Video thumbnail acquisition method and apparatus, electronic device, and storage medium

Abstract

The application discloses a video thumbnail acquisition method and device, electronic equipment and a storage medium, relates to the technical field of video management and streaming media, and is applied to a service node and comprises the following steps: acquiring the play times of each uncut video in a current sampling period by playing logs according to a placeholder preconfigured in streaming media information; acquiring node heat values of each uncut video according to the play times, historical play heat and heat decay coefficients of each uncut video, and screening and acquiring a hot video set; and sending the node heat values to a management node, so that a thumbnail of a target hot video is generated. The technical scheme of the embodiment of the application ensures the real-time performance of the management node in generating a hot video thumbnail through the rapid positioning of uncut videos; the screening and acquisition of hot videos avoid the occupation of more storage resources by preview pictures of non-popular videos; and in addition, the automatic acquisition mode of the video thumbnail reduces the labor cost consumed in the acquisition process.

Description

Technical Field

[0001] This invention relates to the field of video management and streaming media technology, and in particular to a method, apparatus, electronic device and storage medium for obtaining video thumbnails. Background Technology

[0002] With the increasing popularity of internet TV, or OTT TV (Over the Top TV), users' demand for diversified video content is growing, and their demand for real-time and personalized video content is increasing significantly. Users need to quickly locate content nodes and accurately jump to them when playing videos.

[0003] To ensure users accurately locate video content, preview images (i.e., thumbnails) need to be accurately presented to users when they fast forward or rewind. In existing technologies, the generation of video thumbnails relies on static screenshots or manual annotation, which requires extracting keyframe images from all videos in advance and storing them as separate files for preview display when users fast forward or rewind.

[0004] However, this method of obtaining video thumbnails has poor real-time performance and cannot obtain preview information of popular videos in a timely manner; at the same time, the pre-generated keyframe images occupy a lot of storage space and require a continuous investment of human resources for maintenance and updates; in addition, preview images of non-popular video content may occupy a lot of storage and transmission resources, resulting in resource waste. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for obtaining video thumbnails, in order to solve the problem of low efficiency in obtaining video thumbnails.

[0006] According to one aspect of the present invention, a method for obtaining video thumbnails is provided, applied to a business node in a cluster, comprising:

[0007] Based on the pre-configured placeholders in the streaming media information, the playback count of each unsliced ​​video in the current sampling period is obtained through the playback log;

[0008] The node popularity value of each unsliced ​​video is obtained based on the number of times it has been played, its historical popularity, and its popularity decay coefficient; wherein, the popularity decay coefficient is related to the content type of the current unsliced ​​video.

[0009] Based on the node popularity value of each unsliced ​​video, the unsliced ​​videos are sorted to filter and obtain a set of hot videos based on the sorting results;

[0010] The node popularity value of each hot video in the hot video set is sent to the management node, so that the management node can calculate the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

[0011] The step of obtaining the playback count of each unsegmented video in the current sampling period through the playback log based on the pre-configured placeholders in the streaming media information includes: obtaining the playback count of each unsegmented video in the current sampling period through the playback log based on the pre-configured placeholders in the metadata of the streaming media information.

[0012] The step of obtaining the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video is played, its historical playback popularity, and its popularity decay coefficient includes: obtaining the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video is played, its historical playback popularity, its popularity decay coefficient, its sampling period duration, and its playback gain coefficient; wherein, the playback gain coefficient is related to the content type of the current unsliced ​​video.

[0013] According to another aspect of the present invention, a method for obtaining video thumbnails is provided, applied to a management node of a cluster, comprising:

[0014] Based on the node popularity value of each hot video in the hot video set issued by each business node, the comprehensive popularity value of each hot video is calculated and obtained.

[0015] Based on the comprehensive popularity value, historical calculated popularity value, and popularity decay factor of each of the aforementioned trending videos, the current calculated popularity value of each of the aforementioned trending videos is obtained; wherein, the popularity decay factor is related to the playback time of the current trending video;

[0016] The target hot video is obtained based on the current calculated popularity value of each hot video, and the target hot video is cached and sliced ​​in real time to obtain a thumbnail of the target hot video.

[0017] The step of obtaining the current calculated popularity value of each of the aforementioned hot videos based on their comprehensive popularity value, historical calculated popularity value, and popularity decay factor includes: obtaining the current calculated popularity value of each of the aforementioned hot videos based on their comprehensive popularity value, historical calculated popularity value, popularity decay factor, sampling period duration, playback weight coefficient, and popularity amplification coefficient.

[0018] The step of performing real-time caching and slicing processing on the target hot video to obtain a thumbnail of the target hot video includes: obtaining parallel slicing rules based on the resource utilization rate of the management node and the content type of the target hot video, and performing real-time caching and slicing processing on the target hot video according to the parallel slicing rules to obtain a thumbnail of the target hot video.

[0019] According to one aspect of the present invention, a video thumbnail acquisition device is provided, applied to a service node of a cluster, comprising:

[0020] The playback count acquisition module is used to obtain the playback count of each unsliced ​​video in the current sampling period by means of the playback log based on the pre-configured placeholders in the streaming media information.

[0021] The node popularity acquisition module is used to acquire the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video has been played, its historical playback popularity, and its popularity decay coefficient; wherein, the popularity decay coefficient is related to the content type of the current unsliced ​​video.

[0022] The hot video acquisition module is used to sort the unsliced ​​videos according to the node popularity value of each unsliced ​​video, so as to filter and obtain a set of hot videos based on the sorting results;

[0023] The node popularity sending module is used to send the node popularity value of each hot video in the hot video set to the management node, so that the management node can calculate and obtain the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

[0024] According to one aspect of the present invention, a video thumbnail acquisition device is provided, applied to a management node of a cluster, comprising:

[0025] The comprehensive popularity acquisition module is used to calculate and acquire the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node.

[0026] The current popularity acquisition module is used to acquire the current calculated popularity value of each of the aforementioned trending videos based on their comprehensive popularity value, historical calculated popularity value, and popularity decay factor; wherein, the popularity decay factor is related to the playback period of the current trending video.

[0027] The cache slicing execution module is used to obtain target hot videos based on the current calculated popularity value of each hot video, and to perform real-time cache slicing processing on the target hot videos to obtain thumbnails of the target hot videos.

[0028] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the video thumbnail acquisition method described in Embodiment 1 or Embodiment 2 of the present invention, or to perform the video thumbnail acquisition method described in Embodiment 3 or Embodiment 4 of the present invention.

[0029] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the video thumbnail acquisition method described in Embodiment 1 or Embodiment 2 of the present invention, or to implement the video thumbnail acquisition method described in Embodiment 3 or Embodiment 4 of the present invention.

[0030] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the video thumbnail acquisition method described in Embodiment 1 or Embodiment 2 of the present invention, or implements the video thumbnail acquisition method described in Embodiment 3 or Embodiment 4 of the present invention.

[0031] The technical solution of this invention involves the following steps: First, the service node obtains the playback count of each unsliced ​​video in the current sampling period through playback logs based on pre-configured placeholders in the streaming media information. This enables rapid acquisition of unsliced ​​video playback data, ensuring the real-time generation of trending video thumbnails by the subsequent management node. Then, based on the playback count, historical playback popularity, and popularity decay coefficient of the unsliced ​​video, the node popularity value of the unsliced ​​video is obtained. Based on this, a set of trending videos is selected, ensuring the selection of trending videos and avoiding the waste of storage and transmission resources caused by preview images of non-popular video content. Finally, the node popularity values ​​of each trending video in the trending video set are sent to the management node, enabling the management node to generate thumbnails of the target trending videos based on the node popularity values. This achieves automatic acquisition of trending video thumbnails, avoiding the manpower and storage costs associated with pre-generation methods.

[0032] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of a video thumbnail acquisition method provided in Embodiment 1 of the present invention;

[0035] Figure 2 This is a flowchart of another video thumbnail acquisition method provided in Embodiment 2 of the present invention;

[0036] Figure 3 This is a flowchart of another video thumbnail acquisition method provided in Embodiment 3 of the present invention;

[0037] Figure 4 This is a flowchart of another video thumbnail acquisition method provided in Embodiment 4 of the present invention;

[0038] Figure 5 This is a schematic diagram of the structure of a video thumbnail device provided in Embodiment 5 of the present invention;

[0039] Figure 6 This is a schematic diagram of the structure of another video thumbnail device provided according to Embodiment Six of the present invention;

[0040] Figure 7 This is a schematic diagram of the structure of an electronic device that implements the video thumbnail acquisition method of this invention. Detailed Implementation

[0041] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0042] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0043] Example 1

[0044] Figure 1 This is a flowchart of a video thumbnail acquisition method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where a business node obtains the number of video playbacks based on pre-configured placeholders and filters and obtains popular videos based on the number of video playbacks. This method can be executed by the video thumbnail acquisition device in Embodiment 5. The video thumbnail acquisition device can be implemented in hardware and / or software and can be configured in an electronic device (e.g., a cluster of business servers). Figure 1 As shown, the method includes:

[0045] S101. Based on the pre-configured placeholders in the streaming media information, obtain the playback count of each unsliced ​​video in the current sampling period through the playback log.

[0046] In a cluster used to manage video playback and storage, service nodes are responsible for the actual operation of video storage and playback, connecting to user-side clients and responding to playback and download requests from those clients. Management nodes are responsible for overall resource coordination, status monitoring, and service scheduling to ensure stable cluster operation. HLS (HTTP Live Streaming) is a transmission protocol used for video-on-demand and live streaming. Streaming media information is video information based on the streaming media protocol, which consists of metadata and media data. Metadata refers to the attribute information of the video being played, such as video ID (Identity Document), duration, size, and category tags. Media data refers to the files that actually store the video content.

[0047] A URI (Uniform Resource Identifier) ​​is a standard way to identify internet resources. It is used to uniquely identify resources on the network and indicate how to access them. URIs can be pre-segmented in metadata or media data. This way, the assembled URI will have pre-reserved placeholders of a specified type. These placeholders indicate the location of the information to be assembled, thereby enabling the dynamic assembly of information such as content, channel, and picture quality, improving the flexibility and management efficiency of playback links. Metadata can include ".m3u8" files, and media data can include ".ts" files.

[0048] Placeholders may include content provider (CP) identifiers, content type identifiers, intro identifiers, subset identifiers, and resolution identifiers; among them, the content provider identifier indicates the source of the video content, for example, from video platform A; the content type identifier indicates the content category of the video, for example, movie, TV series, and short video; the intro identifier indicates the series name; the subset identifier indicates the season and / or episode number; and the resolution identifier indicates the resolution level.

[0049] For example, when a user requests to play content such as "Season 2, Episode 3 of Drama B, 1080p resolution, from video platform A", the aforementioned "content provider identifier, content type identifier, intro identifier, subset identifier, and resolution identifier" will be replaced with "video platform A, TV series, Drama B, Season 2, Episode 3, 1080p resolution" respectively. By using pre-configured placeholders in the streaming media information, it is ensured that only one valid record is retained for the same content on the same channel. Taking the above type of placeholder as an example, the number of plays for each video can be counted according to the dimensions of "channel-type-intro-subset". Optionally, in this embodiment of the invention, the number and type of pre-configured placeholders in the streaming media information are not specifically limited.

[0050] If a video has already been sliced, meaning a thumbnail for the video preview has been generated, then the video is a sliced ​​video and obviously does not need to be sliced ​​again. If a video has not been sliced, meaning a thumbnail for the video preview has not been generated, then the video is an unsliced ​​video and obviously needs to be sliced. According to a preset sampling period (e.g., 5 minutes), metadata or media data in the playback log is collected, and the playback count of each video is aggregated through placeholders in the metadata or media data. The playback count of the unsliced ​​videos is retained according to the slice identifier.

[0051] For example, based on the placeholders of the above types, the output format can be "Content Provider-Content Type-Intro-Subset-Play Count", with only one record retained for each playback content. Specifically, compared to collecting playback logs from the Service Node System (SNS), collecting playback logs from the local scheduling system (SLB) of the Service Cluster Cache (Cache Service) master node not only reduces the number of interactions and transmission bandwidth but also significantly reduces the amount of data collected, thus improving data collection efficiency.

[0052] In addition, by setting the time granularity to collect playback logs, if a collection failure occurs during the collection process (e.g., network interruption), an exponential backoff strategy can be used to retry. For example, it can retry up to 3 times (i.e., the retry threshold), with intervals of 3 seconds, 6 seconds, and 12 seconds (i.e., different retry intervals). If the collection fails 3 times, an error log is recorded and a collection failure alarm is sent. At the same time, the collection task of the current cycle is terminated to avoid blocking subsequent collection tasks.

[0053] Optionally, in this embodiment of the invention, obtaining the number of times each unsliced ​​video is played in the current sampling period through the playback log based on the pre-configured placeholders in the streaming media information includes: obtaining the number of times each unsliced ​​video is played in the current sampling period through the playback log based on the pre-configured placeholders in the metadata of the streaming media information.

[0054] Specifically, if placeholders are pre-configured in media data, due to the severe fragmentation of media data, user playback behavior may only be concentrated in a few segments (e.g., the opening and climax segments). As a result, the collected segment data cannot represent the true popularity of the entire video, and the popularity representation is prone to distortion. At the same time, the cost of media data aggregation is high. For example, a movie may have thousands of segments, and hundreds of thousands of user visits will generate hundreds of millions of segment data. Complex correlation algorithms are needed to restore the popularity of the entire series, and the computational cost increases exponentially with the number of segments, thus greatly increasing the cost of popularity calculation.

[0055] Furthermore, the resource consumption during media data collection increases exponentially, resulting in excessively high collection costs. Simultaneously, collecting media data from user playback requires gathering all relevant segmented data, leading to a lengthy collection cycle and consequently, poor timeliness in generating subsequent thumbnails. Therefore, pre-configuring placeholders in the metadata not only avoids distortion in popularity representation and reduces the collection and computational resources consumed in the playback count aggregation process, but also improves the efficiency of playback data collection.

[0056] S102. Obtain the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video is played, its historical playback popularity, and its popularity decay coefficient; wherein, the popularity decay coefficient is related to the content type of the current unsliced ​​video.

[0057] Because the number of videos to be sliced ​​in the playback log within a sampling period is too large, if all videos to be sliced ​​are used as the object of popularity calculation, a large number of cold data will be mistakenly calculated as hot videos, which will greatly increase the computational overhead of low-usage data. Therefore, each video to be sliced ​​can be sorted by the number of times it has been played, its historical playback popularity, and its popularity decay coefficient. Among them, the historical playback popularity refers to the popularity value obtained by the current video to be sliced ​​in the previous sampling period, which reflects the historical popularity performance of the current video to be sliced ​​in the current business node. If the video to be sliced ​​is appearing for the first time and there is no historical playback popularity, it can be configured with a pre-set initial popularity value.

[0058] The heat decay coefficient reflects the degree of decay in historical playback heat. The larger the heat decay coefficient, the faster the cold data is eliminated, and it is used to numerically correct historical playback heat. The number of plays reflects the playback status of the video to be sliced ​​in the current sampling period. The node heat value reflects the heat performance of the video to be sliced ​​in the current business node. The current number of plays and historical playback heat are both positively correlated with the node heat value, that is, the more plays and the greater the historical playback heat, the greater the node heat value. For example, the sum of the historical playback heat after heat decay coefficient correction and the number of plays can be used as the node heat value.

[0059] Specifically, the popularity decay coefficient is related to the content type of the currently unsliced ​​video; for example, for short videos, the popularity decay coefficient is a larger value (e.g., 0.08) to accelerate the response speed of trending videos and promptly eliminate old videos; for film and television comprehensive videos, the popularity decay coefficient is a smaller value (e.g., 0.03) to effectively protect long-tail videos with old popularity while boosting new trending videos; for news and information videos, the popularity decay coefficient is a moderate value (e.g., 0.06) to balance the timeliness of trending topics with the stability of content.

[0060] S103. Sort the unsliced ​​videos according to the node popularity value of each unsliced ​​video, and filter and obtain a set of hot videos according to the sorting results.

[0061] Based on the node popularity value of each unsliced ​​video, the videos are sorted. The most popular videos are then selected as the hot video set to filter out cold data from the videos to be sliced, reducing the computational burden of popularity metrics. Specifically, for filtering the hot video set, a hot number threshold can be set, adding a specified number of unsliced ​​videos based on the popularity ranking. Alternatively, a hot percentage threshold can be set, adding a specified percentage of unsliced ​​videos based on the popularity ranking. Furthermore, both the hot number threshold and the hot percentage threshold can be used as sorting criteria, with the larger or smaller value used as the filtering condition to avoid an excessively large or small number of hot videos in the hot video set.

[0062] S104. Send the node popularity value of each hot video in the hot video set to the management node, so that the management node can calculate and obtain the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

[0063] As described in the above technical solution, the node popularity value reflects the popularity value of the current video in a business node. Since the video often responds to the playback needs of different users through different business nodes, the management node will summarize and sum the node popularity values ​​of the same video in different business nodes, that is, summarize them into a comprehensive popularity value. Then, based on the comprehensive popularity value and the historical calculated popularity value, the current calculated popularity value is obtained. Based on the current calculated popularity value, each target hot video is cached and sliced ​​in real time to obtain a thumbnail of the target hot video.

[0064] Thumbnails are small preview images in a video, used to quickly reflect the video content. For example, the cover image displayed in a video file, and the preview image shown when dragging the progress bar to fast forward or rewind, are both thumbnails. In addition, when business nodes send node popularity values ​​to management nodes, the node popularity values ​​can be compressed using Huffman coding to reduce the amount of data transmitted between business nodes and management nodes. At the same time, the node popularity values ​​can also be encrypted using AES-256 (Advanced Encryption Standard 256-bit) encryption to ensure data transmission security.

[0065] The technical solution of this invention involves the following steps: First, the service node obtains the playback count of each unsliced ​​video in the current sampling period through playback logs based on pre-configured placeholders in the streaming media information. This enables rapid acquisition of unsliced ​​video playback data, ensuring the real-time generation of trending video thumbnails by the subsequent management node. Then, based on the playback count, historical playback popularity, and popularity decay coefficient of the unsliced ​​video, the node popularity value of the unsliced ​​video is obtained. Based on this, a set of trending videos is selected, ensuring the selection of trending videos and avoiding the waste of storage and transmission resources caused by preview images of non-popular video content. Finally, the node popularity values ​​of each trending video in the trending video set are sent to the management node, enabling the management node to generate thumbnails of the target trending videos based on the node popularity values. This achieves automatic acquisition of trending video thumbnails, avoiding the manpower and storage costs associated with pre-generation methods.

[0066] Example 2

[0067] Figure 2 This is a flowchart of a video thumbnail acquisition method provided in Embodiment 2 of the present invention. The relationship between this embodiment and the above embodiments is that the node popularity value is related to the number of playbacks, historical playback popularity, popularity attenuation coefficient, sampling period duration, and playback gain coefficient, such as... Figure 2 As shown, the method specifically includes:

[0068] S201. Based on the pre-configured placeholders in the streaming media information, obtain the playback count of each unsliced ​​video in the current sampling period through the playback log.

[0069] S202. Obtain the node popularity value of each unsliced ​​video based on the number of times it has been played, its historical playback popularity, popularity attenuation coefficient, sampling period duration, and playback gain coefficient; wherein, the playback gain coefficient is related to the content type of the current unsliced ​​video.

[0070] Playback gain coefficient is used to correct for playback counts, reflecting the correlation between playback counts and playback popularity to balance the popularity level. The playback gain coefficient is also related to content type; for example, for short videos, a larger playback gain coefficient (e.g., 120) is used to accelerate the response time of trending videos and promptly eliminate older videos. For film and television videos, a smaller playback gain coefficient (e.g., 80) is used to boost new trending videos while effectively protecting older, more popular videos. For news and information videos, a moderate playback gain coefficient (e.g., 100) is used to balance the timeliness of trending topics with content stability. Furthermore, when correcting historical playback popularity using the popularity decay coefficient, it is also related to the sampling period length; the longer the sampling period, the greater the decay of the corrected historical playback popularity.

[0071] In particular, the node popularity value of an unsliced ​​video can also be calculated using the following equation:

[0072] ;

[0073] in, This represents the node popularity value of an unsliced ​​video. Indicates historical playback popularity; Indicates the coefficient of thermal decay; Indicates the duration of the sampling period; Indicates the playback gain coefficient; Indicates the number of plays in the current sampling period; historical popularity decay item. This indicates that the historical playback popularity is based on an exponential function, which varies with the sampling period. The increase is followed by a rapid decay; the current heat gain term This means that by utilizing the properties of the logarithmic function, the rate of increase in popularity gain gradually slows down as the number of plays increases. This can both amplify the difference in actual play counts and effectively suppress fraudulent activities.

[0074] At the same time, through The function ensures that the popularity value is non-negative, adding the attenuated historical playback popularity to the current popularity gain to obtain the node popularity value, thus achieving a dynamic balance between historical and real-time popularity. In addition, the popularity attenuation coefficient can also be configured according to the playback time period of the video to be sliced; for example, the popularity attenuation coefficient is increased by 20% during prime time (e.g., 7-10 pm) to accelerate the elimination of cold data and quickly capture real-time popularity data; the popularity attenuation coefficient is reduced by 10% during non-prime time to retain the popularity of long-tail videos, thereby achieving dynamic configuration of the popularity attenuation coefficient and improving the accuracy of popularity calculation results.

[0075] The aforementioned popularity calculation method not only boasts high computational efficiency, requiring only one exponential decay and one logarithmic gain, but also exhibits good operational stability. In high-concurrency scenarios with a query rate of 10,000 queries per second (QPS), the CPU utilization rate is stably controlled within 1%. Furthermore, it features accurate hot and cold data differentiation, with a cold data rejection rate of 35% and a hot data response cycle of only 12 sampling periods, enabling rapid identification and recommendation of trending videos. In addition, the constructed exponential dynamic decay algorithm possesses excellent anti-spam capabilities; the characteristics of the logarithmic function suppress abnormal playback growth, ensuring the authenticity of popularity data.

[0076] Accordingly, when the original data is 100MB, after preprocessing using the above-mentioned exponential dynamic decay algorithm, 30% of the cold data can be eliminated, reducing the amount of data of the hot video to be transmitted to 70% of the original data. After compression using a compression algorithm based on a 3:1 compression ratio, the data is compressed to one-third of the preprocessed data size. Combining the above preprocessing and compression processes, the original 100MB data only needs to be transmitted as 20MB after the above processing, reducing network bandwidth traffic by 70% overall.

[0077] S203. Sort the unsliced ​​videos according to the node popularity value of each unsliced ​​video, and filter and obtain a set of hot videos according to the sorting results.

[0078] S204. Send the node popularity value of each hot video in the hot video set to the management node, so that the management node can calculate and obtain the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

[0079] The technical solution of this invention obtains the node popularity value of each unsliced ​​video based on the number of times it is played, its historical playback popularity, popularity attenuation coefficient, sampling period duration, and playback gain coefficient. Then, the historical playback popularity is corrected by the popularity attenuation coefficient and sampling period duration, and the number of plays is corrected by the playback gain coefficient. Finally, the attenuated historical playback popularity is added to the current popularity gain to obtain the node popularity value, achieving a dynamic balance between historical and real-time popularity and ensuring the accuracy of the node popularity value calculation results.

[0080] Example 3

[0081] Figure 3 This is a flowchart of a video thumbnail acquisition method provided in Embodiment 3 of the present invention. This embodiment is applicable to situations where the management node calculates target hotspot video thumbnails based on the node heat values ​​issued by each business node. This method can be executed by the video thumbnail acquisition device in Embodiment 6. This video thumbnail acquisition device can be implemented in hardware and / or software and can be configured in an electronic device (e.g., a cluster management server). Figure 3 As shown, the method includes:

[0082] S301. Calculate and obtain the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node.

[0083] For a given video, it may be played on multiple business nodes within the current sampling period. Therefore, based on the hot video sets reported by each business node, all hot videos are aggregated, and the node popularity values ​​of the same hot video on different business nodes are accumulated. The accumulated result is used as the comprehensive popularity value of the current hot video.

[0084] S302. Based on the comprehensive popularity value, historical calculated popularity value, and popularity decay factor of each of the aforementioned trending videos, obtain the current calculated popularity value of each of the aforementioned trending videos; wherein, the popularity decay factor is related to the playback time of the current trending video.

[0085] Historical calculated popularity value refers to the calculated popularity value obtained in the previous sampling period. It reflects the historical popularity performance of the current hot video in the cluster. If the current hot video is appearing for the first time, there is no historical calculated popularity value, and it can be configured to a pre-set initial value. The popularity decay factor is used to correct the historical calculated popularity value. It reflects the steepness of the popularity curve. The larger the value of the popularity decay factor, the faster the video ranking drops and the higher the update frequency of the popularity ranking. The popularity decay factor can be configured to different values ​​according to the playback time of the current hot video (that is, the playback time of the current sampling period).

[0086] For example, when video playback occurs during off-peak hours (e.g., from 0:00 to 9:00 daily), the popularity decay factor is configured to a smaller value (e.g., 1.5) to avoid frequent changes in the rankings due to a smaller number of videos during off-peak periods. When video playback occurs during regular hours (e.g., from 9:00 to 18:00 daily), the popularity decay factor is configured to a moderate value (e.g., 1.8) to accommodate users' fragmented viewing needs during the day, quickly showcasing newly released videos while retaining recently popular ones. When video playback occurs during peak hours (e.g., from 18:00 to 24:00 daily), the popularity decay factor is configured to a larger value (e.g., 2.0) to further highlight the latest popular videos that users are more interested in and quickly eliminate older videos to improve content freshness. Therefore, the sum of the historical calculated popularity value (adjusted based on the popularity decay factor) and the overall popularity value can be used as the current calculated popularity value.

[0087] S303. Obtain the target hot video based on the current calculated popularity value of each hot video, and perform real-time caching and slicing processing on the target hot video to obtain a thumbnail of the target hot video.

[0088] Based on the current calculated popularity value of each trending video, from high to low, the trending videos are stored in a hierarchical linked list in an orderly manner. The number of hierarchical linked lists N and the amount of content stored in each linked list are configurable. If data conflicts occur, the original data will trigger a downgrade or removal process. The hierarchical linked lists of each channel are independent and isolated from each other, effectively preventing data interference. The linked lists are divided into L1-LN levels according to the popularity value from high to low, with L1 level storing the data with the highest popularity value, serving as the highest priority level of trending content.

[0089] Furthermore, through an intelligent upgrade and downgrade strategy, when data that has been downgraded to a lower-level list experiences a rebound in popularity within a certain period of time, a re-grading process will be automatically triggered. This accurately prevents high-popularity content from being mistakenly eliminated due to short-term fluctuations in popularity, ensuring the scientific and reasonable nature of video popularity grading. During the insertion of new videos, they are accurately inserted into the corresponding level list of the relevant channel based on their popularity value. If the list is full, the new video will be compared with the video with the lowest popularity value in that level.

[0090] If a new video's popularity value is greater than the lowest-popularity video in the linked list, the lowest-popularity video is downgraded and inserted into the new video; if the new video's popularity value is less than or equal to the lowest-popularity video, it is inserted into the next level of the linked list, and this process is repeated. When all levels of the linked list are full, the video with the lowest popularity value in the lowest-level linked list is removed to maintain the dynamic balance of the hierarchical linked lists. Therefore, depending on the video capacity of each hierarchical linked list, all popular videos can be used as target popular videos, or only a portion of all popular videos can be used as target popular videos.

[0091] Real-time cached slicing refers to a slicing technique based on HLS streaming media that involves real-time caching and frame extraction. It receives the HLS video stream, caches it in real time, and extracts video frames from the cache to generate still images, thus completing real-time video stream processing. Subsequently, the management node performs operations such as cache management, frame extraction and slicing management, and image processing and merging. Cache management includes cache strategy and cache manager. The cache strategy includes cache path, cache size, and cache data storage strategy. The cache path and size can be set via parameters, and the cache data storage strategy uses a First-In-First-Out (FIFO) queue.

[0092] The cache manager's main functions include writing, reading, and cleaning the cache. Cache writing is responsible for saving downloaded TS fragments to the cache according to their sequence numbers. Cache reading is responsible for retrieving the corresponding TS fragments from the cache; if the data in the cache is insufficient for frame extraction, it waits for the new data to be downloaded before reading. Cache cleaning includes periodic cleaning and real-time cleaning. Periodic cleaning cleans expired TS fragments according to the caching policy and data retention time (e.g., retaining 24 hours of data). Real-time cleaning monitors cache usage and automatically cleans up the oldest cached and expired data when the cache reaches a preset size.

[0093] Frame extraction and image slicing management employs zero-copy technology, directly transferring the cached video data to the Graphics Processing Unit (GPU). The GPU-accelerated video decoder decodes the TS segments to obtain video frames. Based on the set frame extraction strategy, the desired video frames are selected from the decoded video stream. The frame extraction strategy can be as follows: for videos longer than 10 minutes, extract one frame every 10 seconds; for videos 5-10 minutes long, extract one frame every 5 seconds; and for videos shorter than 5 minutes, extract one frame every 2 seconds.

[0094] Image processing and merging includes processing extracted video frame data, converting image formats to a specified format, and losslessly compressing the images to a size of 192×108px to improve image quality and loading speed. Twenty-five small images are sequentially merged into a single 5×5 image of 960×540px. If the number of images is insufficient, they are filled with pure black RGB (0,0,0) pixels. The thumbnails generated based on these image processing and merging rules firstly reduce the number of network requests after merging, significantly reducing bandwidth usage compared to the combined size of the 25 small images through lossless compression. Secondly, memory fragmentation is reduced, as a single large image occupies contiguous memory space, avoiding memory fragmentation caused by the scattered storage of 25 small images.

[0095] Furthermore, the real-time sliding preview offers high smoothness. In high-frequency previews, users can smoothly drag the progress bar to fast forward and rewind. The merged large image can load 25 images at once, compared to the need to request each small image individually. The screen update delay during sliding is significantly reduced, achieving an instant display effect. In particular, choosing the 5×5 merging method to generate thumbnails achieves the optimal balance between file size, loading speed, and positioning accuracy compared to the 4×4 and 6×6 merging methods. It is suitable for scenarios such as short video browsing and long video retrieval, and can significantly improve the smoothness of user interaction and system processing efficiency. Specifically, it achieves a balance between compression efficiency and data redundancy in terms of file size. The 4×4 merging method has fewer small images and fewer overlapping pixel areas, resulting in lower compression efficiency than the 5×5 merging method. If the total number of frames is not divisible by 16, the proportion of black pixels is higher, wasting storage space.

[0096] While the 6x6 merging method produces more small images, the resulting large image is also larger, reducing the compression algorithm's ability to optimize for excessively large images. Furthermore, the total frame count has poor divisibility with 36, resulting in more filler pixels and an actual file size exceeding that of the 5x5 merging method. In contrast, the 5x5 merging method has more overlapping pixel areas, improving lossless compression efficiency (20%-30% smaller file size than the sum of 25 small images); the total frame count has better divisibility with 25, and fewer filler pixels.

[0097] In addition, in terms of loading speed, the 5×5 merging method balances single-load volume and memory efficiency; the 4×4 merging method only contains 16 small images per large image. If the total number of video frames is the same, more large images need to be generated (for example, if there are 535 total frames, the 4×4 merging method requires 34 large images, and the 5×5 merging method requires 22 images), which increases the number of index files and requires handling more file requests during loading, thus reducing efficiency.

[0098] The 6x6 merging method, with each large image containing 36 smaller images, results in excessively large file sizes (especially when there are many fill pixels), leading to longer network transmission times during loading and higher memory consumption, which can cause stuttering. In contrast, the 5x5 merging method, with 25 smaller images, has a moderate merging quantity, keeping the size of each large image small after compression. It also reduces the number of loading operations (for example, 535 frames only require 22 large images), and the contiguous storage of large images in memory avoids fragmentation, resulting in the optimal loading speed.

[0099] Meanwhile, in terms of positioning accuracy, the 5×5 merging method has more efficient grid coordinate mapping; compared with the 5×5 merging method, the 4×4 merging method has a lower grid density at the same frame extraction interval, which leads to less precise positioning; although the 6×6 merging method has a high grid density, the coordinate calculation is complex, and the indexing system needs to process more coordinate nodes, resulting in decreased retrieval efficiency. In contrast, in the 5×5 merging method, the positioning error of each small image coordinate is less than or equal to the frame extraction interval.

[0100] Specifically, after generating thumbnails of target trending videos, the management node can either upload them to the image server or save them locally. When a business node needs to display a thumbnail of a video, it sends a thumbnail display request to the image server or the management node. The image server or the management node then sends the thumbnail of the video to the current business node so that the business node can display the thumbnail corresponding to the current user's action.

[0101] Optionally, in this embodiment of the invention, the real-time caching and slicing process of the target hot video to obtain a thumbnail of the target hot video includes: obtaining parallel slicing rules based on the resource utilization rate of the management node and the content type of the target hot video, and performing real-time caching and slicing process of the target hot video according to the parallel slicing rules to obtain a thumbnail of the target hot video.

[0102] Specifically, the scheduling and execution of image tiling tasks can be handled by the scheduling manager of the management node. The scheduling manager monitors the resource utilization of the central processing unit and the graphics processing unit, such as key indicators like core utilization, memory utilization, and I / O wait time. Based on preset thresholds (e.g., core utilization less than 80%, memory utilization less than 80%, and I / O wait time less than 60ms), it determines whether there are sufficient CPU and GPU resources to execute image tiling tasks and configures the number of concurrent threads based on the CPU and GPU resource utilization.

[0103] The scheduler then reads a tiling task in descending order of priority and submits it to the GPU for execution. The selected tiling task is encapsulated into a GPU-recognizable task format, such as video frame data, frame extraction strategy, and image processing parameters. A new thread is started to asynchronously submit the task to the GPU for execution and returns immediately so that the CPU can continue to process other tasks. At the same time, the task submission time and task ID are recorded to ensure subsequent tracking and result feedback.

[0104] In actual image slicing, image slicing tasks can be divided into short-term image slicing tasks and long-term image slicing tasks. Long-term image slicing tasks refer to image slicing tasks for videos with a relatively long duration (e.g., greater than or equal to the video duration threshold), such as movies or TV series. The time taken for 100,000 image slicing tasks is usually greater than 1 minute, and the minimum playback duration is usually greater than 30 minutes. Short-term image slicing tasks refer to image slicing tasks for short videos or news compilations with a relatively short duration (e.g., less than the video duration threshold). The time taken for 100,000 image slicing tasks is usually less than 1 minute, and the minimum playback duration is usually less than 30 minutes.

[0105] To prevent long-running image slicing tasks from consuming all slicing threads for extended periods, the ratio of short-running to long-running image slicing tasks can be determined based on the content types of all target trending videos. Then, the thread allocation ratio for short-running and long-running image slicing tasks can be determined based on this ratio. For example, a 7:3 thread allocation ratio can be used. According to actual test results, this allocation strategy can reduce the average processing time of long-running image slicing tasks by 25%, while ensuring that the response latency of short-running image slicing tasks is kept below 50ms, achieving a dual optimization of resource utilization and task timeliness.

[0106] The technical solution of this invention calculates the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node; it then obtains the current calculated popularity value of each hot video based on the comprehensive popularity value, historical calculated popularity value, and popularity decay factor; finally, it identifies the target hot video based on the current calculated popularity value and performs real-time caching and slicing processing on the target hot video to obtain its thumbnail. This achieves the calculation and acquisition of the overall popularity performance of each hot video in the cluster, ensuring the selection and acquisition of hot videos and avoiding the waste of storage and transmission resources caused by preview images of non-popular video content; simultaneously, it achieves automatic acquisition of hot video thumbnails, avoiding the manpower and storage costs of pre-generation methods.

[0107] Example 4

[0108] Figure 4 This is a flowchart of a video thumbnail acquisition method provided in Embodiment 4 of the present invention. The relationship between this embodiment and the above embodiments is that the current calculated popularity value is related to the comprehensive popularity value, the historical calculated popularity value, the popularity decay factor, the sampling period duration, the playback weight coefficient, and the popularity amplification coefficient, such as... Figure 4 As shown, the method specifically includes:

[0109] S401. Calculate and obtain the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node.

[0110] S402. Based on the comprehensive popularity value, historical calculated popularity value, popularity decay factor, sampling period duration, playback weight coefficient, and popularity amplification coefficient of each of the aforementioned hot videos, obtain the current calculated popularity value of each of the aforementioned hot videos.

[0111] The playback weighting coefficient is used to correct the overall popularity value, balancing video update speed and stability. Furthermore, when correcting historical calculated popularity values ​​using the popularity decay factor, it is also related to the sampling period; the longer the sampling period, the greater the decay in the corrected historical playback popularity. The popularity amplification coefficient is used to address the difficulty in clearly distinguishing between popular and unpopular videos when new videos have low popularity values. In particular, the current calculated popularity value of trending videos can also be calculated using the following equation:

[0112] ;

[0113] in, This indicates the current calculated heat value; This indicates the historical popularity value, preserving the continuity of historical popularity. Indicates the duration of the sampling period. The design cleverly avoids the problem of the denominator being too small due to the current calculation of heat value, while ensuring that the heat value does not decay in the first time interval; Indicates the factor of declining popularity; This represents the playback weight coefficient; This indicates the overall popularity value; This represents the thermal amplification factor.

[0114] S403. Obtain the target hot video based on the current calculated popularity value of each hot video, and perform real-time caching and slicing processing on the target hot video to obtain a thumbnail of the target hot video.

[0115] The technical solution of this invention obtains the current calculated popularity value of each hot video based on its comprehensive popularity value, historical calculated popularity value, popularity decay factor, sampling period duration, playback weight coefficient, and popularity amplification coefficient. This balances video update speed and stability by adjusting the comprehensive popularity value using the playback weight coefficient. Simultaneously, the adjustment of historical calculated popularity values ​​using the sampling period duration ensures a true reflection of the popularity over time. Furthermore, the popularity amplification coefficient addresses the problem of difficulty in clearly distinguishing between hot and cold videos when the popularity value of a new video is low.

[0116] Example 5

[0117] Figure 5 This is a structural block diagram of a video thumbnail acquisition device provided in Embodiment 5 of the present invention. The device is applied to a service node in a cluster and specifically includes:

[0118] The playback count acquisition module 501 is used to obtain the playback count of each unsliced ​​video in the current sampling period by means of the playback log based on the pre-configured placeholders in the streaming media information.

[0119] The node popularity acquisition module 502 is used to acquire the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video has been played, its historical playback popularity, and its popularity decay coefficient; wherein, the popularity decay coefficient is related to the content type of the current unsliced ​​video.

[0120] The hot video acquisition module 503 is used to sort the unsliced ​​videos according to the node popularity value of each unsliced ​​video, so as to filter and obtain a set of hot videos according to the sorting result;

[0121] The node popularity sending module 504 is used to send the node popularity value of each hot video in the hot video set to the management node, so that the management node can calculate and obtain the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

[0122] The technical solution of this invention involves the following steps: First, the service node obtains the playback count of each unsliced ​​video in the current sampling period through playback logs based on pre-configured placeholders in the streaming media information. This enables rapid acquisition of unsliced ​​video playback data, ensuring the real-time generation of trending video thumbnails by the subsequent management node. Then, based on the playback count, historical playback popularity, and popularity decay coefficient of the unsliced ​​video, the node popularity value of the unsliced ​​video is obtained. Based on this, a set of trending videos is selected, ensuring the selection of trending videos and avoiding the waste of storage and transmission resources caused by preview images of non-popular video content. Finally, the node popularity values ​​of each trending video in the trending video set are sent to the management node, enabling the management node to generate thumbnails of the target trending videos based on the node popularity values. This achieves automatic acquisition of trending video thumbnails, avoiding the manpower and storage costs associated with pre-generation methods.

[0123] Optionally, the playback count acquisition module 501 is specifically used to obtain the playback count of each unsliced ​​video in the current sampling period through the playback log based on the placeholders pre-configured in the metadata of the streaming media information.

[0124] Optionally, the node popularity acquisition module 502 is specifically used to acquire the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video is played, its historical playback popularity, popularity attenuation coefficient, sampling period duration, and playback gain coefficient; wherein, the playback gain coefficient is related to the content type of the current unsliced ​​video.

[0125] The above-described apparatus can execute the video thumbnail acquisition method provided in Embodiment 1 or Embodiment 2 of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the video thumbnail acquisition method provided in any embodiment of the present invention.

[0126] Example 6

[0127] Figure 6 This is a structural block diagram of a video thumbnail acquisition device provided in Embodiment Six of the present invention. The device is applied to the management node of a cluster and specifically includes:

[0128] The comprehensive popularity acquisition module 601 is used to calculate and acquire the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node.

[0129] The current popularity acquisition module 602 is used to acquire the current calculated popularity value of each of the hot videos based on the comprehensive popularity value, historical calculated popularity value and popularity decay factor of each hot video; wherein, the popularity decay factor is related to the playback time of the current hot video;

[0130] The cache slicing execution module 603 is used to obtain target hot videos based on the current calculated popularity value of each hot video, and to perform real-time cache slicing processing on the target hot videos to obtain thumbnails of the target hot videos.

[0131] The technical solution of this invention calculates the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node; it then obtains the current calculated popularity value of each hot video based on the comprehensive popularity value, historical calculated popularity value, and popularity decay factor; finally, it identifies the target hot video based on the current calculated popularity value and performs real-time caching and slicing processing on the target hot video to obtain its thumbnail. This achieves the calculation and acquisition of the overall popularity performance of each hot video in the cluster, ensuring the selection and acquisition of hot videos and avoiding the waste of storage and transmission resources caused by preview images of non-popular video content; simultaneously, it achieves automatic acquisition of hot video thumbnails, avoiding the manpower and storage costs of pre-generation methods.

[0132] Optionally, the current popularity acquisition module 602 is specifically used to acquire the current calculated popularity value of each of the aforementioned hot videos based on the comprehensive popularity value, historical calculated popularity value, popularity decay factor, sampling period duration, playback weight coefficient, and popularity amplification coefficient of each hot video.

[0133] Optionally, the cache slicing execution module 603 is used to obtain parallel slicing rules based on the resource utilization rate of the management node and the content type of the target hot video, and to perform real-time cache slicing processing on the target hot video according to the parallel slicing rules to obtain a thumbnail of the target hot video.

[0134] The above-described apparatus can execute the video thumbnail acquisition method provided in Embodiment 3 or Embodiment 4 of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the video thumbnail acquisition method provided in any embodiment of the present invention.

[0135] Example 7

[0136] Figure 7 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, electronic devices, blade electronic devices, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0137] like Figure 7 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0138] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0139] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the video thumbnail acquisition method.

[0140] In some embodiments, the video thumbnail acquisition method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on a heterogeneous hardware accelerator via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by a processor, one or more steps of the video thumbnail acquisition method described above may be performed. Alternatively, in other embodiments, the processor may be configured to perform the video thumbnail acquisition method by any other suitable means (e.g., by means of firmware).

[0141] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0143] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. 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 fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0144] To provide user interaction, the systems and techniques described herein can be implemented on a heterogeneous hardware accelerator, which includes: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the heterogeneous hardware accelerator. Other types of devices can also be used to provide user interaction; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback); and input from the user can be received in any form (including sound input, voice input, or haptic input).

[0145] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0146] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0147] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0148] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for obtaining video thumbnails, characterized in that, The business nodes used in the cluster include: Based on the pre-configured placeholders in the streaming media information, the playback count of each unsliced ​​video in the current sampling period is obtained through the playback log; The node popularity value of each unsliced ​​video is obtained based on the number of times it has been played, its historical playback popularity, and its popularity decay coefficient. The popularity decay coefficient is related to the content type of the current unsliced ​​video. The historical playback popularity refers to the popularity value of the current unsliced ​​video calculated in the previous sampling period. The step of obtaining the node popularity value of each unsliced ​​video based on its playback count, historical playback popularity, and popularity decay coefficient specifically includes: The node heat value of the unsliced ​​video is calculated using the following equation: ； in, This represents the node popularity value of an unsliced ​​video. Indicates historical playback popularity; Indicates the coefficient of thermal decay; Indicates the duration of the sampling period; Indicates the playback gain coefficient; This indicates the number of times the video was played in the current sampling period; the playback gain coefficient reflects the correlation between the number of plays and the playback popularity, and the playback gain coefficient is related to the content type of the currently unsliced ​​video. Based on the node popularity value of each unsliced ​​video, the unsliced ​​videos are sorted to filter and obtain a set of hot videos based on the sorting results; The node popularity value of each hot video in the hot video set is sent to the management node, so that the management node can calculate the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

2. The video thumbnail acquisition method according to claim 1, characterized in that, The step of obtaining the playback count of each unsliced ​​video segment in the current sampling period through playback logs based on pre-configured placeholders in the streaming media information includes: Based on the pre-configured placeholders in the metadata of the streaming media information, the playback count of each unsliced ​​video in the current sampling period is obtained through the playback log.

3. A method for obtaining video thumbnails, characterized in that, The management nodes used in the cluster include: Based on the node popularity values ​​of each hot video in the hot video set issued by each business node, the comprehensive popularity value of each hot video is calculated; wherein, the node popularity value of each hot video is calculated by the following equation: ； in, The node popularity value represents the trending video. Indicates historical playback popularity; Indicates the coefficient of thermal decay; Indicates the duration of the sampling period; Indicates the playback gain coefficient; This indicates the number of times the video has been played in the current sampling period; the playback gain coefficient reflects the correlation between the number of plays and the playback popularity, and the playback gain coefficient is related to the content type of the current trending video; the historical playback popularity refers to the popularity value of the current trending video calculated in the previous sampling period. Based on the comprehensive popularity value, historical calculated popularity value, and popularity decay factor of each of the aforementioned trending videos, the current calculated popularity value of each of the aforementioned trending videos is obtained; wherein, the popularity decay factor is related to the playback time of the current trending video; The target hot video is obtained based on the current calculated popularity value of each hot video, and the target hot video is cached and sliced ​​in real time to obtain a thumbnail of the target hot video.

4. The video thumbnail acquisition method according to claim 3, characterized in that, The step of obtaining the current calculated popularity value of each of the aforementioned trending videos based on their comprehensive popularity value, historical calculated popularity value, and popularity decay factor includes: The current calculated popularity value of each of the aforementioned trending videos is obtained based on their comprehensive popularity value, historical calculated popularity value, popularity decay factor, sampling period duration, playback weight coefficient, and popularity amplification coefficient.

5. The video thumbnail acquisition method according to claim 3, characterized in that, The step of performing real-time caching and slicing processing on the target hot video to obtain a thumbnail of the target hot video includes: Based on the resource utilization rate of the management node and the content type of the target hot video, parallel slicing rules are obtained, and the target hot video is cached and sliced ​​in real time according to the parallel slicing rules to obtain a thumbnail of the target hot video.

6. A video thumbnail acquisition device, characterized in that, The business nodes used in the cluster include: The playback count acquisition module is used to obtain the playback count of each unsliced ​​video in the current sampling period by means of the playback log based on the pre-configured placeholders in the streaming media information. The node popularity acquisition module is used to acquire the node popularity value of each unsliced ​​video based on the number of times each unsliced ​​video has been played, its historical playback popularity, and its popularity decay coefficient; wherein, the popularity decay coefficient is related to the content type of the current unsliced ​​video; and the historical playback popularity refers to the popularity value of the current unsliced ​​video calculated in the previous sampling period. The node popularity acquisition module is specifically used to calculate the node popularity value of the unsliced ​​video using the following equation: ； in, This represents the node popularity value of an unsliced ​​video. Indicates historical playback popularity; Indicates the coefficient of thermal decay; Indicates the duration of the sampling period; Indicates the playback gain coefficient; This indicates the number of times the video was played in the current sampling period; the playback gain coefficient reflects the correlation between the number of plays and the playback popularity, and the playback gain coefficient is related to the content type of the currently unsliced ​​video. The hot video acquisition module is used to sort the unsliced ​​videos according to the node popularity value of each unsliced ​​video, so as to filter and obtain a set of hot videos based on the sorting results; The node popularity sending module is used to send the node popularity value of each hot video in the hot video set to the management node, so that the management node can calculate and obtain the current calculated popularity value of each hot video based on the node popularity value, and generate a thumbnail of the target hot video based on the current calculated popularity value.

7. A video thumbnail acquisition device, characterized in that, The management nodes used in the cluster include: The comprehensive popularity acquisition module is used to calculate and obtain the comprehensive popularity value of each hot video based on the node popularity value of each hot video in the hot video set issued by each business node; wherein, the node popularity value of the hot video is calculated and obtained by the following equation: ； in, The node popularity value represents the trending video. Indicates historical playback popularity; Indicates the coefficient of thermal decay; Indicates the duration of the sampling period; Indicates the playback gain coefficient; This indicates the number of times the video has been played in the current sampling period; the playback gain coefficient reflects the correlation between the number of plays and the playback popularity, and the playback gain coefficient is related to the content type of the current trending video; the historical playback popularity refers to the popularity value of the current trending video calculated in the previous sampling period. The current popularity acquisition module is used to acquire the current calculated popularity value of each of the aforementioned trending videos based on their comprehensive popularity value, historical calculated popularity value, and popularity decay factor; wherein, the popularity decay factor is related to the playback period of the current trending video. The cache slicing execution module is used to obtain target hot videos based on the current calculated popularity value of each hot video, and to perform real-time cache slicing processing on the target hot videos to obtain thumbnails of the target hot videos.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the video thumbnail acquisition method according to any one of claims 1 or 2, or to perform the video thumbnail acquisition method according to any one of claims 3-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the video thumbnail acquisition method according to any one of claims 1 or 2, or the video thumbnail acquisition method according to any one of claims 3-5.

Images