WO2020119670A1

WO2020119670A1 - Video transcoding method and device

Info

Publication number: WO2020119670A1
Application number: PCT/CN2019/124232
Authority: WO
Inventors: 李庆文
Original assignee: 阿里巴巴集团控股有限公司
Priority date: 2018-12-11
Filing date: 2019-12-10
Publication date: 2020-06-18
Also published as: CN111314706A; CN111314706B

Abstract

Disclosed in embodiments of the present application are a video transcoding method and device. The method comprises: obtaining a source video; determining path information for transcoding the source video to a target video, the path information comprising a transcoding path and a transcoding mode between nodes in the transcoding path; and transcoding the source video on the basis of the determined path information. The technical solution provided by the present application can improve the video transcoding efficiency.

Description

Video transcoding method and device

This application requires the priority of the Chinese patent application filed on December 11, 2018 with the application number 201811510054.7 and the invention titled "a video transcoding method and device", the entire contents of which are incorporated by reference in this application.

Technical field

This application relates to the field of Internet technology, in particular to a video transcoding method and device.

Background technique

With the continuous development of Internet technology, more and more video playback platforms have emerged. In order to provide users with videos with different picture quality, the video playback platform usually needs to transcode the source video to generate multiple videos with different resolutions and different bit rates.

At present, for some multi-level dependent transcoding scenarios, for example, the scenario of producing high frame rate video, it is usually necessary to perform different transcoding tasks separately through multiple transcoding machines. For example, before transcoding the source video, you need to convert the source video at a high frame rate to generate an intermediate result, and then transcode the intermediate result to generate multiple videos with different resolutions and different bit rates . In such a transcoding scenario that relies on intermediate results, after the transcoding task that generates the intermediate results is completed, it is usually necessary to upload the intermediate results to an external storage platform. The intermediate result is read multiple times in the external storage platform. These uploading processes and the process of reading from the external storage platform are both time-consuming, resulting in low efficiency of video transcoding.

Therefore, there is an urgent need to provide a faster video transcoding method.

Summary of the invention

The purpose of the embodiments of the present application is to provide a video transcoding method and device, which can improve the efficiency of video transcoding.

In order to achieve the above object, an embodiment of the present application provides a video transcoding method, the method includes: acquiring a source video; determining path information from the source video to a target video; wherein, the path information includes transcoding A transcoding method between the code path and the nodes in the transcoding path; based on the acquired path information, transcoding the source video.

In order to achieve the above object, the embodiments of the present application also provide a video transcoding device. The device includes: a video acquisition unit for acquiring a source video; and a path determination unit for determining transcoding from the source video to a target video Path information; wherein, the path information includes a transcoding path and a transcoding method between the nodes in the transcoding path; a transcoding unit is used to perform the source video based on the obtained path information Transcoding.

In order to achieve the above object, the embodiments of the present application also provide a video transcoding device. The device includes a memory and a processor. The memory is used to store a computer program. When the computer program is executed by the processor, the foregoing Video transcoding method.

It can be seen from the above that, after obtaining the source video, for the target video to be output, the technical solution provided in this application can determine the path information of transcoding from the source video to the target video. Wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path. In this way, the source video and other intermediate nodes in the transcoding path can be transcoded in turn according to the transcoding path, through the transcoding method between the nodes in the transcoding path, and the output Target video. In this way, the entire transcoding process can be completed in one transcoding machine, reducing the uploading process and the process of reading from an external storage platform, thereby reducing the time for video transcoding and improving the efficiency of video transcoding.

BRIEF DESCRIPTION

In order to more clearly explain the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some of the embodiments described in this application. For those of ordinary skill in the art, without paying any creative labor, other drawings can be obtained based on these drawings.

FIG. 1 is a schematic diagram of a video transcoding method in an embodiment of this application;

2 is a schematic diagram of a directed acyclic transcoding architecture in the implementation of this application;

3 is a schematic structural diagram of a video transcoding device according to an embodiment of the present application;

4 is a schematic structural diagram of another video transcoding device in an embodiment of the present application.

detailed description

In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described The embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the scope of protection of this application.

The present application provides a video transcoding method, which can be applied to a terminal device having an image processing function. The terminal device may be, for example, a desktop computer, a notebook computer, a tablet computer, a workstation, or the like. In addition, the method can also be applied to a business server of a video playing website. The business server may be an independent server or a server cluster composed of multiple servers.

Please refer to FIG. 1, the video transcoding method provided in this application includes the following steps.

S11: Obtain the source video.

In this embodiment, by transcoding the source video, multiple videos with different resolutions and different code rates can be generated.

In this embodiment, the method of acquiring the source video may include reading the source video from the storage path or receiving the source video from another terminal device according to the provided storage path.

S13: Determine path information of transcoding from the source video to the target video; wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path.

In this embodiment, in the actual application scenario, for the target video to be output, transcoding from the source video to the target video may require a transcoding process between multiple nodes. For example, for some multi-level dependent transcoding scenes, for example, scenes that produce high frame rate video, you need to first convert the source video to a high frame rate source video through Frame Rate Conversion (FRC), that is, generate An intermediate result, then transcode the intermediate result to generate a video with the specified resolution, and finally transcode the video to generate the target video with the specified video format and specified resolution. In this way, in order to output the target video with the specified video format and specified resolution, two intermediate nodes need to be passed through. The entire transcoding process can be divided into four levels, which are the first level with the source video as the root node. The source video with high frame rate is the second level of the child node, the third level with the video with the specified resolution as the third level node, and the fourth level with the target video with the specified resolution and the specified video format as the leaf node . Among them, the output of the node of the second level depends on the node of the first level, the output of the node of the third level depends on the node of the second level, and the output of the node of the fourth level depends on the node of the third level. In order to implement such a complex multi-layer dependent transcoding process in the terminal device, the path information of transcoding from the source video to the target video may be determined first. Wherein, the path information may include a transcoding path and a transcoding method between each node in the transcoding path. In this way, based on the determined path information, a transcoding process of the source video can be implemented in a terminal device to obtain the target video.

In one embodiment, the transcoding process from the source video to the target video may be a multi-layer dependent transcoding process that already exists in practical applications, then multiple existing transcodings can be separated from the existing In the task, the dependencies between the transcoding tasks are obtained, and the inputs and outputs in these transcoding tasks can be used as the nodes in a transcoding path. In this way, a transcoding path for transcoding from the source video to a target video and various nodes in the transcoding path can be obtained. The transcoding method between each node can also be obtained directly from each separate transcoding task. In this way, the path information of transcoding from the source video to the target video can be determined. In this embodiment, the transcoding method corresponds to the video transcoding parameters required for transcoding one video to another video, and the parameter values of the video transcoding parameters may be based on the video parameters and audio parameters of the two videos The parameter value is determined. The transcoding parameters may include, for example, fidelity, resolution, transmission code rate, and so on. After these transcoding parameters are set, the video can be transcoded, so that the transcoded video conforming to these transcoding parameters can be obtained.

In one embodiment, considering that the transcoding process from the source video to the target video may not be found in the existing multi-level dependent transcoding process, for this reason, in practical applications, depth Learning methods to build a path recognition model for identifying transcoding path information. For example, the path information corresponding to the video group composed of the source video and the target video may be identified by a support vector machine (SVM). Wherein, the source video and the target video may be used as a path start node and a path end node in the path identification model, respectively. Specifically, when constructing the path recognition model, a training sample set may be obtained in advance, and the training sample set may be used to train the path recognition model, so that the path recognition model can recognize the input video group Corresponding path information. The training sample set may include a sample video group whose corresponding transcoding path conforms to the path information and a sample video group whose corresponding transcoding path does not conform to the path information. The sample video group may include sample videos corresponding to the path start node and the path end node, respectively. In this way, during the training process, the sample video groups in the training sample set may be sequentially input into the path recognition model. An initial neural network can be constructed in the path recognition model, and initial prediction parameters can be preset in the neural network. After processing the input sample video group through the initial prediction parameters, a prediction result of the sample video group can be obtained, and the prediction result can be used to characterize whether the transcoding path corresponding to the sample video group conforms to the The path information. Specifically, when processing the sample video group, the path recognition model may first separately extract a first feature vector corresponding to the parameter information of the source video and a second feature vector corresponding to the parameter information of the target video. Elements in the first feature vector may be parameter values of various parameters of the source video, for example, parameter values of video parameters or audio parameters, and video parameters may include video resolution, video bit rate, video frame rate, and video Format and so on. Similarly, the elements in the first feature vector may be parameter values of various parameters of the target video. In this way, the path identification model can read the parameter value of each parameter in the source video corresponding to the path start node in the sample video group, and the parameter value of each parameter in the target video corresponding to the path end node, and follow In the order of reading, the parameter values form the first feature vector and the second feature vector. In practical applications, since the number of parameters is usually large, the dimension of the extracted feature vector is also large, which will consume more resources to process the feature vector. In view of this, in this embodiment, a convolutional neural network (Convolutional Neural Network, CNN) can also be used to process the sample video group, so as to obtain feature vectors with smaller dimensions for subsequent identification processing.

In this embodiment, after processing the data of the input sample video group through the neural network, a probability value vector of the sample video group can be obtained. The probability value for the specified path information may be included in the probability value vector. The probability value vector may include two probability values, and these two probability values respectively represent the probability that the transcoding path conforms to the specified path information and the probability that it does not conform to the specified path information. For example, after inputting a set of corresponding sample video groups whose transcoding paths conform to the specified path information, a probability value vector of (0.4, 0.8) can be obtained through the path identification model, where 0.4 indicates that the transcoding path conforms to the specified path The probability of information, 0.8 means the probability that the transcoding path does not conform to the specified path information. Since the initial prediction parameters in the path recognition model may not be set accurately enough, the probability results predicted by the path recognition model may be inconsistent with the actual situation. For example, the above input is a sample video group whose transcoding path conforms to the specified path information, but in the obtained probability vector, the probability that the transcoding path matches the specified path information is only 0.4, and that the transcoding path does not match the specified path information. The probability is 0.8. In this case, it indicates that the prediction result is incorrect. At this time, the initial prediction parameter in the path recognition model may be adjusted according to the difference between the prediction result and the correct result. Specifically, the sample video group may have a theoretical probability value result. For example, the theoretical probability value result of the transcoding path conforming to the specified path information may be (1, 0), where 1 represents the probability value that the transcoding path conforms to the specified path information. At this time, the predicted probability value result can be subtracted from the theoretical probability value result to obtain the difference between the two, and then the obtained difference can be used to adjust the initial prediction parameters of the neural network. After the prediction parameters are processed again for the sample video group, the obtained prediction result is consistent with the correct result. In this way, after training a large number of training samples, the path recognition model can distinguish whether the transcoding path corresponding to the sample video group conforms to the specified transcoding path, thereby identifying the path that matches the actual transcoding path corresponding to the sample video group information.

S15: Transcode the source video based on the obtained path information.

In this embodiment, after determining the path information from the source video to the target video to be output, the source video can be converted into a terminal device based on the determined path information Coding process to get the target video. Specifically, the source video may be transcoded according to the transcoding path included in the path information through a transcoding method between nodes in the transcoding path, so that the target video may be obtained. For example, the transcoding path includes four nodes, which are a root node, a child node, a three-level node, and a four-level node in the transcoding order. The root node is the source video, and the four-level node is the target video. Then, you can first transcode the root node through the transcoding method between the root node and the child node to obtain the video corresponding to the child node. Then, by transcoding between the child node and the third-level node, the video corresponding to the child node is transcoded to obtain the video corresponding to the third-level node. Finally, by transcoding between the third-level node and the fourth-level node, the video corresponding to the third-level node is transcoded to obtain the target video. In this way, the entire multi-level dependent transcoding process is completed in one terminal device, reducing the uploading process of the video corresponding to the intermediate node and reading these video processes from the external storage platform, thereby reducing the time for video transcoding, Improve the efficiency of video transcoding.

In this embodiment, in some complex transcoding scenarios, the video to be output often includes at least two kinds of target videos, so that multiple transcoding paths for these target videos will appear. When there are overlapping paths in these transcoding paths, in order to avoid repeating the transcoding process in these overlapping paths, you can first determine whether there are overlapping paths in the transcoding paths for each target video. If there are, you can first follow the overlap The path transcodes the source video to obtain an intermediate node, and then transcodes the intermediate node according to non-overlapping paths in each transcoding path. Specifically, for the two target videos, the first target video and the second target video, the first path information for determining the transcoding from the source video to the first target video, and the first path information for transcoding from the source video to the second target video are determined. After the second path information, if there is an overlapping path between the first path information corresponding to the first target video and the second path information corresponding to the second target video, the source video may be first transcoded according to the overlapping path to obtain After the intermediate node, the intermediate node is transcoded according to the non-overlapping paths in the first path information and the second path information, respectively. In order to realize the above process, a transcoding structure including a directed acyclic graph (DAG) of these nodes can be constructed according to the up-down dependencies between the nodes in the path information for these target videos.

For example, in an application scenario where the source video is transcoded into a target video with Dolby sound effects, a variety of target videos with different Dolby sound effects need to be output. The transcoding paths in the path information corresponding to these target videos include partially overlapping paths. Then, the corresponding DAG transcoding structure can be constructed according to the upper and lower dependencies between the nodes in these path information to merge these overlapping paths, and the subsequent output can be directly completed in a terminal device according to this DAG transcoding structure. Complex transcoding process for multiple target videos. As shown in FIG. 2, among the nodes for the final output target video, the paths corresponding to the nodes Node11 with Dolby audio 11, Node12 with Dolby audio 12 and Node13 with Dolby audio 13 all include The path from root to root Node1 with Dolby Audio 1. Then, the overlapping partial paths can be merged, so that the terminal device can first transcode the source video according to the partial overlapping path to obtain the intermediate node Node1, and then respectively follow the non-overlapping paths to the intermediate node Node1 Perform transcoding to get nodes Node11, Node12 and Node13. Similarly, for other nodes of the final output target video, the Node 21 with Dolby Audio 21 and the Node 22 with Dolby Audio 22 can also be combined with overlapping paths in the above manner, so that a DAG transcoding structure can be constructed Transcoding path. In FIG. 2, the root node root is the first level, the nodes Node1 and Node2 form the second level, and the nodes Node11, Node12, Node13, Node21, and Node22 form the third level, and there is a dependency relationship between each level. The first level is used as an input, and the second level is output, and then the second level can be used as an input, and the third level can be output. In this embodiment, after the construction of the DAG transcoding structure is completed, the construction of the DAG transcoding structure can be extended horizontally and vertically according to the increased transcoding service requirements. As shown by the dashed box in Figure 2, in order to increase the output of the Dolby audio 211, Dolby audio 212, Dolby audio 213, Dolby audio 221 and Dolby audio 311 video transcoding business requirements According to the determined path information of the video to be output, the node Node3 can be added horizontally, the nodes Node211, Node212, Node213 and Node221, and the nodes Node3 and Node311 can be added vertically. In this way, the transcoding path of the DAG transcoding structure can be constructed, and complex transcoding service requirements can be realized in one terminal device. Moreover, during the transcoding process, these intermediate nodes are directly stored locally in the terminal device, so that there is no need to upload the video corresponding to the intermediate node to the external storage device, nor to read the video from the external storage device multiple times for subsequent conversion Code process.

In this embodiment, the functions implemented in the above method steps may be implemented by a computer program, and the computer program may be stored in a computer storage medium. Specifically, the computer storage medium may be coupled with the processor, so that the processor can read the computer program in the computer storage medium. When the computer program is executed by the processor, the following functions can be realized:

S11: Obtain the source video;

S13: Determine path information of transcoding from the source video to the target video; wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path;

S15: Transcode the source video based on the determined path information.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

According to the transcoding path, transcoding the source video through a transcoding method between nodes in the transcoding path to obtain the target video.

In one embodiment, when the computer program is executed by the processor, when the video to be output includes at least two kinds of target videos, the following steps are further implemented:

When there is an overlapping path between the first path information corresponding to the first target video and the second path information corresponding to the second target video, transcode the source video according to the overlapping path to obtain an intermediate node, and then respectively The non-overlapping paths in the first path information and the second path information transcode the intermediate node.

Input a video group composed of the source video and the target video into a path recognition model, and determine path information of transcoding from the source video to the target video; wherein, the source video and the target video are respectively As a path start node and a path end node in the path identification model.

Input a video group composed of the source video and the target video into a path recognition model to extract the first feature vector corresponding to the parameter information of the source video and the parameter information of the target video respectively through the feature recognition model A corresponding second feature vector, and determining the predicted value corresponding to the vector group formed by the first feature vector and the second feature vector through the path recognition model;

The path information characterized by the predicted value is used as the path information for transcoding from the source video to the target video.

Obtaining a training sample set, the training sample set including a corresponding sample video group whose transcoding path conforms to the path information and a corresponding sample video group whose transcoding path does not conform to the path information; the sample video group includes Sample videos corresponding to the path start node and the path end node respectively;

Input a sample video group in the training sample set into a path recognition model, where the path recognition model includes initial prediction parameters;

Processing the input sample video group through the initial prediction parameter to obtain a prediction result of the sample video group, where the prediction result is used to characterize whether a transcoding path corresponding to the sample video group conforms to the path information ;

If the prediction result is incorrect, adjust the initial prediction parameters in the path recognition model according to the difference between the prediction result and the correct result, so that the sample video After the group was processed, the predicted results were consistent with the correct results.

Referring to FIG. 3, this application also provides a video transcoding device, which includes:

The video obtaining unit 100 is used to obtain the source video;

The path determining unit 200 is configured to determine path information for transcoding from the source video to the target video; wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path;

The transcoding unit 300 is configured to transcode the source video based on the acquired path information.

In one embodiment, the transcoding unit is further configured to transcode the source video according to the transcoding path and through a transcoding method between nodes in the transcoding path to obtain the target video.

In one embodiment, when the video to be output includes at least two target videos,

The transcoding unit is further configured to transcode the source video according to the overlapping path when there is an overlapping path between the first path information corresponding to the first target video and the second path information corresponding to the second target video To obtain an intermediate node, and then transcode the intermediate node according to the non-overlapping paths in the first path information and the second path information, respectively.

In one embodiment, the path determination unit is further configured to input a video group composed of the source video and the target video into a path identification model, and determine path information for transcoding from the source video to the target video ; Wherein the source video and the target video are used as the path start node and path end node in the path recognition model, respectively;

The video transcoding device provided in the embodiments of the present specification, in which the specific functions of each unit module can be explained in comparison with the aforementioned method embodiments in the present specification, and can achieve the technical effects of the aforementioned method embodiments, which will not be repeated here.

Referring to FIG. 4, the present application also provides a video transcoding device. The device includes a memory and a processor. The memory is used to store a computer program. When the computer program is executed by the processor, the following steps are implemented:

S11: Obtain the source video;

S15: Transcode the source video based on the determined path information.

In this embodiment, the memory may include a physical device for storing information, usually the information is digitized and then stored on a medium using electrical, magnetic, or optical methods. The memory described in this embodiment may further include: devices that use electrical energy to store information, such as RAM, ROM, etc.; devices that use magnetic energy to store information, such as hard disks, floppy disks, magnetic tapes, magnetic core memories, bubble memories, and U disks ; Devices that use optical methods to store information, such as CDs or DVDs. Of course, there are other types of memory, such as quantum memory, graphene memory, and so on.

In this embodiment, the processor may be implemented in any suitable manner. For example, the processor may employ, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gate, switch, dedicated integration Circuit (Application Specific Integrated Circuit, ASIC), programmable logic controller and embedded microcontroller form, etc.

It can be seen from the above that the technical solution provided in this application and the technical solution provided in this application, after acquiring the source video, can determine the path information of transcoding from the source video to the target video for the target video to be output. Wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path. In this way, the source video and other intermediate nodes in the transcoding path can be transcoded in turn according to the transcoding path, through the transcoding method between the nodes in the transcoding path, and the output Target video. In this way, the entire transcoding process can be completed in one transcoding machine, reducing the uploading process and the process of reading from an external storage platform, thereby reducing the time for video transcoding and improving the efficiency of video transcoding.

In the 1990s, the improvement of a technology can be clearly distinguished from the improvement in hardware (for example, the improvement of circuit structures such as diodes, transistors, and switches) or the improvement in software (the improvement of the process flow). However, with the development of technology, the improvement of many methods and processes can be regarded as a direct improvement of the hardware circuit structure. Designers almost get the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be realized by hardware physical modules. For example, a programmable logic device (Programmable Logic Device, PLD) (such as a field programmable gate array (Field Programmable Gate Array, FPGA)) is such an integrated circuit, and its logic function is determined by the user programming the device. Designers can program themselves to "integrate" a digital system on a PLD without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is also mostly implemented with "logic compiler" software, which is similar to the software compiler used in program development and writing, but before compilation The original code must also be written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), and HDL is not only one kind, but there are many kinds, such as ABEL (Advanced Boolean Expression) Language , AHDL (AlteraHardwareDescriptionLanguage), Confluence, CUPL (CornellUniversityProgrammingLanguage), HDCal, JHDL (JavaHardwareDescriptionLanguage), Lava, Lola, MyHDL, PALASM, RHDL (RubyHardwareDescription) It is VHDL (Very-High-Speed Integrated Circuit Hardware Description) and Verilog2. Those skilled in the art should also be clear that by simply programming the method flow in the above hardware description languages and programming into the integrated circuit, the hardware circuit that implements the logic method flow can be easily obtained.

Those skilled in the art also know that, in addition to implementing the video image transcoding device in a pure computer-readable program code manner, it is entirely possible to make the video image transcoding device into logic gates, switches, application specific integrated circuits, Program the logic controller and embedded microcontroller to achieve the same function. Therefore, such a video image transcoding device can be regarded as a hardware component, and the device for implementing various functions included therein can also be regarded as a structure within the hardware component. Or even, the means for realizing various functions can be regarded as both a software module of an implementation method and a structure within a hardware component.

It can be known from the description of the above embodiments that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product can be stored in a storage medium, such as ROM/RAM, magnetic disk , Optical discs, etc., including several instructions to enable a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods described in various embodiments of the present application or some parts of the embodiments.

The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the video image transcoding device, reference may be made to the introduction of the foregoing method embodiments for comparison and explanation.

The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application may also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media including storage devices.

Although the present application has been described through the embodiments, those of ordinary skill in the art know that there are many variations and changes in the present application without departing from the spirit of the application, and it is hoped that the appended claims include these variations and changes without departing from the spirit of the application.

Claims

A video transcoding method, characterized in that the method includes:

Get the source video;

Determining the path information of transcoding from the source video to the target video; wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path;

Transcode the source video based on the determined path information.
The method of claim 1, transcoding the source video based on the determined path information includes:

According to the transcoding path, transcoding the source video through a transcoding method between nodes in the transcoding path to obtain the target video.
The method according to claim 1, when the video to be output includes at least two target videos, transcoding the source video based on the determined path information includes:

When there is an overlapping path between the first path information corresponding to the first target video and the second path information corresponding to the second target video, transcode the source video according to the overlapping path to obtain an intermediate node, and then respectively The non-overlapping paths in the first path information and the second path information transcode the intermediate node.
The method according to claim 1, wherein the path information is determined in the following manner:

Input a video group composed of the source video and the target video into a path recognition model, and determine path information of transcoding from the source video to the target video; wherein, the source video and the target video are respectively As a path start node and a path end node in the path identification model.
The method according to claim 4, wherein the path information is determined in the following manner:

Input a video group composed of the source video and the target video into a path recognition model to extract the first feature vector corresponding to the parameter information of the source video and the parameter information of the target video respectively through the feature recognition model A corresponding second feature vector, and determining the predicted value corresponding to the vector group formed by the first feature vector and the second feature vector through the path recognition model;

The path information characterized by the predicted value is used as the path information for transcoding from the source video to the target video.
The method according to claim 4, wherein the path recognition model is determined in the following manner:

Obtaining a training sample set, the training sample set including a corresponding sample video group whose transcoding path conforms to the path information and a corresponding sample video group whose transcoding path does not conform to the path information; the sample video group includes Sample videos corresponding to the path start node and the path end node respectively;

Input a sample video group in the training sample set into a path recognition model, where the path recognition model includes initial prediction parameters;

Processing the input sample video group through the initial prediction parameter to obtain a prediction result of the sample video group, where the prediction result is used to characterize whether a transcoding path corresponding to the sample video group conforms to the path information ;

If the prediction result is incorrect, adjust the initial prediction parameters in the path recognition model according to the difference between the prediction result and the correct result, so that the sample video After the group was processed, the predicted results were consistent with the correct results.
A video transcoding device, characterized in that the device includes:

Video acquisition unit for acquiring source video;

A path determining unit, configured to determine path information for transcoding from the source video to the target video; wherein, the path information includes a transcoding path and a transcoding method between nodes in the transcoding path;

The transcoding unit is configured to transcode the source video based on the acquired path information.
The device according to claim 7, wherein the transcoding unit is further configured to transcode the source video according to the transcoding path and through a transcoding method between nodes in the transcoding path Code to get the target video.
The apparatus according to claim 7, wherein when the video to be output includes at least two target videos,

The transcoding unit is further configured to transcode the source video according to the overlapping path when there is an overlapping path between the first path information corresponding to the first target video and the second path information corresponding to the second target video To obtain an intermediate node, and then transcode the intermediate node according to the non-overlapping paths in the first path information and the second path information, respectively.
The apparatus according to claim 7, wherein the path determining unit is further configured to input a video group composed of the source video and the target video into a path recognition model, and determine that transcoding from the source video is Path information of the target video; wherein, the source video and the target video are respectively used as a path start node and a path end node in the path identification model.
A video transcoding device, characterized in that the device includes a memory and a processor, and the memory is used to store a computer program, and when the computer program is executed by the processor, any of claims 1 to 6 is implemented One of the methods.