Video Delivery Method, Device, Electronic Device, and Storage Medium

Abstract

The present disclosure provides a video delivery method, apparatus, electronic device, and storage medium. The video delivery method includes: receiving a video delivery request from a target device, the video delivery request including target device type information of the target device; based on the target device type information, querying target service quality feature information at each preset level among multiple preset levels corresponding to the target device type, wherein the target device type is one of multiple device types, and the target service quality feature information corresponding to the multiple device types is predicted and stored using a service quality prediction model, wherein the service quality prediction model is used to predict the target service quality feature information corresponding to each device type using device performance information corresponding to the device type; based on the target service quality feature information at each preset level and a target service quality feature information threshold, determining at least one preset level matching the target device; and delivering a video corresponding to the at least one preset level to the target device.

Description

Technical Field

[0001] The present disclosure relates to the field of video technology, and more specifically, to a video distribution method, device, electronic device, and storage medium. Background Art

[0002] Video resolution and the choice of codec are important considerations for audio and video content. Higher resolutions provide users with clearer audio and video quality, but also increase the bit rate. More complex codecs also have higher compression rates. At the same bit rate, audio and video files are smaller, which helps improve fluency and reduce bandwidth costs. However, higher resolutions and more advanced codecs also increase the decoding complexity of the playback terminal. When the decoding rendering speed is slower than the video frame rate, frame loss will occur, resulting in an unsmooth picture on the playback terminal and a stuttering experience for the user. Therefore, in order to ensure the user's clarity and smoothness experience while balancing bandwidth costs, it is necessary to provide audio and video streams with various resolutions and codecs of various complexities for playback by the playback terminal.

[0003] Given that different device types have different decoding capabilities, for example, whether they support software / hardware decoding and large differences in decoding speeds, it is usually necessary to manually maintain black and white lists to specify which devices are suitable or not suitable for each resolution and codec, thereby managing the delivery rules for different playback terminals. However, manually maintaining black and white lists consumes a lot of manpower costs, and the lists are usually generated after performance has been damaged, resulting in a serious lag in solving the problem. Summary of the Invention

[0004] The present disclosure provides a video downloading method, device, electronic device and storage medium to at least solve the problems in the above-mentioned related technologies.

[0005] According to a first aspect of an embodiment of the present disclosure, a video delivery method is provided, comprising: receiving a video delivery request from a target device, wherein the video delivery request includes target device type information of the target device; based on the target device type information, querying target service quality feature information corresponding to the target device type, wherein the target device type is one of a plurality of device types, and the target service quality feature information corresponding to the plurality of device types is predicted and stored by a service quality prediction model, wherein the service quality prediction model is used to use device performance information corresponding to each device type to predict the target service quality feature information corresponding to the device type, wherein the target service quality information includes target service quality feature information of each preset level among a plurality of preset levels, and the preset level is a level corresponding to a preset video resolution value and a preset codec mode; based on the target service quality feature information of each preset level among the plurality of preset levels and a target service quality feature information threshold corresponding to the preset level, determining at least one preset level that matches the target device; and delivering the video corresponding to the at least one preset level to the target device.

[0006] Optionally, the service quality prediction model may be trained by using the device performance information corresponding to the multiple device types as training data, and using the real data of the target service quality characteristics of the multiple device types at the multiple preset levels as training targets.

[0007] Optionally, the video delivery method may further include: updating the service quality prediction model at predetermined time intervals, and storing target service quality feature information corresponding to the multiple device types predicted by the updated service quality prediction model.

[0008] Optionally, the video download request may also include real-time status data of the target device; wherein, querying the target service quality characteristic information corresponding to the target device type based on the target device type information may include: based on the target device type information and the real-time status data, querying the target service quality characteristic information corresponding to the target device type and the real-time status data, wherein the target service quality characteristic information corresponding to the multiple device types and multiple status data is predicted and stored by the service quality prediction model, and the service quality prediction model is used to use the device performance information corresponding to each device type and each status data under the device type to predict the target service quality characteristic information of each status data corresponding to the device type.

[0009] Optionally, the multiple status data may be multiple preset status data ranges; wherein, the querying of target service quality characteristic information corresponding to the target device type and the real-time status data based on the information of the target device type and the real-time status data may include: determining the target device type of the target device and the target status data range to which the real-time status data belongs based on the target device type information and the real-time status data; and querying the target service quality characteristic information corresponding to the target status data range under the target device type.

[0010] Optionally, the service quality prediction model can be trained by using the multiple pieces of device performance information of the multiple device types and the actual status data of the multiple device types as training data, and using the actual data of the target service quality characteristics of the multiple device types at the multiple preset levels under the multiple status data as training targets.

[0011] Optionally, the video delivery method may further include: updating the service quality prediction model at predetermined time intervals, and storing target service quality characteristic information corresponding to the multiple device types under the multiple state data predicted by the updated service quality prediction model.

[0012] Optionally, determining at least one preset level that matches the target device based on the target service quality information at each preset level among the multiple preset levels and the target service quality information threshold of the corresponding preset level may include: comparing the target service quality characteristic information at each preset level among the multiple preset levels with the target service quality characteristic information threshold of the corresponding preset level, and determining the preset level whose target service quality characteristic information does not exceed the target service quality characteristic information threshold as the preset level that matches the target device.

[0013] Optionally, before receiving a video download request from the target device, it may also include: based on multiple service quality feature data of the multiple device types at the multiple preset levels and the target experience quality feature data, selecting the service quality feature that is most relevant to the target experience quality feature from the multiple service quality features as the target service quality feature.

[0014] Optionally, the target quality of service characteristic may be a frame loss rate.

[0015] According to a second aspect of an embodiment of the present disclosure, a video delivery apparatus is provided, comprising: a receiving unit configured to receive a video delivery request from a target device, wherein the video delivery request includes target device type information of the target device; a querying unit configured to query target service quality feature information corresponding to the target device type based on the target device type information, wherein the target device type is one of multiple device types, and the target service quality feature information corresponding to the multiple device types is predicted and stored by a service quality prediction model, wherein the service quality prediction model is configured to use device performance information corresponding to each device type to predict the target service quality feature information corresponding to the device type, wherein the target service quality feature information includes target service quality feature information of each preset level among multiple preset levels, wherein the preset level is a level corresponding to a preset video resolution value and a preset codec mode; a determining unit configured to determine at least one preset level matching the target device based on the target service quality feature information at each preset level among the multiple preset levels and a target service quality parameter information threshold corresponding to the preset level; and a delivery unit configured to deliver a video corresponding to the at least one preset level to the target device.

[0016] Optionally, the service quality prediction model may be trained by using the device performance information corresponding to the multiple device types as training data, and using the real data of the target service quality characteristics of the multiple device types at the multiple preset levels as training targets.

[0017] Optionally, the video delivery device may further include an updating unit configured to update the service quality prediction model at predetermined time intervals and store target service quality feature information corresponding to the multiple device types predicted by the updated service quality prediction model.

[0018] Optionally, the video download request may also include real-time status data of the target device; wherein, the query unit may be configured to: based on the information of the target device type and the real-time status data, query the target service quality characteristic information corresponding to the target device type and the real-time status data, wherein the target service quality characteristic information corresponding to the multiple device types and multiple status data is predicted and stored through the service quality prediction model, and the service quality prediction model is used to use the device performance information corresponding to each device type and each status data under the device type to predict the target service quality characteristic information of each status data corresponding to the device type.

[0019] Optionally, the multiple status data may be multiple preset status data ranges; wherein the query unit may be configured to: determine the target device type of the target device and the target status data range to which the real-time status data belongs based on the target device type information and the real-time status data; and query the target service quality characteristic information corresponding to the target status data range under the target device type.

[0020] Optionally, the service quality prediction model can be trained by using the multiple pieces of device performance information of the multiple device types and the actual status data of the multiple device types as training data, and using the actual data of the target service quality characteristics of the multiple device types at the multiple preset levels under the multiple status data as training targets.

[0021] Optionally, the video sending device may also include an updating unit configured to update the service quality prediction model at predetermined time intervals, and store the target service quality characteristic information corresponding to the multiple device types under the multiple state data predicted by the updated service quality prediction model.

[0022] Optionally, the determination unit can be configured to compare the target service quality characteristic information at each preset level of the multiple preset levels with the target service quality characteristic information threshold of the corresponding preset level, and determine the preset level at which the target service quality characteristic information does not exceed the target service quality characteristic information threshold as the preset level that matches the target device.

[0023] Optionally, the video delivery apparatus may further include a selection unit. Before receiving a video delivery request from a target device, the selection unit is configured to select, based on the multiple quality of service feature data of the multiple device types at the multiple preset levels and the target quality of experience feature data, a quality of service feature that is most relevant to the target quality of experience feature from the multiple quality of service features as the target quality of service feature.

[0024] Optionally, the target quality of service characteristic may be a frame loss rate.

[0025] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory storing computer-executable instructions, wherein the computer-executable instructions, when executed by the at least one processor, prompt the at least one processor to execute the video downloading method according to the present disclosure.

[0026] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by at least one processor, the at least one processor is prompted to execute the video downloading method according to the present disclosure.

[0027] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, comprising computer instructions, which, when executed by at least one processor, implement the video downloading method according to the present disclosure.

[0028] The technical solutions provided by the embodiments of the present disclosure bring at least the following beneficial effects:

[0029] According to the present disclosure, a video sending method and a video sending device are provided, which can establish an artificial intelligence model for fitting the online service quality (QoS) characteristic information of the device of the corresponding device type based on the various device performance information corresponding to the device type, and make a level sending decision based on the fitting result obtained by the artificial intelligence model, so as to realize the automatic decision of what level of video to send to the target device requesting the video, thereby replacing the original level sending decision by maintaining a blacklist by using the method of realizing the automated level sending decision based on the various device performance information corresponding to the device type, while saving labor costs, the level sending decision can be made more scientifically, timely and efficiently, thereby improving the user experience.

[0030] In addition, the video distribution method according to the present invention can also combine the device performance information corresponding to the device type with the real-time status data of the device to make level distribution decisions through an artificial intelligence model, so that not only the device performance of the device can be taken into account, but also the current status of the device can be taken into account to implement level distribution decisions. Therefore, the level distribution decision can be implemented at the device granularity, better providing personalized level distribution for each device and improving user experience.

[0031] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute an improper limitation of the present disclosure.

[0033] Figure 1 This is an exemplary application scenario of the video delivery method according to an exemplary embodiment of the present disclosure;

[0034] Figure 2 is a flowchart illustrating a video delivery method according to an exemplary embodiment of the present disclosure;

[0035] Figure 3 is a flowchart illustrating an implementation of a level issuance decision according to an exemplary embodiment of the present disclosure;

[0036] Figure 4 is a schematic diagram illustrating an implementation of a level-delivery decision according to an exemplary embodiment of the present disclosure;

[0037] Figure 5 is a block diagram illustrating a video distribution apparatus according to an exemplary embodiment of the present disclosure;

[0038] Figure 6 is a block diagram of an electronic device 600 according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0039] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0040] It should be noted that the terms "first," "second," and the like in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the numbers used in this manner are interchangeable where appropriate so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation methods described in the following examples do not represent all implementation methods consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with certain aspects of the present disclosure as detailed in the appended claims.

[0041] It should be noted that the phrase "at least one of the items" in this disclosure includes three types of parallel situations: "any one of the items", "a combination of any multiple items of the items", and "all of the items". For example, "including at least one of A and B" includes the following three parallel situations: (1) including A; (2) including B; (3) including A and B. For another example, "performing at least one of step 1 and step 2" includes the following three parallel situations: (1) performing step 1; (2) performing step 2; and (3) performing steps 1 and 2.

[0042] In related technologies, given that different device types (e.g., machine models, device models, etc.) have different decoding capabilities, such as whether they support software / hard decoding, large differences in decoding speeds, etc., it is usually necessary to manually maintain black and white lists to specify the device types that are suitable for each resolution and encoding method, or the device types that are not suitable for the devices being sent. For example, the resolution types may include 540p, 576p, 720p, 1080p, etc., and the codecs may include H.264, H.265, VVC, etc., which may require manual maintenance of a blacklist, for example, at the level of 720p resolution and H.265 codec, the device types that are not suitable for sending videos of this level; at the level of 1080p resolution and VVC codec, the device types that are not suitable for sending videos of this level; and so on; and / or a whitelist, for example, at the level of 720p resolution and H.265 codec, the device types that are suitable for sending videos of this level; at the level of 1080p resolution and VVC codec, the device types that are suitable for sending videos of this level; and so on.

[0043] To adapt the appropriate level to each device type, the Quality of Service (QoS) and Quality of Experience (QoE) performance of each online device type at each level can be evaluated. If the QoS and QoE of a device corresponding to a certain device type degrade, it indicates that videos corresponding to the QoS and QoE degradation levels are not suitable for delivery to devices of that type. Therefore, the device type is added to a blacklist for the corresponding level. This method of managing and maintaining the blacklist requires significant labor costs and regular offline data analysis for maintenance. Because devices corresponding to low-end device types have weak decoding capabilities and are not suitable for decoding high-level videos, regular queries of low-end device types are required to expand the blacklist. Each new level requires A / B testing to confirm the frame loss rate and blacklist, which is inefficient. Furthermore, this method of managing and maintaining the blacklist typically adds blacklists after performance has already been compromised, resulting in significant lag.

[0044] To solve the above technical problems, the present disclosure provides a video delivery method that can establish an artificial intelligence model for fitting the online quality of service (QoS) characteristic information of the device type based on the device performance information corresponding to the device type, and make a level delivery decision based on the fitting results obtained by the artificial intelligence model, so as to realize the automatic decision of what level of video to deliver to the target device requesting the video. Thus, by using the method of making automated level delivery decisions based on the various device performance information corresponding to the device type to replace the original level delivery decision method by maintaining a blacklist, while saving labor costs, it can make level delivery decisions more scientifically, timely and efficiently, thereby improving the user experience. In addition, the video delivery method according to the present disclosure can also combine the device performance information corresponding to the device type with the real-time status data of the device to make level delivery decisions through the artificial intelligence model, so that it can take into account not only the device performance of the device, but also the current status of the device to make level delivery decisions. Therefore, the level delivery decision can be made at the device granularity, better providing personalized level delivery for each device, and improving the user experience.

[0045] Below, we will refer to Figures 1 to 6 The video downloading method and apparatus thereof according to the present disclosure will be described in detail.

[0046] Figure 1 This is an exemplary application scenario of the video delivery method according to an exemplary embodiment of the present disclosure.

[0047] Reference Figure 1 The exemplary application scenario may include, but is not limited to, a server 100 and user terminals 110 and 120. Here, the server 100 can provide a video delivery service. It can be a single server, a server cluster composed of several servers, or a cloud computing platform or a virtualization center. The user terminals 110 and 120 can send video delivery services to the server 100 and receive video streams of various levels suitable for the user terminals 110 and 120 provided by the server 100. Here, the user terminals 110 and 120 are merely exemplary and are obviously not limited to two. There can be tens of thousands of user terminals using the video service. The user terminals 110 and 120 can be implemented as devices such as smartphones, tablet computers, and personal computers.

[0048] In related technologies, the server 100 can query the level combination (which may include one or more levels) suitable for the user terminals 110 and 120 from the black and white lists maintained by the server 100 based on the device types reported by the user terminals 110 and 120, and send the video stream corresponding to the level combination suitable for the user terminals 110 and 120 to the user terminals 110 and 120, and the user terminals 110 and 120 select to play a video stream at one level from the level combination. Here, the level combination suitable for the user terminals 110 and 120 means that, considering the decoding capabilities of the device types of the user terminals 110 and 120 themselves, at least one level is selected for the user terminals 110 and 120 so that the playback will not cause freezes, frame drops, and other phenomena that affect the user experience. However, as mentioned above, maintaining black and white lists consumes a large amount of manpower costs and has a serious delay.

[0049] In the present disclosure, in order to solve the above-mentioned problems of the related art, a method for implementing level-down decision-making based on various device performance information and using an artificial intelligence model is proposed. Specifically, the server 100 can build and maintain an artificial intelligence model for fitting the online quality of service (QoS) characteristic information of the device of the corresponding device type based on the various device performance information device types corresponding to each device type, save the fitting results of each device type obtained by the artificial intelligence model to a storage device (for example, but not limited to, a remote dictionary server (Remote Dictionary Server, Redis)), and regularly update the artificial intelligence model through the online real QoS performance characteristic data of each device type obtained offline, and update the fitting results of each device type obtained by the updated artificial intelligence model to the storage device. When the server 100 receives a video download request from the user terminals 110 and 120, it can query the storage device for fitting results corresponding to the device types of the user terminals 110 and 120 based on the device types of the user terminals 110 and 120 (for example, the device type reported with the video download request), and determine the level combination suitable for the user terminals 110 and 120 based on the fitting results and the preset fitting result thresholds of each level, and download the video stream corresponding to the level combination suitable for the user terminals 110 and 120 to the user terminals 110 and 120, and the user terminals 110 and 120 select to play a video stream at a level from the level combination. According to the present disclosure, by using an automated level download decision-making method based on various device performance information corresponding to the device type to replace the original level download decision-making method by maintaining a blacklist, while saving labor costs, the level download decision can be made more scientifically, timely and efficiently, thereby improving the user experience.

[0050] In addition, in the present disclosure, the server 100 is also capable of combining the device performance information corresponding to the device type with the real-time status data of the device to make a level distribution decision through an artificial intelligence model. That is, when the server 100 receives a video distribution request from the user terminals 110 and 120, it can query the storage device for the fitting results corresponding to the device type and real-time status data of the user terminals 110 and 120 based on the device type and real-time status data of the user terminals 110 and 120 (for example, the device type and real-time status data reported with the video distribution request), and determine the level combination suitable for the user terminals 110 and 120 based on the fitting results and the preset fitting result thresholds of each level, and distribute the video stream corresponding to the level combination suitable for the user terminals 110 and 120 to the user terminals 110 and 120, and the user terminals 110 and 120 select to play a video stream at a level from the level combination. Through this solution, not only the device performance but also the current status of the device can be taken into account to implement level distribution decisions. Therefore, the level distribution decision can be made at the device granularity, better providing personalized level distribution for each device and improving user experience.

[0051] Figure 2 is a flow chart showing a video delivery method according to an exemplary embodiment of the present disclosure. The video delivery method according to an exemplary embodiment of the present disclosure can be applied to a video service platform, for example, Figure 1 The server 100 described in .

[0052] Reference Figure 2 In step 201, a video download request may be received from a target device, wherein the video download request may include information about the target device type of the target device. Here, the target device may be implemented as follows: Figure 1 The user terminals 110 and 112 described in

[15] are shown in Figure 15. When a user uses a video service application installed on a target device, they can initiate a video download request to the server by opening the video service application, clicking on the video thumbnails displayed in the video service application, or clicking on the video service module in the video service application. When the target device initiates a video download request to the server, it can also include its own device type information. Alternatively, when the server receives a video download request from a target device, it can request the device type information from the target device through a specific channel.

[0053] In step 202, target quality of service feature information corresponding to the target device type may be queried based on the target device type information. Here, the target quality of service feature information may include target quality of service feature information of each preset level among multiple preset levels.

[0054] According to an exemplary embodiment of the present disclosure, the preset level may be a level corresponding to a preset video resolution value and a preset codec mode. For example, the preset video resolution value may include, but is not limited to, 540p, 576p, 720p, 1080p, etc., and the preset codec mode may include, but is not limited to, H.264, H.265, VVC, etc. According to an exemplary embodiment of the present disclosure, the preset level may be a combination of a certain video resolution value and a certain codec mode. For example, the preset level may include, but is not limited to, H.265_720p, H.265_1080p, VVC_720p, VVC_1080p, etc. According to another exemplary embodiment of the present disclosure, the preset level may also be a combination of a certain video resolution range value and a certain codec mode. For example, the preset level may include, but is not limited to, H.265_540p_576p (i.e., videos whose codec mode is H.265 and whose resolution is in the range of 540p to 576p belong to this level), H.265_547p_720p (i.e., videos whose codec mode is H.265 and whose resolution is in the range of 577p to 720p belong to this level), H.265_721p_1080p (i.e., videos whose codec mode is H.265 and whose resolution is in the range of 721p to 1080p belong to this level), and H.265_1080p+ (i.e., videos whose codec mode is H.265 and whose resolution is in the range above 1080p belong to this level).

[0055] According to an exemplary embodiment of the present disclosure, the quality of service (QoS) characteristics may include first screen time consumption, freeze rate, start failure rate, frame loss rate, etc. The target quality of service feature may be a quality of service feature pre-selected from a plurality of quality of service features. For example, the quality of service feature may be selected based on demand, experience, etc. According to an exemplary embodiment of the present disclosure, the consumer service quality performance of each device type at each level is obtained, and based on the relationship between each quality of service feature and the quality of experience (QoE) feature, the quality of service feature that has the greatest impact on the quality of experience is found as the target quality of service feature. An artificial intelligence model is used to estimate the target quality of service feature information, and a level-delivery decision is executed based on the estimated target quality of service feature information, so that such a level-delivery decision result can bring the best user experience to the user and achieve the best quality of experience. Specifically, based on multiple service quality characteristics (for example, first screen time consumption, freeze rate, start failure rate, frame loss rate, etc.) data of multiple device types at multiple preset levels and target experience quality characteristics (for example, average device playback time, etc.) data, the service quality characteristics that are most relevant to the target experience quality characteristics (that is, the service quality characteristics that have the greatest impact on the target experience quality characteristic data) can be selected from the multiple service quality characteristics as the target service quality characteristics (for example, frame loss rate). Here, the multiple service quality characteristic data and the target experience quality characteristic data can be obtained offline. In the present disclosure, there is no limitation on the method of finding the service quality characteristics that have the greatest impact on the experience quality (that is, the service quality characteristics that are most relevant to the experience quality) based on the multiple service quality characteristic data and the target experience quality characteristic data, and any feasible analysis method can be used.

[0056] According to an exemplary embodiment of the present disclosure, the target quality of service characteristic information may be implemented as, but not limited to, information indicating the numerical value of the target quality of service characteristic, such as a target quality of service characteristic evaluation value. According to an exemplary embodiment of the present disclosure, assuming that the target quality of service characteristic is the frame loss rate, a higher frame loss rate evaluation value at a certain preset level indicates a greater frame loss rate at the preset level, and a lower frame loss rate evaluation value at a certain preset level indicates a lower frame loss rate at the preset level.

[0057] According to an exemplary embodiment of the present disclosure, target service quality feature information at each of multiple preset levels corresponding to multiple device types may be stored in a storage device associated with the server. Here, the storage device may be implemented as, but not limited to, Redis.

[0058] According to an exemplary embodiment of the present disclosure, target service quality feature information corresponding to a plurality of device types is predicted and stored through a service quality prediction model, and the service quality prediction model is used to use the device performance information corresponding to each device type to predict the target service quality feature information corresponding to the device type. Specifically, through practice and technical cognition, there is a strong correlation between device performance information (e.g., device performance score) and service quality features dominated by device type (e.g., frame loss rate). Therefore, based on multiple device performance information, a service quality prediction model is established to fit the online real target service quality feature data, thereby obtaining the target service quality feature information of the device type at each preset level, and the obtained target service quality feature information of the device type at each preset level is stored for level distribution decision-making. Here, multiple device performances may include, but are not limited to, device hardware performance such as central processing unit (CPU), graphics processing unit (GPU), input / output (IO), memory (MEMORY). Device performance information can be implemented as a comprehensive performance score of device type performance.

[0059] According to an exemplary embodiment of the present disclosure, the service quality prediction model may be implemented by an artificial intelligence model, such as, but not limited to, a regression model. The regression model may include, but is not limited to, a stepwise regression model, an extreme gradient boosting tree (eXtreme Gradient Boosting, XGBoost) model, a light gradient boosting machine (LGBM) model, and a neural network (NN) model.

[0060] According to an exemplary embodiment of the present disclosure, a service quality prediction model needs to be pre-trained, and the trained service quality prediction model is used to predict target service quality characteristics for a device type at each preset level. According to an exemplary embodiment of the present disclosure, the input of the service quality prediction model can be device performance information corresponding to a device type (which can be multiple pieces of device performance information), and the output can be target service quality characteristics for that device type at each preset level. Therefore, the service quality prediction model can be trained by using the multiple pieces of device performance information for multiple device types as training data and using real data of target service quality characteristics for multiple device types at multiple preset levels as training targets (i.e., training labels). Here, the real data of target service quality characteristics for multiple device types at multiple preset levels can be obtained offline. In addition, the value of a loss function can be calculated based on the target service quality characteristics for the device type at each preset level predicted by the service quality prediction model and the real data of the target service quality characteristics for the device type at each preset level. The parameters of the service quality prediction model can be adjusted using the value of the loss function, thereby training the service quality prediction model. This disclosure does not limit the loss function, and any available loss function can be used to calculate the loss, such as, but not limited to, the cross-entropy loss function. In addition, according to an exemplary embodiment of the present disclosure, it is possible to attempt to train a variety of different regression models, and test the trained multiple regression models, and select the regression model with the best test results to implement the service quality prediction model, so that the accuracy of the level distribution decision can be improved by using the regression model that best suits the current video service scenario.

[0061] According to an exemplary embodiment of the present disclosure, the target service quality feature information at each preset level in the multiple preset levels corresponding to the stored multiple device types can be updated regularly. For example, the stored target service quality feature information of a certain device type can be updated, or the target service quality feature information of a new device type can be added, etc., so as to ensure the timeliness and accuracy of the level sending decision, so that the result of the level sending decision is more in line with the current video sending scenario. Specifically, the service quality prediction model can be updated regularly (at predetermined time intervals), and the target service quality feature information at each preset level in the multiple preset levels corresponding to the multiple device types predicted by the updated service quality prediction model can be stored. For example, the real data of the target service quality features of multiple device types at multiple preset levels can be obtained regularly offline, and the service quality prediction model can be retrained using the obtained real data, thereby updating the service quality prediction model.

[0062] In step 203, at least one preset level matching the target device may be determined based on target service quality feature information at each preset level among the plurality of preset levels and target service quality feature information thresholds corresponding to the preset levels.

[0063] According to an exemplary embodiment of the present disclosure, the target service quality feature information at each preset level in a plurality of preset levels is compared with the target service quality feature information threshold of the corresponding preset level, and based on the comparison result, the preset level that matches the target device is determined. For example, the service quality feature is usually a feature that measures whether the service quality is deteriorating (for example, first screen time consumption, freeze rate, broadcast failure rate, frame loss rate). Therefore, the higher the service quality feature information, the more deteriorated the service quality is. Therefore, when the target service quality feature information is compared with its threshold, if the target service quality feature information at a certain preset level exceeds its threshold, it means that the degree of service quality degradation at the preset level has exceeded the allowable range of degradation. If the video stream corresponding to the preset level is sent to the target device for playback, the quality of experience will be deteriorated. If the target service quality feature information at a certain preset level does not exceed its threshold, it means that the degree of service quality degradation at the preset level is within the allowable range of degradation. If the video stream corresponding to the preset level is sent to the target device for playback, the quality of experience will not be deteriorated. Therefore, the preset level at which the target service quality feature information does not exceed the target service quality feature information threshold may be determined as the preset level that matches the target device.

[0064] According to an exemplary embodiment of the present disclosure, target service quality feature information thresholds of various preset levels can be pre-set. For example, target service quality feature information thresholds of various preset levels can be adjusted by methods such as AB testing to achieve the best experience quality.

[0065] At step 204, a video corresponding to at least one preset level may be delivered to the target device. If multiple preset levels are determined in step 203, videos corresponding to level delivery combinations of the multiple preset levels may be delivered to the target device, which then selects a preset level from the level delivery combinations to play the video.

[0066] Furthermore, according to exemplary embodiments of the present disclosure, device performance information can be combined with the device's real-time status data to make level-delivery decisions using an artificial intelligence model. This allows level-delivery decisions to be made at the device level, enabling personalized level-delivery for each device. For example, real-time device status data may include, but is not limited to, real-time data on status characteristics such as device power, temperature, memory remaining, and CPU usage.

[0067] According to exemplary embodiments of the present disclosure, target quality of service (QoS) feature information corresponding to multiple device types and multiple types of status data can be pre-stored. When a target device sends a video download request to a server, it can report not only its device type information but also its real-time status data. Upon receiving the target device's device type information and real-time status data, the server can query the target QoS feature information for each preset level corresponding to its device type and real-time status data. Based on the query results and the target QoS feature information thresholds for each preset level, the server determines a level combination suitable for the target device. The server then downloads the video stream corresponding to the appropriate level combination to the target device, which then selects a video stream from the level combination to play. According to exemplary embodiments of the present disclosure, the server establishes and maintains a QoS prediction model for predicting target QoS feature information based on device performance information and device status data. Target QoS feature information corresponding to multiple device types and multiple types of status data is predicted and stored using the QoS prediction model. The QoS prediction model uses the device performance information corresponding to each device type and each type of status data for that device type to predict the target QoS feature information for each type of status data corresponding to that device type.

[0068] According to an exemplary embodiment of the present disclosure, the various status data items are various preset status data ranges. Therefore, based on the target device type information and the real-time status data, the target device type of the target device and the target status data range to which the real-time status data belongs can be determined; and the target quality of service feature information corresponding to the target status data range under the target device type can be queried.

[0069] For example, the input of the service quality prediction model can be multiple pieces of device performance information and device status data of a device type, or multiple pieces of device performance information and the preset data range to which the device status data belongs, and the output can be the target service quality feature information of the device type under each state data (or the preset data range to which it belongs). Therefore, the service quality prediction model can be trained by using multiple pieces of device performance information of multiple device types and real device status data (or the preset data range to which it belongs) under multiple device types as training data, and using offline acquired real data of target service quality features of multiple device types at multiple preset levels under multiple preset device status data ranges as training targets (i.e., training labels). In addition, the value of the loss function can be calculated based on the target service quality feature information of the device type at each preset level under various state data (or the preset data range to which it belongs) predicted by the service quality prediction model, and the real data of the target service quality features of the device type at each preset level under various state data (or the preset data range to which it belongs), and the parameters of the service quality prediction model can be adjusted according to the value of the loss function, thereby training the service quality prediction model. The present disclosure does not limit the loss function, and any available loss function may be used to calculate the loss, such as, but not limited to, the cross-entropy loss function. In addition, according to an exemplary embodiment of the present disclosure, multiple different regression models may be trained and tested, and the regression model with the best test results may be selected to implement the service quality prediction model, thereby improving the accuracy of the level issuance decision by using the regression model that best suits the current video service scenario.

[0070] According to an exemplary embodiment of the present disclosure, the stored target service quality feature information of various device types at various preset levels under various state data (e.g., the preset data range to which they belong) can be updated regularly. For example, the stored target service quality feature information can be updated, or the target service quality feature information of new device types can be added, so as to ensure the timeliness and accuracy of the level distribution decision, so that the result of the level distribution decision is more in line with the current video distribution scenario. Specifically, the service quality prediction model can be updated regularly (at predetermined time intervals), and the target service quality feature information of various device types at various preset levels under various state data (e.g., the preset data range to which they belong) predicted by the updated service quality prediction model can be stored. For example, the real data of the target service quality features of various device types at various preset levels under various state data (e.g., multiple preset data ranges) can be obtained offline regularly, and the service quality prediction model can be retrained using the obtained real data to update the service quality prediction model.

[0071] Figure 3FIG. 4 is a flowchart illustrating an implementation of a level issuance decision according to an exemplary embodiment of the present disclosure.

[0072] Reference Figure 3 In step 301, multiple service quality feature data and target experience quality feature data for multiple device types at multiple preset levels can be obtained offline. Here, the multiple device types may include multiple electronic device types (e.g., smartphones, tablets, etc.), the multiple preset levels may include a combination of multiple preset video resolutions and multiple preset codecs, the multiple service quality features may include, but are not limited to, first screen time, freeze rate, start failure rate, frame loss rate, etc., and the target experience quality features may include, but are not limited to, average device playback time, etc.

[0073] In step 302, the frame loss rate that is most relevant to the target quality of experience characteristic can be selected from multiple quality of service characteristics as the target quality of service characteristic. Specifically, based on offline acquisition of multiple quality of service characteristic data at multiple preset levels for multiple device types and target quality of experience characteristic data, the quality of service characteristic that has the greatest impact on the quality of experience can be identified based on the relationship between each quality of service characteristic and the quality of experience characteristic. In an exemplary embodiment of the present disclosure, the quality of service characteristic that is most relevant to the quality of experience characteristic (average device playback duration) can be determined to be the frame loss rate.

[0074] In step 303, a service quality prediction model can be established based on multiple pieces of device performance information to fit the frame loss rate, obtain frame loss rate evaluation values for each device type at each level, and store them in Redis. Here, the multiple device performances may include device hardware performance such as the central processing unit (CPU), graphics processing unit (GPU), input / output (IO), and memory. The device performance information can be implemented as a comprehensive performance score of the device performance. In addition, since the frame loss rate is selected as the target service quality feature in step 302, the target service quality feature information predicted by the service quality prediction model can be implemented as a frame loss rate evaluation value. Here, the frame loss rate evaluation value can be an evaluation value indicating the degree of degradation of the frame loss rate. For example, a higher frame loss rate evaluation value indicates a higher frame loss rate. In addition, the service quality prediction model can be implemented as a regression model. Multiple different regression models can be trained and tested, and the regression model with the best test results can be selected to implement the service quality prediction model. In addition, the service quality prediction model can be updated regularly, and the frame loss rate evaluation values of each device type at each level predicted by the updated service quality prediction model can be updated to Redis.

[0075] In step 304, upon receiving a video delivery request initiated by a device, a level delivery combination is determined based on the frame loss rate evaluation value for each level of the corresponding device type compared with the frame loss rate evaluation threshold for each level. The corresponding video is then delivered to the device. The frame loss rate evaluation threshold for each level can be pre-set. For example, the frame loss rate evaluation thresholds for various levels can be adjusted through methods such as A / B testing to achieve the optimal quality of experience. For each level, the frame loss rate evaluation value for that level is compared with its threshold. If the frame loss rate evaluation value for that level exceeds the threshold, it indicates that playing a video at that level on the device will cause significant frame loss. Therefore, the video at that level is filtered out and not delivered. If the frame loss rate evaluation value for that level does not exceed the threshold, it indicates that playing a video at that level on the device will not cause significant frame loss. Therefore, the video at that level can be delivered. Ultimately, the determined deliverable levels are defined as level delivery combinations, and the videos corresponding to these level delivery combinations are delivered to the device.

[0076] Figure 4 FIG. 1 is a schematic diagram illustrating an implementation of a level-delivery decision according to an exemplary embodiment of the present disclosure.

[0077] Reference Figure 4 The client can report device type information to the server. According to an exemplary embodiment of the present disclosure, the server may include a delivery interface module, a broadcast control module, and a device type classification service module. When the delivery interface module in the server receives the device type information reported by the client, the delivery interface module may transmit the device type information to the broadcast control module, which in turn transmits the device type information to the device type classification service module. Based on the device type information, the device type grading service module can query the storage device for target service quality characteristic information for each level corresponding to the device type, such as the h265_0_360 frame loss rate evaluation value (h265 represents the codec mode, 0_360 represents the video resolution range, and the frame loss rate evaluation value represents the target service quality characteristic information), h265_361_480 frame loss rate evaluation value, h265_481_540 frame loss rate evaluation value, h265_541_576 frame loss rate evaluation value, h265_577_720 frame loss rate evaluation value, and h265_720+ frame loss rate evaluation value. The device type grading service module transmits the target service quality characteristic information for each level corresponding to the device type to the broadcast control module. Based on the received target service quality characteristic information for each level and in combination with the online threshold, the broadcast control module determines whether to issue the level, filters out the level combination, and transmits the video corresponding to the filtered level combination to the delivery interface module. The sending interface module sends the video corresponding to the filtered level combination to the client.

[0078] Figure 5 FIG. 4 is a block diagram illustrating a video distribution apparatus according to an exemplary embodiment of the present disclosure.

[0079] Reference Figure 5 According to an exemplary embodiment of the present disclosure, a video sending device 500 may include a receiving unit 501 , a query unit 502 , a determining unit 503 , and a sending unit 504 .

[0080] The receiving unit 501 may receive a video sending request from a target device, wherein the video sending request may include information about the target device type of the target device. Here, the target device may be implemented as follows: Figure 1 The user terminals 110 and 112 described in

[15] are shown in Figure 15. When a user uses a video service application installed on a target device, they can initiate a video download request to the server by opening the video service application, clicking on the video thumbnails displayed in the video service application, or clicking on the video service module in the video service application. When the target device initiates a video download request to the server, it can also include its own device type information. Alternatively, when the server receives a video download request from a target device, it can request the device type information from the target device through a specific channel.

[0081] Based on the target device type information, the query unit 502 may query target quality of service (QoS) feature information corresponding to the target device type. The target QoS feature information may include target QoS feature information for each of a plurality of preset levels. According to an exemplary embodiment of the present disclosure, the preset level may be a level corresponding to a preset video resolution value and a preset codec mode.

[0082] According to an exemplary embodiment of the present disclosure, the target service quality feature may be a service quality feature pre-selected from a plurality of service quality features. For example, the service quality feature may be selected based on demand, experience, etc. According to an exemplary embodiment of the present disclosure, the video delivery device 500 may further include a selection unit (not shown). The selection unit (not shown) may obtain the consumer service quality performance of each device type at each level, and based on the relationship between each service quality feature and the quality of experience (QoE) feature, find the service quality feature that has the greatest impact on the quality of experience as the target service quality feature. An artificial intelligence model may be used to estimate the target service quality feature information, and a level delivery decision may be executed based on the estimated target service quality feature information, so that such a level delivery decision result can bring the best user experience to the user and achieve the best quality of experience. Specifically, the selection unit (not shown) can select the service quality feature that is most relevant to the target experience quality feature (i.e., the service quality feature that has the greatest impact on the target experience quality feature data) from the multiple service quality features based on multiple service quality feature data (e.g., first screen time consumption, freeze rate, start failure rate, frame loss rate, etc.) of multiple device types at multiple preset levels and target experience quality feature data (e.g., average device playback time, etc.), as the target service quality feature (e.g., frame loss rate). Here, the multiple service quality feature data and the target experience quality feature data can be obtained offline. In the present disclosure, there is no limitation on the method of finding the service quality feature that has the greatest impact on the experience quality (i.e., the service quality feature that is most relevant to the experience quality) based on the multiple service quality feature data and the target experience quality feature data, and any feasible analysis method can be used.

[0083] According to an exemplary embodiment of the present disclosure, target service quality feature information at each of a plurality of preset levels corresponding to a plurality of device types may be stored in a storage device associated with a server.

[0084] According to an exemplary embodiment of the present disclosure, target service quality feature information corresponding to multiple device types is predicted and stored by a service quality prediction model. The service quality prediction model is used to use the device performance information corresponding to each device type to predict the target service quality feature information corresponding to the device type. According to an exemplary embodiment of the present disclosure, the service quality prediction model needs to be trained in advance, and the trained service quality prediction model is used to predict the target service quality feature information of the device type at each preset level. According to an exemplary embodiment of the present disclosure, the input of the service quality prediction model can be device performance information corresponding to a device type (which can be multiple pieces of device performance information), and the output can be the target service quality feature information of the device type at each preset level. Therefore, the service quality prediction model can be trained by using the multiple pieces of device performance information of the multiple device types as training data and using the real data of the target service quality features of the multiple device types at multiple preset levels as training targets (i.e., training labels). Here, the real data of the target service quality features of the multiple device types at multiple preset levels can be obtained offline. In addition, the value of the loss function can be calculated based on the target service quality feature information of the device type at each preset level predicted by the service quality prediction model and the actual data of the target service quality feature of the device type at each preset level, and the parameters of the service quality prediction model can be adjusted by the value of the loss function, thereby training the service quality prediction model. The present disclosure does not limit the loss function, and any available loss function can be used to calculate the loss, such as, but not limited to, the cross entropy loss function. In addition, according to an exemplary embodiment of the present disclosure, it is possible to attempt to train a variety of different regression models, and test the trained multiple regression models, and select the regression model with the best test effect to implement the service quality prediction model, so that the accuracy of the level issuance decision can be improved by the regression model that best suits the current video service scenario.

[0085] According to an exemplary embodiment of the present disclosure, the video delivery device 500 may further include an updating unit (not shown). The updating unit (not shown) may periodically update the target service quality feature information at each preset level in the plurality of preset levels corresponding to the stored plurality of device types. For example, the stored target service quality feature information of a certain device type may be updated, or the target service quality feature information of a new device type may be added, etc., thereby ensuring the timeliness and accuracy of the level delivery decision, so that the result of the level delivery decision is more in line with the current video delivery scenario. Specifically, the updating unit (not shown) may periodically (at predetermined time intervals) update the service quality prediction model and store the target service quality feature information at each preset level in the plurality of preset levels corresponding to the plurality of device types predicted by the updated service quality prediction model. For example, the updating unit (not shown) may periodically obtain real data of the target service quality features of the plurality of device types at the plurality of preset levels offline, and use the obtained real data to retrain the service quality prediction model, thereby updating the service quality prediction model.

[0086] The determining unit 503 may determine at least one preset level that matches the target device based on target service quality feature information at each of the multiple preset levels and target service quality feature information thresholds for the multiple preset levels.

[0087] According to an exemplary embodiment of the present disclosure, the determination unit 503 compares the target service quality feature information at each of the multiple preset levels with the target service quality feature information threshold of the corresponding preset level, and determines the preset level that matches the target device based on the comparison result. The determination unit 503 may determine the preset level for which the target service quality feature information does not exceed the target service quality feature information threshold as the preset level that matches the target device.

[0088] The sending unit 504 can send videos corresponding to at least one preset level to the target device. If the determining unit 503 determines multiple preset levels, the sending unit 504 can send videos corresponding to level sending combinations of the multiple preset levels to the target device, and the target device can select a preset level from the level sending combinations to play the video.

[0089] Furthermore, according to exemplary embodiments of the present disclosure, device performance information can be combined with the device's real-time status data to make level distribution decisions using an artificial intelligence model. This allows level distribution decisions to be made at the device level, better enabling personalized level distribution for each device. Target quality of service (QoS) characteristic information corresponding to various device types and various status data can be pre-stored. When a target device sends a video distribution request to a server, it can report not only its device type information but also its real-time status data. After receiving the target device's device type information and real-time status data at the receiving unit 501, the query unit 502 can query the target QoS characteristic information for each preset level corresponding to the device type and real-time status data. The determination unit 503 can determine a level combination suitable for the target device based on the query results and the target QoS characteristic information thresholds for each preset level. The distribution unit 504 can distribute the video stream corresponding to the level combination suitable for the target device to the target device, and the target device can select a video stream at a level from the level combination to play.

[0090] According to an exemplary embodiment of the present disclosure, target service quality characteristic information corresponding to multiple device types and multiple status data is predicted and stored through a service quality prediction model, and the service quality prediction model is used to use the device performance information corresponding to each device type and each status data under the device type to predict the target service quality characteristic information of each status data corresponding to the device type.

[0091] According to an exemplary embodiment of the present disclosure, the multiple status data are multiple preset status data ranges. Therefore, the query unit 502 can determine the target device type of the target device and the target status data range to which the real-time status data belongs based on the target device type information and the real-time status data, and query the target quality of service feature information corresponding to the target status data range under the target device type.

[0092] For example, the input of the service quality prediction model can be multiple pieces of device performance information and device status data of a device type, or multiple pieces of device performance information and the preset data range to which the device status data belongs, and the output can be the target service quality feature information of the device type under each corresponding state data (or the preset data range to which it belongs). Therefore, the service quality prediction model can be trained by using multiple pieces of device performance information of multiple device types and real device status data (or the preset data range to which it belongs) under multiple device types as training data, and using offline acquired real data of target service quality features of multiple device types at multiple preset levels under multiple preset device status data ranges as training targets (i.e., training labels). In addition, the value of the loss function can be calculated based on the target service quality feature information of the device type at each preset level under various state data (or the preset data range to which it belongs) predicted by the service quality prediction model, and the real data of the target service quality features of the device type at each preset level under various state data (or the preset data range to which it belongs), and the parameters of the service quality prediction model can be adjusted according to the value of the loss function, thereby training the service quality prediction model. The present disclosure does not limit the loss function, and any available loss function may be used to calculate the loss, such as, but not limited to, the cross-entropy loss function. In addition, according to an exemplary embodiment of the present disclosure, multiple different regression models may be trained and tested, and the regression model with the best test results may be selected to implement the service quality prediction model, thereby improving the accuracy of the level issuance decision by using the regression model that best suits the current video service scenario.

[0093] According to an exemplary embodiment of the present disclosure, an update unit (not shown) may periodically update stored target service quality feature information for various device types under various state data (e.g., the preset data ranges to which they belong) at various preset levels. For example, the update unit may update the stored target service quality feature information or add target service quality feature information for new device types, thereby ensuring the timeliness and accuracy of level-delivery decisions and making the results of level-delivery decisions more consistent with the current video delivery scenario. Specifically, the update unit (not shown) may periodically (at predetermined intervals) update the service quality prediction model and store target service quality feature information for various device types under various state data (e.g., the preset data ranges to which they belong) at various preset levels, as predicted by the updated service quality prediction model. For example, the update unit (not shown) may periodically obtain real data of target service quality features for various device types under various state data (e.g., the preset data ranges to which they belong) at various preset levels offline, and use the obtained real data to retrain the service quality prediction model, thereby updating the service quality prediction model.

[0094] Figure 6 is a block diagram of an electronic device 600 according to an exemplary embodiment of the present disclosure.

[0095] Reference Figure 6 The electronic device 600 includes at least one memory 601 and at least one processor 1502, wherein the at least one memory 601 stores a set of computer-executable instructions. When the computer-executable instruction set is executed by the at least one processor 602, the video delivery method according to the exemplary embodiment of the present disclosure is executed.

[0096] As an example, the electronic device 600 may be a PC, a tablet device, a personal digital assistant, a smart phone, or other device capable of executing the above-mentioned instruction set. Here, the electronic device 600 is not necessarily a single electronic device, but may also be any device or circuit collection capable of executing the above-mentioned instructions (or instruction set) individually or in combination. The electronic device 600 may also be part of an integrated control system or system manager, or may be configured as a portable electronic device interconnected with a local or remote (e.g., via wireless transmission) interface.

[0097] In electronic device 600, processor 602 may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, the processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0098] The processor 602 can execute instructions or codes stored in the memory 601, wherein the memory 601 can also store data. Instructions and data can also be sent and received over the network via the network interface device, wherein the network interface device can use any known transmission protocol.

[0099] The memory 601 may be integrated with the processor 602, for example, by placing RAM or flash memory within an integrated circuit microprocessor or the like. Furthermore, the memory 601 may comprise a separate device, such as an external disk drive, a storage array, or any other storage device usable by a database system. The memory 601 and the processor 602 may be operatively coupled or may communicate with each other, for example, via an I / O port, a network connection, or the like, such that the processor 602 can access files stored in the memory.

[0100] In addition, the electronic device 600 may further include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.) All components of the electronic device 600 may be connected to each other via a bus and / or a network.

[0101] According to an exemplary embodiment of the present disclosure, a computer-readable storage medium may also be provided, wherein when the instructions in the computer-readable storage medium are executed by at least one processor, the at least one processor is prompted to perform the video delivery method according to the present disclosure. Examples of computer-readable storage media here include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-RLTH , BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or ultra-fast digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device configured to store the computer program and any associated data, data files and data structures in a non-transitory manner and provide the computer program and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the above-mentioned computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. In addition, in one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0102] According to an exemplary embodiment of the present disclosure, a computer program product may also be provided, including computer instructions, which may be executed by at least one processor to implement the video sending method according to the exemplary embodiment of the present disclosure.

[0103] According to the present disclosure, a video sending method and a video sending device are provided, which can establish an artificial intelligence model for fitting the online service quality characteristic information of the device of the corresponding device type based on the various device performance information corresponding to the device type, and make a level sending decision based on the fitting result obtained by the artificial intelligence model, so as to realize the automatic decision of what level of video to send to the target device requesting the video, thereby replacing the original level sending decision by maintaining a blacklist by using the method of realizing the automated level sending decision based on the various device performance information corresponding to the device type, while saving labor costs, the level sending decision can be made more scientifically, timely and efficiently, thereby improving the user experience.

[0104] In addition, the video distribution method disclosed herein can also combine the device performance information corresponding to the device type with the real-time status data of the device to make level distribution decisions through an artificial intelligence model, thereby bringing the level distribution decisions down to the device granularity, better providing personalized level distribution for each device, and improving the user experience.

[0105] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

[0106] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A video delivery method, characterized in that: include: receiving a video delivery request from a target device, wherein the video delivery request includes target device type information of the target device; Based on the target device type information, querying target quality of service feature information corresponding to the target device type, wherein the target device type is one of a plurality of device types, and the target quality of service feature information corresponding to the plurality of device types is predicted and stored using a quality of service prediction model, the quality of service prediction model being configured to predict the target quality of service feature information corresponding to each device type using device performance information corresponding to the device type, wherein the target quality of service information includes target quality of service feature information for each of a plurality of preset levels, the preset levels being levels corresponding to preset video resolution values and preset codec modes; Determining at least one preset level that matches the target device based on target service quality feature information of each preset level among the multiple preset levels and a target service quality feature information threshold corresponding to the preset level; The video corresponding to the at least one preset level is sent to the target device.

2. The video distribution method according to claim 1, wherein: The service quality prediction model is trained by using the device performance information corresponding to the multiple device types as training data and using the real data of the target service quality characteristics of the multiple device types at the multiple preset levels as training targets.

3. The video distribution method according to claim 2, wherein: Also includes: The service quality prediction model is updated at predetermined time intervals, and target service quality feature information corresponding to the multiple device types predicted by the updated service quality prediction model is stored.

4. The video distribution method according to claim 1, wherein: The video delivery request also includes real-time status data of the target device; The querying of target quality of service feature information corresponding to the target device type based on the target device type information includes: Based on the information of the target device type and the real-time status data, the target service quality characteristic information corresponding to the target device type and the real-time status data is queried, wherein the target service quality characteristic information corresponding to the multiple device types and the multiple status data is predicted and stored through the service quality prediction model, and the service quality prediction model is used to use the device performance information corresponding to each device type and each status data under the device type to predict the target service quality characteristic information of each status data corresponding to the device type.

5. The video distribution method according to claim 4, wherein: The multiple status data are multiple preset status data ranges; The querying of target service quality feature information corresponding to the target device type and the real-time status data based on the target device type information and the real-time status data includes: Determining, based on the target device type information and the real-time status data, a target device type of the target device and a target status data range to which the real-time status data belongs; Query target quality of service feature information corresponding to the target status data range under the target device type.

6. The video distribution method according to claim 4, wherein: The service quality prediction model is trained by using multiple pieces of device performance information of the multiple device types and the real status data of the multiple device types as training data, and using the real data of the target service quality characteristics of the multiple device types at the multiple preset levels under the multiple status data as training targets.

7. The video distribution method according to claim 6, wherein: Also includes: The service quality prediction model is updated at predetermined time intervals, and target service quality feature information corresponding to the multiple device types under the multiple state data predicted by the updated service quality prediction model is stored.

8. The video distribution method according to claim 1, wherein: The determining, based on the target service quality information at each preset level among the multiple preset levels and the target service quality information threshold corresponding to the preset level, at least one preset level matching the target device includes: The target service quality characteristic information at each preset level among the multiple preset levels is compared with the target service quality characteristic information threshold of the corresponding preset level, and the preset level whose target service quality characteristic information does not exceed the target service quality characteristic information threshold is determined as the preset level matching the target device.

9. The video distribution method according to claim 1, wherein: Before receiving the video download request from the target device, it also includes: Based on the multiple quality of service feature data of the multiple device types at the multiple preset levels and the target quality of experience feature data, a quality of service feature that is most relevant to the target quality of experience feature is selected from the multiple quality of service features as the target quality of service feature.

10. The video distribution method according to claim 1, wherein: The target quality of service characteristic is frame loss rate.

11. A video distribution device, characterized in that: include: a receiving unit configured to receive a video delivery request from a target device, wherein the video delivery request includes target device type information of the target device; a query unit configured to query target service quality feature information corresponding to the target device type based on the target device type information, wherein the target device type is one of a plurality of device types, and the target service quality feature information corresponding to the plurality of device types is predicted and stored by a service quality prediction model, the service quality prediction model being configured to use device performance information corresponding to each device type to predict the target service quality feature information corresponding to the device type, wherein the target service quality feature information includes target service quality feature information for each preset level among a plurality of preset levels, the preset levels being levels corresponding to preset video resolution values and preset codec modes; a determining unit configured to determine at least one preset level matching the target device based on target service quality feature information at each preset level of the multiple preset levels and a target service quality parameter information threshold corresponding to the preset level; The sending unit is configured to send the video corresponding to the at least one preset level to the target device.

12. An electronic device, characterized in that: include: at least one processor; at least one memory storing computer-executable instructions, When the computer-executable instructions are executed by the at least one processor, they prompt the at least one processor to execute the video downloading method according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that When the instructions in the computer-readable storage medium are executed by at least one processor, the at least one processor is prompted to execute the video downloading method according to any one of claims 1 to 10.

Images