Video push method, device, computer equipment and storage medium

Abstract

The present disclosure provides a video pushing method, apparatus, computer equipment and storage medium, including: obtaining a video to be pushed; determining a reference video whose similarity detection result with the video to be pushed meets preset conditions; based on historical playback information of the reference video and similarity detection results between the reference video and the video to be pushed, determining a target highlight segment of the video to be pushed, and pushing the video to be pushed based on the target highlight segment.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a video push method, apparatus, computer equipment and storage medium. Background Art

[0002] When pushing videos, in order to improve the push effect of the video, the highlight clips of the video are generally selected for push. The highlight clips of the video can be understood as the more exciting or attractive clips in the video. As more and more video resources are available on the Internet, the competition among video resources is becoming more and more fierce. The selection of highlight clips directly determines the consumption of video resources in some scenes. Therefore, it is particularly important to select highlight clips. Summary of the invention

[0003] The embodiments of the present disclosure at least provide a video push method, apparatus, computer equipment and storage medium.

[0004] In a first aspect, an embodiment of the present disclosure provides a video push method, including:

[0005] Get the video to be pushed;

[0006] Determine a reference video whose similarity detection result with the video to be pushed meets a preset condition;

[0007] Based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, the target highlight segment of the video to be pushed is determined, and the video to be pushed is pushed based on the target highlight segment.

[0008] In a possible implementation manner, determining a reference video whose similarity detection result with the video to be pushed meets a preset condition includes:

[0009] Dividing the video to be pushed into multiple video segments;

[0010] For any video segment, a reference video whose similarity detection result with the video segment meets a preset condition is determined.

[0011] In a possible implementation manner, dividing the video to be pushed into a plurality of video segments includes:

[0012] Divide the video to be pushed into multiple video segments according to preset duration; or,

[0013] Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or,

[0014] Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

[0015] In a possible implementation manner, the similarity detection result between the reference video and the video to be pushed includes time information of a video segment in the video to be pushed that is similar to the reference video, and the historical playback information includes a historical playback count and a historical completion count;

[0016] The determining of a target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed includes:

[0017] For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the reference moment is a reference video whose corresponding time information includes the video moment;

[0018] Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment;

[0019] Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

[0020] In a possible implementation manner, determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment includes:

[0021] Determining a completion rate threshold based on a first predicted completion rate corresponding to each video moment;

[0022] Based on the completion rate threshold, a target highlight segment of the video to be pushed is determined.

[0023] In a possible implementation manner, determining the target highlight segment of the video to be pushed based on the completion rate threshold includes:

[0024] Determine a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold;

[0025] Determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment;

[0026] The target highlight segment is determined based on the second predicted completion rates respectively corresponding to the candidate highlight segments.

[0027] In a possible implementation manner, after determining, based on the completion rate threshold, a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold, the method further includes:

[0028] The time intervals between adjacent candidate highlight segments are determined, and the adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval are merged, and the merged candidate highlight segments are used as updated candidate highlight segments.

[0029] In a possible implementation manner, merging adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval includes:

[0030] Among the adjacent candidate highlight segments whose corresponding time intervals are smaller than the preset time interval, determining the start time of the candidate highlight segment with an earlier time in the video to be pushed and the end time of the candidate highlight segment with a later time in the video to be pushed;

[0031] The video segment between the start time and the end time is used as a candidate highlight segment after merging.

[0032] In a second aspect, the embodiment of the present disclosure further provides a video push device, including:

[0033] The acquisition module is used to obtain the video to be pushed;

[0034] A first determination module is used to determine a reference video whose similarity detection result with the video to be pushed meets a preset condition;

[0035] The second determination module is used to determine the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, and push the video to be pushed based on the target highlight segment.

[0036] In a possible implementation manner, when determining a reference video whose similarity detection result with the video to be pushed satisfies a preset condition, the first determination module is configured to:

[0037] Dividing the video to be pushed into multiple video segments;

[0038] For any video segment, a reference video whose similarity detection result with the video segment meets a preset condition is determined.

[0039] In a possible implementation manner, when dividing the video to be pushed into a plurality of video segments, the first determining module is configured to:

[0040] Divide the video to be pushed into multiple video segments according to preset duration; or,

[0041] Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or,

[0042] Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

[0043] In a possible implementation manner, the similarity detection result between the reference video and the video to be pushed is time information of a video segment in the video to be pushed that is similar to the reference video, and the historical playback information includes a historical playback count and a historical completion count;

[0044] The second determination module, when determining the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, is used to:

[0045] For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the reference moment is a reference video whose corresponding time information includes the video moment;

[0046] Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment;

[0047] Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

[0048] In a possible implementation manner, when determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment, the second determination module is used to:

[0049] Determining a completion rate threshold based on a first predicted completion rate corresponding to each video moment;

[0050] Based on the completion rate threshold, a target highlight segment of the video to be pushed is determined.

[0051] In a possible implementation manner, when determining the target highlight segment of the video to be pushed based on the completion rate threshold, the second determination module is configured to:

[0052] Determine a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold;

[0053] Determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment;

[0054] The target highlight segment is determined based on the second predicted completion rates respectively corresponding to the candidate highlight segments.

[0055] In a possible implementation manner, after determining, based on the completion rate threshold, a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold, the second determination module is further configured to:

[0056] The time intervals between adjacent candidate highlight segments are determined, and the adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval are merged, and the merged candidate highlight segments are used as updated candidate highlight segments.

[0057] In a possible implementation manner, when merging adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval, the second determination module is configured to:

[0058] Among the adjacent candidate highlight segments whose corresponding time intervals are smaller than the preset time interval, determining the start time of the candidate highlight segment with an earlier time in the video to be pushed and the end time of the candidate highlight segment with a later time in the video to be pushed;

[0059] The video segment between the start time and the end time is used as a candidate highlight segment after merging.

[0060] In a third aspect, an embodiment of the present disclosure further provides a computer device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are performed.

[0061] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are executed.

[0062] The video push method, device, computer equipment and storage medium provided by the embodiments of the present disclosure can estimate the target highlight segment in the video to be pushed based on the historical playback information of a reference video similar to the video to be pushed after obtaining the video to be pushed, and push the video to be pushed based on the target highlight segment. In this way, by estimating the historical playback information of each video playback moment of the video to be pushed through the historical playback information of the reference video, the target highlight segment with better estimated historical playback information can be selected, and the video to be pushed can be further pushed based on the target highlight segment, thereby improving the push effect of the video to be pushed.

[0063] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0065] Figure 1 A flow chart of a video push method provided by an embodiment of the present disclosure is shown;

[0066] Figure 2 An overall flow chart of a video push method provided by an embodiment of the present disclosure is shown;

[0067] Figure 3 A schematic diagram of the architecture of a video push device provided by an embodiment of the present disclosure is shown;

[0068] Figure 4 A schematic diagram of the structure of a computer device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0070] In the related art, when determining the highlight segment in a video, there are generally two main methods:

[0071] Method A: Based on neural network model.

[0072] Specifically, a neural network model can be pre-trained to detect highlight segments in a video. However, when training such a neural network model, it is generally based on sample videos and highlight segment annotation information in the sample videos. Since the highlight segment annotation information in the sample videos is manually annotated, different people may annotate different highlight segments when viewing the same sample video. Therefore, in this method, since the annotation information itself is subjective, the accuracy of the trained neural network model is low.

[0073] Method B: Based on playback data.

[0074] Specifically, after a video has been played for a period of time, the highlight clips in the video can be screened out based on the video viewing curve, video barrage, etc. However, this method is only applicable to videos with a certain historical playback time, and is not applicable to videos that do not contain playback data or have less playback data.

[0075] Based on the above research, the present disclosure provides a video push method, device, computer equipment and storage medium, which can estimate the target highlight segment in the video to be pushed based on the historical playback information of the reference video similar to the video to be pushed after obtaining the video to be pushed, and push the video to be pushed based on the target highlight segment. In this way, by estimating the historical playback information of each video playback moment of the video to be pushed through the historical playback information of the reference video, the target highlight segment with better estimated historical playback information can be screened out, and the video to be pushed can be further pushed based on the target highlight segment, thereby improving the push effect of the video to be pushed.

[0076] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0077] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0078] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0079] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.

[0080] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0081] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0082] To facilitate understanding of this embodiment, a video push method disclosed in the embodiment of the present disclosure is first introduced in detail. The executor of the video push method provided in the embodiment of the present disclosure is generally a computer device with certain computing capabilities. The computer device includes, for example: a terminal device or a server or other processing device. The terminal device can be a personal computer, a smart phone, a tablet computer, etc.

[0083] See also Figure 1 As shown, it is a flowchart of the video push method provided by the embodiment of the present disclosure, and the method includes steps 101 to 103, wherein:

[0084] Step 101: Get the video to be pushed.

[0085] Step 102: Determine a reference video whose similarity detection result with the video to be pushed meets a preset condition.

[0086] Step 103: Based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, determine the target highlight segment of the video to be pushed, and push the video to be pushed based on the target highlight segment.

[0087] The following is a detailed description of the above steps.

[0088] For step 101,

[0089] In different scenarios, the method of obtaining the video to be pushed may be different. For example, if the execution subject of the video push method provided by the present disclosure is a server, the method of obtaining the video to be pushed may refer to obtaining the video to be pushed uploaded by each user terminal. Accordingly, after obtaining the video to be pushed, the server may determine the target highlight segment of the video to be pushed according to the method provided by the present disclosure, and push the target highlight segment to each user terminal.

[0090] If the subject of executing the video push provided by the present disclosure is a user terminal, then obtaining the video to be pushed may refer to obtaining the video to be pushed locally. Accordingly, after the user terminal determines the target highlight segment of the video to be pushed according to the method provided by the present disclosure, the time information of the target highlight segment in the video to be pushed and the video to be pushed can be uploaded to the server, and the target highlight segment corresponding to the time information can be pushed to each user terminal through the server.

[0091] Since the purpose of determining the target highlight segment is to push the video, if the quality of the video to be pushed is relatively poor, the push effect may be affected accordingly. Therefore, in a possible implementation, after obtaining the video to be pushed, the quality of the video to be pushed can be tested first, and after the quality of the video to be pushed passes the test, the target highlight segment of the video to be pushed can be determined.

[0092] The quality detection of the video to be pushed may refer to the detection of the video itself to be pushed, for example, the picture clarity, whether it is equipped with subtitles, picture jitter, etc. may be detected.

[0093] In a possible implementation, the quality of the video to be pushed can be detected by a pre-trained quality detection model. The quality detection model can be trained by sample videos and quality marks of the sample videos, and the quality marks of the sample videos can be used to mark the sample videos as having quality problems.

[0094] For step 102,

[0095] In a possible implementation, when determining a reference video whose similarity detection result with the video to be pushed meets a preset condition, similarity detection may be performed between the video to be pushed and each candidate video in the video library.

[0096] The candidate videos in the video library may refer to videos whose popularity information meets the popularity conditions, for example, videos whose playback volume exceeds a preset playback volume, and / or the number of comments exceeds a preset number of comments, and / or the completion rate exceeds a preset completion rate.

[0097] The completion rate may refer to the ratio between the number of complete playbacks and the total number of playbacks.

[0098] Since there are large differences between videos with high popularity in different periods, for example, videos with high popularity last month are type A videos, and videos with high popularity this month are type B videos, in order to meet the requirements for video types in each period, the candidate videos in the video library can also be videos within a preset time from the current moment, for example, videos from the past three years.

[0099] When performing similarity detection between the video to be pushed and each candidate video in the video library, visual similarity detection can be performed illustratively. Specifically, the visual features of the video to be pushed can be extracted based on a pre-trained visual feature extraction network, and then the similarity between the visual features of the video to be pushed and the visual features of each candidate video can be calculated, and the similarity detection result can include the similarity.

[0100] In practical applications, similarity detection may also be performed through other modal features, such as audio features or portrait features, etc., which is not limited in the present disclosure.

[0101] In actual applications, since the video to be pushed is often relatively long, if the video to be pushed is directly tested for similarity with each candidate video, the detection speed may be slow. Therefore, in order to improve the detection speed, the video to be pushed can be divided into multiple video segments, and then for any video segment, the video segment can be tested for similarity with each candidate video to determine a reference video whose similarity detection result with the video segment meets preset conditions.

[0102] When dividing the video to be pushed into multiple video segments, any one of the following methods may be used:

[0103] Divide the video to be pushed into multiple video segments according to preset duration; or,

[0104] Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or,

[0105] Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

[0106] The preset duration can be determined in combination with the push status of historical highlight clips. For example, the average duration of historical highlight clips with good push effects can be used as the preset duration. The push effect can be determined by the number of push playbacks. The number of push playbacks can refer to the number of playbacks of the original video entered by clicking on the pushed highlight clip. The historical highlight clip with good push effect can refer to the historical highlight clip with a push playback number exceeding the preset number, for example.

[0107] The performing scene detection on the video to be pushed may, for example, be performed on the video to be pushed through a pre-trained scene detection network; the dividing the video to be pushed into a plurality of video segments based on the scene detection result may, for example, determine the time point of scene transition according to the scene detection result, and divide the video to be pushed according to the time point of scene transition.

[0108] The performing shot detection on the video to be pushed may, for example, be performed on the video to be pushed by a pre-trained shot detection network; based on the shot detection result, the video to be pushed is divided into a plurality of video segments, for example, the time point of the shot conversion may be determined according to the shot detection result, and the video to be pushed may be divided according to the time point of the shot conversion.

[0109] The reference video whose similarity detection result meets the preset conditions may exemplarily refer to the candidate videos among the above-mentioned candidate videos, whose corresponding similarity exceeds the preset similarity; or the similarities may be sorted from large to small, and the candidate videos ranked in the top K positions may be used as the reference videos, where K is a preset positive integer.

[0110] Exemplarily, after the video to be pushed is divided into M video segments, for any video segment, a candidate video whose similarity with the video segment exceeds a preset similarity can be used as a reference video corresponding to the video segment; or based on the similarity with the video segment, each candidate video can be sorted from large to small, and the top K candidate videos can be used as reference videos corresponding to the video segment. Accordingly, there are at most M*K reference videos corresponding to the video to be pushed.

[0111] For step 103,

[0112] The similarity detection result between the reference video and the video to be pushed meets the preset conditions, that is, it can be understood that the reference video is similar to the video to be pushed, but the reference video may only be similar to a partial time period of the video to be pushed. Therefore, in a possible implementation, the similarity detection result between the reference video and the video to be pushed includes the time information of the video clips in the video to be pushed that are similar to the reference video.

[0113] Exemplarily, if reference video A is similar to the video to be pushed, it is possible that the reference video A is similar to the 20th to 30th seconds of the video to be pushed, so the similarity detection result between reference video A and the video to be pushed includes the 20th to 30th seconds.

[0114] The historical playback information of the reference video may include, for example, the number of historical playbacks and the number of historical completions, where the number of historical completions may refer to the number of historical complete playbacks.

[0115] In a possible implementation manner, when determining the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, the following steps may be performed:

[0116] Step a1: for any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the reference moment is a reference video whose corresponding time information includes the video moment.

[0117] Step a2: Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment.

[0118] Step a3: determining a target highlight segment of the video to be pushed based on a first predicted completion rate corresponding to each video moment.

[0119] The video moment of the video to be pushed may be a preset moment in the video to be pushed, for example, every 2 seconds may be a video moment; or the video moment of the video to be pushed may be a moment determined according to the duration of the video to be pushed and the number of preset moments, the number of preset moments being the number of video moments of the video to be pushed, for example, if the duration of the video to be pushed is 5 minutes and the number of preset moments is 5, then the 1st minute, 2nd minute, 3rd minute, 4th minute and 5th minute of the video to be pushed are the video moments respectively.

[0120] Since the similarity detection result between the reference video and the video to be pushed includes the time information of the video segments in the video to be pushed that are similar to the reference video, the reference video corresponding to each video moment can be determined based on the time information of the video segments in the video to be pushed that are similar to the reference video.

[0121] For example, it is shown in Table 1 below:

[0122] Table 1

[0123]

[0124] Video 1 is similar to the 10th to 20th seconds of the video to be pushed, video 2 is similar to the 20th to 30th seconds of the video to be pushed, and video 3 is similar to the 20th to 50th seconds of the video to be pushed. Then the reference video corresponding to the 10th second of the video moment is video 1, the reference videos corresponding to the 20th second of the video moment are video 1, video 2, and video 3, the reference videos corresponding to the 30th second of the video moment are video 2 and video 3, the reference video corresponding to the 40th second of the video moment is video 3, and the reference video corresponding to the 50th second of the video moment is video 3.

[0125] In step a2, in a possible implementation, for any video moment, when determining the first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment, the ratio of the sum of the historical number of playbacks of the reference videos corresponding to the video moment to the sum of the historical number of completions of the reference videos corresponding to the video moment can be used as the first predicted completion rate corresponding to the video moment.

[0126] For example, when calculating the first predicted completion rate threshold corresponding to each video moment, it can be calculated by the following formula:

[0127]

[0128] Among them, r i represents the first predicted completion rate of the i-th video moment, m represents the number of reference videos corresponding to the i-th video moment, and f j represents the number of historical completions of the jth reference video, ω j Indicates the historical playback count of the j-th video.

[0129] For example, continuing from Table 1 above, if the historical playback times and historical completion times corresponding to each reference video in Table 1 are as shown in Table 2 below:

[0130] Table 2

[0131] Reference Video Video 1 Video 2 Video 3 Time Information 10 to 20 seconds 20 to 30 seconds 20 to 50 seconds Historical playback times a b c Historical completion times m n p

[0132] Correspondingly, the first predicted completion rate corresponding to the 10th second of the video moment is m / n, the first predicted completion rate corresponding to the 20th second of the video moment is (m+n+p) / (a+b+c), the first predicted completion rate corresponding to the 30th second of the video moment is (n+p) / (b+c), and the first predicted completion rate corresponding to the 40th and 50th seconds of the video moment is p / c.

[0133] In one possible implementation, in step a3, when determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment, a predicted completion rate curve corresponding to the video to be pushed can be drawn based on the first predicted completion rate corresponding to each video moment, and then the video segment in the predicted completion rate curve that exceeds a preset completion rate threshold can be used as the target highlight segment.

[0134] However, since the first predicted completion rate is determined based on the historical playback information of the reference video, the historical playback information of different types of videos may differ greatly. If a uniformly specified predicted completion rate is used, the target highlight segments may not be screened out, or too many target highlight segments may be screened out.

[0135] Therefore, in one possible implementation, different preset completion rate thresholds may be set in advance for different types of videos to be pushed, or in another possible implementation, the completion rate threshold may be determined based on the predicted completion rate of the current video to be pushed.

[0136] Specifically, a completion rate threshold may be determined based on the first predicted completion rate corresponding to each video moment, and then the target highlight segment of the video to be pushed may be determined based on the completion rate threshold.

[0137] When determining the completion rate threshold based on the first predicted completion rate corresponding to each video moment, the first predicted completion rates corresponding to each video moment can be sorted in ascending order, and a preset proportion of the first predicted completion rates can be used as the completion rate threshold.

[0138] Exemplarily, the preset ratio may be 80%. Then, after sorting the first predicted completion rates corresponding to the video moments in ascending order, the 80% predicted completion rate may be used as the completion rate threshold.

[0139] After determining the completion rate threshold, when determining the target highlight segment of the video to be pushed based on the completion rate threshold, the following steps may be performed exemplarily:

[0140] Step b1: determining candidate highlight segments whose corresponding first predicted completion rates are higher than the completion rate threshold.

[0141] Step b2: determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment.

[0142] Step b3: determining the target highlight segment based on the second predicted completion rates corresponding to the candidate highlight segments.

[0143] In one possible implementation, when determining the second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment contained in the candidate highlight segment, the mean / median of the first predicted completion rate of each video moment contained in the candidate highlight segment can be exemplarily used as the second predicted completion rate corresponding to the candidate highlight segment.

[0144] When determining the target highlight segment based on the second predicted completion rates corresponding to each of the candidate highlight segments, it is exemplified that the candidate highlight segments can be sorted from large to small based on the corresponding second predicted completion rates, and the candidate highlight segments ranked first H are used as the target highlight segments (H is a preset positive integer), or the candidate highlight segments whose corresponding second predicted completion rates are greater than a preset value can be used as the target highlight segments.

[0145] After determining the candidate highlight segments based on the completion rate threshold, the determined candidate highlight segments may be shorter in duration. Therefore, in a possible implementation, in order to ensure the continuity of the target highlight segments finally determined, after determining the candidate highlight segments based on the completion rate threshold, some candidate highlight segments may be merged.

[0146] Specifically, the time intervals between adjacent candidate highlight segments may be determined, and adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval may be merged, and the merged candidate highlight segments may be used as updated candidate highlight segments.

[0147] Exemplarily, if candidate highlight segment 1 is the video segment of the 10th to 50th seconds of the video to be pushed, candidate highlight segment 2 is the video segment of the 70th to 100th seconds of the video to be pushed, and the video segment of the 51st to 69th seconds of the video to be pushed is not the candidate highlight segment, if the preset time interval is 30 seconds, the time interval between candidate highlight segment 1 and candidate highlight segment 2 is 20 seconds, which is less than the preset time interval, then candidate highlight segment 1 and candidate highlight segment 2 can be merged.

[0148] Specifically, when merging adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval, exemplarily, the start time of the candidate highlight segment that is earlier in the video to be pushed and the end time of the candidate highlight segment that is later in the video to be pushed can be determined among the adjacent candidate highlight segments whose corresponding time intervals are less than the preset time interval; and then the video segment between the start time and the end time is used as the merged candidate highlight segment.

[0149] Continuing with the above example, among the adjacent candidate highlight segments to be merged, candidate highlight segment 1 is an earlier candidate highlight segment in the video to be pushed, and its start time is the 10th second. Candidate highlight segment 2 is a later candidate highlight segment in the video to be pushed, and its end time is the 100th second. After merging candidate highlight segment 1 and candidate highlight segment 2, the merged candidate highlight segments are from the 10th to the 100th second.

[0150] It should be noted that, when merging candidate highlight segments, at least two candidate highlight segments may be merged, and the merged candidate highlight segment may be used as an updated candidate highlight segment.

[0151] After the candidate highlight segments are merged, it is necessary to recalculate the second predicted completion rate of the merged candidate highlight segments. For example, the mean / median of the first predicted completion rates corresponding to each video moment contained in the merged candidate highlight segments can be used as the second predicted completion rate of the merged candidate highlight segments. The merged candidate highlight segments are no substantially different from the candidate highlight segments selected directly based on the first predicted completion rate, and can also participate in the screening of step b3.

[0152] In a possible implementation, in order to ensure the scene integrity or shot integrity of the target highlight segment, after determining the target highlight segment based on the above method, the scene recognition result and / or shot recognition result of the video to be pushed may also be combined.

[0153] Exemplarily, the starting scene segment where the starting moment of the target highlight segment is located and the ending scene segment where the ending moment of the target highlight segment is located can be determined, and the segment between the starting moment of the starting scene segment and the ending moment of the ending scene segment can be used as the updated target highlight segment.

[0154] After determining the target highlight segment, the video to be pushed can be pushed based on the target highlight segment. For example, each target highlight segment can be pushed to each user terminal for display. After triggering the target highlight segment, the user terminal can jump to the video to be pushed where the target highlight segment is located and play it, thereby improving the playback effect of the video to be pushed by displaying the target highlight segment.

[0155] The following is an introduction to the overall process of the above video push method in conjunction with the overall flow chart. Figure 2 As shown, it is an overall flow chart of a video push method provided by the present disclosure, which includes the following steps:

[0156] Step 1: Get the video to be pushed.

[0157] Step 2: Divide the video to be pushed into multiple video segments.

[0158] Step 3: Find reference videos similar to each video clip.

[0159] Step 4: Determine the first predicted completion rate of each video moment.

[0160] Step 5: Determine the completion rate threshold.

[0161] Step 6: Screen candidate highlight segments and merge similar candidate highlight segments.

[0162] Step 7: Sort candidate highlight segments.

[0163] Step 8: Determine the target highlight segment.

[0164] The detailed description of the above steps refers to the above embodiments and will not be repeated here.

[0165] The video push method provided by the embodiment of the present disclosure can estimate the target highlight segment in the video to be pushed based on the historical playback information of the reference video similar to the video to be pushed after obtaining the video to be pushed, and push the video to be pushed based on the target highlight segment. In this way, by estimating the historical playback information of each video playback moment of the video to be pushed through the historical playback information of the reference video, the target highlight segment with better estimated historical playback information can be selected, and the video to be pushed can be further pushed based on the target highlight segment, thereby improving the push effect of the video to be pushed.

[0166] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0167] Based on the same inventive concept, a video pushing device corresponding to the video pushing method is also provided in the embodiment of the present disclosure. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned video pushing method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0168] Reference Figure 3 , which is a schematic diagram of the architecture of a video push device provided by an embodiment of the present disclosure, the device comprises: an acquisition module 301, a first determination module 302, and a second determination module 303; wherein,

[0169] The acquisition module 301 is used to acquire the video to be pushed;

[0170] A first determination module 302 is used to determine a reference video whose similarity detection result with the video to be pushed meets a preset condition;

[0171] The second determination module 303 is used to determine the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, and push the video to be pushed based on the target highlight segment.

[0172] In a possible implementation manner, when determining a reference video whose similarity detection result with the video to be pushed satisfies a preset condition, the first determination module 302 is configured to:

[0173] Dividing the video to be pushed into multiple video segments;

[0174] For any video segment, a reference video whose similarity detection result with the video segment meets a preset condition is determined.

[0175] In a possible implementation manner, when dividing the video to be pushed into a plurality of video segments, the first determining module 302 is configured to:

[0176] Divide the video to be pushed into multiple video segments according to preset duration; or,

[0177] Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or,

[0178] Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

[0179] In a possible implementation manner, the similarity detection result between the reference video and the video to be pushed is time information of a video segment in the video to be pushed that is similar to the reference video, and the historical playback information includes a historical playback count and a historical completion count;

[0180] The second determination module 303, when determining the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, is used to:

[0181] For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the reference moment is a reference video whose corresponding time information includes the video moment;

[0182] Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment;

[0183] Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

[0184] In a possible implementation manner, when determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment, the second determination module 303 is configured to:

[0185] Determining a completion rate threshold based on a first predicted completion rate corresponding to each video moment;

[0186] Based on the completion rate threshold, a target highlight segment of the video to be pushed is determined.

[0187] In a possible implementation manner, when determining the target highlight segment of the video to be pushed based on the completion rate threshold, the second determination module 303 is configured to:

[0188] Determine a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold;

[0189] Determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment;

[0190] The target highlight segment is determined based on the second predicted completion rates respectively corresponding to the candidate highlight segments.

[0191] In a possible implementation manner, after determining, based on the completion rate threshold, a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold, the second determination module 303 is further configured to:

[0192] The time intervals between adjacent candidate highlight segments are determined, and the adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval are merged, and the merged candidate highlight segments are used as updated candidate highlight segments.

[0193] In a possible implementation manner, when merging adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval, the second determination module 303 is configured to:

[0194] Among the adjacent candidate highlight segments whose corresponding time intervals are smaller than the preset time interval, determining the start time of the candidate highlight segment with an earlier time in the video to be pushed and the end time of the candidate highlight segment with a later time in the video to be pushed;

[0195] The video segment between the start time and the end time is used as a candidate highlight segment after merging.

[0196] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0197] Based on the same technical concept, the embodiment of the present disclosure also provides a computer device. Figure 4 As shown, it is a schematic diagram of the structure of a computer device 400 provided in an embodiment of the present disclosure, including a processor 401, a memory 402, and a bus 403. Among them, the memory 402 is used to store execution instructions, including a memory 4021 and an external memory 4022; the memory 4021 here is also called an internal memory, which is used to temporarily store the operation data in the processor 401, and the data exchanged with the external memory 4022 such as a hard disk. The processor 401 exchanges data with the external memory 4022 through the memory 4021. When the computer device 400 is running, the processor 401 communicates with the memory 402 through the bus 403, so that the processor 401 executes the following instructions:

[0198] Get the video to be pushed;

[0199] Determine a reference video whose similarity detection result with the video to be pushed meets a preset condition;

[0200] Based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, the target highlight segment of the video to be pushed is determined, and the video to be pushed is pushed based on the target highlight segment.

[0201] In a possible implementation manner, in the instructions executed by the processor 401, determining a reference video whose similarity detection result with the video to be pushed meets a preset condition includes:

[0202] Dividing the video to be pushed into multiple video segments;

[0203] For any video segment, a reference video whose similarity detection result with the video segment meets a preset condition is determined.

[0204] In a possible implementation manner, in the instruction executed by the processor 401, dividing the video to be pushed into multiple video segments includes:

[0205] Divide the video to be pushed into multiple video segments according to preset duration; or,

[0206] Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or,

[0207] Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

[0208] In a possible implementation, in the instruction executed by the processor 401, the similarity detection result between the reference video and the video to be pushed includes time information of a video segment in the video to be pushed that is similar to the reference video, and the historical playback information includes a historical playback count and a historical completion count;

[0209] The determining of a target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed includes:

[0210] For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the reference moment is a reference video whose corresponding time information includes the video moment;

[0211] Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment;

[0212] Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

[0213] In a possible implementation manner, in the instructions executed by the processor 401, determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment includes:

[0214] Determining a completion rate threshold based on a first predicted completion rate corresponding to each video moment;

[0215] Based on the completion rate threshold, a target highlight segment of the video to be pushed is determined.

[0216] In a possible implementation manner, in the instructions executed by the processor 401, determining the target highlight segment of the video to be pushed based on the completion rate threshold includes:

[0217] Determine a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold;

[0218] Determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment;

[0219] The target highlight segment is determined based on the second predicted completion rates respectively corresponding to the candidate highlight segments.

[0220] In a possible implementation manner, the instructions executed by the processor 401, after determining, based on the completion rate threshold, a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold, further include:

[0221] The time intervals between adjacent candidate highlight segments are determined, and the adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval are merged, and the merged candidate highlight segments are used as updated candidate highlight segments.

[0222] In a possible implementation manner, in the instruction executed by the processor 401, merging adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval includes:

[0223] Among the adjacent candidate highlight segments whose corresponding time intervals are smaller than the preset time interval, determining the start time of the candidate highlight segment with an earlier time in the video to be pushed and the end time of the candidate highlight segment with a later time in the video to be pushed;

[0224] The video segment between the start time and the end time is used as a candidate highlight segment after merging.

[0225] The embodiment of the present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the video push method described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0226] The embodiments of the present disclosure also provide a computer program product, which carries a program code. The instructions included in the program code can be used to execute the steps of the video pushing method described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0227] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0228] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0229] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0230] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0231] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0232] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A video push method, characterized in that: include: Get the video to be pushed; Determine a reference video whose similarity detection result with the video to be pushed meets a preset condition, wherein the reference video includes a video whose heat information meets a heat condition; Based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, determining the target highlight segment of the video to be pushed, and pushing the video to be pushed based on the target highlight segment; The similarity detection result between the reference video and the video to be pushed includes time information of video segments in the video to be pushed that are similar to the reference video, and the historical playback information includes the number of historical playbacks and the number of historical completions; Wherein, based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, determining the target highlight segment of the video to be pushed includes: For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the video moment is a reference video whose corresponding time information includes the video moment; Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment; Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

2. The method according to claim 1, characterized in that The determining of a reference video whose similarity detection result with the video to be pushed meets a preset condition includes: Dividing the video to be pushed into multiple video segments; For any video segment, a reference video whose similarity detection result with the video segment meets a preset condition is determined.

3. The method according to claim 2, characterized in that The step of dividing the video to be pushed into a plurality of video segments includes: Divide the video to be pushed into multiple video segments according to preset duration; or, Performing scene detection on the video to be pushed, and dividing the video to be pushed into multiple video segments based on the scene detection result; or, Shot detection is performed on the video to be pushed, and based on the shot detection result, the video to be pushed is divided into multiple video segments.

4. The method according to claim 1, characterized in that: The step of determining the target highlight segment of the video to be pushed based on the first predicted completion rate corresponding to each video moment includes: Determining a completion rate threshold based on a first predicted completion rate corresponding to each video moment; Based on the completion rate threshold, a target highlight segment of the video to be pushed is determined.

5. The method according to claim 4, characterized in that The determining, based on the completion rate threshold, a target highlight segment of the video to be pushed includes: Determine a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold; Determining a second predicted completion rate corresponding to the candidate highlight segment based on the first predicted completion rate corresponding to each video moment included in the candidate highlight segment; The target highlight segment is determined based on the second predicted completion rates respectively corresponding to the candidate highlight segments.

6. The method according to claim 5, characterized in that After determining, based on the completion rate threshold, a candidate highlight segment whose corresponding first predicted completion rate is higher than the completion rate threshold, the method further includes: The time intervals between adjacent candidate highlight segments are determined, and the adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval are merged, and the merged candidate highlight segments are used as updated candidate highlight segments.

7. The method according to claim 6, characterized in that The merging of adjacent candidate highlight segments whose corresponding time intervals are less than a preset time interval includes: Among the adjacent candidate highlight segments whose corresponding time intervals are smaller than the preset time interval, determining the start time of the candidate highlight segment with an earlier time in the video to be pushed and the end time of the candidate highlight segment with a later time in the video to be pushed; The video segment between the start time and the end time is used as a candidate highlight segment after merging.

8. A video push device, characterized in that: include: The acquisition module is used to obtain the video to be pushed; A first determination module is used to determine a reference video whose similarity detection result with the video to be pushed meets a preset condition, wherein the reference video includes a video whose heat information meets the heat condition; A second determination module is used to determine a target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, and push the video to be pushed based on the target highlight segment; The similarity detection result between the reference video and the video to be pushed includes time information of video segments in the video to be pushed that are similar to the reference video, and the historical playback information includes the number of historical playbacks and the number of historical completions; The second determination module, when determining the target highlight segment of the video to be pushed based on the historical playback information of the reference video and the similarity detection result between the reference video and the video to be pushed, is further used to: For any video moment of the video to be pushed, determine a reference video corresponding to the video moment, wherein the reference video corresponding to the video moment is a reference video whose corresponding time information includes the video moment; Determine a first predicted completion rate corresponding to the video moment based on the historical number of playbacks and the historical number of completions of the reference video corresponding to the video moment; Based on the first predicted completion rate corresponding to each video moment, a target highlight segment of the video to be pushed is determined.

9. A computer device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the video push method according to any one of claims 1 to 7 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the video pushing method according to any one of claims 1 to 7 are executed.

Images