A video transcoding method, apparatus, device, and storage medium
By predicting the future status and popularity of videos, high-compression transcoding is applied only to visible and popular videos, solving the problem of wasted video transcoding resources and improving user experience and resource utilization efficiency.
Patent Information
- Application Number
- CN202211337551.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-28
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2042-10-28
AI Technical Summary
Existing video transcoding methods result in videos being hidden or deleted by submitters, preventing normal distribution and wasting transcoding resources.
By predicting the future state of the video to be transcoded, if the state is visible and the popularity is high, high compression rate transcoding is performed; otherwise, high compression rate transcoding is delayed or canceled.
It effectively saves transcoding resources, avoids high-compression transcoding of invisible videos, improves user viewing experience and reduces resource waste.
Smart Images

Figure CN115665417B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to computer technology, and more particularly to a video transcoding method, apparatus, device, and storage medium. Background Technology
[0002] With the rapid development of computer technology, there is a need to transcode user-uploaded videos to reduce bitrate and improve the user viewing experience. Currently, to ensure a good viewing experience, high compression rates may be used for transcoding. Efficient transcoding, which saves transcoding resources, is a problem that the industry is researching. Summary of the Invention
[0003] This disclosure provides a video transcoding method, apparatus, device, and storage medium to save transcoding resources and avoid wasting transcoding resources.
[0004] In a first aspect, embodiments of this disclosure provide a video transcoding method, including:
[0005] Obtain the target video to be transcoded;
[0006] Predict the future state of the target video to obtain the state prediction result corresponding to the target video;
[0007] If the state prediction result is a visible state, then the target video is subjected to popularity prediction to determine the first popularity prediction result corresponding to the target video;
[0008] If the first popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
[0009] Secondly, embodiments of this disclosure also provide a video transcoding apparatus, comprising:
[0010] The target video acquisition module is used to acquire the target video to be transcoded;
[0011] The video state prediction module is used to predict the future video state of the target video and obtain the state prediction result corresponding to the target video.
[0012] The popularity prediction module is used to predict the popularity of the target video if the state prediction result is a visible state, and to determine the first popularity prediction result corresponding to the target video.
[0013] The video transcoding module is used to perform high compression rate transcoding on the target video if the first popularity prediction result is a popular video.
[0014] Thirdly, embodiments of this disclosure also provide an electronic device, the electronic device comprising:
[0015] One or more processors;
[0016] Storage device for storing one or more programs.
[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the video transcoding method as described in any of the embodiments of this disclosure.
[0018] Fourthly, embodiments of this disclosure also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the video transcoding method as described in any of the embodiments of this disclosure.
[0019] In this embodiment, the future video state of the target video to be transcoded is predicted to obtain the state prediction result corresponding to the target video. If the state prediction result is a visible state, it indicates that the target video can be distributed normally in the future and will not be deleted or hidden by users. At this time, popularity prediction of the target video is allowed, and when the first popularity prediction result corresponding to the target video is a hot video, the target video is transcoded with a high compression rate. This allows for high compression rate transcoding of videos that will be visible in the future and have high popularity, thereby greatly saving transcoding resources while ensuring the user's viewing experience. It also avoids high compression rate transcoding of future non-visible videos, thus avoiding the waste of transcoding resources due to the inability to distribute the video. Attached Figure Description
[0020] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.
[0021] Figure 1 This is a schematic flowchart of a video transcoding method provided in an embodiment of this disclosure;
[0022] Figure 2 This is an example diagram of the network structure of a time series prediction model involved in an embodiment of this disclosure;
[0023] Figure 3 This is a schematic flowchart of another video transcoding method provided in this embodiment of the disclosure;
[0024] Figure 4 This is a schematic diagram of the structure of a video transcoding device provided in an embodiment of this disclosure;
[0025] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation
[0026] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.
[0027] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.
[0028] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.
[0029] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.
[0030] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".
[0031] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
[0032] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.
[0033] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.
[0034] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.
[0035] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.
[0036] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.
[0037] In implementing this disclosure, the applicant discovered that existing video transcoding methods, after high-compression transcoding, often result in the video being hidden or deleted by the submitter, preventing its proper distribution and wasting transcoding resources. To address this, the technical solution of this disclosure predicts the future state of the video and performs high-compression transcoding on videos that will be visible and popular in the future. This avoids high-compression transcoding of future unvisible videos, thus preventing the waste of transcoding resources due to undistributable videos and saving transcoding resources.
[0038] Figure 1 This is a flowchart illustrating a video transcoding method provided in an embodiment of the present disclosure. This embodiment is applicable to the situation of high compression rate transcoding of submitted videos. The method can be executed by a video transcoding device, which can be implemented in the form of software and / or hardware. Optionally, it can be implemented by an electronic device, such as a mobile terminal, a PC, or a server.
[0039] like Figure 1 As shown, the video transcoding method specifically includes the following steps:
[0040] S110, Obtain the target video to be transcoded.
[0041] The target video can refer to the video that requires high-compression transcoding. High-compression transcoding can refer to video transcoding using high-compression encoding methods, such as H.265 encoding.
[0042] Specifically, videos that have not yet undergone any high-compression level transcoding can be used as target videos. For example, newly uploaded videos by contributors can be used as target videos for transcoding.
[0043] For example, S110 may include: acquiring a newly uploaded submission video; performing low-compression transcoding on the submission video; and using the submission video as the target video for high-compression transcoding.
[0044] Submitted videos can refer to original videos created by the submitter (i.e., the video creator). Low-compression transcoding refers to video transcoding using low-compression encoding methods, such as H.264 encoding. The bitrate of videos encoded using low-compression methods is higher than that of videos encoded using high-compression methods.
[0045] Specifically, upon receiving a newly uploaded video from a contributor, the video can be directly transcoded at a low compression rate, such as H.264. This allows for the delivery of the low-compression transcoded video before high-compression transcoding, ensuring normal playback. After low-compression transcoding, the submitted video can be used as the target video for high-compression transcoding.
[0046] S120. Predict the future state of the target video and obtain the state prediction result corresponding to the target video.
[0047] The video status can include visible and invisible states. A visible state means the video is visible to the viewer, allowing them to watch it normally. An invisible state means the video is not visible to the viewer, meaning they cannot watch it. Invisible states can be set by the uploader through methods such as deletion or hiding the video. Visible and invisible states are opposites. A visible state indicates a video that can be distributed normally. An invisible state indicates a video that cannot be distributed normally. The status prediction result can be either visible or invisible.
[0048] Specifically, the future state of a target video can be predicted based on one or more prediction methods to obtain the corresponding state prediction result. For example, a neural network intelligent model (such as a binary classification network model) and / or the target contributor's historical submitted video state information can be used to predict whether the target video will be in an invisible state in the future, obtaining the target invisibility probability, and determining whether the state prediction result is an invisible state based on the target invisibility probability. Alternatively, a neural network intelligent model and / or the target contributor's historical submitted video state information can be used to predict whether the target video will be in a visible state in the future, obtaining the target visibility probability, and determining whether the state prediction result is a visible state based on the target visibility probability. The implementation processes of these two methods are similar. The following only describes the first method of video state prediction using invisibility probability; the other method of video state prediction using visibility probability will not be elaborated upon.
[0049] For example, S120 may include: predicting whether a target video will be in an invisible state in the future based on a video prediction network model, and determining a first invisibility probability corresponding to the target video; determining a second invisibility probability corresponding to the target video based on the number of videos submitted by the target video contributor within a preset historical time period and the number of normal videos currently in a visible state; determining a target invisibility probability corresponding to the target video based on the first invisibility probability and / or the second invisibility probability; and determining a state prediction result corresponding to the target video based on an invisibility probability threshold and a target invisibility probability.
[0050] The video prediction network model can be a neural network model used to predict the probability that a video will be in an invisible state in the future. The video prediction network model can be pre-trained based on sample data and label results to ensure the accuracy of the prediction. The target video contributor refers to the user who uploaded the target video. The preset historical time period can be a pre-set historical time period prior to the current moment. For example, the preset historical time period could be the 365 days prior to the current day. The target invisibility probability can be the probability that the target video will be in an invisible state in the future. The higher the invisibility probability, the greater the likelihood that the contributor has set the video to an invisible state. The invisibility probability threshold can be a pre-set minimum probability that the video will be in an invisible state. For example, the invisibility probability threshold could be 50%.
[0051] Specifically, in this embodiment, the target video information can be input into a pre-trained video prediction network model, so that the video prediction network model can predict whether the target video will be in an invisible state in the future based on the target video information, determine the first invisibility probability of the target video being in an invisible state in the future and output it, so that the first invisibility probability corresponding to the target video can be accurately predicted based on the video prediction network model.
[0052] This embodiment can also obtain the number of videos submitted by the target video contributor within a preset historical time period, as well as the current video status of each submitted video, and calculate the number of normal videos that are currently visible among all submitted videos based on the video status. The ratio of the number of normal videos to the number of submitted videos can be determined as the visibility probability, and the difference between 1 and the visibility probability can be used as the second invisibility probability. Alternatively, the difference between the number of submitted videos and the number of normal videos can be determined as the abnormal number, and the ratio of the abnormal number of videos to the number of submitted videos can be determined as the second invisibility probability.
[0053] This embodiment can directly determine the first or second invisibility probability as the target invisibility probability corresponding to the target video, or it can fuse the first and second invisibility probabilities to determine the target invisibility probability corresponding to the target video, thereby further improving the accuracy of invisibility probability prediction. For example, the first and second invisibility probabilities can be averaged or weighted, and the result can be used as the target invisibility probability; alternatively, the minimum probability between the first and second invisibility probabilities can be used as the target invisibility probability. By comparing the target invisibility probability with an invisibility probability threshold, if the target visibility probability is less than the threshold, the predicted state of the target video is determined to be visible; otherwise, the predicted state is determined to be invisible.
[0054] S130. If the state prediction result is a visible state, then perform a popularity prediction on the target video and determine the first popularity prediction result corresponding to the target video.
[0055] The initial popularity prediction result can include either trending or unpopular videos. Trending videos are those with high view counts, indicating popularity among users. Unpopular videos are those with low view counts, indicating low popularity among users. The initial popularity prediction result can refer to the popularity prediction result made before the target video undergoes its first high-rate transcoding.
[0056] Specifically, when the predicted state of the target video is "visible," it indicates that the video uploader will not delete or hide the video in the future, and it can be distributed normally. In this case, popularity prediction for the target video is allowed. For example, the target video information can be input into a pre-trained popularity prediction model. The popularity prediction model can directly predict the first popularity prediction result. Alternatively, it can output the prediction probability that the target video is a hot video. If the output prediction probability is greater than 0.5, the first popularity prediction result corresponding to the target video can be determined as a hot video; otherwise, it is a cold video.
[0057] It should be noted that if the state prediction result is "invisible", it means that the contributor is very likely to delete or hide the target video in the future, making it impossible for viewers to watch. In this case, high compression rate transcoding of the target video can be paused to avoid wasting resources caused by high compression rate transcoding of a future invisible video.
[0058] S140. If the first popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
[0059] Specifically, if the first popularity prediction result is a trending video, a high compression ratio encoding method can be used to transcode the target video to obtain a video file with a lower bitrate. This reduces bandwidth consumption during video transmission, thereby lowering bandwidth costs. Furthermore, this embodiment only performs high compression ratio transcoding on trending videos that are currently visible and have high popularity, rather than all uploaded videos. This improves the overall viewability of transcoded videos, reduces the number of transcoded videos that cannot be distributed normally, and minimizes transcoding resource waste. Under certain resource conditions, the bitrate of the video viewed by the user can be reduced, ensuring a better viewing experience.
[0060] The technical solution of this disclosure involves predicting the future video state of the target video to be transcoded, thereby obtaining a state prediction result corresponding to the target video. If the state prediction result indicates a visible state, it means that the target video can be distributed normally in the future and will not be deleted or hidden by users. In this case, popularity prediction of the target video is allowed, and when the first popularity prediction result corresponding to the target video is a hot video, the target video is transcoded with a high compression rate. This allows for high compression rate transcoding of videos that will be visible in the future and have high popularity, thus greatly saving transcoding resources while ensuring the user's viewing experience. It also avoids high compression rate transcoding of future non-visible videos, thereby avoiding the waste of transcoding resources.
[0061] Based on the above technical solutions, the video prediction network model can be a time series prediction model that predicts based on user submission behavior sequences, thereby further improving the accuracy of video state prediction. The time series prediction model can be pre-trained based on sample data and label results.
[0062] When the video prediction network model is a time series prediction model, based on the video prediction network model, predicting whether the target video will be in a non-visible state in the future and determining the first non-visible probability corresponding to the target video may include: obtaining the target video feature information, the historical video sequence feature information corresponding to the target video contributor, and the target video contributor feature information; inputting the target video feature information, the historical video sequence feature information, and the target video contributor feature information into the time series prediction model to predict whether the target video will be in a non-visible state in the future; and determining the first non-visible probability corresponding to the target video based on the output of the time series prediction model.
[0063] The target video feature information may include, but is not limited to, the video quality information and video bitrate information corresponding to the target video. Historical video sequence feature information may refer to the historical video feature information corresponding to each historical video submitted by the target video contributor. For example, historical video sequence feature information may include the historical video feature information corresponding to each historical video sorted sequentially by submission time. Historical video feature information may include, but is not limited to, the video quality information, video bitrate information, video frame rate information, video duration information, video width and height information, submission time, time interval since the last submission, number of videos submitted that day, and whether the video was deleted or hidden after a preset hour (e.g., 1 hour). Target video contributor feature information may include, but is not limited to, the contributor's average video deletion / hiding rate over a historical time period, such as the average video deletion / hiding rate over the past 7 days, the average video deletion / hiding rate over the past 30 days, etc.
[0064] Specifically, by inputting the target video feature information, historical video sequence feature information, and target video contributor feature information into a pre-trained time series prediction model, the time series prediction model can predict the first invisibility probability of the target video being in an invisible state in the future based on the input information, and output the first invisibility probability. Thus, the time series prediction model can be used to more accurately determine the first invisibility probability of the target video being in an invisible state in the future.
[0065] For example, a time series prediction model may include: a sequence network sub-model, a deep network sub-model, and a fully connected network sub-model. The sequence network sub-model can be a network model used to process sequence data composed of multiple videos. For example, the network structure of the sequence network sub-model may include, but is not limited to: a recurrent neural network (RNN) sub-model, a translational transformer sub-model, and a convolutional neural network (CNN) sub-model. The RNN sub-model may be, but is not limited to, a gated recurrent unit (GRU) sub-model. The deep neural network (DNN) sub-model can be a network model used to process single video data. The fully connected network sub-model can be used to perform fully connected linear activation processing on the outputs of the sequence network sub-model and the deep network sub-model to obtain the output first non-visible probability. Figure 2 An example diagram of the network structure of a time series prediction model is given.
[0066] For example, inputting target video feature information, historical video sequence feature information, and target video contributor feature information into a time series prediction model to predict whether the target video will be in a non-visible state in the future can include:
[0067] The historical video sequence feature information is input into the sequence network sub-model for time series information processing to obtain the first feature information; the target video feature information and the target video contributor feature information are input into the deep network sub-model for depth information processing to obtain the second feature information; the first feature information and the second feature information are input into the fully connected network sub-model for fully connected linear processing to obtain the first non-visible probability that the target video will be in a non-visible state in the future.
[0068] Specifically, such as Figure 2 As shown, historical video sequence feature information (x1, x2…xn) is input into the sequence network sub-model for time series information processing to obtain the first feature information. Simultaneously, the target video feature information and the target video contributor feature information can be input into the deep network sub-model (DNN) for depth information processing to obtain the second feature information. The first and second feature information are then input into the fully connected network sub-model (cancat) for fully connected and linear processing to obtain the first invisibility probability (rate) of the target video being in an invisible state in the future. By utilizing the sequence network sub-model and the deep network sub-model to process different forms of feature information respectively, the accuracy of information prediction can be further improved.
[0069] Based on the above technical solutions, S120 may include: obtaining the number of videos submitted by the target video contributor within a preset historical time period; if the number of submitted videos is greater than or equal to a video number threshold, then predicting the future video state of the target video and obtaining the state prediction result corresponding to the target video.
[0070] The preset historical time period can be a pre-defined historical period preceding the current moment. For example, the preset historical time period could refer to the last 365 days. The video quantity threshold can be a pre-defined minimum number of submitted videos that allow for a predicted video status. For example, the video quantity threshold could be set to 10.
[0071] Specifically, the system can obtain the number of videos submitted by the target video contributor within a preset historical time period and check whether this number is greater than or equal to a video quantity threshold to determine whether to allow video state prediction. If the number of submitted videos is greater than or equal to the threshold, it indicates that the contributor's activity level is high, and the future video state of the target video can be accurately predicted. In this case, prediction of the future video state is allowed, ensuring the accuracy of the prediction results. If the number of submitted videos is less than the threshold, it indicates that accurate video state prediction is not possible. In this case, high-compression transcoding of the target video can be paused to allocate limited transcoding resources to more valuable videos for high-compression transcoding, thereby reducing the video bitrate for users under certain resource conditions and improving the user viewing experience.
[0072] Figure 3 This is a flowchart illustrating another video transcoding method provided in this disclosure. Based on the above-disclosed embodiments, this disclosure provides a detailed description of delayed transcoding with high compression rates and cyclic transcoding with multiple high compression rate levels. Explanations of terms that are the same as or corresponding to those in the above-disclosed embodiments are not repeated here.
[0073] like Figure 3 As shown, the video transcoding method specifically includes the following steps:
[0074] S310, Obtain the target video to be transcoded;
[0075] S320. Obtain the number of videos submitted by the target video contributor within a preset historical time period, and check whether the number of submitted videos is greater than or equal to the video number threshold. If yes, proceed to step S330; otherwise, proceed to step S360.
[0076] S330. Predict the future state of the target video and obtain the state prediction result corresponding to the target video. If the state prediction result is a visible state, proceed to S340. If the state prediction result is a non-visible state, proceed to step S360.
[0077] S340. Perform popularity prediction on the target video and determine the first popularity prediction result corresponding to the target video. If the first popularity prediction result is a hot video, then proceed to step S350. If the first popularity prediction result is a cold video, then proceed to step S360.
[0078] S350: Perform high compression transcoding on the target video and execute S390.
[0079] S360: Get the current video playback count of the target video.
[0080] Specifically, when the number of submitted videos is less than the video quantity threshold, the predicted state of the target video is "invisible," or the predicted first popularity of the target video is "cold video," it indicates that high-compression transcoding of the target video needs to be paused. It should be noted that pausing high-compression transcoding does not mean that high-compression transcoding will not be performed on the target video at all; rather, it delays the triggering of high-compression transcoding so that it can still be performed when the target video's value reaches a certain level, thereby ensuring a better user viewing experience.
[0081] Since the target video has already undergone low-compression transcoding, when a viewer sends a request to play the target video, the low-compression target video can be distributed to the viewer, and the current video playback count, i.e., the current number of times the video has been played, can be counted in real time so as to trigger subsequent high-compression transcoding operations based on the current video playback count.
[0082] S370. If the current video playback count is greater than or equal to the target playback count threshold, then re-predict the popularity of the target video and determine the second popularity prediction result corresponding to the target video. If the second popularity prediction result is a hot video, then proceed to step S380. If the second popularity prediction result is a cold video, then proceed to step S391.
[0083] The target playback threshold can refer to the maximum number of video views reached when the video is currently hidden or deleted. The target playback threshold can be a fixed value, such as 50, or a dynamically determined value for different videos. For example, the target player threshold can be dynamically determined based on the target non-visible probability and target visible probability of the target video. The sum of the target non-visible probability and target visible probability is 1. The second popularity prediction result can refer to the result obtained from the popularity prediction operation triggered by the video playback volume. The first popularity prediction result can refer to the result obtained from the popularity prediction operation triggered by the predicted future video state.
[0084] Specifically, when the current video view count is greater than or equal to the target view count threshold, it indicates that the target video is still visible, meaning the uploader has not yet hidden or deleted it. In this case, a second popularity prediction can be performed on the target video to obtain a second popularity prediction result. If the current video view count is consistently less than the target view count threshold, it indicates that the uploader immediately hid or deleted the target video after uploading. In this case, high-rate transcoding is unnecessary, and the target view count threshold can effectively filter out videos that are immediately hidden or deleted after uploading, avoiding the waste of transcoding resources.
[0085] For example, when the number of submitted videos is less than the video number threshold, the method further includes: setting a preset threshold as the target playback threshold for the target video. Specifically, when the number of submitted videos is less than the video number threshold, there is no need to predict the future video status. In this case, the target playback threshold can be directly set to a fixed preset threshold, such as 50, for different videos.
[0086] When the state prediction result corresponding to the target video is an invisible state, or the first popularity prediction result corresponding to the target video is a cold video, the method further includes: obtaining the target invisibility probability corresponding to the target video; and determining the target playback threshold corresponding to the target video based on the correspondence between the invisibility probability and the initial playback threshold and the target invisibility probability.
[0087] The initial playback threshold refers to the minimum number of views required for a video to first meet the triggering condition. The invisibility probability is positively correlated with the initial playback threshold. The initial playback threshold increases with the invisibility probability; that is, videos with a higher invisibility probability will have a higher initial playback count when high compression rate transcoding is allowed, thus further avoiding resource waste.
[0088] Specifically, the process for determining the target's invisibility probability can be found in the description of the above embodiments. Based on the pre-set correspondence between the invisibility probability and the initial playback threshold, the initial playback threshold corresponding to the target's invisibility probability can be determined, and this initial playback threshold can be used as the target playback threshold for determining the playback volume. By dynamically determining the initial playback threshold corresponding to the target video based on the invisibility probability, resource waste caused by high-rate transcoding can be further avoided.
[0089] S380. If the second popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
[0090] Specifically, when the second popularity prediction result is a hot video, it indicates that the target video has reached the high compression rate transcoding requirement. At this time, high compression rate transcoding of the target video can be allowed, thereby achieving a delay in the timing of high compression rate transcoding and further ensuring the user's viewing experience.
[0091] S390. Detect whether there is a high compression ratio level that has not yet been transcoded. If so, proceed to step S391.
[0092] The compression ratio transcoding level can include multiple high compression ratio levels. For example, when transcoding using the high compression encoding method H.265, resolution levels such as 540p, 720p, and 1080p, as well as different compression rates, can be selected. Different resolution levels and compression rates correspond to different high compression ratio levels. Different high compression ratio levels can correspond to different transcoding optimization levels, so that each high compression ratio level can be transcoded sequentially based on transcoding priority.
[0093] Specifically, after the target video is predicted to be a hot video and undergoes one high-compression transcoding, for example, after S350 or S380, it can be checked whether there are still any high compression rate bits that have not been transcoded, in order to determine whether to continue the high compression rate transcoding operation. If there are no high compression rate bits that have not yet been transcoded, it means that all high compression rate bits have been transcoded, and the transcoding operation can be ended at this time.
[0094] S391. Based on the playback threshold sequence, update the target playback threshold, and return to S360 based on the updated target playback threshold.
[0095] Among them, the playback threshold sequence refers to the playback thresholds arranged from smallest to largest.
[0096] Specifically, when the second task prediction result is a cold video, or when there are still un-transcoded high compression ratio levels, the popularity prediction operation can be triggered again based on the current video playback volume of the target video. For example, the popularity prediction operation is triggered when the current video playback volume reaches a certain value again, until the popularity prediction result is a hot video, at which point the target video is transcoded with a high compression ratio. For example, the target playback volume threshold can be updated based on the playback volume threshold sequence, and the operation of step S360 can be returned based on the updated target playback volume threshold. Specifically, in the playback volume threshold sequence, the next playback volume threshold of the current target playback volume threshold can be used as the updated target playback volume threshold, so that the playback volume can continue to be judged based on the next playback volume threshold, and this process can be repeated cyclically until the current video playback volume never meets the transcoding condition, or there are no un-transcoded high compression ratio levels. For example, if the playback volume threshold sequence is {50, 100, 200, 500, 1000…}, and the current target playback volume threshold is 50, then the updated target playback volume threshold is 100.
[0097] The technical solution of this disclosure, when the state prediction result is an invisible state, the first popularity prediction result is a cold video, or the number of submitted videos is less than the video number threshold, detects whether the current video playback volume of the target video is greater than or equal to the target playback volume threshold. If so, the popularity of the target video is re-predicted to determine the second popularity prediction result corresponding to the target video. When the second popularity prediction result is a hot video, the target video is transcoded with a high compression rate, thereby delaying the timing of high compression rate transcoding, further ensuring the user viewing experience. Furthermore, the target playback volume threshold can effectively filter out videos that are set to private or deleted after uploading, further avoiding the waste of transcoding resources.
[0098] Figure 4 This is a schematic diagram of the structure of a video transcoding device provided in an embodiment of the present disclosure, as shown below. Figure 4 As shown, the device specifically includes: a target video acquisition module 410, a video status prediction module 420, a popularity prediction module 430, and a video transcoding module 440.
[0099] The system includes a target video acquisition module 410 for acquiring a target video to be transcoded; a video state prediction module 420 for predicting the future video state of the target video and obtaining a state prediction result corresponding to the target video; a popularity prediction module 430 for predicting the popularity of the target video if the state prediction result is a visible state, and determining a first popularity prediction result corresponding to the target video; and a video transcoding module 440 for transcoding the target video with a high compression rate if the first popularity prediction result is a hot video.
[0100] The technical solution provided in this disclosure predicts the future video state of the target video to be transcoded, obtaining a state prediction result corresponding to the target video. If the state prediction result is a visible state, it indicates that the target video can be distributed normally in the future and will not be deleted or hidden by users. In this case, popularity prediction is allowed for the target video. When the first popularity prediction result corresponding to the target video is a hot video, the target video is transcoded with a high compression rate. This allows for high compression rate transcoding of videos that will be visible in the future and have high popularity, thereby greatly saving transcoding resources while ensuring the user's viewing experience. It also avoids high compression rate transcoding of future non-visible videos, thus avoiding the waste of transcoding resources.
[0101] Based on the above technical solution, the video state prediction module 420 is specifically used for:
[0102] Obtain the number of videos submitted by the target video contributor within a preset historical time period; if the number of submitted videos is greater than or equal to a video number threshold, predict the future video state of the target video and obtain the state prediction result corresponding to the target video.
[0103] Based on the above technical solutions, the video state prediction module 420 includes:
[0104] The first non-visible probability determination unit is used to predict whether the target video will be in a non-visible state in the future based on the video prediction network model, and determine the first non-visible probability corresponding to the target video.
[0105] The second non-visible probability determination unit is used to determine the second non-visible probability of the target video based on the number of videos submitted by the target video contributor within a preset historical time period and the number of normal videos that are currently visible.
[0106] A target non-visibility probability determination unit is used to determine the target non-visibility probability corresponding to the target video based on the first non-visibility probability and / or the second non-visibility probability;
[0107] The state prediction result determination unit is used to determine the state prediction result corresponding to the target video based on the non-visibility probability threshold and the non-visibility probability of the target.
[0108] Based on the above technical solutions, the video prediction network model is a time series prediction model;
[0109] The first non-visible probability determination unit includes:
[0110] The feature information acquisition subunit is used to acquire the target video feature information corresponding to the target video, the historical video sequence feature information corresponding to the target video contributor, and the target video contributor feature information;
[0111] The feature information input subunit is used to input the target video feature information, the historical video sequence feature information, and the target video contributor feature information into the time series prediction model to predict whether the target video will be in an invisible state in the future.
[0112] The first non-visible probability subunit is used to determine the first non-visible probability corresponding to the target video based on the output of the time series prediction model.
[0113] Based on the above technical solutions, the time series prediction model includes: a sequence network sub-model, a deep network sub-model, and a fully connected network sub-model;
[0114] The feature information input subunit is specifically used for:
[0115] The historical video sequence feature information is input into the sequence network sub-model for time series information processing to obtain first feature information; the target video feature information and the target video contributor feature information are input into the deep network sub-model for depth information processing to obtain second feature information; the first feature information and the second feature information are input into the fully connected network sub-model for fully connected linear processing to obtain the first invisibility probability that the target video will be in an invisible state in the future.
[0116] Based on the above technical solutions, the device also includes:
[0117] The current video playback volume acquisition module is used to acquire the current video playback volume of the target video if the state prediction result is a non-visible state, the first popularity prediction result is a cold video, or the number of submitted videos is less than the video number threshold.
[0118] The second popularity prediction result determination module is used to re-predict the popularity of the target video if the current video playback volume is greater than or equal to the target playback volume threshold, and determine the second popularity prediction result corresponding to the target video.
[0119] The video transcoding module 440 is further configured to: if the second popularity prediction result is a hot video, then transcode the target video with a high compression rate.
[0120] Based on the above technical solutions, the device also includes:
[0121] The target playback threshold determination module is used to obtain the target invisibility probability corresponding to the target video when the state prediction result is an invisibility state or the first popularity prediction result is a cold video; and determine the target playback threshold corresponding to the target video based on the correspondence between the invisibility probability and the initial playback threshold and the target invisibility probability; wherein the invisibility probability and the initial playback threshold are positively correlated.
[0122] Based on the above technical solutions, the device also includes:
[0123] The target playback threshold update module is used to update the target playback threshold based on the playback threshold sequence if the second popularity prediction result is a cold video, or if there are still un-transcoded high compression ratio levels after high compression ratio transcoding of the target video, and return to perform the operation of obtaining the current video playback volume of the target video based on the updated target playback threshold.
[0124] Based on the above technical solutions, the target video acquisition module 410 is specifically used for:
[0125] Get the newly uploaded video submission; perform low-compression transcoding on the submitted video, and use the submitted video as the target video for high-compression transcoding.
[0126] The video transcoding apparatus provided in this disclosure can execute the video transcoding method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the video transcoding method.
[0127] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.
[0128] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Reference is made below. Figure 5 It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 5 The diagram below shows the structure of the terminal device or server 500. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.
[0129] like Figure 5 As shown, electronic device 500 may include a processing unit (e.g., central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. An edit / output (I / O) interface 505 is also connected to bus 504.
[0130] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.
[0131] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.
[0132] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
[0133] The electronic device provided in this embodiment and the video transcoding method provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.
[0134] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the video transcoding method provided in the above embodiments.
[0135] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.
[0136] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.
[0137] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.
[0138] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to:
[0139] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: acquire a target video to be transcoded; predict the future video state of the target video to obtain a state prediction result corresponding to the target video; if the state prediction result is a visible state, perform a popularity prediction on the target video to determine a first popularity prediction result corresponding to the target video; if the first popularity prediction result is a hot video, perform high compression rate transcoding on the target video.
[0140] Computer program code for performing the operations of this disclosure can be written in one or more programming languages or a combination thereof, including but not limited to object-oriented programming languages such as Java, Smalltalk, and C++, as well as conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0141] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0142] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".
[0143] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.
[0144] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0145] According to one or more embodiments of this disclosure, [Example 1] provides a video transcoding method, including:
[0146] Obtain the target video to be transcoded;
[0147] Predict the future state of the target video to obtain the state prediction result corresponding to the target video;
[0148] If the state prediction result is a visible state, then the target video is subjected to popularity prediction to determine the first popularity prediction result corresponding to the target video;
[0149] If the first popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
[0150] According to one or more embodiments of this disclosure, [Example 2] provides a video transcoding method, which further includes:
[0151] Optionally, predicting the future state of the target video to obtain a state prediction result corresponding to the target video includes:
[0152] Get the number of videos submitted by the target video contributor within a preset historical time period;
[0153] If the number of submitted videos is greater than or equal to the video number threshold, then the future video state of the target video is predicted to obtain the state prediction result corresponding to the target video.
[0154] According to one or more embodiments of this disclosure, [Example 3] provides a video transcoding method, which further includes:
[0155] Optionally, predicting the future state of the target video to obtain a state prediction result corresponding to the target video includes:
[0156] Based on a video prediction network model, the system predicts whether the target video will be in a non-visible state in the future, and determines the first non-visible probability corresponding to the target video.
[0157] The second non-visible probability corresponding to the target video is determined based on the number of videos submitted by the target video contributor within a preset historical time period and the number of normal videos that are currently visible.
[0158] The target non-visible probability corresponding to the target video is determined based on the first non-visible probability and / or the second non-visible probability;
[0159] Based on the non-visibility probability threshold and the target non-visibility probability, the state prediction result corresponding to the target video is determined.
[0160] According to one or more embodiments of this disclosure, [Example 4] provides a video transcoding method, which further includes:
[0161] Optionally, the video prediction network model is a time series prediction model;
[0162] The video prediction network model predicts whether the target video will be in a non-visible state in the future, and determines the first non-visible probability corresponding to the target video, including:
[0163] Obtain the target video feature information corresponding to the target video, the historical video sequence feature information corresponding to the target video contributor, and the target video contributor feature information;
[0164] The target video feature information, the historical video sequence feature information, and the target video contributor feature information are input into the time series prediction model to predict whether the target video will be in an invisible state in the future.
[0165] Based on the output of the time series prediction model, the first non-visible probability corresponding to the target video is determined.
[0166] According to one or more embodiments of this disclosure, Example 5 provides a video transcoding method, which further includes:
[0167] Optionally, the time series prediction model includes: a sequence network sub-model, a deep network sub-model, and a fully connected network sub-model;
[0168] The step of inputting the target video feature information, the historical video sequence feature information, and the target video contributor feature information into the time series prediction model to predict whether the target video will be in a non-visible state in the future includes:
[0169] The historical video sequence feature information is input into the sequence network sub-model for time series information processing to obtain the first feature information;
[0170] The target video feature information and the target video contributor feature information are input into the deep network sub-model for depth information processing to obtain the second feature information;
[0171] The first feature information and the second feature information are input into the fully connected network sub-model for fully connected linear processing to obtain the first non-visible probability that the target video will be in a non-visible state in the future.
[0172] According to one or more embodiments of this disclosure, [Example Six] provides a video transcoding method, further comprising:
[0173] Optionally, the method further includes:
[0174] If the state prediction result is a non-visible state, the first popularity prediction result is a cold video, or the number of submitted videos is less than the video number threshold, then obtain the current video playback count of the target video;
[0175] If the current video playback count is greater than or equal to the target playback count threshold, then the target video is re-predicted for popularity, and a second popularity prediction result corresponding to the target video is determined.
[0176] If the second popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
[0177] According to one or more embodiments of this disclosure, [Example Seven] provides a video transcoding method, further comprising:
[0178] Optionally, when the state prediction result is a non-visible state, or the first heat prediction result is a cold video, the method further includes:
[0179] Obtain the probability that the target is not visible in the target video;
[0180] Based on the correspondence between the non-visibility probability and the initial playback threshold and the target non-visibility probability, the target playback threshold corresponding to the target video is determined;
[0181] Among them, the non-visible probability is positively correlated with the initial playback threshold.
[0182] According to one or more embodiments of this disclosure, [Example Eight] provides a video transcoding method, further comprising:
[0183] Optionally, the method further includes:
[0184] If the second popularity prediction result is a cold video, or if there are still un-transcoded high compression levels after high compression rate transcoding of the target video, then the target playback threshold is updated based on the playback threshold sequence, and the operation of obtaining the current video playback volume of the target video is returned based on the updated target playback threshold.
[0185] According to one or more embodiments of this disclosure, [Example Nine] provides a video transcoding method, further comprising:
[0186] Optionally, obtaining the target video to be transcoded includes:
[0187] Get newly uploaded submitted videos;
[0188] The submitted video is transcoded at a low compression rate, and then used as the target video for high compression rate transcoding.
[0189] According to one or more embodiments of this disclosure, [Example 10] provides a video transcoding apparatus, including:
[0190] The target video acquisition module is used to acquire the target video to be transcoded;
[0191] The video state prediction module is used to predict the future video state of the target video and obtain the state prediction result corresponding to the target video.
[0192] The popularity prediction module is used to predict the popularity of the target video if the state prediction result is a visible state, and to determine the first popularity prediction result corresponding to the target video.
[0193] The video transcoding module is used to perform high compression rate transcoding on the target video if the first popularity prediction result is a popular video.
[0194] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.
[0195] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.
[0196] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.
Claims
1. A video transcoding method, characterized in that, include: Obtain the target video to be transcoded; Obtain the target video feature information, the historical video sequence feature information, and the target video contributor feature information corresponding to the target video; input the target video feature information, the historical video sequence feature information, and the target video contributor feature information into a time series prediction model to predict whether the target video will be in an invisible state in the future; Based on the output of the time series prediction model, determine the first non-visible probability corresponding to the target video; Based on the number of videos submitted by the target video contributor within a preset historical time period and the number of normal videos currently visible, a second non-visible probability is determined for the target video. The target video contributor refers to the user who submitted and uploaded the target video. The target non-visible probability corresponding to the target video is determined based on the first non-visible probability and / or the second non-visible probability; Based on the non-visibility probability threshold and the non-visibility probability of the target, the state prediction result corresponding to the target video is determined; If the state prediction result is a visible state, then the target video is subjected to popularity prediction to determine the first popularity prediction result corresponding to the target video; If the first popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
2. The video transcoding method according to claim 1, characterized in that, After obtaining the target video to be transcoded, the process also includes: The number of videos submitted by the target video contributor within a preset historical time period is obtained, where the target video contributor refers to the user who submitted and uploaded the target video; If the number of submitted videos is greater than or equal to the video number threshold, then the future video state of the target video is predicted to obtain the state prediction result corresponding to the target video.
3. The video transcoding method according to claim 1, characterized in that, The time series prediction model includes: a sequence network sub-model, a deep network sub-model, and a fully connected network sub-model; The step of inputting the target video feature information, the historical video sequence feature information, and the target video contributor feature information into a time series prediction model to predict whether the target video will be in a non-visible state in the future includes: The historical video sequence feature information is input into the sequence network sub-model for time series information processing to obtain the first feature information; The target video feature information and the target video contributor feature information are input into the deep network sub-model for depth information processing to obtain the second feature information; The first feature information and the second feature information are input into the fully connected network sub-model for fully connected linear processing to obtain the first non-visible probability that the target video will be in a non-visible state in the future.
4. The video transcoding method according to claim 1 or 2, characterized in that, The method further includes: If the state prediction result is an invisible state, the first popularity prediction result is a cold video, or the number of videos submitted by the target video contributor within a preset historical time period is less than the number of videos threshold, then the current video playback count of the target video is obtained. The target video contributor refers to the user who submitted and uploaded the target video. If the current video playback count is greater than or equal to the target playback count threshold, then the target video is re-predicted for popularity, and a second popularity prediction result corresponding to the target video is determined. If the second popularity prediction result is a hot video, then the target video is transcoded with a high compression rate.
5. The video transcoding method according to claim 4, characterized in that, When the state prediction result is a non-visible state, or the first heat prediction result is a cold video, the method further includes: Obtain the non-visible probability of the target video; Based on the correspondence between the non-visibility probability and the initial playback threshold and the target non-visibility probability, the target playback threshold corresponding to the target video is determined; Among them, the non-visible probability is positively correlated with the initial playback threshold.
6. The video transcoding method according to claim 4, characterized in that, The method further includes: If the second popularity prediction result is a cold video, or if there are still un-transcoded high compression levels after high compression rate transcoding of the target video, then the target playback threshold is updated based on the playback threshold sequence, and the operation of obtaining the current video playback volume of the target video is returned based on the updated target playback threshold.
7. The video transcoding method according to claim 1, characterized in that, The process of obtaining the target video to be transcoded includes: Get newly uploaded submitted videos; The submitted video is transcoded at a low compression rate, and then used as the target video for high compression rate transcoding.
8. A video transcoding device, characterized in that, include: The target video acquisition module is used to acquire the target video to be transcoded; The video state prediction module is used to obtain the target video feature information, the historical video sequence feature information, and the target video contributor feature information corresponding to the target video; and input the target video feature information, the historical video sequence feature information, and the target video contributor feature information into the time series prediction model to predict whether the target video will be in an invisible state in the future. Based on the output of the time series prediction model, a first invisibility probability corresponding to the target video is determined; based on the number of videos submitted by the target video contributor within a preset historical time period and the number of normal videos currently visible, a second invisibility probability corresponding to the target video is determined, wherein the target video contributor refers to the user who submitted and uploaded the target video; based on the first invisibility probability and / or the second invisibility probability, a target invisibility probability corresponding to the target video is determined; based on the invisibility probability threshold and the target invisibility probability, a state prediction result corresponding to the target video is determined. The popularity prediction module is used to predict the popularity of the target video if the state prediction result is a visible state, and to determine the first popularity prediction result corresponding to the target video. The video transcoding module is used to perform high compression rate transcoding on the target video if the first popularity prediction result is a popular video.
9. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the video transcoding method as described in any one of claims 1-7.
10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the video transcoding method as described in any one of claims 1-7.
Citation Information
Patent Citations
Video transcoding method and device, server and medium
CN114257815A
