Method and device for detecting highlights of live broadcast resources and computing equipment

Abstract

The invention discloses a method and a device for detecting highlights of live broadcast resources and computing equipment. The method comprises the following steps: acquiring playing effect index values of a target live broadcast resource at different sampling times; generating a plurality of sliding time windows with different window lengths, wherein each sliding time window divides a resource interval of the target live broadcast resource into a plurality of time segments; calculating an index average value of each time slice according to the playing effect index value in each time slice; determining a correction coefficient of each sliding time window; obtaining an effect evaluation value of the time slice according to the index average value of the time slice and the correction coefficient of the sliding time window to which the time slice belongs; and screening out a target time fragment according to the effect evaluation value, and determining a wonderful fragment in the target live broadcast resource according to the target time fragment. According to the scheme, a complicated content recognition process is not needed, so that the highlight fragment recognition efficiency is high; and compared with a fixed time length detection method, the method has higher detection precision.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a method, device, computing equipment, computer storage medium and computer program product for detecting highlights of live broadcast resources. Background Art

[0002] With the continuous development of science and technology and society, the emergence of live broadcasting has greatly enriched people's work and life. In order to facilitate live broadcast playback, secondary editing, data analysis, etc., it is usually necessary to detect the highlights of live broadcast resources.

[0003] There are currently two commonly used methods for detecting highlights of live broadcast resources:

[0004] The first is a fixed duration detection method, which is to find the highest point of indicators such as the number of viewers during the live broadcast, and use the highest point as a benchmark to expand forward and backward by a fixed length to obtain a highlight segment. For example, in the prior art, if it is determined that the number of viewers in the live broadcast room is the largest at time T1, the segment corresponding to the period [T-2, T+3] is taken as the highlight period.

[0005] The second is the content detection method, which is to score the image frames in the live video through image detection algorithms or AI models, and then determine the exciting time periods based on the content scores.

[0006] However, the inventors discovered during the implementation process that the prior art has the following defects: the fixed duration detection method has insufficient accuracy in detecting highlight clips and cannot adapt to the detection of highlight periods in different scenarios; and the content detection method has a huge amount of calculation, consumes a lot of system resources, and has low detection efficiency. Summary of the invention

[0007] In view of the above problems, the present application is proposed to provide a method, apparatus, computing device, computer storage medium and computer program product for detecting highlights of live broadcast resources that overcome the above problems or at least partially solve the above problems.

[0008] According to a first aspect of the present application, a method for detecting highlights of a live broadcast resource is provided, comprising:

[0009] Get the playback effect index values ​​of the target live broadcast resource at different sampling times;

[0010] Generate multiple sliding time windows with different window lengths, each sliding time window divides the resource interval of the target live broadcast resource into multiple time slices;

[0011] Calculate the average value of the index of each time slice according to the playback effect index value in each time slice;

[0012] Determine the correction factor for each sliding time window;

[0013] For each time slice, the effect evaluation value of the time slice is obtained according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs;

[0014] The target time slices are selected according to the effect evaluation values ​​of each time slice, and the highlight segments in the target live broadcast resource are determined according to the target time slices.

[0015] In an optional implementation, determining the correction coefficient of each sliding time window includes:

[0016] Assign a fixed correction factor to the sliding time window with the smallest window length;

[0017] For any non-minimum sliding time window, the correction coefficient of the sliding time window compared with the previous sliding time window is calculated based on the window length and adjustment parameters of the sliding time window, and the correction coefficient of the sliding time window is calculated based on the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared with the previous sliding time window.

[0018] In an optional implementation, the calculating, according to the window length and adjustment parameter of the sliding time window, a correction coefficient of the sliding time window compared to the previous sliding time window includes:

[0019] Calculate the sum of the adjustment parameter, the window length of the sliding time window, and -1;

[0020] The ratio of the window length of the sliding time window to the sum is used as a correction coefficient of the sliding time window compared with the previous sliding time window.

[0021] In an optional implementation, the adjustment parameter is determined by:

[0022] Count the clip lengths of live highlights clip resources uploaded in the past;

[0023] Determining a reference window length according to the number of clip resources corresponding to different clip lengths;

[0024] Determine the first clipping resource quantity of the reference window length and the second clipping resource quantity of the reference window length; wherein the reference window length=reference window length+1;

[0025] The adjustment parameter is generated according to the first clipping resource quantity and the second clipping resource quantity.

[0026] In an optional implementation, before generating a plurality of sliding time windows with different window lengths, the method further includes:

[0027] Count the clip lengths of live highlights clip resources uploaded in the past;

[0028] Generate a window length value range according to the clip length;

[0029] The generating a plurality of sliding time windows with different window lengths includes: generating a plurality of sliding time windows with different window lengths according to the window length value range.

[0030] In an optional implementation, the method further includes: acquiring system performance parameters;

[0031] The generating a window length value range according to the clipping length includes: generating a window length value range according to the clipping length and the system performance parameter.

[0032] In an optional implementation, the step of selecting a target time slice according to the effect evaluation value of each time slice, and determining a highlight segment in the target live broadcast resource according to the target time slice comprises:

[0033] Sort the time slices in descending order of effect evaluation values;

[0034] Select the time slices with the effect evaluation values ​​in the top N as the target time slices;

[0035] Aggregating the target time slices to obtain an aggregated time period;

[0036] The segment corresponding to the aggregation period is taken as the highlight segment.

[0037] According to a second aspect of the present application, a device for detecting highlights of a live broadcast resource is provided, comprising:

[0038] An indicator acquisition module is used to obtain the playback effect indicator values ​​of the target live broadcast resource at different sampling times;

[0039] A window generation module, used to generate a plurality of sliding time windows with different window lengths, each sliding time window dividing a resource interval of a target live broadcast resource into a plurality of time slices;

[0040] The average value calculation module is used to calculate the average value of the index of each time slice according to the playback effect index value in each time slice;

[0041] A correction coefficient determination module, used to determine the correction coefficient of each sliding time window;

[0042] An evaluation value calculation module is used to obtain, for each time slice, an effect evaluation value of the time slice according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs;

[0043] The segment identification module is used to filter out the target time slices according to the effect evaluation values ​​of each time slice, and determine the wonderful segments in the target live broadcast resource according to the target time slices.

[0044] In an optional implementation, the correction coefficient determination module is used to: assign a fixed correction coefficient to a sliding time window with a minimum window length;

[0045] For any non-minimum sliding time window, the correction coefficient of the sliding time window compared with the previous sliding time window is calculated based on the window length and adjustment parameters of the sliding time window, and the correction coefficient of the sliding time window is calculated based on the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared with the previous sliding time window.

[0046] In an optional implementation, the correction coefficient determination module is used to: calculate the sum of the adjustment parameter, the window length of the sliding time window, and -1;

[0047] The ratio of the window length of the sliding time window to the sum is used as a correction coefficient of the sliding time window compared with the previous sliding time window.

[0048] In an optional implementation, the device further includes an adjustment parameter generation module for counting the clip lengths of historically uploaded live highlight clip resources;

[0049] Determining a reference window length according to the number of clip resources corresponding to different clip lengths;

[0050] Determine the first clipping resource quantity of the reference window length and the second clipping resource quantity of the reference window length; wherein the reference window length=reference window length+1;

[0051] The adjustment parameter is generated according to the first clipping resource quantity and the second clipping resource quantity.

[0052] In an optional implementation, the device further includes: a window length range determination module, which is used to count the clip lengths of the live broadcast highlight clip resources uploaded historically;

[0053] Generate a window length value range according to the clip length;

[0054] The generating a plurality of sliding time windows with different window lengths includes: generating a plurality of sliding time windows with different window lengths according to the window length value range.

[0055] In an optional implementation, the window length range determination module is used to: obtain a system performance parameter; and generate a window length value range according to the clip length and the system performance parameter.

[0056] In an optional implementation, the segment identification module is used to: sort the time slices in descending order of effect evaluation values;

[0057] Select the time slices with the effect evaluation values ​​in the top N as the target time slices;

[0058] Aggregating the target time slices to obtain an aggregated time period;

[0059] The segment corresponding to the aggregation period is taken as the highlight segment.

[0060] According to a third aspect of the present application, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0061] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for detecting highlights of live broadcast resources.

[0062] According to a fourth aspect of the present application, a computer storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for detecting highlights of live broadcast resources.

[0063] According to a fifth aspect of the present application, a computer program product is provided, comprising at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for detecting highlights of live broadcast resources.

[0064] The embodiment of the present application uses multiple sliding time windows with different window lengths to divide the resource interval of the target live resource to obtain time slices of different lengths, and allocates corresponding correction coefficients to the sliding time windows of different window lengths. The correction coefficient of the sliding time window to which the time slice belongs is used to correct the average value of the index of the time slice to obtain the effect evaluation value of the time slice, and then the highlight segment is identified according to the effect evaluation value. With this solution, there is no need to perform a complex content recognition process, so the highlight segment recognition efficiency is high; and compared with the fixed duration detection method, it can have higher detection accuracy and wider application scenarios.

[0065] The embodiment of the present application generates a correction coefficient for any non-minimum sliding time window compared to the previous sliding time window, and then generates a correction coefficient for each sliding time window based on the correction coefficient, thereby improving the accuracy of determining the correction coefficient and further improving the detection accuracy of highlight clips.

[0066] The embodiment of the present application determines the correction coefficient of the sliding time window compared to the previous sliding time window based on the adjustment parameters and the window length, so that the mean values ​​of the indicators in two adjacent sliding time windows of different window lengths can be accurately balanced through the correction coefficient.

[0067] The embodiment of the present application determines adjustment parameters according to the clip length of the live broadcast highlight clip resources uploaded historically, thereby improving the determination accuracy of the adjustment parameters.

[0068] The embodiment of the present application generates a window length value range in advance according to the clip length of the live highlight clip resources uploaded historically, so that the generated multiple sliding time windows can adapt to the actual situation, thereby improving the rationality of the generated sliding time windows.

[0069] The embodiment of the present application combines the historical clipping length and system performance parameters to comprehensively determine the window length value range, thereby further improving the accuracy of determining the window length value range, avoiding the adverse impact of a large window length range on system performance, and avoiding a waste of system resources caused by a small window length range.

[0070] The embodiment of the present application aggregates the target time slices to obtain an aggregated time period, and uses the segments corresponding to the aggregated time period as highlight segments. This improves the continuity of the highlight segments and the user experience on the one hand, and reduces the number of highlight segments and reduces maintenance costs on the other hand.

[0071] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0073] Figure 1 A schematic diagram of an operating environment provided for implementing at least one embodiment of the present application is shown;

[0074] Figure 2 A flowchart of a method for detecting highlights of a live broadcast resource provided in Embodiment 1 of the present application is shown;

[0075] Figure 3 A schematic diagram of displaying a highlight clip provided in the first embodiment of the present application is shown;

[0076] Figure 4 A flow chart of a method for detecting highlights of a live broadcast resource provided in Embodiment 2 of the present application is shown;

[0077] Figure 5 A three-point mean value schematic diagram provided in Example 2 of the present application is shown;

[0078] Figure 6 A flow chart of a method for determining an adjustment parameter provided in Embodiment 2 of the present application is shown;

[0079] Figure 7 A schematic diagram showing the structure of a device for detecting highlights of live broadcast resources provided in Embodiment 3 of the present application is shown;

[0080] Figure 8 A schematic diagram of the structure of a computing device provided in Embodiment 4 of the present application is shown. DETAILED DESCRIPTION

[0081] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0082] It should be noted that the evaluation index data and / or live broadcast highlight clip resources (including but not limited to those used for analyzing data, storing data, displaying data, etc.) involved in the embodiments of the present application are all authorized or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for choosing to authorize or reject.

[0083] Figure 1 The present invention is applicable to an application environment including, but not limited to, a client 2 , a server 4 , and a network 6 .

[0084] in:

[0085] The server 4 may be composed of a single or multiple computing devices. The multiple computing devices may include virtualized computing instances. Virtualized computing instances may include virtual machines, such as simulations of computer systems, operating systems, servers, etc. The computing device may load a virtual machine based on a virtual image and / or other data defining specific software (e.g., operating system, dedicated application, server) for simulation. As the demand for different types of processing services changes, different virtual machines may be loaded and / or terminated on one or more computing devices. A hypervisor may be implemented to manage the use of different virtual machines on the same computing device.

[0086] The server 4 may be configured to communicate with the client 2, etc., via a network 6. The network 6 includes various network devices, such as routers, switches, multiplexers, hubs, modems, bridges, repeaters, firewalls, proxy devices, and / or the like. The network 6 may include physical links, such as coaxial cable links, twisted pair cable links, optical fiber links, combinations thereof, etc., or wireless links, such as cellular links, satellite links, Wi-Fi links, etc.

[0087] The server 4 can provide storage, reading, downloading, writing, querying, deleting and other services, such as providing static resource download services to clients through multiple domain names.

[0088] Client 2 can be running Windows, Android TM ) or IOS and other operating systems, such as smart phones, tablet devices, laptop computers, virtual reality devices, gaming devices, set-top boxes, car terminals, smart TVs. Based on the above operating systems, various applications, such as browsers, can be run.

[0089] Embodiment 1

[0090] Figure 2 The flowchart of a method for detecting highlights of a live broadcast resource provided in the first embodiment of the present application is shown. The method provided in the embodiment of the present application can be applied to a live broadcast platform.

[0091] Specifically, Figure 2 As shown, the method comprises the following steps:

[0092] Step S210, obtaining the playback effect index values ​​of the target live broadcast resource at different sampling times.

[0093] The target live resource is the live resource to be detected for highlights, that is, this application is used to detect highlights in the target live resource. The target live resource can be an entire live stream or a part of a live stream. The target live resource can be video, audio, and / or graphics, etc.

[0094] In order to improve user experience and facilitate data analysis, the live broadcast platform will record the playback effect index values ​​at different sampling times during the live broadcast of the target live broadcast resource. Among them, the playback effect index value is the value of at least one playback effect indicator, which includes but is not limited to: barrage volume, viewing volume, new volume, and / or revenue volume, etc. Among them, the barrage volume is the number of barrages detected at the corresponding sampling time; the viewing volume is the number of viewers in the live broadcast room detected at the corresponding sampling time; the new volume is the number of new entrants to the live broadcast room detected at the corresponding sampling time; the revenue volume is the anchor revenue value detected at the corresponding sampling time.

[0095] In an optional implementation, in order to improve the efficiency of highlight segment detection, when the sampling frequency is high, the playback effect index values ​​of different sampling times can be clustered to obtain the playback effect index value of each sampling unit, and a sampling unit contains multiple sampling times. For example, if the sampling frequency is 10 seconds, that is, the playback effect index value of the target live resource is detected every 10 seconds, and the sampling unit is 1 minute, then the playback effect index value of multiple sampling times in the same minute is accumulated to obtain the playback effect index value of the sampling unit.

[0096] Step S220, generating a plurality of sliding time windows with different window lengths, each sliding time window dividing the resource interval of the target live broadcast resource into a plurality of time slices.

[0097] Generate multiple sliding time windows, each with a different window length. The sliding time window can slide according to a preset step size and a preset direction. For example, a sliding time window A with a window length of 3 minutes (indicating that the interval between the window start point and the end point is 3 minutes), a sliding time window B with a window length of 4 minutes (indicating that the interval between the window start point and the end point is 4 minutes), a sliding time window C with a window length of 5 minutes (indicating that the interval between the window start point and the end point is 5 minutes)... A sliding time window J with a window length of 10 minutes (indicating that the interval between the window start point and the end point is 10 minutes) can be generated.

[0098] For each sliding time window, the sliding time window can divide the resource interval of the target live resource into multiple time slices. Among them, the resource interval of the target live resource is the time interval between the start time and the end time of the target live resource. For any sliding time window, the interval of the sliding time window in the time window in the initial state is a time slice, and then the sliding time window slides according to the preset direction and the preset step size, then the interval in the sliding time window is used as another time slice, and so on until the sliding time window reaches the end position and can no longer slide, so as to realize the division of the resource interval by the sliding time window, and the length of the time slices obtained by the same sliding time window is the same, which is equal to the window length of the sliding time window. For example, the resource interval is [1, 5], and the sliding time window A divides the resource interval into time slices [1, 3], [2, 4], and [3, 5]; the sliding time window B divides the resource interval into time slices [1, 4] and [2, 5].

[0099] Step S230, calculating the average index value of each time slice according to the playback effect index value in each time slice.

[0100] For any time slice, obtain the playback effect index value at the sampling time in the time slice, and then calculate the average value of the playback effect index value in the time slice, and use the average value as the index average value of the time slice.

[0101] In an optional implementation, the embodiment of the present application can generate an average value of the indicator under any indicator dimension, and obtain the highlight clip under the indicator dimension through subsequent steps S240-S260. For example, this step calculates the average value of the viewing volume indicator of the time slice, and then obtains the highlight clip under the viewing volume indicator dimension; it can also calculate the average value of the barrage indicator, and then obtain the highlight clip under the barrage indicator dimension, and so on.

[0102] In another optional implementation, the average values ​​of indicators under multiple indicator dimensions can be combined to obtain a comprehensive indicator average value, for example, the comprehensive indicator average value is obtained by weighting different indicators and weighted summing up the corresponding indicator average values, and then the highlights under the comprehensive indicator dimension are obtained through subsequent steps.

[0103] In the process of calculating the indicator average value, the cumulative value of the playback effect index value within the time slice can be calculated, and the ratio of the cumulative value to the length of the time slice can be used as the indicator average value; or, the cumulative value of the playback effect index value of each sampling unit within the time slice can be calculated, and then the ratio of the cumulative value to the length of the time slice can be used as the indicator average value.

[0104] Step S240, determining the correction coefficient of each sliding time window.

[0105] In the actual implementation process, the playback effect index value during resource playback often has a certain degree of continuity. If the average index values ​​of the time slices divided by different sliding time windows are directly compared, the wonderful clips will be uneven and there will be a large split. In view of this, the embodiment of the present application allocates a corresponding correction coefficient to each sliding time window, that is, each sliding time window has a corresponding correction coefficient. The correction coefficient is positively correlated to the window length of the sliding time window, that is, the larger the window length of the sliding time window, the larger the corresponding correction coefficient.

[0106] Step S250, for each time slice, obtain the effect evaluation value of the time slice according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs.

[0107] For any time slice, the average value of the index of the time slice is corrected by using the correction coefficient of the sliding time window to which the time slice belongs to obtain the effect evaluation value of the time slice. For example, the product of the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs can be used as the effect evaluation value of the time slice.

[0108] Step S260, screen out target time slices according to the effect evaluation values ​​of each time slice, and determine the highlight segments in the target live broadcast resource according to the target time slices.

[0109] Select N time slices with high effect evaluation values ​​as target time slices. For example, sort the time slices in descending order of effect evaluation values, and select the time slices with the top N effect evaluation values ​​as target time slices. Then, use the segments corresponding to the target time slices as the highlights in the target live broadcast resource.

[0110] In an optional implementation, the segment corresponding to each target time slice may be used as a highlight segment, thereby simplifying the implementation process of the method.

[0111] In another optional implementation, there may be overlaps between the N target time slices obtained. In order to enhance the user's viewing experience of the highlight segments, this implementation aggregates the target time slices to obtain an aggregated time period, and uses the segments corresponding to the aggregated time period as highlight segments. Specifically, multiple target time slices with time overlaps and / or slice intervals below a preset threshold can be aggregated into one aggregated time period. For example, if the target time slices are [3, 5] and [4, 7], then the two target time slices are aggregated to obtain an aggregated time period [3, 7], and the segments corresponding to the aggregated time period [3, 7] are identified as highlight segments, thereby streamlining the number of highlight segments and improving the continuity of the highlight segments.

[0112] In an optional implementation, if step S230 calculates the average values ​​of indicators under different indicator dimensions, then step S250 obtains the effect evaluation values ​​under different indicator dimensions, and step S260 obtains the highlights under any indicator dimension according to the average value of the effects under the indicator dimension. The indicator value curve under different indicator dimensions can be displayed in a visual manner, and the highlights under the indicator dimension can be marked in the curve. The resulting highlights can be displayed to the user. Figure 3 As shown, the highlight clips (highlight clips) identified under different indicator dimensions can be displayed, and the clip length of the highlight clip and the corresponding clip interval can be displayed. The user can select a highlight clip to play, and the position of the currently playing highlight clip on the timeline is marked in the playback timeline, thereby improving the user experience.

[0113] It can be seen that the method for detecting the highlights of live broadcast resources provided by the embodiment of the present application uses multiple sliding time windows with different window lengths to divide the resource interval of the target live broadcast resource to obtain time slices of different lengths, and allocates corresponding correction coefficients to the sliding time windows of different window lengths, and uses the correction coefficient of the sliding time window to which the time slice belongs to correct the average value of the index of the time slice to obtain the effect evaluation value of the time slice, and then identifies the highlights according to the effect evaluation value. With this solution, there is no need to perform a complex content recognition process, so that the efficiency of highlight recognition is high; and compared with the fixed duration detection method, it can have higher detection accuracy and wider application scenarios.

[0114] Embodiment 2

[0115] Figure 4 The flowchart of a method for detecting highlights of a live broadcast resource provided in the second embodiment of the present application is shown. The method provided in the embodiment of the present application can be applied to a live broadcast platform.

[0116] Specifically, Figure 4 As shown, the method comprises the following steps:

[0117] Step S410: Count the clip lengths of the live broadcast highlight clip resources uploaded in the past.

[0118] In the actual implementation process, the live broadcast platform provides users (including audience users and anchor users) with a clip upload entrance, through which users can upload their own edited live broadcast highlights. In the embodiment of the present application, the live broadcast highlights edited and uploaded by users are called live broadcast highlight clip resources.

[0119] Determine the clip lengths of multiple live highlight clip resources uploaded historically, where the clip lengths are resource durations of the live highlight clip resources, and count the number of live highlight clip resources corresponding to different clip lengths.

[0120] Step S420: Generate a window length value range according to the clip length.

[0121] By implementing step S410, the clip length distribution of each live highlight clip resource uploaded in history (i.e., the number of live highlight clip resources corresponding to different clip lengths) can be obtained, and the window length value range is determined according to the clip length distribution. For example, if it is determined according to the clip length distribution that most of the clip lengths are within the interval [3, 10], the window length value range is determined to be [3, 10]. The window length value range can be pre-generated and stored as configuration data.

[0122] Optionally, system performance parameters may be obtained, which may include hardware performance parameters or CPU utilization parameters. Then, a window length value range is generated according to the clip length and the system performance parameters, that is, a window length value range is generated according to the clip length distribution and the system performance parameters. Specifically, a reference window length value range may be determined according to the clip length distribution, and then a performance score may be generated according to the performance parameters. The reference window length value range is corrected using the performance score to obtain a final window length value range. For example, if the reference window length value range is [3, 10] and the performance score is 90%, then the window length value range may be [3, 9], thereby using the performance score to narrow the window length value range based on the reference window length value range. The lower the performance score, the smaller the generated window length value range. In this way, the window length range can take into account the system performance, avoid the system processing efficiency being reduced due to the window length range being set too long, and avoid the system resource waste caused by the window length range being set too short.

[0123] In an optional implementation, the window length value range may be updated at corresponding intervals or after receiving an update instruction, that is, step S410 and step S420 are re-executed when an update condition is met to obtain an updated window length value range.

[0124] Step S430: generating adjustment parameters according to the clip length.

[0125] The adjustment parameters are the basis for the subsequent generation of correction coefficients. Specifically, if [i, i+m-1] is the time slice corresponding to the highlight clip when the window length is m, the average index value of the time slice is called avg_m, and the index value of the i+mth unit is a i+m , when n*avg_m≤a i+m When ≤avg_m, it indicates that a i+M If it is close to the average value avg_m of the indicator during the exciting period, then a i+m It should also be part of the highlights, that is, a i+m As an extension point of the wonderful period, n is an adjustment parameter, n < 1 but close to 1, for example, n can be 0.8.

[0126] by Figure 5 For example: the sampling unit is minutes, and the playback effect index values ​​of the i, i+1, i+2, and i+3 minutes are a i 、a i+1 、a i+2 、a i+3 [i, i+2] is the time slice corresponding to the highlight clip when the window length is 3 minutes. The average value of the index of the [i, i+2] time slice is avg_3 (also called the three-point average in Table 1). The index value a of the i+3th minute (called the fourth minute in Table 1) is i+3 The specific conditions are as follows:

[0127] Table 1

[0128]

[0129] In this application, it is assumed that [i, i+2] is the time slice corresponding to the 3-minute highlight clip, so a i+3 >avg_3 does not hold; a i+3 =avg_3, indicating that the three-point mean (the average value of the index in the interval [i, i+2]) is the same as the four-point mean (the average value of the index in the interval [i, i+3]), then the fourth minute should be included in the highlights; n*avg_3≤a i+3 When , it indicates that the fourth minute is close to the average value of the highlight segment, the fourth minute is taken as the extension point of the highlight segment; a i+3 When < n*avg_3, it indicates that the fourth minute is significantly lower than the average value of the indicator of the highlight period and should not be used as the extension point of the highlight segment. Therefore, this application uses n*avg_m as the judgment condition for whether to use the i+mth as the extension point of the highlight segment [i, i+m-1].

[0130] In the embodiment of the present application, in order to improve the rationality of the adjustment parameter configuration, this step specifically generates the adjustment parameter according to the clip length obtained in step S410. Figure 6 The steps shown below get the adjustment parameters:

[0131] S431, determining a reference window length according to the number of clip resources corresponding to different clip lengths.

[0132] For each clip length, the number of live highlight clip resources corresponding to the clip length is counted, and this number is the number of clip resources corresponding to the clip length. According to the number of clip resources corresponding to each clip length, the reference window length is determined from the clip lengths that appear. For example, the clip length with the largest number of corresponding clip resources is used as the reference window length.

[0133] S432, determining the first clipping resource quantity of the reference window length and the second clipping resource quantity of the reference window length; wherein the reference window length=the reference window length+1.

[0134] The number of clip resources corresponding to the reference window length is the first number of clip resources, and the reference window length = reference window length + 1, where 1 represents the window length difference between two adjacent sliding time windows. That is, the reference window length is the next window length of the reference window length, and the number of clip resources corresponding to the reference window length is the second number of clip resources.

[0135] For example, the number of clip resources of three minutes is the largest, so three minutes is the benchmark window length, four minutes is the reference window length, the number of clip resources corresponding to three minutes is the first number of clip resources, and the number of clip resources corresponding to four minutes is the second number of clip resources.

[0136] S433: Generate an adjustment parameter according to the first clipping resource quantity and the second clipping resource quantity.

[0137] The adjustment parameter is specifically generated according to the ratio of the second clipping resource quantity to the first clipping resource quantity.

[0138] In an optional implementation, the adjustment parameter may be updated at corresponding intervals, that is, when the update period is reached, step S410 and step S430 are re-executed to obtain updated adjustment parameters.

[0139] Among them, step S410 to step S430 are executed asynchronously with subsequent steps.

[0140] Step S440, obtaining the playback effect index values ​​of the target live broadcast resource at different sampling times.

[0141] Step S450, generating a plurality of sliding time windows with different window lengths according to the window length value range, each sliding time window divides the resource interval of the target live broadcast resource into a plurality of time slices.

[0142] A plurality of sliding time windows with different window lengths are generated according to a window length value range, so that the window length of the generated sliding time window is within the window length value range, and after the sliding time windows are sorted according to the window length of the sliding time window, the length difference between two adjacent sliding time windows is the same, for example, the length difference may be 1 sampling unit.

[0143] Step S460, calculating the average index value of each time slice according to the playback effect index value in each time slice.

[0144] Step S470: Determine the correction coefficient of each sliding time window in combination with the adjustment parameters.

[0145] From the above description, it can be seen that the present application uses n*avg_m as the determination condition of whether to use the i+mth sampling unit as the extension point of the highlight segment [i, i+m-1]. Then a i+3 =n*avg_m is the critical condition, then the average value of the indicator avg_m+1 with a window length of m+1 is as follows:

[0146] avg _m+1 =(n*avg _m +m*avg _m ) / (m+1) (Formula 1)

[0147] Formula 1 is transformed into the following formula 2:

[0148] avg _m+1 / avg _m =(n+m) / (m+1) (Formula 2)

[0149] The correction coefficient of window length m+1 compared to window length m is shown in Formula 3:

[0150] k m+1→m =(m+1) / (n+m) (Formula 3)

[0151] That is, the following formula 4 can make the average values ​​of the indicators of window length m+1 and window length m consistent:

[0152] avg _m+1 *k m+1→m =avg _m (Formula 4)

[0153] Therefore, the correction coefficient k of the window length m+1 compared with the window length m can be calculated by the above formula 3:m+1→m , and accordingly we can get the correction coefficient k of the window length m compared to the window length m-1 m→m-1 . That is, we get the following formula 5:

[0154] k m→m-1 =m / (n+m-1) (Formula 5)

[0155] Among them, k m→m-1 It represents the correction coefficient of window length m compared to window length m-1, and can also be called the correction coefficient of sliding time window m compared to its previous sliding time window m-1, where m is the window length of sliding time window m, and m-1 is the window length of its previous sliding time window m-1 (after sorting the sliding time windows according to their window lengths, the sliding time window located before the sliding time window is called the previous sliding time window of the sliding time window).

[0156] Therefore, for any sliding time window m, the correction coefficient k compared with the previous sliding time window m-1 is calculated according to the above formula 5: m→m-1 Specifically, the sum of the adjustment parameter n, the window length m of the sliding time window, and -1 (corresponding to the denominator in Formula 5) is calculated, and the ratio of the window length m of the sliding time window to the sum (n+m-1) is used as the correction coefficient of the sliding time window compared to the previous sliding time window.

[0157] The present application calculates the correction coefficient of each sliding time window based on the correction coefficient of each sliding time window compared to the previous sliding time window. Specifically, a fixed correction coefficient is assigned to the sliding time window with the smallest window length, such as a fixed correction coefficient of 1; for any non-minimum sliding time window, the correction coefficient of the sliding time window compared to the previous sliding time window is calculated based on the window length and adjustment parameters of the sliding time window; and then the correction coefficient of the sliding time window is calculated based on the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared to the previous sliding time window. For example, the correction coefficient of the sliding time window can be obtained by the following formula 6:

[0158] P m =P m-1 *k m→m-1 (Formula 6)

[0159] Among them, P m is the correction coefficient of the sliding time window m; P m-1 k is the correction coefficient of the previous sliding time window m-1 of the sliding time window m; m→m-1is the correction coefficient of the sliding time window m compared to the previous sliding time window m-1. That is, the correction coefficient of the sliding time window is obtained by multiplying the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared to the previous sliding time window.

[0160] Combining the above formula 5 and formula 6, the general calculation formula 7 of the correction coefficient can be obtained:

[0161]

[0162] Among them, P m represents the correction coefficient of the sliding time window m; m represents the window length of the sliding time window m; Indicates the correction coefficient of the sliding time window with the smallest window length, which is also a fixed correction coefficient; m min Indicates the window length of the sliding time window with the minimum window length; n represents the adjustment parameter.

[0163] Step S480, for each time slice, obtain the effect evaluation value of the time slice according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs.

[0164] Step S490, select the target time slice according to the effect evaluation value of each time slice, and determine the highlight segment in the target live broadcast resource according to the target time slice.

[0165] It can be seen that the highlight clip detection method of the live broadcast resource provided in the embodiment of the present application generates a window length value range in advance according to the clip length of the live broadcast highlight clip clip resources uploaded historically, so that the window length of the generated multiple sliding time windows can be adapted to the actual situation, thereby improving the rationality of the generated sliding time windows; and also determines the adjustment parameters according to the clip length of the live broadcast highlight clip clip resources uploaded historically, thereby improving the determination accuracy of the adjustment parameters; and determines the correction coefficient of the sliding time window compared with the previous sliding time window according to the adjustment parameters and the window length, so that the correction coefficient can be used to accurately balance the indicator mean within two adjacent sliding time windows of different window lengths, and finally generates the correction coefficient of each sliding time window in combination with the correction coefficient, thereby improving the determination accuracy of the correction coefficient and further improving the detection accuracy of the highlight clips.

[0166] Embodiment 3

[0167] Figure 7 FIG. 2 shows a schematic diagram of a device for detecting highlights of a live broadcast resource provided in Embodiment 3 of the present application. Figure 7As shown, the device 700 includes: an indicator acquisition module 710, a window generation module 720, an average value calculation module 730, a correction coefficient determination module 740, an evaluation value calculation module 750, and a segment identification module 760.

[0168] The indicator acquisition module 710 is used to obtain the playback effect indicator values ​​of the target live broadcast resource at different sampling times;

[0169] A window generation module 720, configured to generate a plurality of sliding time windows with different window lengths, each sliding time window dividing a resource interval of a target live broadcast resource into a plurality of time slices;

[0170] The average value calculation module 730 is used to calculate the average value of the index of each time slice according to the playback effect index value in each time slice;

[0171] A correction coefficient determination module 740, used to determine a correction coefficient for each sliding time window;

[0172] An evaluation value calculation module 750 is used to obtain, for each time slice, an effect evaluation value of the time slice according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs;

[0173] The segment identification module 760 is used to filter out target time segments according to the effect evaluation values ​​of each time segment, and determine the wonderful segments in the target live broadcast resource according to the target time segment.

[0174] In an optional implementation, the correction coefficient determination module 740 is used to: assign a fixed correction coefficient to the sliding time window with the smallest window length;

[0175] For any non-minimum sliding time window, the correction coefficient of the sliding time window compared with the previous sliding time window is calculated based on the window length and adjustment parameters of the sliding time window, and the correction coefficient of the sliding time window is calculated based on the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared with the previous sliding time window.

[0176] In an optional implementation, the correction coefficient determination module 740 is used to: calculate the sum of the adjustment parameter, the window length of the sliding time window, and -1;

[0177] The ratio of the window length of the sliding time window to the sum is used as a correction coefficient of the sliding time window compared with the previous sliding time window.

[0178] In an optional implementation, the device further includes an adjustment parameter generation module (not shown in the figure) for counting the clip lengths of the live highlight clip resources uploaded in the past;

[0179] Determining a reference window length according to the number of clip resources corresponding to different clip lengths;

[0180] Determine the first clipping resource quantity of the reference window length and the second clipping resource quantity of the reference window length; wherein the reference window length=reference window length+1;

[0181] The adjustment parameter is generated according to the first clipping resource quantity and the second clipping resource quantity.

[0182] In an optional implementation, the device further includes: a window length range determination module (not shown in the figure), which is used to count the clip lengths of the live broadcast highlight clip resources uploaded in the past;

[0183] Generate a window length value range according to the clip length;

[0184] The generating a plurality of sliding time windows with different window lengths includes: generating a plurality of sliding time windows with different window lengths according to the window length value range.

[0185] In an optional implementation, the window length range determination module is used to: obtain a system performance parameter; and generate a window length value range according to the clip length and the system performance parameter.

[0186] In an optional implementation, the segment identification module 760 is used to: sort the time slices in descending order of effect evaluation values;

[0187] Select the time slices with the effect evaluation values ​​in the top N as the target time slices;

[0188] Aggregating the target time slices to obtain an aggregated time period;

[0189] The segment corresponding to the aggregation period is taken as the highlight segment.

[0190] It can be seen that the highlight segment detection device for live broadcast resources provided by the embodiment of the present application uses multiple sliding time windows with different window lengths to divide the resource interval of the target live broadcast resource to obtain time slices of different lengths, and allocates corresponding correction coefficients to the sliding time windows of different window lengths, and uses the correction coefficient of the sliding time window to which the time slice belongs to correct the average value of the index of the time slice to obtain the effect evaluation value of the time slice, and then identifies the highlight segment according to the effect evaluation value. With this solution, there is no need to perform a complex content recognition process, and the highlight segment recognition efficiency is high; and it can have higher detection accuracy than the fixed time length detection method.

[0191] Embodiment 4

[0192] Figure 8 A schematic diagram of the structure of a computing device provided in Embodiment 4 of the present application is shown. The specific embodiments of the present application do not limit the specific implementation of the computing device.

[0193] like Figure 8 As shown, the computing device may include: a processor (processor) 802 , a communication interface (Communications Interface) 804 , a memory (memory) 806 , and a communication bus 808 .

[0194] The processor 802, the communication interface 804, and the memory 806 communicate with each other via a communication bus 808. The communication interface 804 is used to communicate with other devices such as a client or other server network elements. The processor 802 is used to execute a program 810, which can specifically execute the relevant steps in the above-mentioned embodiment of the method for detecting a highlight segment of a live broadcast resource of a computing device.

[0195] Specifically, the program 810 may include program codes, which include computer operation instructions.

[0196] The processor 802 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0197] The memory 806 is used to store the program 810. The memory 806 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory. The program 810 can be specifically used to enable the processor 802 to execute the steps in the above method embodiment.

[0198] Embodiment 5

[0199] Embodiment 5 of the present application provides a non-volatile computer storage medium, which stores at least one executable instruction or computer program, and the executable instruction or computer program can enable the processor to perform operations corresponding to the method for detecting highlights of live broadcast resources in any of the above-mentioned method embodiments.

[0200] Embodiment 6

[0201] Embodiment 6 of the present application provides a computer program product, which includes at least one executable instruction or computer program, and the executable instruction or computer program can enable a processor to perform operations corresponding to the method for detecting highlights of live broadcast resources in any of the above-mentioned method embodiments.

[0202] In summary, according to the computing device, computer storage medium and computer program product provided by this embodiment, a plurality of sliding time windows with different window lengths are used to divide the resource interval of the target live broadcast resource to obtain time slices of different lengths, and corresponding correction coefficients are assigned to sliding time windows of different window lengths. The correction coefficient of the sliding time window to which the time slice belongs is used to correct the average value of the index of the time slice to obtain the effect evaluation value of the time slice, and then the highlight segment is identified according to the effect evaluation value. With this solution, there is no need to perform a complex content recognition process, and the highlight segment recognition efficiency is high; and it can have higher detection accuracy than the fixed time length detection method.

[0203] The algorithm or display provided here are not inherently related to any specific computer, virtual system or other equipment. Various general systems can also be used together with the teaching based on this. According to the above description, it is obvious to construct the structure required for this type of system. In addition, the present application embodiment is not directed to any specific programming language yet. It should be understood that various programming languages ​​can be utilized to realize the content of the present application described here, and the above description of specific languages ​​is to disclose the best mode of implementation of the present application.

[0204] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0205] Similarly, it should be understood that in order to streamline the present application and help understand one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, the various features of the embodiments of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following intention: the claimed application requires more features than the features clearly stated in each claim. More specifically, as reflected in the claims below, the inventive aspects are less than all the features of the single embodiment disclosed above. Therefore, the claims following the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself serves as a separate embodiment of the present application.

[0206] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and in addition they may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0207] In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present application and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0208] The various component embodiments of the present application can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all functions of some or all components according to the embodiments of the present application. The application can also be implemented as a device or apparatus program (e.g., computer program and computer program product) for executing a part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0209] It should be noted that the above embodiments illustrate the present application rather than limit the present application, and that those skilled in the art may design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be constructed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of multiple such elements. The present application may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A method for detecting highlights of live broadcast resources, characterized in that: include: Get the playback effect index values ​​of the target live broadcast resource at different sampling times; Generate a plurality of sliding time windows with different window lengths, each sliding time window dividing the resource interval of the target live broadcast resource into a plurality of time slices; Calculate the average value of the index of each time slice according to the playback effect index value in each time slice; Determine the correction factor for each sliding time window; For each time slice, the effect evaluation value of the time slice is obtained according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs; The target time slices are screened out according to the effect evaluation values ​​of the respective time slices, and the highlight segments in the target live broadcast resource are determined according to the target time slices.

2. The method according to claim 1, characterized in that Determining the correction coefficient of each sliding time window includes: Assign a fixed correction factor to the sliding time window with the smallest window length; For any non-minimum sliding time window, calculate the correction coefficient of the sliding time window compared with the previous sliding time window based on the window length and adjustment parameters of the sliding time window; calculate the correction coefficient of the sliding time window based on the correction coefficient of the previous sliding time window and the correction coefficient of the sliding time window compared with the previous sliding time window.

3. The method according to claim 2, characterized in that The step of calculating the correction coefficient of the sliding time window compared to the previous sliding time window according to the window length and the adjustment parameter of the sliding time window includes: Calculate the sum of the adjustment parameter, the window length of the sliding time window, and -1; The ratio of the window length of the sliding time window to the sum is used as a correction coefficient of the sliding time window compared with the previous sliding time window.

4. The method according to claim 3, characterized in that The adjustment parameters are determined in the following manner: Count the clip lengths of live highlights clip resources uploaded in the past; Determining a reference window length according to the number of clip resources corresponding to different clip lengths; Determine the first clipping resource quantity of the reference window length and the second clipping resource quantity of the reference window length; wherein the reference window length=reference window length+1; The adjustment parameter is generated according to the first clipping resource quantity and the second clipping resource quantity.

5. The method according to any one of claims 1 to 4, characterized in that Before generating a plurality of sliding time windows with different window lengths, the method further comprises: counting the clip lengths of the live highlight clip resources uploaded in the past; generating a window length value range according to the clip lengths; The generating a plurality of sliding time windows with different window lengths includes: generating a plurality of sliding time windows with different window lengths according to the window length value range.

6. The method according to claim 5, characterized in that The method further comprises: obtaining system performance parameters; The generating a window length value range according to the clipping length includes: generating the window length value range according to the clipping length and the system performance parameter.

7. The method according to any one of claims 1 to 6, characterized in that The step of selecting a target time slice according to the effect evaluation value of each time slice, and determining a highlight segment in the target live broadcast resource according to the target time slice comprises: Sorting the time slices in descending order of effect evaluation values; Select the time slices with the effect evaluation values ​​in the top N as the target time slices; Aggregating the target time slices to obtain an aggregated time period; The segment corresponding to the aggregation period is taken as the highlight segment.

8. A device for detecting highlights of live broadcast resources, characterized in that: include: An indicator acquisition module is used to obtain the playback effect indicator values ​​of the target live broadcast resource at different sampling times; A window generation module, used to generate a plurality of sliding time windows with different window lengths, each sliding time window dividing the resource interval of the target live broadcast resource into a plurality of time slices; The average value calculation module is used to calculate the average value of the index of each time slice according to the playback effect index value in each time slice; A correction coefficient determination module, used to determine the correction coefficient of each sliding time window; An evaluation value calculation module is used to obtain, for each time slice, an effect evaluation value of the time slice according to the average value of the index of the time slice and the correction coefficient of the sliding time window to which the time slice belongs; The segment identification module is used to filter out the target time slice according to the effect evaluation value of each time slice, and determine the wonderful segments in the target live broadcast resource according to the target time slice.

9. A computing device, characterized in that include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the method for detecting highlights of live broadcast resources as described in any one of claims 1-7.

10. A computer storage medium, characterized in that: The storage medium stores at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the method for detecting highlights of live broadcast resources as described in any one of claims 1-7.

11. A computer program product, characterized in that It includes at least one executable instruction, and the executable instruction enables the processor to execute the operation corresponding to the method for detecting highlights of live broadcast resources as described in any one of claims 1-7.

Images