CN114693812A

CN114693812A - Video processing method and device

Info

Publication number: CN114693812A
Application number: CN202210311587.2A
Authority: CN
Inventors: 侯芬; 何钧; 张希文
Original assignee: Shanghai Bilibili Technology Co Ltd
Current assignee: Shanghai Bilibili Technology Co Ltd
Priority date: 2022-03-28
Filing date: 2022-03-28
Publication date: 2022-07-01
Also published as: WO2023185175A1

Abstract

The application provides a video processing method and a video processing device, wherein the method comprises the steps of inputting object characteristics of a target object corresponding to a target video into an object characteristic decision model under the condition that the target video is not played, and obtaining a first video decision result; playing the target video under the condition that the first video decision result does not meet a decision condition; determining a second video decision result according to a preset video processing strategy in a target playing time period; and under the condition that the second video decision result meets the decision condition, performing video processing on the target video. The method designs a transcoding decision model and a transcoding decision strategy, a video transcoding decision is made in advance through the transcoding decision model before the video is opened, and the decision is complemented in time through the transcoding decision strategy after the video is opened, so that the timely and effective video transcoding is realized.

Description

Video processing method and device

Technical Field

The present application relates to the field of video processing technologies, and in particular, to a video processing method. The application also relates to a video processing apparatus, a computing device, and a computer-readable storage medium.

Background

Video coding is an important technical means in the video field. The uncoded video may be bulky, causing significant stress on both storage and transmission of the video. Therefore, in the storage and transmission of video, etc., compression of video data is generally achieved by video coding.

However, in terms of decision of whether to encode the video or not, namely transcoding, the currently adopted strategy is simpler, one strategy is to transcode all videos indiscriminately, and the other strategy is to make a transcoding decision through artificial subjective judgment; but both strategies can have the situation of resource waste or insufficient coding.

Disclosure of Invention

In view of this, the present application provides a video processing method. The application also relates to a video processing device, a computing device and a computer readable storage medium, which are used for solving the technical problems of resource waste or insufficient coding in video transcoding in the prior art.

According to a first aspect of embodiments of the present application, there is provided a video processing method, including:

under the condition that a target video is not played, inputting the object characteristics of a target object corresponding to the target video into an object characteristic decision model to obtain a first video decision result;

under the condition that the first video decision result does not meet the decision condition, playing the target video;

determining a second video decision result according to a preset video processing strategy within the target playing time period;

and under the condition that the second video decision result meets the decision condition, performing video processing on the target video.

According to a second aspect of embodiments of the present application, there is provided a video processing apparatus including:

the first result obtaining module is configured to input the object features of the target object corresponding to the target video into an object feature decision model to obtain a first video decision result under the condition that the target video is not played;

the video playing module is configured to play the target video under the condition that the first video decision result does not meet a decision condition;

the second result obtaining module is configured to determine a second video decision result according to a preset video processing strategy in the target playing time period;

a video processing module configured to perform video processing on the target video if the second video decision result satisfies the decision condition.

According to a third aspect of embodiments herein, there is provided a computing device comprising a memory, a processor and computer instructions stored on the memory and executable on the processor, the processor implementing the steps of the video processing method when executing the instructions.

According to a fourth aspect of embodiments of the present application, there is provided a computer-readable storage medium storing computer instructions which, when executed by a processor, implement the steps of the video processing method.

The video processing method comprises the steps that under the condition that a target video is not played, the object characteristics of a target object corresponding to the target video are input into an object characteristic decision model, and a first video decision result is obtained; under the condition that the first video decision result does not meet the decision condition, playing the target video; determining a second video decision result according to a preset video processing strategy in a target playing time period; and under the condition that the second video decision result meets the decision condition, performing video processing on the target video.

Specifically, before the target video is not played, the video processing method makes a video transcoding decision in advance through the object features of the target object corresponding to the target video and a pre-trained object feature decision model; and under the condition that the video transcoding decision is that the video is not transcoded, after the target video is played, carrying out video transcoding decision again through a preset video processing strategy so as to solve resource waste caused by transcoding all the target videos, and under the condition that the video needs transcoding according to the strategy, effectively transcoding the target videos, avoiding the occurrence of insufficient coding and realizing accurate transcoding of the target videos.

Drawings

Fig. 1 is an exemplary illustration of a video processing method in a specific application scenario according to an embodiment of the present application;

fig. 2 is a flowchart of a video processing method according to an embodiment of the present application;

fig. 3 is a process flow diagram of a video processing method applied to a video transcoding scene according to an embodiment of the present application;

fig. 4 is a schematic structural diagram of a video processing apparatus according to an embodiment of the present application;

fig. 5 is a block diagram of a computing device according to an embodiment of the present application.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit of this application and is therefore not limited to the specific implementations disclosed below.

The terminology used in the one or more embodiments of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of the present application. As used in one or more embodiments of the present application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present application refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It will be understood that, although the terms first, second, etc. may be used herein in one or more embodiments of the present application to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first aspect may be termed a second aspect, and, similarly, a second aspect may be termed a first aspect, without departing from the scope of one or more embodiments of the present application. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.

First, the noun terms to which one or more embodiments of the present application relate are explained.

Data mining: the method is based on subjects such as machine learning, pattern recognition, statistics and databases, potential information is mined from data, and decision makers are helped to solve actual problems.

And (3) machine learning: it is a discipline that aims at how to improve the performance of the system itself by means of calculation, using experience, and the main content studied is about the algorithm that generates "models" from data on a computer, i.e. "learning algorithm".

Statistics: statistics is the science about understanding the overall quantitative features and quantitative relationships of objective phenomena. The method is scientific in methodology for recognizing the regularity of objective phenomena quantity by collecting, sorting and analyzing statistical data.

Characteristic engineering: the method is a means for extracting features from original data to the maximum extent for use by models and algorithms, and comprises data preprocessing, feature selection, dimension extension, feature extension and the like.

Video transcoding: and coding the video to realize video data compression. Common encoding standards include JPEG, MJPEG, H264, H265, AV1, and the like.

And (3) transcoding decision: and whether to perform transcoding decision on the video.

Bandwidth gain: the reduction of bandwidth charging brought by video transcoding.

XGboost model: the XGBoost is a method based on a Tree structure and combined with ensemble learning, and the basic Tree structure is a Classification Regression Tree (CART).

In the present application, a video processing method is provided, and the present application relates to a video processing apparatus, a computing device, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.

Referring to fig. 1, fig. 1 is an exemplary illustration of a video processing method in a specific application scenario according to an embodiment of the present application.

The specific application scenario of fig. 1 includes a client 102 and a server 104.

In particular, a user (e.g., an uploader of videos) sends a video to be played to the server 104 via the client 102.

After receiving the video to be played, the server 104 determines a user of the video to be played, obtains a video percentage that is over ten thousand of the video played in all videos of the user, transcodes the video to be played under the condition that the video percentage is greater than a preset percentage threshold, and sends the transcoded video to a client (including but not limited to the client 102) for playing; the preset duty ratio threshold may be calculated according to historical data, and the following embodiment will describe a specific calculation process.

Acquiring data of multiple dimensions of the user, such as fan number, forwarding amount, praise amount and the like, when the video ratio is smaller than a preset ratio threshold; and carrying out data processing on the data of the multiple dimensions to obtain the user characteristics of the user. Inputting the user characteristics of the user into a pre-trained object characteristic decision model to obtain a label corresponding to the user characteristics, and determining a video transcoding decision result according to the label, wherein the video transcoding decision result is not transcoded under the condition that the label is 0; under the condition that the label is 1, the video transcoding decision result is transcoding; and transcoding the video to be played and sending the transcoded video to the client for playing under the condition that the video transcoding decision result is transcoding.

And under the condition that the video decision result is not transcoded, deciding whether the video is transcoded according to a preset decision strategy. Specifically, the video is played, and a video segment is divided every three minutes or five minutes and the like within a preset time period (for example, within 48 hours) after the video is played, and when the playing amount on the next video segment is increased by more than 1000 compared with the playing amount on the previous video segment, the video is transcoded, and the video which has been played is subjected to complementary transcoding; or calculating a playing amount threshold value, namely a heat threshold value, on each video segment according to statistics, transcoding the video under the condition that the playing amounts of three or four continuous video segments exceed the playing amount threshold value, and performing complementary transcoding on the played video; and sending the transcoded video to the client for playing.

If the video is not transcoded after the preset time period is finished after the video is played, acquiring the playing amount, the forwarding amount, the praise amount and the like of the video after the video is played in the preset time period, and performing data processing on the playing amount, the forwarding amount, the praise amount and the like of the video after the video is played to obtain the video playing characteristics of the video after the video is played. Inputting the video playing characteristics into a pre-trained playing characteristic decision model to obtain a label corresponding to the playing characteristics, and determining a video transcoding decision result according to the label, wherein the video transcoding decision result is not transcoded under the condition that the label is 0; under the condition that the label is 1, the video transcoding decision result is transcoding; and then, under the condition that the video transcoding decision result is transcoding, transcoding the video, performing complement transcoding on the already played video, and sending the transcoded video to the client for playing.

The video processing method provided by the embodiment of the application realizes a transcoding decision scheme consisting of two models and a decision strategy sequence by using a machine learning and statistical method, and realizes timely and effective transcoding decision of the video. Specifically, before the video is played, a video transcoding decision is made in advance to trigger transcoding by utilizing a plurality of dimensional data (such as fan number, forwarding amount, praise amount and the like) and video information (such as video playing amount over ten-thousand proportion and the like) of a user to which the video belongs and combining a first machine learning model (namely an object feature decision model); when the video transcoding decision of the first machine learning model is not transcoding, playing the video, and performing transcoding decision by using real-time heat data (such as playing amount on a video segment) in combination with a preset transcoding decision strategy within preset video playing time (such as 48 hours); and under the condition that the video transcoding is not carried out through the preset transcoding decision strategy, the video transcoding decision result is determined by combining the data (such as video playing amount, forwarding amount, praise amount and the like) of the video after being played for several days with the playing characteristic decision model. The available data are different due to the fact that the video is in different stages, the timeliness of video transcoding is sequentially decreased by the two models and the decision strategy, the accuracy of video transcoding is sequentially increased, whether the video is transcoded or not is decided by sequentially executing the two models and the decision strategy, the timeliness of transcoding can be achieved, the accuracy of transcoding can also be achieved, the video can be timely and effectively transcoded, and the maximization of the difference value between the income and the cost is achieved.

Referring to fig. 2, fig. 2 is a flowchart illustrating a video processing method according to an embodiment of the present application, which specifically includes the following steps:

step 202: and under the condition that the target video is not played, inputting the object characteristics of the target object corresponding to the target video into an object characteristic decision model to obtain a first video decision result.

The target video may be understood as the video to be played, and may be any type, any duration, and any format of video, such as sports video, entertainment video, movie with duration of two hours, and the like.

The target object corresponding to the target video can be understood as an uploader of the target video; the object feature of the target object may be understood as an object feature formed by performing data processing on object attribute information of the target object, where the object attribute information includes, but is not limited to, fan number, forwarding amount, praise amount, and the like.

In practical application, some object attribute information cannot be directly used, for example, a video partition only has one partition number, and the partition number itself has no meaning, but the video partition itself belongs to a hot partition or a cold partition; if the partition number is directly used as the object feature, the video partition has no meaning, so that some data processing is performed on the video partition, and a score is configured for the video partition to indicate that the video partition is a hot partition or a cold partition, so that a practical object feature is formed.

Therefore, after the object attribute information of the target object is obtained, data processing is performed on the object attribute information to determine the object feature of the target object, and then the first video decision result can be quickly and accurately obtained according to the object feature in combination with the object feature decision model. The specific implementation mode is as follows:

the step of inputting the object characteristics of the target object corresponding to the target video into the object characteristic decision model to obtain a first video decision result includes:

acquiring object attribute information of a target object corresponding to the target video;

performing data processing on the object attribute information to obtain object characteristics of the target object;

and inputting the object characteristics of the target object into an object characteristic decision model to obtain a first video decision result.

For detailed description of the object attribute information of the target object, reference may be made to the above embodiments, which are not described herein again; the object feature decision model includes, but is not limited to, the XGBoost model.

Performing data processing on the object attribute information to obtain object characteristics of the target object; it can be understood that if some object attribute information is directly used as an object feature, the use effect is not obvious, such as the partition number of the video partition, etc., then the object attribute information of the video partition may be processed and processed, for example, a metric value capable of distinguishing cold and hot conditions is set for the video partition, so that the object attribute information of the video partition becomes a meaningful object feature.

When the method is used specifically, the object characteristics of a target object are input into a pre-trained object characteristic decision model to obtain a first video decision result of the target video; in practical application, the specific application scenes of the video processing method are different, and the first video decision result is also different, for example, if the video processing method is applied to a video transcoding scene, the first video decision result can be video transcoding or video non-transcoding; when the video processing method is applied to a video fast-forward scene, the first video decision result can be video fast-forward or video not fast-forward and the like.

For convenience of understanding, the following embodiments are described by taking the application of the video processing method to a video transcoding scene as an example, but the application of the video processing method to other realizable scenes, such as the video fast-forwarding scene described above, is not limited.

In addition, before the object feature decision model is used for deciding whether the video is transcoded or not, the object feature decision model needs to be obtained through pre-training so as to ensure the rapidness and the accuracy of video transcoding decisions of subsequent applications. The specific implementation mode is as follows:

the training steps of the object feature decision model are as follows:

obtaining sample videos, and determining a sample object corresponding to each sample video and a video playing amount;

determining a training sample according to the object attribute information of the sample object;

determining a sample label corresponding to the training sample according to the video playing amount;

and training the object feature decision model according to the training samples and the sample labels.

The sample video can be understood as a plurality of videos which are played in history and have any type, any playing time and any format; the sample object of the sample video can be understood as an uploader of the sample video; the video playing amount of the sample video can be understood as how many times the sample video is played, that is, how many times the sample video is watched; the object attribute information of the sample object is the same as the object attribute information of the target object in the above embodiment, and may be understood as the number of fans, forwarding amount, praise amount, and the like.

In specific implementation, a plurality of sample videos are obtained, and then a sample object and a video playing amount corresponding to each sample video are determined; determining training samples according to the object attribute information of the sample object, and determining a sample label corresponding to each training sample according to the video playing amount; and training the object characteristic decision model according to the training samples and the training labels.

Determining a sample label corresponding to a training sample according to the video playing amount, wherein the sample label is set to 1 when the video playing amount is greater than or equal to a preset playing amount threshold (for example, 500 samples), and determining the training sample corresponding to the sample label to be a positive sample; and under the condition that the video playing amount is smaller than the preset playing amount threshold value, setting the sample label to be 0, and determining that the training sample corresponding to the sample label is a negative sample.

For example, if the video playback amount corresponding to a sample video is 1000, and if the preset playback amount threshold is 500, it may be determined that the object attribute information of the sample object corresponding to the sample video is a training sample, and the sample label determined for the training sample is 1 by using the video playback amount corresponding to the sample video, that is, the training sample is a positive sample.

In practical application, some invalid information may exist in the object attribute information of the sample object, and in order to improve the training effect of the object feature decision model, the object attribute information of the sample object is subjected to data processing to realize feature construction; therefore, the follow-up reasonable object features can be used as training samples based on the constructed reasonable object features, the object feature decision model can be trained, and the model accuracy is improved. The specific implementation mode is as follows:

the determining a training sample according to the object attribute information of the sample object includes:

carrying out data processing on the object attribute information of the sample object to obtain object characteristics of the sample object;

determining object features of the sample object as a training sample.

For a specific implementation manner of performing data processing on the attribute information of the sample object to obtain the object feature of the sample object, reference may be made to the detailed description of the above embodiments, which is not described herein again.

Specifically, after the object attribute information of the sample object is obtained, data processing is performed on the attribute information of the sample object to obtain object features of the sample object; subsequently, the object features of the sample object can be used as training samples, and the training of the object feature decision model is realized by combining the sample labels of the training samples determined according to the video playing amount, so that the use effect of the object feature decision model is improved.

Before the object feature decision model is used, if the video playing amount uploaded by the target object history corresponding to the target video is relatively high, the probability that the target video is played in a large amount after being uploaded is very high, and therefore before the target video is not played, whether the video needs transcoding or not can be determined according to the historical video data of the target object corresponding to the target video and the current playing situation of other videos, so that the situation that the watching experience is influenced due to untimely transcoding under the situation that the target video is played in a large amount after being online is avoided.

For example, from the training samples, a positive sample video is determined, then historical data of an object of the positive sample video under a certain characteristic is determined, and a characteristic threshold value under the characteristic is determined; in particular use, the feature of the target object of the target video may be obtained, and the feature is compared with a feature threshold obtained under a positive sample to determine whether the target video needs to be transcoded. Specifically, the specific manner of obtaining the feature threshold is as follows:

after the sample label corresponding to the training sample is determined according to the video playing amount, the method further comprises the following steps:

determining a positive sample video in the sample videos according to the sample label;

determining a positive sample object corresponding to the positive sample video, and determining a target feature according to historical video data of the positive sample object;

and determining a corresponding characteristic threshold according to the target characteristic.

The target characteristics can be any characteristics, such as the number of praise at video points is over ten thousand, or the number of play at video points is over ten thousand; determining the target characteristics according to historical video data of the positive sample object under the condition that the target characteristics are that the video praise amount is over ten thousand, wherein the target characteristics can be understood as obtaining the praise amount of each video in all videos historically uploaded by the positive sample object, and determining the video praise amount of all video points of the positive sample object according to the praise amount of each video, namely the video praise amount is over ten thousand; when the target feature is that the video playing amount is over ten thousand percent, the target feature is determined according to the historical video data of the positive sample object, and it can be understood that the playing amount of each video in all videos historically uploaded by the positive sample object is obtained, and the video playing amount of the positive sample object is over ten thousand percent, that is, the video playing amount is over ten thousand percent, is determined according to the playing amount of each video.

Specifically, taking a target feature as an example of the video playing amount over ten thousand, determining a corresponding feature threshold according to the target feature, which can be understood as obtaining the video playing amount over ten thousand of all positive sample objects, and determining statistics such as a mean value, a minimum value, a maximum value and the like; and training the weight of each statistic, and finally weighting and summing the statistics to obtain a final threshold (namely a characteristic threshold). In practical application, the training of the weight is to traverse the parameter combination of the weight, test a group of weight values finally obtained on a test set, for example, the weight of the minimum value and the average value is to be trained, and if the step length is selected to be 0.1, the parameter combination is 0.1 and 0.9; 0.2, 0.8; 0.3, 0.7, etc.

In the embodiment of the description, the occupation ratio that the video playing amount of an uploader of a currently known transcoded video is over ten thousand in all videos in history or the occupation ratio that the video praise amount of all videos in history is over ten thousand is obtained; subsequently, calculating a corresponding characteristic threshold according to the occupation ratio of ten thousand of video playing amount of each uploader or the occupation ratio of ten thousand of video praise of each uploader; when whether the target video is transcoded is judged subsequently, whether the target video needs to be transcoded or not can be quickly determined according to the comparison between the video playing amount of the uploader corresponding to the target video with the ten-thousand ratio or the video approval amount with the corresponding characteristic threshold value. The specific implementation mode is as follows:

before the inputting the object feature of the target object into the object feature decision model and obtaining the first video decision result, the method further includes:

determining a target feature of a target object corresponding to the target video and a feature value of the target feature;

obtaining a fourth video decision result according to the association relation between the characteristic value of the target characteristic and the characteristic threshold value;

and under the condition that the fourth video decision result meets a decision condition, performing video processing on the target video according to the fourth video decision result.

The target feature of the target object is the same as the target feature obtained by the feature threshold, for example, if the obtained feature threshold is a feature threshold in which the video playing amount is over ten thousand, the target feature is the video playing amount is over ten thousand; and the characteristic value of the target characteristic is a specific ratio.

Specifically, under the condition that a target video is not played, acquiring a target feature of a target object corresponding to the target video and a feature value of the target feature, namely, a video playing amount over ten thousand percent and a ratio; comparing the ratio of the video amount exceeding ten thousand percent with the obtained characteristic threshold value, and obtaining a fourth video decision result according to a specific comparison result, namely transcoding or not transcoding the video; finally, under the condition that the fourth video decision result is determined to be transcoding, determining that the fourth video decision result meets decision conditions; at this time, the target video may be transcoded according to the fourth video decision result.

In practical application, a fourth video decision result is obtained according to the correlation between the feature value of the target feature and the feature threshold, wherein the fourth video decision result is determined as video transcoding when the feature value of the target feature is compared with the feature threshold and is greater than or equal to the feature threshold; and under the condition that the characteristic value of the target characteristic is smaller than the characteristic threshold value, determining that the fourth video decision result is that the video is not transcoded.

For example, if the feature threshold is 70%, the percentage of the video amount that exceeds the ten-thousand ratio is 75%, and the percentage of the video amount that exceeds the ten-thousand ratio is 75% compared with the feature threshold 70%, it may be determined that the percentage of the video amount that exceeds the ten-thousand ratio is greater than the feature threshold, and at this time, it may be determined that the fourth video decision result is video transcoding.

In the embodiment of the present specification, in order to ensure timeliness of transcoding a target video, before transcoding the target video, a comparison may be performed between a target feature of a target object corresponding to the target video and a feature threshold calculated according to a target feature of a historical transcoded video, so as to quickly make a decision as to whether the target video is transcoded, so as to realize that, when a historical video playing amount of the target object is too large, a probability that the target video uploaded by the target object will be subsequently played in a large amount is higher by default, so that transcoding playing may be performed directly, and timeliness of transcoding and subsequent video playing profits are improved.

Step 204: and under the condition that the first video decision result does not meet the decision condition, playing the target video.

Specifically, the video processing method provided in the embodiment of the present specification has different specific application scenarios and different first video decision results, and determines whether the first video decision result satisfies the decision condition or not; for example, when the video processing method is applied to a video transcoding scene, the first video decision result may be understood as video transcoding introduced in the above embodiment or video not to be transcoded, and the corresponding video decision condition may be understood as a video transcoding condition; if the video processing method is applied to a video recommendation scene, the first video decision result can be understood as video recommendation or video non-recommendation, and the corresponding video decision condition can be understood as a video recommendation condition, etc.

Then, when the first video decision result is that the video is not transcoded, it may be determined that the first video decision result does not satisfy the decision condition; at this time, the original video data of the target video is played without transcoding.

When the first video decision result is video transcoding, it can be determined that the first video decision result satisfies the decision condition; at this time, the target video is directly transcoded and the transcoded target video is played.

Step 206: and determining a second video decision result according to a preset video processing strategy in the target playing time period.

The target playing time period can be set according to actual application, which is not limited in the embodiment of the present application; for example, the target playing period is set to 48 hours or 50 hours.

In addition, the preset video processing strategy can also be set according to practical application, for example, the video can be sliced, and whether transcoding is required to be performed on the video is determined according to the playing amount of the sliced video; and determining whether the video needs to be transcoded according to the overall playing amount or the praise amount of all the played videos in the target playing time period.

In the embodiment of the application, the video is segmented, whether transcoding of the video is needed is determined according to the playing amount of the segmented video, the video is used as a preset video processing strategy, and a second video decision result determined according to the preset video processing strategy in a target playing time period is introduced in detail. The specific implementation mode is as follows:

the determining a second video decision result according to a preset video processing strategy includes:

and obtaining at least two video clips according to a preset division rule, and determining a second video decision result according to the playing amount of the at least two video clips.

The preset division rule can be set according to practical application, and the embodiment of the application is not limited at all. For example, the preset division rule is to divide one video segment every three minutes, or divide one video segment every five minutes, etc.

For convenience of understanding, the following embodiments are specifically described by taking an example in which a preset division rule is used to divide a video clip every three minutes.

In specific implementation, the target video is played, one video segment is divided every three minutes, and under the condition that at least two video segments are divided, a second video decision result is determined according to the playing amount of the at least two divided video segments.

According to the video processing method provided by the embodiment of the application, under the condition that the target video starts to be played, at least two video segments can be obtained according to the preset division rule, and the second video decision result can be rapidly determined according to the playing amount of the divided at least two video segments, so that the transcoding timeliness of the target video is guaranteed.

After the at least two video segments are divided, the specific implementation modes of determining the second video decision result include at least two modes according to the playing amounts of the divided at least two video segments, one mode can compare the playing amounts of two adjacent video segments, and the second video decision result is quickly determined according to the growth condition of the playing amounts. The specific implementation mode is as follows:

the determining a second video decision result according to the playing amounts of the at least two video segments includes:

determining the play amount difference value of any two adjacent video clips in the at least two video clips;

and determining a second video decision result according to the incidence relation between the playing quantity difference value and the difference threshold value.

The difference threshold may be set according to practical applications, for example, the difference threshold is 500 or 1000.

If the at least two video segments include video segment 1, video segment 2, and video segment 3, any two adjacent video segments may be understood as video segment 1 and video segment 2, or video segment 2 and video segment 3.

Calculating the play amount difference value of any two adjacent video clips after determining any two adjacent video clips; for example, any two adjacent video segments include a video segment 1 and a video segment 2, where the playing amount of the video segment 1 is 100, the playing amount of the video segment 2 is 1100, and then the difference between the playing amounts of the video segment 1 and the video segment 2 is 1000; and finally, according to the incidence relation between the playing quantity difference value and a preset difference value threshold value, quickly determining a second video decision result so as to ensure the timeliness of video transcoding.

In specific implementation, the determining a second video decision result according to the association relationship between the playback quantity difference and the difference threshold includes:

determining a second video decision result as video transcoding under the condition that the difference value of the playing amount is greater than or equal to the difference value threshold; or alternatively

And under the condition that the difference value of the playing amount is smaller than the difference value threshold value, determining that the second video decision result is that the video is not transcoded.

Following the above example, taking the difference of the playback amounts as 1000 and the difference threshold as 1000 as an example, it may be determined that the difference of the playback amounts 1000 is equal to the difference threshold 1000, and then it may be determined that the second video decision result is video transcoding; if the difference of the playing amounts is 900 and the difference threshold is 1000, it may be determined that the difference of the playing amounts 900 is smaller than the difference threshold 1000, and it may be determined that the second video decision result is that the video is not transcoded.

In practical application, a target video may start to be played, that is, a video segment is divided every three minutes, and the current playing amount of the video segment is obtained, after a second video segment is divided and the current playing amount of the video segment is obtained, if it is determined that the increase amplitude of the second video segment is larger than that of the first video segment, video transcoding may be performed on the target video; by analogy, real-time playing and timely video transcoding judgment can be achieved.

According to the video processing method provided by the embodiment of the application, the second video decision result can be quickly and accurately determined according to the incidence relation between the playing quantity difference value and the difference threshold value of any two adjacent video clips in the playing process of the target video; that is, under the condition that the playing amount of the latter video segment is increased greatly compared with the playing amount of the former video segment, it can be determined that the probability that the target video is played in a large amount is high, and at this time, the target video can be subjected to video transcoding.

Alternatively, the playing amount of a preset number of consecutive video segments may be compared with the calculated heat threshold, and the second video decision result may be quickly determined according to the relationship between the playing amount and the heat threshold. The specific implementation mode is as follows:

determining the playing amount of each video clip in the at least two video clips;

determining the heat threshold of the at least two video clips according to the playing amount of each video clip;

and determining a second video decision result according to the playing amount of each of the at least two video clips and the heat threshold.

The at least two video segments in the embodiment of the present application may be understood as all video segments divided according to a preset division rule within a target playing time period.

After the playing amount of each video clip is obtained, the heat threshold value for the target playing time period can be calculated according to the playing amounts of all the video clips; specifically, the calculation method of the heat threshold is the same as that of the above feature threshold, and is not described herein again. For example, the minimum value, the average value, and the like of the playing amount of the video clip in the target playing time period are obtained, and the final heat threshold is obtained by calculating according to the above manner.

After the heat thresholds of at least two video segments are determined according to the playing amount of each video segment, the second video decision result can be rapidly determined according to the playing amounts of the at least two video segments and the heat thresholds. The specific implementation mode is as follows:

determining a second video decision result according to the playing amount of each of the at least two video segments and the heat threshold, including:

acquiring a preset number of continuous video clips from the at least two video clips;

determining the playing amount of each video clip in the continuous video clips;

and determining a second video decision result according to the playing quantity of each video clip in the continuous video clips and the incidence relation of the heat threshold.

The preset number may be set according to practical applications, for example, the preset number may be set to 2, 3, or 4.

Taking the preset number as 2 as an example, in specific implementation, 2 continuous video segments are obtained from at least two video segments, and the playing amount of each video segment in the two continuous video segments is determined; and determining a second video decision result according to the association relation between the playing amount of each of the two continuous video clips and the heat threshold.

Still continuing the above example, two consecutive video segments include video segment 1 and video segment 2, where the playing amount of video segment 1 is 100, the playing amount of video segment 2 is 1100, and the heat threshold is 800, and then the second video decision result can be quickly obtained according to the association relationship between the playing amount of video segment 1100 and the playing amount of video segment 2 1100 and the heat threshold 800, respectively.

The specific implementation manner of determining the second video decision result according to the playing amount of each video clip in the continuous video clips and the heat threshold is as follows:

the determining a second video decision result according to the play amount of each video clip in the continuous video clips and the incidence relation of the heat threshold includes:

determining a second video decision result as video transcoding under the condition that the playing amount of each video segment in the continuous video segments is greater than or equal to a heat threshold; or

And under the condition that the playing amount of any one of the continuous video clips is smaller than the heat threshold, determining that the second video decision result is that the video is not transcoded.

Still continuing the above example, if the playing volume of the video segment 1 is 1000, the playing volume of the video segment 2 is 1500, and the heat threshold is 800;

then, comparing the playing amount of each video segment in the continuous video segments with the heat threshold, it may be determined that the playing amount 1000 of the video segment 1 in the continuous video segments and the playing amount 1500 of the video segment 2 are both greater than or equal to the heat threshold 800, and at this time, it may be determined that the second video decision result is video transcoding. If any or all of the video segments 1 and 2 are smaller than the heat threshold 800, it may be determined that the second video decision result is that the video is not transcoded.

Step 208: and under the condition that the second video decision result meets the decision condition, performing video processing on the target video.

In specific implementation, the video processing method provided by the embodiment of the application has different specific application scenes, and has different processing contents for performing video processing on the target video; for example, when the video processing method provided in the embodiment of the present application is applied to a video transcoding scene, and when the second video decision result satisfies the decision condition, performing video processing on the target video includes:

and under the condition that the second video decision result is video transcoding, determining that the second video decision result meets the decision condition, and transcoding the target video.

If the decision condition is the video transcoding condition, after the second video decision result is determined in the above manner, if the second video decision result is video transcoding, it can be determined that the second video decision result meets the video transcoding condition, and at this time, video transcoding can be performed on the target video.

According to the video processing method provided by the embodiment of the application, before the target video is not played, a video transcoding decision is made in advance through the object features of the target object corresponding to the target video and the combination of a pre-trained object feature decision model; and under the condition that the video transcoding decision is that the video is not transcoded, after the target video is played, carrying out video transcoding decision again through a preset video processing strategy so as to solve resource waste caused by transcoding all the target videos, and under the condition that the video needs transcoding according to the strategy, effectively transcoding the target videos, avoiding the occurrence of insufficient coding and realizing accurate transcoding of the target videos.

And if the second video decision result does not meet the decision condition, judging whether the target video has transcoding value or not again according to the playing condition of the played target video after the target video is played for a period of time and judging whether video transcoding is needed or not in order to maximize the difference value between the bandwidth benefit and the computational power and the storage cost under the condition of limited resources. The specific implementation mode is as follows:

after determining the second video decision result according to the preset video processing strategy, the method further includes:

when the target playing time period is over and the second video decision result does not meet the decision condition, acquiring video playing characteristics of the target video within preset playing time;

inputting the video playing characteristics into a playing characteristic decision model to obtain a third video decision result;

and under the condition that the third video decision result meets the decision condition, performing video processing on the target video.

The preset playing time can be set according to time, and the preset playing time is greater than or equal to the target playing time period. For example, the target playing time period is 48 hours, and the preset playing time may be 50 hours or 60 hours. The playing characteristic decision model is a pre-trained machine learning model, including but not limited to an XGBoost model, and the specific training process can be referred to the following detailed description of the playing characteristic decision model training.

Taking a target playing time period of 48 hours and a preset playing time of 50 hours as an example, when the target playing time period is over and the second video decision result does not satisfy the decision condition, the video playing characteristics of the target video within the preset playing time are obtained, which can be understood as that, when the second video decision result obtained by the above arbitrary preset video processing strategy still does not satisfy the decision condition within 48 hours of playing the target video, the target video is continuously played, and when the target video is played for 50 hours, the video playing characteristics of the target video within 50 hours of playing are obtained.

In order to ensure the accuracy of video playing characteristics and the rapid identification of a subsequent playing characteristic decision model; the video playing characteristic is obtained by processing data according to the obtained video playing attribute information of the target video within the preset playing time period. The specific implementation mode is as follows:

the acquiring the video playing characteristics of the target video within the preset playing time period includes:

acquiring video playing attribute information of the target video within the preset playing time period;

and performing data processing on the video playing attribute information to obtain the video playing characteristics of the target video.

The video playing attribute information includes, but is not limited to, a playing amount, a forwarding amount, a praise amount, and the like after the target video is played.

In practical applications, the specific processing manner for performing data processing on the video playing attribute information to obtain the video playing characteristics of the target video is the same as the specific processing manner for performing data processing on the object attribute information to obtain the object characteristics of the target object, and is not described herein again.

Specifically, after the video playing characteristics of the target video are obtained, the video playing characteristics can be input into the playing characteristic decision model to obtain a third video decision result; under the condition that the third video decision result meets the decision condition, video processing can be carried out on the target video; such as transcoding video.

In practical application, since the video decision results obtained by presetting the video processing strategy and the playing characteristic decision model are all after the target video is played, if the obtained video decision result is video transcoding, in order to avoid missing video transcoding, under the condition that the target video which is not played is transcoded according to the video decision result, the target video which is played before the video transcoding is determined is subjected to complementary transcoding.

In this embodiment of the present specification, in order to maximize a difference between a bandwidth benefit and a computational power and a storage cost under a condition that a resource is limited under a condition that a first video decision result obtained through the object feature decision model and a second video decision result obtained according to a preset video processing policy do not satisfy a video transcoding condition, after a target video is played for a period of time, whether the target video has a transcoding value or not is judged again according to a playing condition of the played target video, and whether video transcoding is required or not is determined again, so as to ensure that accurate and timely transcoding of the target video can be achieved.

Before the play characteristic decision model is used, the play characteristic decision model is trained in advance so as to improve the accuracy and the effectiveness of the result of the play characteristic decision model. The specific training mode of the play characteristic decision model is as follows:

the training steps of the play characteristic decision model are as follows:

acquiring sample videos, and determining video playing attribute information corresponding to each sample video;

determining a training sample according to video playing attribute information corresponding to the sample video;

determining a sample label corresponding to the training sample according to the video playing amount in the video playing attribute information;

and training the play characteristic decision model according to the training sample and the sample label.

For specific introduction of the sample videos and the video playing attribute information corresponding to each sample video, reference may be made to the above embodiments.

In addition, in the training of the play characteristic decision model, in order to improve the training effect of the play characteristic decision model, data processing is also performed on the video play attribute information to obtain the video play characteristics of the standard sample video, and the play characteristic decision model is trained. The specific implementation mode is as follows:

the determining a training sample according to the video playing attribute information corresponding to the sample video includes:

performing data processing on video playing attribute information corresponding to the sample video to obtain video playing characteristics of the sample video;

and determining the video playing characteristics of the sample video as a training sample.

In specific implementation, for a specific training process of the playing feature decision model and a data processing process of the video playing attribute information corresponding to the sample video, reference may be made to the specific implementation process of the object feature decision model in the above embodiments, which is not described herein again.

According to the video processing method provided by the embodiment of the application, before the target video is not played, a video transcoding decision is made in advance through the object characteristics of the target object corresponding to the target video and a pre-trained object characteristic decision model; when the video transcoding decision is that the video is not transcoded, after the target video is played, the video transcoding decision can be determined again through a preset video processing strategy within a target playing time period, and when the video transcoding decision determined according to the preset video processing strategy is that the video is not transcoded, the video transcoding decision can be further determined by combining a playing characteristic decision model after the target video is played for a period of time according to the playing condition of the played video; the method and the device aim to solve resource waste caused by transcoding all target videos, effectively transcode the target videos under the condition that the videos are determined to be transcoded according to the strategy, avoid the condition of insufficient coding and realize accurate transcoding of the target videos.

The following description further describes the video processing method with reference to fig. 3, by taking an application of the video processing method provided in the present application in a video transcoding scene as an example. Fig. 3 shows a processing flow chart of a video processing method applied to a video transcoding scene, which specifically includes the following steps:

step 302: and acquiring the video to be transcoded.

Step 304: and determining an uploader of the video to be transcoded, and acquiring the historical video playing amount of the uploader.

Step 306: and determining that the video playing amount of the uploader exceeds the ten-thousand occupation ratio according to the historical video playing amount of the uploader.

Step 308: and judging whether the video playing amount of the uploader is larger than or equal to a preset occupation ratio threshold value or not, if so, executing the step 310, and otherwise, executing the step 312.

The preset ratio threshold is the same as the characteristic threshold in the above embodiments, and is not described herein again.

Step 310: and (6) transcoding the video.

Step 312: and acquiring the object characteristics of the uploader, inputting the object characteristics of the uploader into a first machine learning model, and acquiring a first video transcoding result of the video to be transcoded.

Step 314: determining whether the first video transcoding result satisfies a transcoding condition, if yes, performing step 310, and if not, performing step 316.

The method for acquiring the object features of the uploader is the same as the method for acquiring the object features of the target object in the embodiment; the first machine learning model may be understood as the object feature decision model in the above embodiments.

Step 316: and playing the video to be transcoded, and determining a second video transcoding result according to a preset video transcoding strategy within a target playing time period.

The preset video transcoding policy may be understood as the preset video processing policy in the above embodiment.

Step 318: determining whether the second video transcoding result satisfies a transcoding condition, if yes, performing step 320, and if not, performing step 322.

Step 320: and transcoding the video and complementing the transcoding.

Specifically, when the second video transcoding result meets the transcoding condition, that is, video transcoding is required, the video to be transcoded which is not played is transcoded, and the video to be transcoded which is played before is complemented and transcoded.

Step 322: and in the preset playing time period of playing the video to be transcoded, acquiring the video playing characteristics of the video to be transcoded in the preset playing time period, and inputting the video playing characteristics into a second machine learning model to acquire a third video transcoding result.

The method for acquiring the video playing characteristics of the video to be transcoded is the same as the method for acquiring the video playing characteristics of the target video in the embodiment; the second machine learning model may be understood as the play feature decision model in the above embodiments.

Step 324: and judging whether the third video transcoding result meets the transcoding condition, if so, executing the step 320, and if not, ending the process.

According to the video processing method provided by the embodiment of the application, historical data related to the video is analyzed through a machine learning model and a statistical method, whether the video has transcoding value or not is accurately judged, effective transcoding of the video is achieved, and the difference value of bandwidth income, computing power and storage cost is maximized under the condition that resources are limited. Specifically, a plurality of data sources are utilized, data analysis is carried out through a machine learning model and a statistical method, a transcoding decision strategy and a transcoding decision model (a first machine learning model) are designed, a video transcoding decision is made before a video is opened, video transcoding is complemented in time through another transcoding decision strategy and a transcoding decision model after the video is opened (played), timely and effective video transcoding is achieved, resources can be prevented from being wasted by unnecessary video transcoding, videos with transcoding values are identified as soon as possible, and finally the difference value of income and cost is maximized. The video processing method provided by the application can be used for carrying out data analysis through a machine learning model and a statistical method, designing a plurality of transcoding decision strategies and transcoding decision models, making a video transcoding decision before the video is opened, and supplementing the decision in time after the video is opened, so that timely and effective video transcoding is realized.

Corresponding to the above method embodiment, the present application further provides an embodiment of a video processing apparatus, and fig. 4 shows a schematic structural diagram of a video processing apparatus provided in an embodiment of the present application. As shown in fig. 4, the apparatus includes:

a first result obtaining module 402, configured to, when a target video is not played, input an object feature of a target object corresponding to the target video into an object feature decision model to obtain a first video decision result;

a video playing module 404 configured to play the target video if the first video decision result does not satisfy a decision condition;

a second result obtaining module 406, configured to determine a second video decision result according to a preset video processing policy within the target playing time period;

a video processing module 408 configured to perform video processing on the target video if the second video decision result satisfies the decision condition.

Optionally, the apparatus further comprises:

a third result obtaining module configured to:

when the target playing time period is finished and the second video decision result does not meet the decision condition, acquiring video playing characteristics of the target video within a preset playing time period;

Optionally, the third result obtaining module is further configured to:

Optionally, the second result obtaining module 406 is further configured to:

determining a second video decision result as video transcoding under the condition that the difference value of the playing amount is greater than or equal to the difference value threshold; or

Optionally, the second result obtaining module 406 is further configured to:

Optionally, the video processing module 408 is further configured to:

Optionally, the apparatus further comprises:

a first model training module configured to: training the object feature decision model;

the training steps of the object feature decision model are as follows:

acquiring sample videos, and determining a sample object corresponding to each sample video and a video playing amount;

Optionally, the first model training module is further configured to:

performing data processing on the object attribute information of the sample object to obtain object characteristics of the sample object;

determining object features of the sample object as a training sample.

Optionally, the apparatus further comprises:

a feature threshold acquisition module configured to:

Optionally, the apparatus further comprises:

a fourth result obtaining module configured to:

Optionally, the first result obtaining module 402 is further configured to:

Optionally, the apparatus further comprises:

a second model training module configured to: training the play characteristic decision model;

the training steps of the play characteristic decision model are as follows:

Optionally, the second model training module is further configured to:

According to the video processing device provided by the embodiment of the application, before the target video is not played, a video transcoding decision is made in advance through the object features of the target object corresponding to the target video and the combination of a pre-trained object feature decision model; and under the condition that the video transcoding decision is that the video is not transcoded, after the target video is played, carrying out video transcoding decision again through a preset video processing strategy so as to solve resource waste caused by transcoding all the target videos, and under the condition that the video needs transcoding according to the strategy, effectively transcoding the target videos, avoiding the occurrence of insufficient coding and realizing accurate transcoding of the target videos.

The above is a schematic scheme of a video processing apparatus of the present embodiment. It should be noted that the technical solution of the video processing apparatus belongs to the same concept as the technical solution of the video processing method, and details that are not described in detail in the technical solution of the video processing apparatus can be referred to the description of the technical solution of the video processing method.

FIG. 5 illustrates a block diagram of a computing device 500 provided in accordance with one embodiment of the present description. The components of the computing device 500 include, but are not limited to, a memory 510 and a processor 520. Processor 520 is coupled to memory 510 via bus 530, and database 550 is used to store data.

Computing device 500 also includes access device 540, access device 540 enabling computing device 500 to communicate via one or more networks 560. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. The access device 540 may include one or more of any type of network interface, e.g., a Network Interface Card (NIC), wired or wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the above-described components of computing device 500, as well as other components not shown in FIG. 5, may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device architecture shown in FIG. 5 is for purposes of example only and is not limiting as to the scope of the present description. Other components may be added or replaced as desired by those skilled in the art.

Computing device 500 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), mobile phone (e.g., smartphone), wearable computing device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 500 may also be a mobile or stationary server.

Wherein the processor 520 implements the steps of the video processing method when executing the instructions.

The foregoing is a schematic diagram of a computing device of the present embodiment. It should be noted that the technical solution of the computing device and the technical solution of the video processing method belong to the same concept, and details that are not described in detail in the technical solution of the computing device can be referred to the description of the technical solution of the video processing method.

An embodiment of the present application further provides a computer readable storage medium, which stores computer instructions, and when the instructions are executed by a processor, the instructions implement the steps of the video processing method as described above.

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium belongs to the same concept as the technical solution of the above-mentioned video processing method, and details that are not described in detail in the technical solution of the storage medium can be referred to the description of the technical solution of the above-mentioned video processing method.

The foregoing description of specific embodiments of the present application has been presented. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, etc. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.

It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

The preferred embodiments of the present application disclosed above are intended only to aid in the explanation of the application. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the teaching of this application. The embodiments were chosen and described in order to best explain the principles of the application and its practical applications, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A video processing method, comprising:

determining a second video decision result according to a preset video processing strategy in a target playing time period;

2. The video processing method according to claim 1, wherein after determining the second video decision result according to the preset video processing policy, further comprising:

3. The video processing method according to claim 2, wherein the obtaining the video playing characteristics of the target video within the preset playing time period comprises:

4. The video processing method according to claim 1, wherein the determining a second video decision result according to a preset video processing policy comprises:

5. The video processing method according to claim 4, wherein the determining a second video decision result according to the playing amount of the at least two video segments comprises:

6. The method according to claim 5, wherein determining the second video decision result according to the association relationship between the playback volume difference and the difference threshold comprises:

7. The video processing method according to claim 4, wherein said determining a second video decision result according to the playback amount of the at least two video segments comprises:

8. The video processing method according to claim 7, wherein said determining a second video decision result according to the playing amount of each of the at least two video segments and the heat threshold comprises:

9. The video processing method according to claim 8, wherein determining the second video decision result according to the relationship between the playing amount of each of the consecutive video segments and the heat threshold comprises:

10. The video processing method according to claim 6 or 9, wherein in a case that the second video decision result satisfies the decision condition, performing video processing on the target video comprises:

11. The video processing method according to claim 1, wherein the training of the object feature decision model comprises:

12. The video processing method of claim 11, wherein determining training samples from object attribute information of the sample objects comprises:

determining object features of the sample object as a training sample.

13. The video processing method according to claim 12, wherein after determining the sample label corresponding to the training sample according to the video playback amount, the method further comprises:

14. The video processing method according to claim 13, wherein before inputting the object feature of the target object into the object feature decision model and obtaining the first video decision result, the method further comprises:

15. The video processing method according to claim 1, wherein the inputting the object feature of the target object corresponding to the target video into the object feature decision model to obtain the first video decision result comprises:

16. The video processing method according to claim 2, wherein the training step of the play feature decision model is as follows:

17. The video processing method according to claim 16, wherein the determining a training sample according to the video playing attribute information corresponding to the sample video comprises:

18. A video processing apparatus, comprising:

19. A computing device comprising a memory, a processor, and computer instructions stored on the memory and executable on the processor, wherein the processor implements the steps of the video processing method of any of claims 1-17 when executing the instructions.

20. A computer-readable storage medium storing computer instructions which, when executed by a processor, implement the steps of the video processing method of any of claims 1-17.