Video processing method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a video processing method and device, equipment and a storage medium. The method comprises the steps that a to-be-processed video is acquired, and a video description text is generated based on preset video attribute information of the to-be-processed video; performing feature extraction on the to-be-processed video and the video description text, and fusing the obtained video feature vector and the text feature vector to form a fused feature vector; according to the fusion feature vector, determining an enhancement feedback prediction result of the to-be-processed video, the enhancement feedback prediction result being used for reflecting whether enhancement processing is performed on the to-be-processed video; and under the condition that the enhancement feedback prediction result meets the enhancement processing condition, video enhancement processing is allowed to be performed on the to-be-processed video. By using the method, the enhancement feedback prediction result can be determined for the to-be-processed video, and enhancement processing can be performed on the to-be-processed video only when the enhancement feedback prediction result meets the enhancement processing condition, so that the invalid consumption of enhancement processing computing power resources is reduced, and the enhancement processing cost is reduced.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of computer vision, and particularly relate to a video processing method and device, equipment and storage medium. BACKGROUND

[0002] In a video on demand service, in order to improve the viewing experience of users, the server will apply an enhanced or policy algorithm to the video before delivering the video to the client to improve the video quality or viewing fluency.

[0003] However, not all videos are suitable for enhancement processing. Due to different viewing audiences, the processed video delivered by the server may not bring obvious positive feedback or even negative feedback. As a result, not only the viewing experience of the viewing audience is affected, but also the computing resources of the server are wasted to a great extent. SUMMARY

[0004] Embodiments of the present disclosure provide a video processing method, device, equipment and storage medium to improve the positive feedback of the processed video in the viewing audience and reduce the computing cost of video enhancement processing.

[0005] In a first aspect, the embodiments of the present disclosure provide a video processing method, which comprises:

[0006] obtaining a to-be-processed video and generating a video description text based on preset video attribute information of the to-be-processed video;

[0007] extracting features from the to-be-processed video and the video description text, and fusing the obtained video feature vector and text feature vector to form a fusion feature vector;

[0008] determining an enhancement feedback prediction result of the to-be-processed video according to the fusion feature vector, the enhancement feedback prediction result being used to reflect whether to perform enhancement processing on the to-be-processed video;

[0009] in a case where the enhancement feedback prediction result meets an enhancement processing condition, allowing video enhancement processing to be performed on the to-be-processed video.

[0010] In a second aspect, the embodiments of the present disclosure further provide a video processing device, which comprises:

[0011] an obtaining module configured to obtain a to-be-processed video and generate a video description text based on preset video attribute information of the to-be-processed video;

[0012] a feature extraction module configured to extract features from the to-be-processed video and the video description text, and fuse the obtained video feature vector and text feature vector to form a fusion feature vector;

[0013] A result prediction module is configured to determine an enhanced feedback prediction result of the to-be-processed video according to the fusion feature vector, the enhanced feedback prediction result being used to indicate whether the to-be-processed video is subjected to enhanced processing.

[0014] An enhanced processing module is configured to allow the to-be-processed video to be subjected to video enhancement processing when the enhanced feedback prediction result meets an enhanced processing condition.

[0015] In a third aspect, the embodiments of the present disclosure further provide a computer device, which comprises:

[0016] one or more processors;

[0017] a storage device configured to store one or more programs,

[0018] When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the video processing method provided by any of the embodiments of the present disclosure.

[0019] In a fourth aspect, the embodiments of the present disclosure further provide a computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the video processing method provided by any of the embodiments of the present disclosure.

[0020] In a fifth aspect, the embodiments of the present disclosure further provide a computer program product comprising a computer program, and the computer program is executed by a processor to implement the video processing method provided by any of the embodiments of the present disclosure.

[0021] The technical scheme of the embodiment of the present disclosure specifically discloses a video processing method, device, equipment and storage medium. The method first acquires a to-be-processed video and generates a video description text based on preset video attribute information of the to-be-processed video; performs feature extraction on the to-be-processed video and the video description text, and fuses the obtained video feature vector and text feature vector to form a fusion feature vector; determines an enhanced feedback prediction result of the to-be-processed video according to the fusion feature vector, and the enhanced feedback prediction result is used to reflect whether the to-be-processed video is subjected to enhanced processing; and in the case that the enhanced feedback prediction result meets an enhanced processing condition, the to-be-processed video is allowed to be subjected to video enhancement processing. The above technical scheme of the embodiment can analyze the features of the video to predict the enhanced feedback prediction result of the video after the video is subjected to enhanced processing before the video is subjected to enhanced processing, and limit that only in the case that the enhanced feedback prediction result meets the enhanced processing condition, the to-be-processed video is allowed to be subjected to enhanced processing. The above technical scheme of the embodiment is equivalent to a preprocessing scheme before the video is subjected to enhanced processing, and by determining the enhanced feedback prediction result, it can be determined in advance whether the video is suitable for enhanced processing before the video is subjected to enhanced processing, and only the video suitable for enhanced processing is subjected to enhanced processing, so as to better reduce the invalid consumption of enhanced processing computing resource and reduce the enhanced processing cost. Meanwhile, the video filtered by the technical scheme of the embodiment can improve the viewing experience of the audience, thereby bringing better positive feedback effect. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical scheme of the exemplary embodiments of the present disclosure, the drawings needed in the description of the embodiments will be briefly introduced. Obviously, the drawings introduced are only a part of the drawings of the present disclosure, and not all the drawings. Those skilled in the art can obtain other drawings according to these drawings without creating creative labor.

[0023] Figure 1a A flowchart of a video processing method provided by the embodiment of the present disclosure is shown;

[0024] Figure 1b An implementation process diagram for determining the enhanced feedback prediction result of the to-be-processed video in the video processing method provided by the embodiment of the present disclosure is shown;

[0025] Figure 2 A structural diagram of a video processing device provided by the embodiment of the present disclosure is shown;

[0026] Figure 3 A structural diagram of a computer device provided by the embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0027] Embodiments of the present disclosure will be described below in greater detail with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein; rather, these embodiments are provided so as to more completely and thoroughly understand the present disclosure. It is understood that the drawings and embodiments of the present disclosure are for exemplary purposes only and are not intended to limit the scope of protection of the present disclosure.

[0028] It should be understood that each of the steps recited in the method embodiments of the present disclosure can be performed in different orders and / or in parallel. In addition, the method embodiments can include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0029] The term "comprising" and variations thereof as used herein are open-ended, that is "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Related terms are defined in the following description.

[0030] It should be noted that the terms "first", "second", and the like in the present disclosure are merely used to distinguish different devices, modules or units, and do not imply the order or interdependence of the functions performed by these devices, modules or units. It should be noted that the terms "one", "multiple" in the present disclosure are illustrative and not limiting, and those skilled in the art should understand that "one" or "multiple" should be understood as "one or more" unless otherwise explicitly stated in the context.

[0031] The names of the messages or information exchanged between the devices in the embodiments of the present disclosure are merely for illustrative purposes and are not intended to limit the scope of the messages or information.

[0032] The names of the messages or information exchanged between the devices in the embodiments of the present disclosure are merely for illustrative purposes and are not intended to limit the scope of the messages or information.

[0033] It can be understood that before using the technical solutions disclosed in the embodiments of the present disclosure, the type of personal information involved in the present disclosure, the scope of use, the use scenario, etc. should be informed to the user and the authorization of the user should be obtained in accordance with relevant laws and regulations.

[0034] For example, in response to receiving an active request of a user, prompt information is sent to the user to explicitly prompt the user that the operation requested to be performed will require obtaining and using personal information of the user. Thus, the user can autonomously select whether to provide personal information to the software or hardware, such as an electronic device, an application program, a server, or a storage medium, performing the operation of the technical solution of the present disclosure according to the prompt information.

[0035] As an optional but non-limiting implementation, in response to receiving an active request of a user, the manner of sending prompt information to the user may, for example, be a pop-up window manner, in which the prompt information can be presented in the form of text. In addition, the pop-up window can also carry selection controls for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0036] It can be understood that the above notification and obtaining user authorization process is only illustrative and does not limit the implementation of the present disclosure, and other manners meeting relevant laws and regulations can also be applied to the implementation of the present disclosure.

[0037] Figure 1a A flowchart of a video processing method provided by the embodiment of the present disclosure is provided. The embodiment can be applied to the case of video processing. The method can be executed by a video processing device, which can be implemented by software and / or hardware, and can be configured in a terminal and / or a server to implement the video processing method in the embodiment of the present disclosure.

[0038] It should be noted that one application scenario of the embodiment can be described as follows: in a video on demand service, in order to improve the viewing experience of users, the server often applies various enhancement or strategy algorithms (such as a video quality enhancement algorithm, a gear combination algorithm, etc.) to improve the video quality or viewing fluency. However, the gain that a general algorithm can obtain in different viewing crowds or different types of videos is different. For some crowds or videos, no obvious positive feedback can be obtained, and even negative feedback can occur. Therefore, not only the viewing experience of the viewing audience is affected, but also the computing resources of the enhancement processing are wasted to a great extent.

[0039] Based on this, the embodiment provides a video processing method, which can improve the positive feedback of the processed video in the viewing audience and reduce the computing cost of video enhancement processing. Specifically, as shown in Figure 1a The video processing method provided by the embodiment can include the following steps.

[0040] S101, obtaining a to-be-processed video and generating a video description text based on preset video attribute information of the to-be-processed video.

[0041] In this embodiment, the execution subject can be regarded as a kind of server with video distribution function application, or a terminal capable of responding to the requested video. In this embodiment, when one or more videos have certain enhancement requirements, the video can be subjected to enhancement processing to meet the enhancement requirements. For a video, the enhancement requirements it has can include quality enhancement, noise removal, and deblurring, and thus the target enhancement processing strategies that can be adopted can include quality enhancement algorithms, noise removal algorithms, and deblurring algorithms.

[0042] In this embodiment, the video to be processed can be considered as a video with enhancement processing requirements and it is determined that what kind of enhancement processing can be performed. The number of videos to be processed in this embodiment is not specifically limited, and can be one or more. In this embodiment, the video to be processed can be pre-cached in the original video set. In this embodiment, the video stored in the original video set can be regarded as a video with enhancement processing requirements, and thus the video to be processed that needs to be processed can be directly searched from the original video set. The video in the original video set can be directly obtained from the video creation end, for example, the execution subject can receive the original video created by the video creation end. In addition, the video in the original video set can also be pre-imported from the outside by the execution subject.

[0043] In this embodiment, after determining the object to be processed, a video description text of the video can be further formed. In this embodiment, the video description text can be formed by summarizing the preset video attribute information related to the video. The preset video attribute information can be understood as information for recording the basic attribute information possessed by the video and the video creation related information.

[0044] For example, the basic attribute information can include video duration, video resolution, and video tags, and the video creation related information can include video creator information, whether background music is used, whether a filter is included, and whether there is a transition. The text formed by summarizing the above attribute contents can be regarded as the video description text of the video to be processed.

[0045] S102, feature extraction is performed on the video to be processed and the video description text, and a video feature vector and a text feature vector obtained are fused to form a fusion feature vector.

[0046] In this embodiment, feature extraction refers to the process of converting raw data into feature representations that can be used by machine learning algorithms and models. The raw data can be video data, image data, and text data, etc. Specifically, the feature extraction involved in this embodiment can be text feature extraction and video feature extraction, etc. In this embodiment, different video feature extraction models can be used to extract features from the video to be processed, depending on specific business needs. Similarly, text feature extraction can also be performed on the video description text using text recognition models, etc.

[0047] In this embodiment, the video feature vector of the video to be processed can refer to the feature vector extracted by a deep learning network (such as a convolutional neural network, a self-attention neural network, etc.) that meets the business requirements. The process of determining the text feature vector of the video description text can be considered as the specific implementation of extracting attribute features such as the resolution of the video to be processed, whether the video to be processed uses background music, and whether the video to be processed contains filters contained in the video description text and representing them in vector form.

[0048] In this embodiment, feature vector fusion can be considered as the integration of multiple feature vectors into a composite feature vector through a certain feature fusion method. The specific feature fusion method can be weighted fusion, serial fusion, etc., and this application does not limit the specific fusion method.

[0049] In this embodiment, feature extraction can be performed on the video to be processed and the video description text of the video to be processed by a feature extraction network model that matches the business requirements, so as to obtain video feature vectors and text feature vectors respectively. Then, the video feature vectors and text feature vectors can be fused by a feature fusion method to obtain a fused feature vector.

[0050] S103. Based on the fused feature vector, determine the enhancement feedback prediction result of the video to be processed. The enhancement feedback prediction result is used to indicate whether the video to be processed should be enhanced.

[0051] It should be noted that the videos on this execution entity typically need to participate in some kind of business, such as video push service. Furthermore, business evaluation metrics can be set for the participating business, such as video views or viewing time in the video push service. Based on the evaluation metric values, the effectiveness of the video in providing business feedback can be assessed. For example, higher video views or longer viewing time can be considered as better business feedback provided by the video in that service.

[0052] Furthermore, it can be seen that whether the video is enhanced or not, it can obtain a corresponding business evaluation index value relative to the business evaluation metrics when participating in a certain business. If the business evaluation index value of the enhanced video is higher than that of the unprocessed original video, the enhancement can be considered to have brought positive feedback to the video; if the business evaluation index value of the enhanced video is less than or equal to that of the unprocessed original video, the enhancement can be considered to have brought negative feedback to the video.

[0053] In this embodiment, this step can be used to predict enhanced feedback for the video to be processed by fusing the determined feature vectors, such as the fused feature vector obtained by fusing video feature vectors and text feature vectors. The enhanced feedback prediction result can be understood as predictive information characterizing the positive or negative feedback that the enhanced video can bring relative to a certain business evaluation indicator.

[0054] For example, if the predicted augmentation feedback is negative (e.g., a value not greater than 0), it can be assumed that the video's evaluation index value relative to the business evaluation index after augmentation is lower than the video's inherent evaluation index value. Similarly, if the predicted augmentation feedback is positive (e.g., a positive value), it can be assumed that the video's evaluation index value relative to the business evaluation index after augmentation is higher than the video's inherent evaluation index value.

[0055] In this embodiment, as one way to enhance the prediction results, it can be achieved by constructing an enhanced feedback prediction model. For example, a feature analysis network structure for feature vector analysis can be constructed in the enhanced feedback prediction model. First, the fused vector of video feature vector and text feature vector is analyzed through the constructed feature analysis network structure to further obtain the video's own feature information, such as the pixel feature information of the video frame, which plays a key role in video enhancement difference feedback.

[0056] Following the above description, a difference prediction network structure can be constructed within the enhanced feedback prediction model to perform enhanced difference prediction. This network structure can then be used to simulate the calculation of the difference in evaluation metrics between the video and the service evaluation metrics before and after enhancement processing. It should be noted that the evaluation metric value of a video relative to a certain service evaluation metric may be affected by one or more features of the video itself. For example, video quality can affect the evaluation metric value. Considering that video quality is affected by its own pixel values ​​and resolution, the evaluation metric value can also be considered to be affected by the video's own pixel values ​​and resolution.

[0057] Therefore, this embodiment can construct a network structure capable of evaluating one or more features of a video to simulate the determination of the evaluation index values ​​of the video relative to the business evaluation indicators before and after enhancement processing, and can simulate the calculation of the difference in evaluation index values, so that the calculated difference result is determined as the enhanced feedback prediction result output by the enhanced feedback prediction model. Through continuous training and learning of the constructed enhanced feedback prediction model, this embodiment can ensure that the enhanced feedback prediction model can ultimately output enhanced feedback prediction results that are closer to actual applications.

[0058] Furthermore, commonly used business evaluation metrics such as video views and video playback duration are significantly influenced by the video's length. This embodiment considers constructing multiple difference prediction network structures for different video duration ranges to obtain enhancement result prediction values ​​output by each network structure. These different enhancement difference prediction values ​​can characterize the enhancement feedback prediction results before and after enhancement processing for videos within different duration ranges. This embodiment can find the enhancement feedback prediction result for the video to be processed from the different enhancement feedback prediction results based on the video duration corresponding to the video to be processed.

[0059] S104. If the enhanced feedback prediction result meets the enhancement processing conditions, video enhancement processing is allowed for the video to be processed.

[0060] In this embodiment, the enhancement processing condition can be considered as the determination condition for deciding whether to perform enhancement processing on the video to be processed. This enhancement processing condition can be set to a positive feedback prediction result. This step considers that the execution condition for enhancing the video to be processed is met when the enhancement feedback prediction result is positive feedback. Therefore, this step allows the video to be enhanced using an enhancement algorithm that meets the enhancement requirements.

[0061] Similarly, in this embodiment, if the enhancement feedback prediction result is determined to be negative feedback, it can be considered that the execution conditions for enhancing the video to be processed are not met, and the execution of this step can be stopped.

[0062] This embodiment provides a video processing method that, before video enhancement processing, analyzes video features to predict the enhancement feedback prediction result after enhancement processing. Enhancement processing is only allowed if the enhancement feedback prediction result meets the enhancement processing conditions. This embodiment's technical solution is equivalent to a preprocessing scheme before video enhancement processing. By predicting the enhancement feedback prediction result, it can determine in advance whether a video is suitable for enhancement processing, and only enhance suitable videos, thus reducing the ineffective consumption of computing resources and lowering the cost of enhancement processing. Furthermore, videos filtered through this embodiment's technical solution can improve the viewer's viewing experience, resulting in better positive feedback.

[0063] As a first optional embodiment of this example, based on the above optimization, the enhanced feedback prediction result of the video to be processed can be determined according to the fused feature vector, specifically optimized as follows:

[0064] a1) Input the fused feature vector into the constructed enhanced feedback prediction model to obtain at least one enhanced difference prediction value, each enhanced difference prediction value corresponding to a time range.

[0065] In this embodiment, different enhancement processing algorithms can be pre-set to meet different enhancement processing requirements, such as noise reduction, deblurring, and video enhancement algorithms to improve resolution and image quality. In practical applications, the required enhancement processing algorithm can be selected according to actual business needs.

[0066] In this embodiment, corresponding augmentation feedback prediction models can be pre-determined for different augmentation processing algorithms. The augmentation feedback prediction models corresponding to different augmentation processing algorithms can be considered to have the same network structure. However, during the model training stage, the same initial augmentation feedback prediction model can be trained for different augmentation processing algorithms to make the augmentation feedback prediction models corresponding to different augmentation processing algorithms correspond to different network parameters.

[0067] This step allows us to obtain a pre-built and trained augmented feedback prediction model. For example, the augmented feedback prediction model corresponding to a video quality enhancement processing algorithm can be considered as a model formed after training an initial augmented feedback prediction model on sample videos related to the video quality enhancement processing algorithm. Based on the above description of the augmented feedback prediction model in this embodiment, it can be seen that the augmented difference prediction value determined by the augmented feedback prediction model is equivalent to the prediction of the difference in evaluation index values ​​before and after the video to be processed, relative to the set business evaluation index values.

[0068] Considering that the evaluation index values ​​of business performance indicators require a focus on certain feature information (such as pixel features) in the video to be processed, a feature analysis network structure built using an enhanced feedback prediction model can be used to analyze the fused feature vectors to extract the feature information that can influence changes in the evaluation index values. Then, the network structure built within the enhanced feedback prediction model for enhancing difference prediction can be used to simulate and calculate the difference in evaluation index values ​​before and after the enhancement processing of the video, based on the feature information input to the feature analysis network structure. This difference value can serve as the enhanced difference prediction value, representing the enhanced feedback prediction result of the video to be processed.

[0069] Among them, considering that the evaluation index values ​​of business evaluation indicators are greatly affected by video duration, such as the upper limit of the average playback duration of short videos being much lower than that of long videos, dividing the duration range of different videos cannot guarantee that the enhanced feedback prediction results output by the enhanced feedback prediction model will be suitable for all video durations.

[0070] Based on this, this embodiment can also construct network structures for enhanced difference prediction for different duration ranges in the enhanced feedback prediction model. Thus, the feature output information from the feature analysis network structure processing the fused feature vectors can be input into the corresponding enhanced difference prediction network structures for different duration ranges, and the enhanced difference prediction values ​​can be obtained based on the output of each network structure. The duration range can be divided according to the characteristics of the video playback duration, such as within 10 seconds, between 10 and 20 seconds, between half a minute and one minute, etc., and multiple duration ranges can be defined.

[0071] b1) Determine the target duration range of the video to be processed, and determine the enhanced difference prediction value corresponding to the target duration range as the enhanced feedback prediction result.

[0072] In this embodiment, the multiple enhanced difference prediction values ​​output above can be used to form an enhanced difference prediction list, and each enhanced difference prediction value has a known corresponding duration range. Then, by combining the playback duration of the video to be processed, the duration range in which the video to be processed is located can be determined. In this embodiment, this is denoted as the target duration range. Finally, the enhanced difference prediction value corresponding to the target duration range can be taken from the enhanced difference prediction list as the enhanced feedback prediction result of the video to be processed.

[0073] The first optional embodiment of the above technical solution provides a specific implementation of the enhancement feedback prediction result corresponding to the video to be processed. In this optional embodiment, the enhancement difference prediction value matching the video to be processed can be obtained by processing the fused feature vector through the enhancement feedback prediction model and combining it with the target duration range of the video to be processed. Through this optional technical solution, the enhancement feedback prediction result before and after the enhancement processing of the video to be processed can be effectively determined. The enhancement feedback prediction result can better filter out videos that are not suitable for enhancement processing, so as to achieve a better effect of saving enhancement processing computing resources.

[0074] Based on the above optimizations, the enhanced feedback prediction model includes a feature vector processing layer and at least one enhanced difference prediction layer, with each enhanced difference prediction layer corresponding to a time range.

[0075] In this embodiment, as described above, a network structure for feature analysis is constructed in the enhanced feedback prediction model. In this optional embodiment, the network structure can preferably be a feature vector processing layer. It is also known that network structures for enhanced difference prediction are constructed in the enhanced feedback prediction model for different time ranges, and each enhanced difference prediction network structure can be regarded as an enhanced difference prediction layer.

[0076] Based on the above optimizations, the enhanced feedback prediction model constructed from the fused feature vector input can be further optimized to obtain at least one enhanced difference prediction value as follows:

[0077] b11) The fused feature vector is processed by the feature vector processing layer to obtain the difference prediction feature vector.

[0078] In this embodiment, the feature vector processing layer can extract the feature information of interest for video enhancement difference determination from the fused feature vector. In this embodiment, the feature information determined by the feature vector processing layer can be denoted as the difference and prediction feature vector in vector form.

[0079] b12) Input the difference prediction feature vector into each of the enhanced difference prediction layers to obtain the corresponding enhanced difference prediction value.

[0080] In this embodiment, the network structure included in each enhancement difference prediction layer, combined with the trained network parameters, can be used to simulate the calculation of the evaluation index value relative to the set service evaluation index for the video, and can also simulate the difference in the evaluation index value corresponding to the video before and after enhancement processing. The difference value output by the enhancement difference prediction layer in this step can be determined as the enhancement difference prediction value within the corresponding duration range.

[0081] As we know, one enhanced difference prediction layer corresponds to one time range; therefore, multiple time ranges correspond to multiple enhanced difference prediction layers. Thus, different enhanced difference prediction layers can each output their corresponding enhanced difference prediction values.

[0082] The above technical solution provides a specific implementation for calculating the enhancement difference prediction value using an enhanced feedback prediction model. The determined enhancement difference prediction value provides basic data support for whether the subsequent video to be processed should be enhanced using the enhancement processing algorithm corresponding to the enhancement requirements.

[0083] To better understand the implementation of the method provided in this embodiment for determining the enhanced difference prediction value of the video to be processed, this embodiment describes its implementation process through an example. For example, Figure 1b This diagram illustrates the implementation process of determining the enhanced feedback prediction result of the video to be processed in the video processing method provided in the embodiments of this disclosure.

[0084] Specifically, such as Figure 1b As shown, the video to be processed can generate video description text based on preset video attribute information. Through video feature extraction and text feature extraction, the video feature vector and the text feature vector of the video description text can be obtained. The fused feature vector formed by feature fusion processing of the video feature vector and the text feature vector can be used as input information for the enhanced feedback prediction model.

[0085] Specifically, through Figure 1b As can be seen, the enhanced feedback prediction model includes a feature vector processing layer and at least one enhanced difference prediction layer i (i∈[1,N], where i is the i-th duration range). Each enhanced difference prediction layer corresponds to a duration range. The difference prediction feature vector obtained by processing and fusing the feature vector through the feature vector processing layer can be used as the input content of each enhanced difference prediction layer. The difference prediction feature vector F1 is input into each enhanced difference prediction layer to obtain the corresponding enhanced difference prediction value and form an enhanced difference prediction value list. Finally, the playback duration of the video to be processed in the video description text can be combined to obtain the enhanced difference prediction value corresponding to the duration range matched by the playback duration. This enhanced difference prediction value is used as the enhanced feedback prediction result of the video to be processed to determine whether to perform enhanced processing on the video to be processed.

[0086] Based on the above embodiments, the method provided in this embodiment can be further optimized to include the training implementation of the enhanced feedback prediction model, and its implementation steps may include:

[0087] a2) Determine the sample training set and the constructed initial augmentation feedback prediction model, wherein the sample training set includes at least one sample data set, the sample data set containing sample videos and the actual augmentation difference values ​​of the sample videos.

[0088] In this embodiment, the sample training set may include one or more sample data groups. The sample videos in each sample data group can be considered as videos participating in the initial training of the augmented feedback prediction model. These videos may be historical videos with known values ​​for two evaluation metrics. One known evaluation metric is the original metric data determined relative to the business evaluation metrics from the original video without augmentation, and the other is the augmented metric data determined relative to the business evaluation metrics from the augmented video obtained after augmentation.

[0089] As described above, the actual value of the enhancement difference of the sample videos in the sample data group can be considered as the difference between the enhancement index data and the original index data contained in the two known evaluation index values ​​of the sample video.

[0090] In this embodiment, the initial augmented feedback prediction model can be considered as the initially constructed augmented feedback prediction model, in which the network parameters of the network structure have not yet been adjusted through training. The initial augmented feedback prediction model in this embodiment may include an initial feature vector processing layer and at least one initial augmented difference prediction layer.

[0091] Based on the above optimizations, this embodiment provides a specific implementation of the sample training set, which may include:

[0092] a21) Collect a set number of original videos as sample videos.

[0093] The original video can refer to historical videos that have not undergone enhancement processing, and the number of such videos must be at least greater than 1.

[0094] In this embodiment, a set number of historical videos that have not undergone enhancement processing can be collected from a video database storing video data as sample videos.

[0095] a22) For each sample video, obtain the enhanced video by enhancing the sample video through the target enhancement processing algorithm.

[0096] In this embodiment, the actual difference in enhancement between the sample videos can be determined through a control experiment. The videos participating in the control experiment specifically include the original video and the enhanced video after enhancement processing of the original video.

[0097] a23) After the sample video and the enhanced video are distributed, the original indicator data and enhanced indicator data of the sample video and the enhanced video relative to the set business evaluation indicators are statistically analyzed within a set time.

[0098] In this embodiment, to obtain control data for the control experiment, both the sample video and the enhanced video can be distributed to the viewer's client. After a certain period, the service data generated by the original video and the enhanced video on the viewer's side can be statistically analyzed according to a set service evaluation index. For example, assuming the service evaluation index is video viewing time, the viewing time generated by the original video and the enhanced video within a set time period can be statistically analyzed. Thus, the video viewing time calculated relative to the original video can be recorded as the original index data, and the video viewing time calculated relative to the enhanced video can be recorded as the enhanced index data. The set time can be considered as a period of time after the sample video and the enhanced video are distributed.

[0099] In addition, the business evaluation metric can also be the number of video views. In this case, after the sample video and the enhanced video are distributed, the original number of views of the sample video relative to the set business evaluation metric within a set time period can be counted, and this original number of views can be recorded as the original metric data; and the enhanced video's enhanced video's enhanced video's enhanced video views relative to the set business evaluation metric within a set time period can be counted, and this enhanced video's enhanced video views can be recorded as the enhanced metric data. For example, if the set business evaluation metric is the number of video views, and the set time period is 5 minutes, then the original metric data of the sample video relative to the set business evaluation metric within 5 minutes can be counted as 600 times, and the enhanced metric data of the enhanced video relative to the set business evaluation metric within 5 minutes can be counted as 800 times.

[0100] a24) The difference between the enhanced indicator data and the original indicator data is determined as the actual value of the enhancement difference of the sample video.

[0101] The difference in indicator data refers to the difference between the enhanced indicator data and the original indicator data. Specifically, it is the difference between the enhanced indicator data and the original indicator data. For example, following the above description, taking the set business evaluation indicator as the number of video views as an example, in a control experiment with a set time of 5 minutes, the original indicator data can be 600 times, and the enhanced indicator data can be 800 times; the difference of 200 between 800 and 600 can be determined as the actual value of the enhancement difference of the sample video.

[0102] b2) The sample video feature vector and sample text feature vector determined relative to the sample video are fused, and the obtained sample fused feature vector is input into the initial enhancement feedback prediction model to obtain the initial enhancement difference prediction value corresponding to the sample video.

[0103] In this embodiment, the sample video feature vector and sample text feature vector relative to the sample video can be obtained through pre-extraction. In this step, the sample fusion feature vector after fusing the sample video feature vector and sample text feature vector can be used as the initial enhancement feedback prediction model. Based on the network parameters currently available in the initial enhancement feedback prediction model, the initial enhancement difference prediction value output by the initial enhancement feedback prediction model can be obtained.

[0104] c2) By setting a loss function formula, and combining the initial predicted value of the enhancement difference with the actual value of the enhancement difference, the loss function value is determined.

[0105] In this embodiment, the defined loss function can refer to a pre-set loss function, such as a mean squared error loss function; the actual enhancement difference value, as label information, exists in the sample data group where the currently input sample video is located. In this embodiment, the loss function value can refer to the difference between the initial predicted enhancement difference value and the actual enhancement difference value calculated by the defined loss function. This step can calculate the loss function value between the initial predicted enhancement difference value and the actual enhancement difference value based on the pre-set loss function.

[0106] d2) Update the model parameters in the initial augmented feedback prediction model according to the loss function value, and return to perform the initial augmented difference prediction value again until the training termination condition is met, so as to obtain the trained augmented feedback prediction model.

[0107] In this embodiment, the training termination condition can be set to the current calculated loss function value being less than a set threshold, or the training termination condition can also be set to the number of training iterations reaching a certain threshold.

[0108] In this embodiment, the model parameters (network parameters of the network structure in the model) in the initial augmented feedback prediction model can be updated and adjusted according to the loss function value. Then, the initial augmented difference prediction value can be determined again. After that, the model can be updated multiple times until the training termination condition is met, and the trained augmented feedback prediction model is obtained.

[0109] Based on the above description, it can be seen that by obtaining historical videos from the historical database and the actual values ​​of the enhanced differences of the historical videos to form at least one sample data group, a sample training set is formed based on the at least one sample data group, and the model parameters in the initial enhanced feedback prediction model are updated according to a set loss function until the training termination condition is met, a trained enhanced feedback prediction model is obtained. Based on the trained enhanced feedback prediction model, the enhanced feedback prediction results involved in the video to be processed can be effectively determined.

[0110] As a third optional embodiment of this example, the determination that the enhanced feedback prediction result meets the enhancement processing condition can be specified as follows: when the enhanced difference prediction value of the enhanced feedback prediction result is positive, it is determined that the enhanced feedback prediction result meets the enhancement processing condition.

[0111] In this embodiment, taking video quality enhancement as an example, if the enhancement difference prediction value is positive, it means that the video to be processed after enhancement is clearer and has more views than the video to be processed without enhancement, which is equivalent to the video gaining positive benefits after enhancement.

[0112] Following the above description, if the enhancement difference prediction value is negative, it means that although the image quality of the enhanced video is improved compared to the unenhanced video, the viewing experience is negatively affected, potentially leading to a decrease in video views. In this case, not only are the computational resources for enhancement wasted, but a worse viewing experience also occurs. If the enhancement difference prediction value is 0, it means there is no difference between the enhanced video and the unenhanced video. To avoid wasting computational resources, the enhancement process can be cancelled in this situation.

[0113] The above technical solution limits the enhancement difference prediction value, allowing the video to be processed only when the enhancement difference prediction value of the enhancement feedback prediction result is positive. This effectively avoids the negative feedback that may occur after applying the target enhancement processing algorithm, and better ensures the maximization of positive feedback in the business.

[0114] As a fourth optional embodiment of this example, based on the above embodiments, the video processing method provided in this example may further include:

[0115] a4) The target enhanced video after performing video enhancement processing on the video to be processed is used as the video to be sent.

[0116] In this embodiment, taking the execution entity as the server or request-response end as an example, the target enhanced video formed after enhancement processing can be recorded as the video to be sent, and sent to the audience side that has the need to watch the video to be processed, or randomly sent to multiple audience side clients.

[0117] b4) If the enhanced feedback prediction result does not meet the enhanced processing conditions, the video to be processed will be directly used as the video to be sent.

[0118] In this embodiment, as another determination branch of the enhanced feedback prediction result, when the enhanced difference prediction value is negative or 0, it can be considered that the target enhancement processing algorithm is not suitable for enhancing the video to be processed in this case. Therefore, in this case, the video to be processed can be directly sent as a video to be distributed to the viewer-side client that needs the video to be processed through the added step, or it can be randomly distributed to multiple viewer-side clients.

[0119] The above technical solution uses the target enhanced video after video enhancement processing of the video to be processed as the video to be sent. If the enhancement feedback prediction result does not meet the enhancement processing conditions, the processing of the video to be processed is canceled, and the video to be processed can be directly used as the video to be sent. This effectively avoids the negative feedback that may occur in the video to be processed after the target enhancement processing algorithm is applied.

[0120] Figure 2 This is a schematic diagram of a video processing apparatus provided in an embodiment of the present disclosure. This embodiment is applicable to video processing. The apparatus can be implemented by software and / or hardware and can be configured in a terminal and / or server to implement the video processing method in the embodiments of the present disclosure. Specifically, the apparatus may include: an acquisition module 21, a feature extraction module 22, a result prediction module 23, and an enhancement processing module 24.

[0121] The acquisition module 21 is used to acquire the video to be processed and generate video description text based on the preset video attribute information of the video to be processed.

[0122] Feature extraction module 22 is used to extract features from the video to be processed and the video description text, and to fuse the obtained video feature vector and text feature vector to form a fused feature vector;

[0123] The result prediction module 23 is used to determine the enhancement feedback prediction result of the video to be processed based on the fused feature vector. The enhancement feedback prediction result is used to indicate whether the video to be processed should be enhanced.

[0124] The enhancement processing module 24 allows video enhancement processing to be performed on the video to be processed if the enhancement feedback prediction result meets the enhancement processing conditions.

[0125] This embodiment provides a video processing apparatus that, before performing enhancement processing on a video, analyzes the video's features to predict the enhancement feedback prediction result after enhancement processing. Enhancement processing is only permitted if the enhancement feedback prediction result meets the enhancement processing conditions. This embodiment's technical solution is equivalent to a preprocessing scheme before video enhancement processing. By predicting the enhancement feedback prediction result, it can determine in advance whether a video is suitable for enhancement processing, and only enhance suitable videos, thereby reducing the ineffective consumption of enhancement processing computing resources and lowering enhancement processing costs. Furthermore, videos filtered through this embodiment's technical solution can improve the viewer's viewing experience, resulting in better positive feedback effects.

[0126] Furthermore, the result prediction module 23 may include:

[0127] The prediction unit is used to input the fused feature vector into the constructed enhanced feedback prediction model and obtain at least one enhanced difference prediction value, each enhanced difference prediction value corresponding to a time range.

[0128] The determining unit is used to determine the target duration range of the video to be processed, and to determine the enhanced difference prediction value corresponding to the target duration range as the enhanced feedback prediction result.

[0129] Furthermore, the enhanced feedback prediction model includes a feature vector processing layer and at least one enhanced difference prediction layer, with each enhanced difference prediction layer corresponding to a time range;

[0130] Accordingly, the prediction unit can be specifically used for:

[0131] The fused feature vector is processed by the feature vector processing layer to obtain the difference prediction feature vector;

[0132] The difference prediction feature vector is input into each of the enhanced difference prediction layers to obtain the corresponding enhanced difference prediction value.

[0133] Furthermore, the device also includes a model training module, which is specifically used for:

[0134] A sample training set and an initial augmentation feedback prediction model are determined. The sample training set includes at least one sample data set, which contains sample videos and the actual augmentation difference values ​​of the sample videos.

[0135] The sample video feature vector and sample text feature vector determined relative to the sample video are fused, and the obtained sample fused feature vector is input into the initial enhancement feedback prediction model to obtain the initial enhancement difference prediction value corresponding to the sample video.

[0136] The loss function value is determined by setting a loss function formula and combining the initial predicted value of the enhancement difference with the actual value of the enhancement difference;

[0137] The model parameters in the initial augmented feedback prediction model are updated based on the loss function value, and the initial augmented difference prediction value is repeated until the training termination condition is met, thus obtaining the trained augmented feedback prediction model.

[0138] Furthermore, the sample determination unit included in the model training module can be specifically used for:

[0139] A set number of original videos were collected and used as sample videos.

[0140] For each sample video, an enhanced video is obtained by enhancing the sample video using the target enhancement processing algorithm.

[0141] After the sample video and the enhanced video are distributed, the original indicator data and enhanced indicator data of the sample video and the enhanced video relative to the set business evaluation indicators are statistically analyzed within a set time period.

[0142] The difference between the enhanced indicator data and the original indicator data is determined as the actual value of the enhancement difference of the sample video.

[0143] Furthermore, the enhanced processing module 24 is specifically used for:

[0144] When the enhanced difference prediction value, which is the result of the enhanced feedback prediction, is positive, it is determined that the enhanced feedback prediction result satisfies the enhancement processing condition.

[0145] Furthermore, the video processing device further includes: a video distribution and determination module, used for...

[0146] The target enhanced video, after undergoing video enhancement processing on the video to be processed, will be used as the video to be sent out.

[0147] If the enhanced feedback prediction result does not meet the enhanced processing conditions, the video to be processed will be directly used as the video to be sent.

[0148] The above-described apparatus can execute the methods provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the methods.

[0149] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0150] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Reference is made below. Figure 3 It illustrates a computer device suitable for implementing embodiments of the present disclosure (e.g., Figure 3 The diagram below shows the structure of the terminal device or server 30. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0151] like Figure 3 As shown, the computer device 30 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 31, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 32 or a program loaded from a storage device 38 into a random access memory (RAM) 33. The RAM 33 also stores various programs and data required for the operation of the computer device 30. The processing unit 31, the ROM 32, and the RAM 33 are interconnected via a bus 35. An edit / output (I / O) interface 34 is also connected to the bus 35.

[0152] Typically, the following devices can be connected to I / O interface 34: input devices 36 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 37 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 38 including, for example, magnetic tapes, hard disks, etc.; and communication devices 39. Communication device 39 allows computer device 30 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 A computer device 30 with various devices is shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have instead.

[0153] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 39, or installed from a storage device 38, or installed from a ROM 32. When the computer program is executed by the processing device 31, it performs the functions defined in the methods of embodiments of this disclosure.

[0154] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0155] The computer device provided in this embodiment and the video processing method provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0156] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the video processing method provided in the above embodiments.

[0157] It should be noted that the computer-readable medium described above in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0158] In this disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0159] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0160] The aforementioned computer-readable medium may be included in the aforementioned computer device; or it may exist independently and not assembled into the computer device.

[0161] The aforementioned computer-readable medium carries one or more programs that, when executed by the computer device, cause the computer device to:

[0162] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0163] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0164] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0165] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0166] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0167] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0168] Furthermore, although the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0169] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A video processing method, characterized in that, include: Acquire the video to be processed and generate video description text based on the preset video attribute information of the video to be processed; Feature extraction is performed on the video to be processed and the video description text, and the obtained video feature vector and text feature vector are fused to form a fused feature vector; Based on the fused feature vector, the enhancement feedback prediction result of the video to be processed is determined, and the enhancement feedback prediction result is used to indicate whether the video to be processed should be enhanced. If the enhanced feedback prediction result meets the enhancement processing conditions, video enhancement processing is allowed for the video to be processed.

2. The method according to claim 1, characterized in that, The step of determining the enhanced feedback prediction result of the video to be processed based on the fused feature vector includes: The fused feature vector is input into the constructed enhanced feedback prediction model to obtain at least one enhanced difference prediction value, and each enhanced difference prediction value corresponds to a time range. The target duration range of the video to be processed is determined, and the enhanced difference prediction value corresponding to the target duration range is determined as the enhanced feedback prediction result.

3. The method according to claim 2, characterized in that, The enhanced feedback prediction model includes a feature vector processing layer and at least one enhanced difference prediction layer, with each enhanced difference prediction layer corresponding to a time range. The step of inputting the fused feature vector into the constructed enhanced feedback prediction model to obtain at least one enhanced difference prediction value includes: The fused feature vector is processed by the feature vector processing layer to obtain the difference prediction feature vector; The difference prediction feature vector is input into each of the enhanced difference prediction layers to obtain the corresponding enhanced difference prediction value.

4. The method according to claim 2, characterized in that, The training steps of the enhanced feedback prediction model include: Determine the sample training set and construct the initial augmentation feedback prediction model. The sample training set includes at least one sample data set, which contains sample videos and the actual augmentation difference values ​​of the sample videos. The sample video feature vector and sample text feature vector determined relative to the sample video are fused, and the obtained sample fused feature vector is input into the initial enhancement feedback prediction model to obtain the initial enhancement difference prediction value corresponding to the sample video. The loss function value is determined by setting a loss function formula and combining the initial predicted value of the enhancement difference with the actual value of the enhancement difference; The model parameters in the initial augmented feedback prediction model are updated based on the loss function value, and the initial augmented difference prediction value is repeated until the training termination condition is met, thus obtaining the trained augmented feedback prediction model.

5. The method according to claim 4, characterized in that, The steps for determining the sample training set include: A set number of original videos were collected and used as sample videos. For each sample video, an enhanced video is obtained by enhancing the sample video using the target enhancement processing algorithm. After the sample video and the enhanced video are distributed, the original indicator data and enhanced indicator data of the sample video and the enhanced video relative to the set business evaluation indicators are statistically analyzed within a set time. The difference between the enhanced indicator data and the original indicator data is determined as the actual value of the enhancement difference of the sample video.

6. The method according to any one of claims 1-5, characterized in that, The determination step for whether the enhanced feedback prediction result meets the enhancement processing conditions includes: When the enhanced difference prediction value, which is the result of the enhanced feedback prediction, is positive, it is determined that the enhanced feedback prediction result satisfies the enhancement processing condition.

7. The method according to any one of claims 1-5, characterized in that, Also includes: The target enhanced video, after undergoing video enhancement processing on the video to be processed, will be used as the video to be sent out. If the enhanced feedback prediction result does not meet the enhanced processing conditions, the video to be processed will be directly used as the video to be sent.

8. A video processing apparatus, characterized in that, include: The acquisition module is used to acquire the video to be processed and generate video description text based on the preset video attribute information of the video to be processed; The feature extraction module is used to extract features from the video to be processed and the video description text, and to fuse the obtained video feature vector and text feature vector to form a fused feature vector. The result prediction module is used to determine the enhancement feedback prediction result of the video to be processed based on the fused feature vector. The enhancement feedback prediction result is used to indicate whether the video to be processed should be enhanced. The enhancement processing module allows video enhancement processing to be performed on the video to be processed if the enhancement feedback prediction result meets the enhancement processing conditions.

9. A computer device, characterized in that, The computer device includes: One or more processors; a storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the video processing method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the video processing method as described in any one of claims 1-7.

11. A computer program product comprising a computer program that, when executed by a processor, implements the video processing method according to any one of claims 1-7.

Images