction module includes a keyword set establishment unit and an audience user extraction unit. The keyword set establishment unit is configured to establish a keyword set corresponding to the target user, and analyze the pausing and rewinding behaviors of the target user when watching the characteristic video, and obtain a final keyword set of the target user in combination with the titles of the videos searched and played by the target user in the past. The audience user extraction unit is configured to extract audience users corresponding to the target video from the target user according to the script content corresponding to the target video and the keyword set.

Images