CN112954464A - Video definition selection method and device based on network anomaly prediction - Google Patents
Video definition selection method and device based on network anomaly prediction Download PDFInfo
- Publication number
- CN112954464A CN112954464A CN202110081968.1A CN202110081968A CN112954464A CN 112954464 A CN112954464 A CN 112954464A CN 202110081968 A CN202110081968 A CN 202110081968A CN 112954464 A CN112954464 A CN 112954464A
- Authority
- CN
- China
- Prior art keywords
- data
- playing
- video
- network
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010187 selection method Methods 0.000 title claims abstract description 20
- 238000000034 method Methods 0.000 claims abstract description 41
- 230000002159 abnormal effect Effects 0.000 claims abstract description 26
- 238000003066 decision tree Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 abstract description 4
- 238000010801 machine learning Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 2
- 238000006731 degradation reaction Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/462—Content or additional data management, e.g. creating a master electronic program guide from data received from the Internet and a Head-end, controlling the complexity of a video stream by scaling the resolution or bit-rate based on the client capabilities
- H04N21/4621—Controlling the complexity of the content stream or additional data, e.g. lowering the resolution or bit-rate of the video stream for a mobile client with a small screen
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
- H04N21/4508—Management of client data or end-user data
- H04N21/4532—Management of client data or end-user data involving end-user characteristics, e.g. viewer profile, preferences
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The embodiment of the application discloses a video definition selection method and a video definition selection device based on network anomaly prediction, and the method comprises the following steps: receiving a video data playing request sent by a user through a client to determine target video data, and acquiring playing characteristic data when the user plays the target video data according to the video data playing request and the target video data; inputting the playing characteristic data into a pre-trained network anomaly prediction model to predict network state data; and selecting the playing code rate of the target video data according to a preset selection rule based on the network state data. According to the technical scheme, the network state is predicted according to data such as historical playing and transmission, video definition is selected according to the network state data, accordingly, the video definition is automatically reduced when the network state is abnormal so as to reduce the pause phenomenon during video playing, a video file with higher definition is selected when the network state is normal so as to provide image quality experience of a user, and watching experience of the user is integrally improved.
Description
Technical Field
The embodiment of the application relates to the technical field of computers, in particular to a video definition selection method based on network anomaly prediction, a video definition selection device based on network anomaly prediction, electronic equipment and a storage medium.
Background
With the popularization of mobile internet and smart phones, the viewing of streaming media videos becomes an indispensable part of people's daily entertainment life. The video definition meeting the current network condition is selected during video playing, so that the video downloading time can be shortened and the video playing pause rate can be reduced on the premise of ensuring the video definition, and the watching experience of a user can be optimized. In a short video playing scene, because the video time is short, one video is often only one file, so that selecting an appropriate video bitrate has a great influence on the viewing experience of a user. Meanwhile, short video watching is often in a mobile client scene, and due to factors such as environmental changes of users, abnormal network conditions greatly different from historical network conditions can occur.
In the prior art, the download speed of the current video playing is often estimated only according to the historical download speed of the user, or according to the bandwidth utilization rate and historical basic network data, and by combining simple formula calculation. These methods have the following disadvantages: firstly, simply representing the network condition by using characteristics such as downloading speed and the like, and selecting the definition according to the characteristics as a judgment basis, wherein the abnormal network condition appearing in a short video mobile client scene cannot be predicted; secondly, the used historical information is single, more dimensional characteristic information including the user historical watching behavior and the client information is not utilized for prediction, and the model characteristics are limited; thirdly, the algorithm model is simple, and high-dimensional information extraction cannot be performed on the features; meanwhile, the selection of the model parameters has subjectivity, and different network characteristic information cannot be learned according to the training data set.
Disclosure of Invention
The embodiment of the application provides a video definition selection method based on network anomaly prediction, a video definition selection device based on network anomaly prediction, electronic equipment and a storage medium, so that the definition of a played video can be timely adjusted by combining the network anomaly prediction condition, the video is prevented from being jammed during watching, and the watching experience of a user is improved.
In a first aspect, an embodiment of the present application provides a video sharpness selection method based on network anomaly prediction, including:
receiving a video data playing request sent by a user through a client, and determining target video data based on the video data playing request, wherein the video data playing request carries a target playing code rate;
acquiring playing characteristic data when a user plays the target video data according to the video data playing request and the target video data; the playing characteristic data comprises at least one of video downloading speed, user information, playing information, video information and underlying network information;
inputting the playing characteristic data into a pre-trained network anomaly prediction model to predict network state data; the network state data comprises an abnormal state and a normal state;
and selecting the playing code rate of the target video data according to a preset selection rule based on the network state data.
In a second aspect, an embodiment of the present application provides a video sharpness selecting apparatus based on network anomaly prediction, including:
a play request receiving module: the system comprises a client, a server and a server, wherein the client is used for receiving a video data playing request sent by a user through the client and determining target video data based on the video data playing request, and the video data playing request carries a target playing code rate;
the characteristic data acquisition module: the video playing device is used for acquiring playing characteristic data when a user plays the target video data according to the video data playing request and the target video data; the playing characteristic data comprises at least one of video downloading speed, user information, playing information, video information and underlying network information;
a network state output module: the network state prediction module is used for inputting the playing characteristic data into a pre-trained network abnormity prediction model to predict network state data; the network state data comprises an abnormal state and a normal state;
a definition selection module: and the method is used for selecting the playing code rate of the target video data according to a preset selection rule based on the network state data.
In a third aspect, an embodiment of the present application provides a computer device, including: a memory and one or more processors;
the memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method for video sharpness selection based on network anomaly prediction according to the first aspect.
In a fourth aspect, embodiments of the present application provide a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform a video sharpness selection method based on network anomaly prediction according to the first aspect.
According to the method and the device, a network abnormity prediction model is trained in advance through a large amount of data, target video data and a target playing code rate are determined according to a video data playing request of a user at a client, playing characteristic data when the user plays the target video data are obtained based on the video data playing request and the target video data, the playing characteristic data are input into the network abnormity prediction model to obtain network state data, video definition can be selected based on the network state data, video definition is automatically reduced when the network state is abnormal to reduce the pause phenomenon during video playing, a video file with higher definition is selected when the network state is normal to provide image quality experience of the user, and watching experience of the user is integrally improved.
Drawings
Fig. 1 is a flowchart of a video sharpness selection method based on network anomaly prediction according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of a video sharpness selecting apparatus based on network anomaly prediction according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a computer device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, specific embodiments of the present application will be described in detail with reference to the accompanying drawings. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not limiting of the application. It should be further noted that, for the convenience of description, only some but not all of the relevant portions of the present application are shown in the drawings. Before discussing exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart may describe the operations (or steps) as a sequential process, many of the operations can be performed in parallel, concurrently or simultaneously. In addition, the order of the operations may be re-arranged. The process may be terminated when its operations are completed, but may have additional steps not included in the figure. The processes may correspond to methods, functions, procedures, subroutines, and the like.
The embodiment of the application provides a video definition selection method based on network anomaly prediction, a video definition selection device based on network anomaly prediction, computer equipment and a storage medium. According to the method and the device, a network abnormity prediction model is trained in advance through a large amount of data, target video data and a target playing code rate are determined according to a video data playing request of a user at a client, playing characteristic data when the user plays the target video data are obtained based on the video data playing request and the target video data, the playing characteristic data are input into the network abnormity prediction model to obtain network state data, video definition can be selected based on the network state data, video definition is automatically reduced when the network state is abnormal to reduce the pause phenomenon during video playing, a video file with higher definition is selected when the network state is normal to provide image quality experience of the user, and watching experience of the user is integrally improved.
Fig. 1 is a flowchart of a video sharpness selection method based on network anomaly prediction according to an embodiment of the present application, where the video sharpness selection method based on network anomaly prediction according to the embodiment of the present application may be executed by a video sharpness selection apparatus based on network anomaly prediction, and the video sharpness selection apparatus based on network anomaly prediction may be implemented in hardware and/or software and integrated in a computer device.
The following description will be given taking as an example a video sharpness selection method performed by a video sharpness selection apparatus based on network anomaly prediction. Referring to fig. 1, the video sharpness selection method based on network anomaly prediction includes:
s101: receiving a video data playing request sent by a user through a client, and determining target video data based on the video data playing request, wherein the video data playing request carries a target playing code rate.
In the embodiment of the application, a possible carrier of the user client is a terminal, and the terminal can be any intelligent device, including a smart phone, a desktop computer, a notebook computer, a shielded computer, an intelligent watch, and the like. It is easily understood that, in the embodiment of the present application, the client borne on the smart device is an application layer in a transmission system architecture, and in this embodiment, is generally a video client, such as a live platform, a short video platform, and the like.
As an application scenario of the embodiment of the application, the application scenario may be that when a user requests to play a short video through a short video client, the short video client sends request information to a server, and the server responds to the request information and selects target video data according to a network to which the short video client has access, and sends the target video data to the short video client for playing. The target playing bit rate is used to express the intention of the target video data requested by the user, or the expected playing bit rate or the highest bit rate allowed by the historical bandwidth, the bit rate is also the bit rate of the video, which means the number of bits transmitted per second, and when the bit rate is higher, the more data transmitted per second, the clearer the image quality. Therefore, when a higher playing rate is selected, it also means that a video file with higher definition is selected.
S102: and acquiring playing characteristic data when the user plays the target video data according to the video data playing request and the target video data.
In the embodiment of the application, when a user logs in a client, account authentication of the user is often required. The bearer intelligent device of the client used by the user often has a device model, and when the user makes a video data playing request, the current network type can be obtained at the same time. When a user sends a video data playing request through a client, the video data playing request carries specified target video data, and therefore playing characteristic data is obtained when the video data playing request and the target video data are combined.
It should be noted that, in the embodiment of the present application, since a video data playing request of a user needs to be received first, and after target video data is determined, the target video data may need to be played first, where playing of the target video data is mainly to obtain playing characteristic data. Even in the case where the target video data is played in advance, the play time period for the target video data may be only a moment.
In the embodiment of the present application, the play special data may be understood as network characteristic data when the short video client plays the video data, for example, a file size of the short video data received by the client, a transmission time for receiving the short video data, a download speed, and the like. Specifically, the play feature data in this embodiment may include at least one of video download speed, user information, play information, video information, and underlying network information. The video download speed may be understood as a transmission speed when the client of the user receives the video data, and may be calculated based on target video data of the user, user information, and the like, or may be a value obtained by directly performing actual playing on the target video data for a very short time.
S103: inputting the playing characteristic data into a pre-trained network anomaly prediction model to predict network state data; the network state data includes an abnormal state and a normal state.
In the embodiment of the application, a network anomaly prediction model is constructed in advance, the current actual network state data of the user client is predicted based on the network anomaly prediction model, and the subsequent definition is selected according to the prediction result.
Specifically, in the present application, the network anomaly prediction model is trained in advance by the following method:
acquiring playing characteristic data and network state data when a user plays video data historically; and training a network anomaly prediction model by using the playing characteristic data as input characteristics and the network state data as output characteristics.
In the embodiment of the application, when the user plays video data historically, the short video data may be played through the same client or video platform, for example, the user plays a certain amount of short video data on the same live platform, or the user plays long movie data through the same client or video platform. The format of the played video data includes, but is not limited to, RM, RMVB, AVI, FLV, MP 4. The length and the format of the collected video data played by the user history are not limited.
And the network state data comprises an abnormal state and a normal state, if the abnormal state occurs, the network state data is marked as 1, otherwise, the network state data is marked as 0, and the network state data is used as an expected output value of the model of the supervised training.
In order to ensure that the network anomaly prediction model can input enough types of materials so as to ensure that the trained model can cope with various complex network environments and user behaviors in a later use scene and realize more accurate prediction of whether network anomalies occur, the play characteristic data of the embodiment of the application comprises at least one of video download speed, user information, play information, video information and underlying network information.
The user information comprises at least one of user country, ISP information, equipment model, operating system and network type; the playing information comprises the watching time length, the pause times, the pause time length, the waiting playing time length, the pre-downloading proportion and the cache area proportion of a user; the video information comprises video duration and video code rate; the network information comprises whether the current playing video data is multiplexed with a TCP link, round-trip delay and packet loss rate; the abnormal state of the network state data comprises the network type switching of the user, the overtime of the link request and the disconnection of the link.
As a further preferred embodiment, the using the play feature data as an input feature includes: and standardizing the playing characteristic data, and taking the standardized playing characteristic data as input characteristics.
And the standardized processing procedure comprises the following steps: extracting numerical characteristic data and non-numerical characteristic data in the playing characteristic data; converting the non-numerical characteristic data into numerical characteristic data according to a preset characteristic code; and processing the numerical characteristic data into standard normal distribution data, and taking the standard normal distribution data as input characteristics.
In the above normalization process, the numerical characteristic data is specifically processed into standard normal distribution data according to the following formula:
wherein x is the numerical characteristic data, x*For normal distribution data, μ is the mean of the feature in all sample data, and σ is the standard deviation of the feature in all sample data.
In this embodiment, non-numerical characteristic data, such as user country, network type, device model, ISP (Internet Service Provider) information, etc., is first converted into numerical characteristics.
In the application, the method for training the network anomaly prediction model by using the play characteristic data as the input characteristic and using the network state data as the output characteristic comprises the following steps:
and determining an initialized network anomaly prediction model. The method comprises the steps that a plurality of user historical standard normal distribution data are taken as a plurality of training samples and are sequentially input to an initialized network anomaly prediction model for training, so that the network prediction model obtained by current input standard normal distribution data training is always the superposition of the network prediction model obtained by last input standard normal distribution data training and a current decision tree; the training sample label is recorded once every time the standard normal distribution data is input. The current decision tree comprises currently input standard normal distribution data and decision tree parameters, and the decision tree parameters are obtained through calculation according to a preset loss function and a training sample label.
In the embodiment of the present application, a lifting tree model is taken as an example, and the lifting tree model can be expressed as:
wherein,
representing the mth decision tree; thetamRepresents the mth decision tree parameter; m is the number of decision trees;
representing the input features.
The offline training method of the lifting tree model comprises the following steps:
first, the initial lifting tree is determined as:
when step m is executed, the model is expressed as:
determination of the mth decision tree parameter Θ by empirical risk minimizationmComprises the following steps:
wherein, yiRepresenting the ith training sample label, namely whether the ith training sample label is an abnormal task or not; l (-) is a loss function.
In the embodiment of the machine learning model, the data are collected on line and trained off line, so that the machine learning model can effectively utilize the characteristic information of more dimensions, and the problem of single characteristic information is solved. Meanwhile, a machine learning model is used for fitting a complex high-order function, so that the problems of simple model and poor prediction performance of a common method are solved.
When the network anomaly prediction model is trained, the play characteristic data of the user can be input to predict the network state data. In practical applications, when a user makes a video data playing request through a client, assuming that a short video is requested to be played, since the data volume of the short video is usually small, the short video can be encoded and packetized by a data packet. When a long video is requested to be played, the data can be divided into a plurality of data packets for encoding, packaging and sending due to the fact that the data volume of the long video is usually large.
As a preferred solution of the embodiment of the present application, the trained network anomaly prediction model is further updated periodically, and a new network anomaly prediction model is deployed in a background server after each update, so that the network anomaly prediction model can always conform to the latest training template data, and the problem of the prediction performance degradation of an old model due to the change of a network condition with time is avoided. The method and the device solve the problem that the traditional algorithm model parameters have subjectivity in a data driving mode.
Specifically, the method for updating the network anomaly prediction model comprises the steps of acquiring first playing characteristic data and first network state data of a user during historical playing of video data within a preset time interval at preset time intervals; and training a network anomaly prediction model by using the first playing characteristic data as a new input characteristic and using the first network state data as a new output characteristic. The preset time interval may be, for example, one month. In this embodiment, the first playing characteristic data includes at least one of video downloading speed, user information, playing information, video information, and underlying network information, and the network state data also includes an abnormal state and a normal state.
S104: and selecting the playing code rate of the target video data according to a preset selection rule based on the network state data.
The method and the device for predicting the network state obtain predicted network state data based on a network state prediction model, and select the definition of the target video data aiming at the difference of the network state data, namely select the playing code rate of the target video data.
In an embodiment of the present application, the preset selection rule is as follows: when the predicted network state data is normal, selecting the playing code rate of the target video data to be at least consistent with the target playing code rate, for example, the target playing code rate corresponding to the target video data of the user is 100kb/s, and when the network state data is normal, selecting the playing code rate to be 100kb/s, or selecting the playing code rate to be more than 100 kb/s; and when the predicted network state data is abnormal, selecting the playing code rate of the target video data to be lower than the target playing code rate.
That is, when the network state data is in an abnormal state, it indicates that the network state of the current client may be unstable or there are other network abnormal conditions, and at this time, the playing definition of the target video data is reduced. In the embodiment, the definition of the playing is expressed by the playing code rate, for example, when the target playing code rate is 100kb/s, and when the network state data is in an abnormal state, the playing code rate is reduced by selecting the target video with lower definition, for example, the target playing code rate of 100kb/s is reduced to 90 kb/s. On the other hand, when the network state data is in a normal state, the network is not abnormal at present, and at the moment, the video file with high definition can be selected as far as possible, so that the image quality experience of the user is improved.
The method abstracts whether the network abnormity occurs when the predicted target video data is played into a two-classification problem, fits a complex model structure by training a lifting tree model based on a data driving and machine learning method, and predicts whether the network abnormity occurs when a user watches the target video data according to multi-dimensional characteristic information comprising historical downloading information, user information, playing information, historical network internal data and the like. And predicting a result by utilizing the machine learning model trained offline on line, and if the predicted result is that network abnormality occurs, selecting and issuing a video file with lower definition by the server, so that the downloading time and the pause rate of the video are reduced. And updating the user data at intervals, and retraining the model so as to reduce the influence of the change of the network condition along with the time on the detection performance of the abnormal network.
As shown in fig. 2, an embodiment of the present application further provides a video definition selecting apparatus based on network anomaly prediction, which includes a play request receiving module 201, a feature data collecting module 202, a network state output module 203, and a definition selecting module 204. The playing request receiving module 201 is configured to receive a video data playing request sent by a user through a client, and determine target video data based on the video data playing request, where the video data playing request carries a target playing code rate. The playing characteristic data comprises at least one of video downloading speed, user information, playing information, video information and underlying network information. The characteristic data acquisition module 202 is configured to acquire playing characteristic data when the user plays the target video data according to the video data playing request and the target video data. The network state output module 203 is used for inputting the playing characteristic data into a pre-trained network anomaly prediction model to predict network state data; the network state data includes an abnormal state and a normal state. The definition selection module 204 is configured to select a playing code rate of the target video data according to a preset selection rule based on the network state data.
In the network status output module 203, the applied network anomaly prediction model is obtained by acquiring play characteristic data and network status data when a plurality of users play video data historically, and training by using the play characteristic data as an input characteristic and the network status data as an output characteristic. The playing characteristic data in the network anomaly prediction model comprises at least one of video downloading speed, user information, playing information, video information and underlying network information.
The embodiment further updates the trained network anomaly prediction model regularly, and deploys a new network anomaly prediction model to the server on the background after each update, so that the network anomaly prediction model can always conform to the latest training sample data, and the problem of prediction performance degradation of the old model caused by the change of the network condition along with time is avoided. The method and the device solve the problem that the traditional algorithm model parameters have subjectivity in a data driving mode.
Specifically, the method for updating the network anomaly prediction model comprises the steps of acquiring first playing characteristic data and first network state data of a user during historical playing of video data within a preset time interval at preset time intervals; and training a network anomaly prediction model by using the first playing characteristic data as a new input characteristic and using the first network state data as a new output characteristic. In fact, the first play characteristic data and the first network status data correspond to the same substantial data characteristics as the play characteristic data and the network status data described earlier in this application, only because in this example, the play characteristic data and the network status data generated within the preset time interval are targeted, and also the first play characteristic data includes at least one of video download speed, user information, play information, video information, and underlying network information, and the network status data also includes an abnormal status and a normal status.
According to the method and the device, whether the network abnormity occurs during video playing in a video scene is predicted by utilizing the downloading speed, the user information, the playing information, the video information and the underlying network information during playing of historical video data through a data driving and machine learning method. Based on the abnormal network prediction result, the service selects the file with lower definition to issue, so that the video downloading time is reduced, and the situation of pause during playing is avoided. Compared with a general definition selection method only according to network bandwidth, the method can better adapt to the complex network environment in the mobile client scene. By adopting a model updating method, the machine learning model is retrained by using a new data set at intervals, and the model can better adapt to network states of different time periods.
Referring to fig. 3, fig. 3 shows a schematic structural diagram of a computer device of the present application. As shown in fig. 3, a computer device provided by the embodiment of the present application includes a memory 301 and one or more processors 302; the memory 301 is used for storing one or more programs; when executed by the one or more processors 302, cause the one or more processors to implement a method for video sharpness selection based on network anomaly prediction according to any of the present applications.
Embodiments of the present application also provide a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform the video sharpness selection method based on network anomaly prediction provided in the above embodiments.
Of course, the storage medium provided in the embodiments of the present application contains computer-executable instructions, and the computer-executable instructions are not limited to the video definition selection method based on network anomaly prediction described above, and may also perform related operations in the methods provided in any embodiments of the present application.
The video sharpness selecting apparatus, device and storage medium based on network anomaly prediction provided in the foregoing embodiments may execute the video sharpness selecting method based on network anomaly prediction provided in any embodiment of the present application, and reference may be made to the video sharpness selecting method based on network anomaly prediction provided in any embodiment of the present application without detailed technical details described in the foregoing embodiments.
The foregoing is considered as illustrative of the preferred embodiments of the invention and the technical principles employed. The present application is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present application has been described in more detail with reference to the above embodiments, the present application is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present application, and the scope of the present application is determined by the scope of the claims.
Claims (11)
1. A video definition selection method based on network anomaly prediction is characterized by comprising the following steps:
receiving a video data playing request sent by a user through a client, and determining target video data based on the video data playing request, wherein the video data playing request carries a target playing code rate;
acquiring playing characteristic data when a user plays the target video data according to the video data playing request and the target video data; the playing characteristic data comprises at least one of video downloading speed, user information, playing information, video information and underlying network information;
inputting the playing characteristic data into a pre-trained network anomaly prediction model to predict network state data; the network state data comprises an abnormal state and a normal state;
and selecting the playing code rate of the target video data according to a preset selection rule based on the network state data.
2. The video sharpness selection method of claim 1, wherein the network anomaly training model is obtained by training through the following steps:
acquiring playing characteristic data and network state data when a plurality of users play video data historically;
and training a network anomaly prediction model by using the playing characteristic data as input characteristics and the network state data as output characteristics.
3. The method of claim 2, wherein using the play feature data as an input feature comprises:
and standardizing the playing characteristic data, and taking the standardized playing characteristic data as input characteristics.
4. The method of claim 3, wherein the normalizing the playing feature data and the normalizing the data as the input feature comprises:
extracting numerical characteristic data and non-numerical characteristic data in the playing characteristic data;
converting the non-numerical characteristic data into numerical characteristic data according to a preset characteristic code;
and processing the numerical characteristic data into standard normal distribution data, and taking the standard normal distribution data as input characteristics.
5. The method of claim 4, wherein training a network anomaly prediction model using the play feature data as input features and the network state data as output features comprises:
determining an initialized network anomaly prediction model;
the method comprises the steps that a plurality of user historical standard normal distribution data are taken as a plurality of training samples and are sequentially input to an initialized network anomaly prediction model for training, so that the network prediction model obtained by current input standard normal distribution data training is always the superposition of the network prediction model obtained by last input standard normal distribution data training and a current decision tree; recording a training sample label once every time standard normal distribution data is input;
the current decision tree comprises currently input standard normal distribution data and decision tree parameters, and the decision tree parameters are obtained through calculation according to a preset loss function and a training sample label.
6. The video sharpness selection method of any of claims 2-5, wherein the training of the network anomaly training model further comprises:
acquiring first play characteristic data and first network state data of a user during historical play of video data within a preset time interval at intervals of a preset time interval;
and training a network anomaly prediction model by using the first playing characteristic data as a new input characteristic and using the first network state data as a new output characteristic.
7. The video sharpness selection method of any one of claim 1, wherein the user information includes at least one of a user country, ISP information, equipment model, operating system, and network type; the playing information comprises the watching time length, the pause times, the pause time length, the waiting playing time length, the pre-downloading proportion and the cache area proportion of a user; the video information comprises video duration and video code rate; the network information comprises whether the current playing video data is multiplexed with a TCP link, round-trip delay and packet loss rate; the abnormal state of the network state data comprises the network type switching of the user, the overtime of the link request and the disconnection of the link.
8. The method according to claim 1, wherein the predetermined selection rule is:
when the predicted network state data is normal, selecting the playing code rate of the target video data to be at least consistent with the target playing code rate;
and when the predicted network state data is abnormal, selecting the playing code rate of the target video data to be lower than the target playing code rate.
9. A video sharpness selection apparatus based on network anomaly prediction, comprising:
a play request receiving module: the system comprises a client, a server and a server, wherein the client is used for receiving a video data playing request sent by a user through the client and determining target video data based on the video data playing request, and the video data playing request carries a target playing code rate;
the characteristic data acquisition module: the video playing device is used for acquiring playing characteristic data when a user plays the target video data according to the video data playing request and the target video data; the playing characteristic data comprises at least one of video downloading speed, user information, playing information, video information and underlying network information;
a network state output module: the network state prediction module is used for inputting the playing characteristic data into a pre-trained network abnormity prediction model to predict network state data; the network state data comprises an abnormal state and a normal state;
a definition selection module: and the method is used for selecting the playing code rate of the target video data according to a preset selection rule based on the network state data.
10. A computer device, comprising: a memory and one or more processors;
the memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method for video sharpness selection based on network anomaly prediction according to any one of claims 1-8.
11. A storage medium containing computer-executable instructions for performing the method for video sharpness selection based on network anomaly prediction according to any one of claims 1-8 when executed by a computer processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110081968.1A CN112954464A (en) | 2021-01-21 | 2021-01-21 | Video definition selection method and device based on network anomaly prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110081968.1A CN112954464A (en) | 2021-01-21 | 2021-01-21 | Video definition selection method and device based on network anomaly prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112954464A true CN112954464A (en) | 2021-06-11 |
Family
ID=76235749
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110081968.1A Pending CN112954464A (en) | 2021-01-21 | 2021-01-21 | Video definition selection method and device based on network anomaly prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112954464A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113556571A (en) * | 2021-07-20 | 2021-10-26 | 苏州景昱医疗器械有限公司 | Video processing method and device, electronic equipment and computer readable storage medium |
| CN115052190A (en) * | 2022-06-08 | 2022-09-13 | 上海哔哩哔哩科技有限公司 | Video playing method and device |
| CN118102005A (en) * | 2024-04-28 | 2024-05-28 | 腾讯科技（深圳）有限公司 | Video data processing method, device, equipment and medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106162233A (en) * | 2016-07-08 | 2016-11-23 | 合网络技术（北京）有限公司 | Code check recommends method and device |
| CN107835200A (en) * | 2017-12-19 | 2018-03-23 | 北京奇艺世纪科技有限公司 | A kind of system of selection of video definition, device and electronic equipment |
| CN108769760A (en) * | 2018-05-23 | 2018-11-06 | 中国联合网络通信集团有限公司 | A kind of code rate adjustment method, UE, MEC and network system |
| US20190297356A1 (en) * | 2018-03-23 | 2019-09-26 | At&T Intellectual Property I, L.P. | Methods to estimate video playback buffer |
| CN111242171A (en) * | 2019-12-31 | 2020-06-05 | 中移（杭州）信息技术有限公司 | Model training, diagnosis and prediction method and device for network fault and electronic equipment |
| CN111405327A (en) * | 2020-04-03 | 2020-07-10 | 广州市百果园信息技术有限公司 | Network bandwidth prediction model training method, video data playing method and device |
| CN111447471A (en) * | 2020-03-26 | 2020-07-24 | 广州市百果园信息技术有限公司 | Model generation method, play control method, device, equipment and storage medium |
| CN111601118A (en) * | 2020-05-13 | 2020-08-28 | 广州市百果园信息技术有限公司 | Live video processing method, system, device and terminal |
| CN111669627A (en) * | 2020-06-30 | 2020-09-15 | 广州市百果园信息技术有限公司 | Method, device, server and storage medium for determining video code rate |
-
2021
- 2021-01-21 CN CN202110081968.1A patent/CN112954464A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106162233A (en) * | 2016-07-08 | 2016-11-23 | 合网络技术（北京）有限公司 | Code check recommends method and device |
| CN107835200A (en) * | 2017-12-19 | 2018-03-23 | 北京奇艺世纪科技有限公司 | A kind of system of selection of video definition, device and electronic equipment |
| US20190297356A1 (en) * | 2018-03-23 | 2019-09-26 | At&T Intellectual Property I, L.P. | Methods to estimate video playback buffer |
| CN108769760A (en) * | 2018-05-23 | 2018-11-06 | 中国联合网络通信集团有限公司 | A kind of code rate adjustment method, UE, MEC and network system |
| CN111242171A (en) * | 2019-12-31 | 2020-06-05 | 中移（杭州）信息技术有限公司 | Model training, diagnosis and prediction method and device for network fault and electronic equipment |
| CN111447471A (en) * | 2020-03-26 | 2020-07-24 | 广州市百果园信息技术有限公司 | Model generation method, play control method, device, equipment and storage medium |
| CN111405327A (en) * | 2020-04-03 | 2020-07-10 | 广州市百果园信息技术有限公司 | Network bandwidth prediction model training method, video data playing method and device |
| CN111601118A (en) * | 2020-05-13 | 2020-08-28 | 广州市百果园信息技术有限公司 | Live video processing method, system, device and terminal |
| CN111669627A (en) * | 2020-06-30 | 2020-09-15 | 广州市百果园信息技术有限公司 | Method, device, server and storage medium for determining video code rate |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113556571A (en) * | 2021-07-20 | 2021-10-26 | 苏州景昱医疗器械有限公司 | Video processing method and device, electronic equipment and computer readable storage medium |
| CN115052190A (en) * | 2022-06-08 | 2022-09-13 | 上海哔哩哔哩科技有限公司 | Video playing method and device |
| CN115052190B (en) * | 2022-06-08 | 2024-04-02 | 上海哔哩哔哩科技有限公司 | Video playing method and device |
| CN118102005A (en) * | 2024-04-28 | 2024-05-28 | 腾讯科技（深圳）有限公司 | Video data processing method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112954464A (en) | Video definition selection method and device based on network anomaly prediction | |
| CN114401447B (en) | Video clamping prediction method, device, equipment and medium | |
| US11595458B2 (en) | Adaptive bit rate media streaming based on network conditions received via a network monitor | |
| US10235157B2 (en) | Method and system for analytics-based updating of networked devices | |
| EP2320601B1 (en) | Method and system for adapting a session timeout period | |
| US20220191592A1 (en) | Method and system for user-oriented streaming of content | |
| US20200015121A1 (en) | Systems and methods for mobile network guidance for over-the-top applications | |
| JP7505086B2 (en) | SYSTEM AND METHOD FOR EXTERNAL SYSTEM INTEGRATION - Patent application | |
| CN112422441B (en) | Congestion control method and system based on QUIC transmission protocol | |
| CN107634850B (en) | A kind of application state acquisition methods and its equipment, storage medium, server | |
| US11134023B2 (en) | Network path redirection | |
| CN114040257B (en) | Self-adaptive video stream transmission playing method, device, equipment and storage medium | |
| US20230093630A1 (en) | System and method for adapting to changing constraints | |
| CN103034632A (en) | Information transmitting method and a system | |
| CN114025190B (en) | Multi-code rate scheduling method and multi-code rate scheduling device | |
| CN113596325B (en) | Method and device for capturing images, electronic equipment and storage medium | |
| CN115412326A (en) | Abnormal flow detection method and device, electronic equipment and storage medium | |
| WO2019141905A1 (en) | An apparatus, a method and a computer program for running a neural network | |
| KR102492022B1 (en) | Method, Apparatus and System of managing contents in Multi-channel Network | |
| US9681163B1 (en) | Identify bad files using QoS data | |
| US11032347B2 (en) | Bandwidth usage reduction employing media treading with reductive edging | |
| CN112994934A (en) | Data interaction method, device and system | |
| CN114885215B (en) | Training method of code rate self-adaptive model, video code rate self-adaptive method and device | |
| US11810206B2 (en) | Solver-based media assignment for content moderation | |
| CN118170566A (en) | Message processing method oriented to edge equipment and related equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210611 |