Method, device, electronic device and storage medium for determining next video

Abstract

The present disclosure relates to a method and device for determining a next video, an electronic device and a storage medium. The method comprises: obtaining a multi-modal feature of a tail video segment of a current video, the multi-modal feature being extracted from a plurality of video feature dimensions, the plurality of video feature dimensions comprising at least two of: a visual feature dimension, a text feature dimension, and an audio feature dimension; and determining a next continuous video of the current video according to a comparison result of the multi-modal feature of the tail video segment and multi-modal features of a plurality of candidate videos. The present disclosure can characterize and compare videos from a plurality of video feature dimensions by comparing multi-modal features, which has high accuracy. The multi-modal feature of the tail video segment is compared to ensure that the determined next continuous video is a video that is immediately adjacent to the tail video segment of the current video, thereby solving the problem of needing to view multiple videos to determine the next continuous video of the current video.

Description

Technical Field

[0001] This disclosure relates to video processing technology, and more particularly to a method, apparatus, electronic device, and storage medium for determining the next video. Background Technology

[0002] On short video platforms, a long video is usually cut into multiple different short videos. In order to watch the entire long video to which the current short video belongs, users often need to watch multiple videos that do not belong to the long video to determine the next consecutive video. Moreover, there is often overlap between short videos belonging to the same long video, resulting in a poor continuous viewing experience for users and a bad viewing experience.

[0003] Therefore, in order to improve the user viewing experience, there is an urgent need for a method to determine the next consecutive video in the video the user is currently watching. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides a method, apparatus, electronic device, and storage medium for determining the next video. The technical solution of this disclosure is as follows:

[0005] According to a first aspect of the present disclosure, a method for determining the next video is provided, comprising:

[0006] Obtain the multimodal features of the end video segment of the current video. The multimodal features are extracted from multiple video feature dimensions, including at least two of the following: visual feature dimension, text feature dimension, and audio feature dimension.

[0007] Based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos, the next consecutive video of the current video is determined.

[0008] Optionally, the method further includes:

[0009] Obtain the description information of each of the multiple candidate videos;

[0010] Based on the description information of each of the multiple candidate videos, the multiple candidate videos are clustered to obtain multiple video clusters;

[0011] Obtain the multimodal features of each candidate video included in each of the multiple video clusters, and store the multimodal features of each candidate video included in each of the multiple video clusters in the video feature library of each of the multiple video clusters.

[0012] Based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos, the next consecutive video of the current video is determined, including:

[0013] Obtain the video cluster to which the current video belongs;

[0014] Based on the multimodal features of the tail video segment and the comparison results with multiple candidate videos in the video feature library of the video cluster to which the current video belongs, the next consecutive video of the current video is determined.

[0015] Optionally, the multimodal features of each candidate video included in each of the plurality of video clusters are obtained, and the multimodal features of each candidate video included in each of the plurality of video clusters are stored in the video feature library of each of the plurality of video clusters, including:

[0016] Each candidate video in each of the multiple video clusters is cut into multiple video segments of a preset duration;

[0017] Extract multiple video frames from each video segment;

[0018] Obtain the multimodal features of each video frame;

[0019] The multimodal features of multiple video frames belonging to the same video segment are concatenated to obtain the multimodal features of that video segment.

[0020] The multimodal features of multiple video segments of each candidate video included in each of the multiple video clusters are stored in the video feature library of each of the multiple video clusters.

[0021] Optionally, after dividing each candidate video in each of the plurality of video clusters into multiple video segments of preset duration, the method further includes:

[0022] Obtain the video identifier of each candidate video from the plurality of candidate videos;

[0023] Store the correspondence between the multimodal features of each video segment and the video identifier of its corresponding video;

[0024] Based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos, the next consecutive video of the current video is determined, including:

[0025] Based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple video segments in the video feature library, the multimodal features of the target video segment are determined.

[0026] Based on the correspondence, query the target video identifier corresponding to the multimodal features of the target video segment;

[0027] The target video corresponding to the target video identifier is determined as the next consecutive video of the current video.

[0028] Optionally, the method further includes:

[0029] Determine the next consecutive video of each video in each of the multiple video clusters, and record the temporal relationship of multiple video pairs, where each video pair includes a video and its next consecutive video;

[0030] Based on the temporal relationship of the multiple video pairs, a header video is determined, wherein the header video is not the next consecutive video of any video;

[0031] Based on the temporal relationship of the multiple video pairs, the video identifier of the header video and the identifiers of multiple subsequent videos of the header video are sequentially concatenated to generate the video chain corresponding to the header video.

[0032] Optionally, if there are multiple video chains corresponding to the header video, the method further includes:

[0033] The quality of each of the multiple video chains corresponding to the header video is determined, and the quality includes at least one of the following: video clarity, multimodal feature similarity, similarity of video description information, video chain duration, and manual review score;

[0034] Based on the quality of each of the multiple video chains corresponding to the header video, the multiple video chains corresponding to the header video are filtered to obtain the filtered video chains.

[0035] Optionally, the method further includes:

[0036] Detect the video currently playing on the user's terminal;

[0037] Obtain the video chain containing the currently playing video, and then obtain the next consecutive video of the currently playing video;

[0038] When it is detected that the currently playing video has finished playing, the next consecutive video of the currently playing video is pushed to the user terminal.

[0039] Optionally, the method further includes:

[0040] The start time of the target video segment is determined as the start time of the next consecutive video of the currently playing video;

[0041] When it is detected that the currently playing video has finished playing, the next consecutive video of the currently playing video is pushed to the user terminal, including:

[0042] When it is detected that the currently playing video has finished playing, a playback instruction for the next consecutive video of the currently playing video is generated, so as to instruct the user terminal to play the next consecutive video of the currently playing video from the start playback time;

[0043] The next consecutive video of the currently playing video and the playback instruction are pushed to the user terminal.

[0044] Optionally, based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos, the next consecutive video of the current video is determined, including:

[0045] The multimodal features of the tail video segment are compared with the multimodal features of each of the multiple candidate videos for similarity.

[0046] The target video whose similarity to the multimodal features of the tail video segment exceeds a preset threshold is determined as the next consecutive video of the current video.

[0047] According to a second aspect of the present disclosure, an apparatus for determining a next video is provided, comprising:

[0048] The feature acquisition module is configured to acquire multimodal features of the end video segment of the current video. The multimodal features are extracted from multiple video feature dimensions, which include at least two of the following: visual feature dimension, text feature dimension, and audio feature dimension.

[0049] The comparison and determination module is configured to determine the next consecutive video of the current video based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos.

[0050] Optionally, the device further includes:

[0051] The description information acquisition module is configured to acquire description information for each of the multiple candidate videos.

[0052] The clustering module is configured to cluster the multiple candidate videos according to their respective description information to obtain multiple video clusters;

[0053] The storage module is configured to acquire the multimodal features of each candidate video included in each of the plurality of video clusters, and store the multimodal features of each candidate video included in each of the plurality of video clusters in the video feature library of each of the plurality of video clusters.

[0054] The comparison determination module includes:

[0055] The video cluster unit is configured to obtain the video cluster to which the current video belongs;

[0056] The comparison determination unit is configured to determine the next consecutive video of the current video based on the multimodal features of the tail video segment and the comparison results of multiple candidate videos in the video feature library of the video cluster to which the current video belongs.

[0057] Optionally, the storage module includes:

[0058] The cutting unit is configured to cut each candidate video included in each of the plurality of video clusters into a plurality of video segments of a preset duration;

[0059] The extraction unit is configured to extract multiple video frames from each of the video segments;

[0060] A video frame unit is configured to acquire multimodal features of each video frame;

[0061] The splicing unit is configured to splice the multimodal features of multiple video frames belonging to the same video segment to obtain the multimodal features of that video segment.

[0062] The video segment feature storage unit is configured to store the multimodal features of multiple video segments of each candidate video included in each of the multiple video clusters in the respective video feature library of the multiple video clusters.

[0063] Optionally, the device further includes:

[0064] The video identifier acquisition module is configured to acquire the video identifier of each of the plurality of candidate videos;

[0065] The correspondence storage module is configured to store the correspondence between the multimodal features of each video segment and the video identifier of the video to which it belongs;

[0066] The comparison determination module includes:

[0067] The target video segment determination unit is configured to determine the multimodal features of the target video segment based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple video segments in the video feature library;

[0068] The query unit is configured to query the target video identifier corresponding to the multimodal features of the target video segment based on the correspondence relationship;

[0069] The determining unit is configured to determine the target video corresponding to the target video identifier as the next consecutive video of the current video.

[0070] Optionally, the device further includes:

[0071] The timing module is configured to determine the next consecutive video of each video in each of the plurality of video clusters, and to record the timing relationship of multiple video pairs, each video pair including a video and its next consecutive video;

[0072] The header video determination module is configured to determine the header video based on the temporal relationship of the multiple video pairs, wherein the header video is not the next consecutive video of any video;

[0073] The concatenation module is configured to concatenate the video identifier of the header video and the identifiers of multiple subsequent videos of the header video in sequence according to the temporal relationship of the multiple video pairs, thereby generating a video chain corresponding to the header video.

[0074] Optionally, the device further includes:

[0075] The quality determination module is configured to determine the quality of each of the multiple video chains corresponding to the header video, wherein the quality includes at least one of the following: video clarity, multimodal feature similarity, similarity of video description information, video chain duration, and manual review score;

[0076] The filtering module is configured to filter the multiple video chains corresponding to the header video based on their respective quality, and obtain the filtered video chains.

[0077] Optionally, the device further includes:

[0078] The detection module is configured to detect the video currently playing on the user terminal;

[0079] The video chain acquisition module is configured to acquire a video chain containing the currently playing video, and then acquire the next consecutive video of the currently playing video;

[0080] The push module is configured to push the next consecutive video of the currently playing video to the user terminal when it is detected that the currently playing video has finished playing.

[0081] Optionally, the device further includes:

[0082] The time determination module is configured to determine the start playback time of the target video segment as the start playback time of the next consecutive video of the currently playing video;

[0083] The push module includes:

[0084] The instruction generation unit is configured to generate a playback instruction for the next consecutive video of the currently playing video when it is detected that the currently playing video has finished playing, so as to instruct the user terminal to play the next consecutive video of the currently playing video from the start playback time;

[0085] The push unit is configured to push the next consecutive video of the currently playing video and the playback instruction to the user terminal.

[0086] Optionally, the comparison determination module includes:

[0087] The similarity comparison unit is configured to compare the multimodal features of the tail video segment with the multimodal features of each of the multiple candidate videos;

[0088] The video determination unit is configured to determine the target video whose similarity to the multimodal features of the tail video segment exceeds a preset threshold as the next consecutive video of the current video.

[0089] According to a third aspect of the present disclosure, an electronic device is provided, comprising:

[0090] processor;

[0091] Memory used to store the processor's executable instructions;

[0092] The processor is configured to execute the instructions to implement the method for determining the next video as described in the first aspect.

[0093] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided that, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the method for determining the next video as described in the first aspect.

[0094] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program or computer instructions, which, when executed by a processor, implement the method for determining the next video as described in the first aspect.

[0095] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: Based on the comparison results of the multimodal features of the end video segment of the current video and the multimodal features of multiple candidate videos, the next consecutive video of the current video can be determined. Specifically, comparing multimodal features allows for characterization and comparison of the video from multiple video feature dimensions, resulting in high accuracy. Since the multimodal features of the multiple candidate videos are compared with the multimodal features of the end video segment of the current video, it ensures that the determined next consecutive video is the video immediately adjacent to the end video segment of the current video, thus solving the problem of needing to view multiple videos to determine the next consecutive video of the current video.

[0096] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0097] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0098] Figure 1 This is a flowchart illustrating a method for determining the next video according to an exemplary embodiment;

[0099] Figure 2 This is an example diagram of a video chain in this disclosure;

[0100] Figure 3 This is a block diagram illustrating an apparatus for determining the next video according to an exemplary embodiment;

[0101] Figure 4 This is a block diagram illustrating an apparatus for determining the next video according to an exemplary embodiment. Detailed Implementation

[0102] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0103] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0104] It is understood that this disclosure applies not only to short video playback platforms, but also to other platforms that can play videos.

[0105] Figure 1 This is a flowchart illustrating a method for determining the next video according to an exemplary embodiment, such as... Figure 1 As shown, this method for determining the next continuous video can be used in electronic devices such as computers, mobile phones, and tablets, and includes the following steps.

[0106] In step S11, the multimodal features of the end video segment of the current video are obtained. The multimodal features are extracted from multiple video feature dimensions, including at least two of the following: visual feature dimension, text feature dimension, and audio feature dimension.

[0107] The end segment of the current video refers to the portion of the current video that is counted down to its preset countdown time. For example, if the current video is 30 seconds long and the preset countdown time is 5 seconds, then the end segment of the current video refers to the portion of the video from the 25th second to the 30th second.

[0108] Considering that relying solely on visual signals such as pixel statistics or deep learning intermediate features to determine the next consecutive video is inaccurate when video quality varies or is noisy, the inventors conceived of using multimodal features of video to determine the next consecutive video. These multimodal features are extracted from multiple video feature dimensions, and are obtained by concatenating these multimodal features. These multiple video feature dimensions include at least two of the following: visual feature dimensions, text feature dimensions, and audio feature dimensions.

[0109] In the visual feature dimension, a video feature extractor can be used to obtain the features output from the intermediate layers of the deep learning model as visual features. In the text feature dimension, audio and video text can be obtained. First, an ASR (Automatic Speech Recognition) module is used to recognize the audio text in the video, and an OCR (Optical Character Recognition) module is used to recognize the video text. The text is then bound to a time scale, and a text feature extractor is used to extract vector representations of the audio and video text. In the audio feature dimension, an audio feature extractor can be used to extract the vector representation of the audio.

[0110] The current video segment at the end is divided into 1-second wide segments with a 100-millisecond step, sliding backwards. Multiple video frames are extracted from each segment, and features are extracted from multiple video feature dimensions for each frame. These extracted features are then concatenated to form the multimodal features of that 1-second video segment. It's important to understand that the 1-second and 100-millisecond intervals are optional implementations; other durations can be selected in practice.

[0111] In step S12, the next consecutive video of the current video is determined based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos.

[0112] To determine the next consecutive video from the current video, the multimodal features of the last video segment of the current video are compared with the multimodal features of each of the multiple candidate videos. Thus, the next consecutive video from the current video can be determined based on the comparison results between the multimodal features of any of the candidate videos and the multimodal features of the last video segment of the current video.

[0113] Based on the above technical solution, the next consecutive video of the current video is determined according to the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple candidate videos, including: comparing the similarity between the multimodal features of the tail video segment and the multimodal features of each of the multiple candidate videos; and determining the target video whose similarity with the multimodal features of the tail video segment exceeds a preset threshold as the next consecutive video of the current video.

[0114] The comparison result can be determined by comparing the similarity of the multimodal features of each 1-second segment of the current video with the multimodal features of multiple videos, or by concatenating the multimodal features of each 1-second segment of the current video to obtain the multimodal features of the current video segment, and then directly using the multimodal features of the current video segment for similarity comparison.

[0115] By utilizing the retrieval capabilities of ANN (Approximate Nearest Neighbor), multiple videos with similar multimodal features to the multimodal features of the tail video segment of the current video can be retrieved simultaneously.

[0116] Multimodal features are represented by vectors. The similarity between the target video and the end video segment of the current video can be obtained by calculating the cosine distance between their multimodal features.

[0117] When the similarity between the multimodal features of the target video and the multimodal features of the end video segment of the current video exceeds a preset threshold, the target video is determined as the next consecutive video of the current video. The next consecutive video is a video that follows the current video. For example, if a long video A is segmented into three short videos A1, A2, and A3, where A1 + A2 + A3 = A, then A2 is the next consecutive video of A1, A3 is the next consecutive video of A2, and both A2 and A3 are subsequent videos of A1. It is understandable that sometimes the segmented short videos cannot be accurately aligned; there may be gaps or overlaps between two consecutive videos. A gap refers to a blank space between two videos. For example, if there is a gap between A1 and A2, then A1 + A2 < A - A3.

[0118] The method for determining the next video provided in this exemplary embodiment compares the multimodal features of the end video segment of the current video with the multimodal features of multiple videos for similarity. The target video whose similarity to the multimodal features of the end video segment exceeds a preset threshold is determined as the next consecutive video of the current video. The comparison of multimodal feature similarity allows for characterization and comparison of videos from multiple video feature dimensions, resulting in high accuracy. Since the similarity between the multimodal features of the target video and the multimodal features of the end video segment of the current video exceeds the preset threshold, the target video is similar to the end video segment of the current video in multiple video feature dimensions. Therefore, the target video can be considered the video immediately adjacent to the end video segment of the current video and can be determined as the next consecutive video of the current video. This solves the problem of needing to view multiple videos to determine the next consecutive video of the current video.

[0119] Based on the above technical solution, the method for determining the next video further includes: obtaining description information of each of the multiple candidate videos; clustering the multiple candidate videos according to their description information to obtain multiple video clusters; obtaining multimodal features of each candidate video included in each of the multiple video clusters, and storing the multimodal features of each candidate video included in each of the multiple video clusters in the video feature library of each of the multiple video clusters; determining the next consecutive video of the current video based on the comparison result between the multimodal features of the tail video segment and the multimodal features of the multiple candidate videos, including: obtaining the video cluster to which the current video belongs; determining the next consecutive video of the current video based on the comparison result between the multimodal features of the tail video segment and the video feature library of the video cluster to which the current video belongs.

[0120] Video playback platforms or servers store a vast number of videos. Directly comparing the similarity of the multimodal features of the last video segment with the multimodal features of each individual video would be computationally intensive and resource-intensive. Considering that the descriptive information of the current video and its next consecutive video is usually highly correlated, multiple candidate videos can be clustered using the video's descriptive information, including video title, video tags, and video text. Optionally, video clustering can also be performed using visual content or a combination of different information. Thus, candidate videos belonging to the same video cluster are more likely to be the current video and the next consecutive video than candidate videos not belonging to the same cluster.

[0121] The comparison focuses on the similarity between multimodal features. Therefore, a corresponding video feature library can be constructed for each video cluster in multiple video clusters, and the multimodal features of each candidate video in each video cluster can be stored in the corresponding video feature library.

[0122] Thus, by comparing the multimodal features of the tail video segment with the multimodal features of each of the multiple candidate videos, the video cluster to which the current video belongs can be obtained. Then, the multimodal features of the tail video segment are compared with the multimodal features of each of the multiple candidate videos in the video feature library of the video cluster to which the current video belongs. Specifically, if the current video is already in a video cluster, the video cluster to which the current video belongs is directly obtained; if the current video is a newly obtained video that has not yet been clustered into any video cluster, it is first clustered into a video cluster based on the description information of the current video.

[0123] The method for determining the next video provided in this exemplary embodiment can first cluster the videos using the descriptive information of the candidate videos to obtain multiple video clusters, and then construct a video feature library for each video cluster, storing the multimodal features of each candidate video in the corresponding video feature library. Thus, when comparing the similarity of the multimodal features of the tail video segment with the multimodal features of each of the multiple candidate videos, it is only necessary to compare the similarity of the multimodal features of each of the multiple candidate videos in the video feature library corresponding to the current video with the multimodal features of the tail video segment, thereby saving computational resources and improving work efficiency.

[0124] Based on the above technical solution, the multimodal features of each candidate video included in each of the multiple video clusters are obtained, and the multimodal features of each candidate video included in each of the multiple video clusters are stored in the video feature library of each of the multiple video clusters. This includes: cutting each candidate video included in each of the multiple video clusters into multiple video segments of preset duration; extracting multiple video frames from each video segment; obtaining the multimodal features of each video frame; splicing the multimodal features of multiple video frames belonging to the same video segment to obtain the multimodal features of that video segment; and storing the multimodal features of multiple video segments of each candidate video included in each of the multiple video clusters in the video feature library of each of the multiple video clusters.

[0125] To improve accuracy, the multimodal features of a video can be characterized using multimodal features from multiple video segments. Each candidate video within a video cluster is then divided into multiple video segments of a preset duration; for example, each candidate video can be divided into multiple 1-second video segments.

[0126] Multiple video frames are extracted from each video segment, and the multimodal features of each video frame are obtained. The method for obtaining the multimodal features of each video frame can be referred to the aforementioned method, and will not be repeated here. The multimodal features of multiple video frames belonging to the same video segment are concatenated to obtain the multimodal features of that video segment.

[0127] The multimodal features of multiple video segments of each candidate video included in each of the multiple video clusters are stored in the video feature library of each of the multiple video clusters.

[0128] Therefore, compared to comparing the similarity between the multimodal features of multiple candidate videos and the multimodal features of the end video segment of the current video, comparing the similarity between the multimodal features of video segments of multiple candidate videos and the multimodal features of the end video segment of the previous video can identify the video with better alignment with the end video of the current video.

[0129] Optionally, to further improve the alignment with the current video, the multimodal features of each 1-second segment of the current video can be compared with the multimodal features of video segments from multiple candidate videos to determine the target video.

[0130] Based on the above technical solution, after dividing each candidate video in each of the multiple video clusters into multiple video segments of preset duration, the method further includes: obtaining the video identifier of each candidate video in the multiple candidate videos; storing the correspondence between the multimodal features of each video segment and the video identifier of its corresponding video; determining the next consecutive video of the current video based on the comparison result between the multimodal features of the tail video segment and the multimodal features of the multiple candidate videos, including: determining the multimodal features of the target video segment based on the comparison result between the multimodal features of the tail video segment and the multimodal features of multiple video segments in the video feature library; querying the target video identifier corresponding to the multimodal features of the target video segment based on the correspondence; and determining the target video corresponding to the target video identifier as the next consecutive video of the current video.

[0131] After dividing each video in each video cluster into multiple video segments of a preset duration, the video identifier of each video in the multiple videos can be obtained, and the correspondence between the multimodal features of each video segment and the video identifier of the video to which it belongs can be stored.

[0132] Determining the next consecutive video of the current video based on the comparison results of the multimodal features of the tail video segment and the multimodal features of multiple candidate videos may include: determining the multimodal features of the target video segment based on the comparison results of the multimodal features of the tail video segment and the multimodal features of multiple video segments in the video feature library; then querying the target video identifier corresponding to the multimodal features of the target video segment in the correspondence relationship; and determining the target video corresponding to the target video identifier as the next consecutive video of the current video.

[0133] The method for determining the next video provided in this exemplary embodiment can store the correspondence between multimodal features and their corresponding video identifiers in a video feature library. Since the calculation focuses on the similarity between multimodal features, it first identifies multimodal features with a similarity exceeding a preset threshold, and then determines the video corresponding to that multimodal feature. Therefore, when determining the multimodal features of a target video segment, the video identifier of the target video can be quickly determined directly based on the correspondence. This improves work efficiency.

[0134] Based on the above technical solution, the method for determining the next video further includes: determining the next consecutive video of each video included in each of the multiple video clusters, and recording the temporal relationship of multiple video pairs, where each video pair includes one video and its next consecutive video; determining the header video according to the temporal relationship of the multiple video pairs, wherein the header video is not the next consecutive video of any video; and sequentially concatenating the video identifier of the header video and the identifiers of multiple subsequent videos of the header video according to the temporal relationship of the multiple video pairs to generate a video chain corresponding to the header video. It is understood that videos belonging to the same video chain are not necessarily cut from the same video; they may belong to different videos, but there is continuity between two connected videos. For example, the first and second episodes of a TV series may be two independent videos, but they are continuous, so the second episode of the TV series can be considered the next consecutive video of the first episode.

[0135] Taking each video in a video cluster as the current video, determine the next consecutive video of the current video, and record the temporal relationship of multiple video pairs. A video pair includes a video and its next consecutive video. For example, when determining that the multiple next views of the current video B are C1 and C2, the temporal relationship can be recorded as B→C1 and the temporal relationship B→C2.

[0136] Based on the temporal relationship of each video pair, the header video can be determined. The header video is not the next consecutive video after any other video. After determining the header video, the temporal relationship of each video pair can be queried. The video identifier of the header video and the identifiers of multiple subsequent videos of that header video are concatenated sequentially to generate the video chain corresponding to the header video. Subsequent videos of the header video refer to videos that have a direct or indirect temporal relationship with the header video and belong to the video chain corresponding to the header video.

[0137] For example, given the temporal relationships A→B, B→C1, B→C2, C1→D1, C1→D2, C2→D1, the header video can be identified as A based on these temporal relationships. The video chain corresponding to header video A is as follows: Figure 2 As shown, it contains three video chains: A→B→C1→D1, A→B→C1→D2, and A→B→C2→D1.

[0138] Optionally, each video in multiple video chains corresponding to the same header video can be identified as a video in the same video set.

[0139] Based on the above technical solution, when there are multiple video chains corresponding to the header video, the method for determining the next consecutive video further includes: determining the quality of each of the multiple video chains corresponding to the header video, wherein the quality includes at least one of the following: video clarity, multimodal feature similarity, similarity of video description information, video chain duration, and manual review score; and filtering the multiple video chains corresponding to the header video based on their respective quality to obtain the filtered video chains.

[0140] The quality of multiple video chains corresponding to the same header video may vary. Therefore, it is necessary to filter the multiple video chains corresponding to the header video to remove the video chains with poor quality and retain the video chains with better quality, so as to obtain the filtered video chains.

[0141] The quality of a video chain can include at least one of the following: video clarity, multimodal feature similarity, similarity of video description information, video chain duration, and manual review score. The average video clarity of each video in the same video chain is used as the video clarity of that video chain; the average multimodal similarity between any two connected videos in the same video chain is used as the multimodal feature similarity of that video chain; the video chain duration can be the total duration of all videos in the same video chain and / or the average duration of all videos. The video chain duration can be used to determine whether there are gaps or repetitions between each video in the video chain and the next consecutive video.

[0142] After identifying multiple high-quality video chains based on video clarity, multimodal feature similarity, video chain duration, and video description information similarity, these chains can be manually reviewed to remove those without a beginning or end, ultimately retaining the highest-quality videos. This process allows for filtering of multiple video chains corresponding to the same header video, resulting in a selected set of high-quality video chains.

[0143] Based on the above technical solution, the method for determining the next video further includes: detecting the video currently being played on the user terminal; obtaining a video chain containing the currently played video, and then obtaining the next consecutive video of the currently played video; and when it is detected that the currently played video has finished playing, pushing the next consecutive video of the currently played video to the user terminal.

[0144] The system detects the video currently playing on the user terminal, obtains the video chain containing the currently playing video, and determines the next consecutive video based on the video chain containing the currently playing video. When the currently playing video is detected to have finished playing, the next consecutive video of that video is pushed to the user terminal. If there are multiple video chains containing the currently playing video, the system can select the next consecutive video of the video chain with the highest quality to push to the user terminal. If the user terminal played the previous video in a video chain containing the currently playing video before playing the current video, one or more video chains can be determined based on the currently playing video and its previous video. Therefore, when pushing the next consecutive video, it only needs to determine the highest quality video chain from these one or more video chains and then push the next consecutive video of the currently playing video in that highest quality video chain to the user terminal.

[0145] The method for determining the next video provided in this exemplary embodiment can automatically push the next consecutive video of the currently playing video to the user terminal after the currently playing video has finished playing. In this way, the user does not need to watch multiple videos before determining the next consecutive video, so that they can watch each video in the entire video chain smoothly, which can effectively improve the user experience.

[0146] Based on the above technical solution, the method for determining the next video further includes: determining the start playback time of the target video segment as the start playback time of the next consecutive video of the currently playing video; when it is detected that the currently playing video has finished playing, pushing the next consecutive video of the currently playing video to the user terminal, including: when it is detected that the currently playing video has finished playing, generating a playback instruction for the next consecutive video of the currently playing video to instruct the user terminal to play the next consecutive video of the currently playing video from the start playback time; and pushing the next consecutive video of the currently playing video and the playback instruction to the user terminal.

[0147] When dividing each video in a video cluster into multiple video segments of a preset duration, the start playback time of each video segment is recorded. When storing the correspondence between the multimodal features of each video segment and the video identifier of the video to which it belongs, the correspondence between the multimodal features of each video segment and its start playback time is also stored.

[0148] The start playback time of the target video segment is determined as the start playback time of the next consecutive video of the currently playing video. Optionally, after identifying a target video segment whose multimodal features are more than a preset threshold similar to the end video segment of the current video, the start playback time of the target video segment is obtained from the stored correspondence; and when recording the temporal relationship of multiple video pairs, the start playback time of each next consecutive video is recorded; and when generating the video chain corresponding to the head video, the start playback time of each next consecutive video is recorded in the video chain.

[0149] When the currently playing video is detected to have finished playing, a playback instruction for the next consecutive video is generated. This playback instruction instructs the user terminal to play the next consecutive video from the start time of the current video's next consecutive video. The next consecutive video and the playback instruction are pushed to the user terminal. The user terminal, according to the playback instruction, plays the next consecutive video from the start time of the target video segment within the next consecutive video, automatically aligning the next consecutive video with the currently playing video. Video alignment means displaying the overlapping portion of the two videos only once, connecting the two videos. For example, if video A = A1 + A2 + A3, where A3 is the end segment of video A, and video A' = A3 + A4, then the aligned videos A and A' will play A1, A2, A3, and A4 sequentially, skipping the overlapping A3 segment once.

[0150] In this way, after watching the currently playing video, users can directly watch the next consecutive video and automatically skip the repeated parts. This automatically aligns the end of the currently playing video with the target video segment of the next consecutive video, solving the problem of users having to watch the same content repeatedly and having a poor experience.

[0151] Figure 3 This is a block diagram illustrating an apparatus for determining the next video according to an exemplary embodiment, with reference to... Figure 3 The device includes a feature acquisition module 31 and a comparison and determination module 32.

[0152] The feature acquisition module 31 is configured to acquire the multimodal features of the end video segment of the current video. The multimodal features are extracted from multiple video feature dimensions, including at least two of the following: visual feature dimension, text feature dimension, and audio feature dimension.

[0153] The comparison determination module 32 is configured to determine the next consecutive video of the current video based on the comparison results of the multimodal features of the tail video segment and the multimodal features of multiple candidate videos.

[0154] Optionally, the device further includes:

[0155] The description information acquisition module is configured to acquire description information for each of the multiple candidate videos.

[0156] The clustering module is configured to cluster the multiple candidate videos according to their respective description information to obtain multiple video clusters;

[0157] The storage module is configured to acquire the multimodal features of each candidate video included in each of the plurality of video clusters, and store the multimodal features of each candidate video included in each of the plurality of video clusters in the video feature library of each of the plurality of video clusters.

[0158] The comparison determination module 32 includes:

[0159] The video cluster unit is configured to obtain the video cluster to which the current video belongs;

[0160] The comparison determination unit is configured to determine the next consecutive video of the current video based on the multimodal features of the tail video segment and the comparison results of multiple candidate videos in the video feature library of the video cluster to which the current video belongs.

[0161] Optionally, the storage module includes:

[0162] The cutting unit is configured to cut each candidate video included in each of the plurality of video clusters into a plurality of video segments of a preset duration;

[0163] The extraction unit is configured to extract multiple video frames from each of the video segments;

[0164] A video frame unit is configured to acquire multimodal features of each video frame;

[0165] The splicing unit is configured to splice the multimodal features of multiple video frames belonging to the same video segment to obtain the multimodal features of that video segment.

[0166] The video segment feature storage unit is configured to store the multimodal features of multiple video segments of each candidate video included in each of the multiple video clusters in the respective video feature library of the multiple video clusters.

[0167] Optionally, the device further includes:

[0168] The video identifier acquisition module is configured to acquire the video identifier of each of the plurality of candidate videos;

[0169] The correspondence storage module is configured to store the correspondence between the multimodal features of each video segment and the video identifier of the video to which it belongs;

[0170] The comparison determination module 32 includes:

[0171] The target video segment determination unit is configured to determine the multimodal features of the target video segment based on the comparison results between the multimodal features of the tail video segment and the multimodal features of multiple video segments in the video feature library;

[0172] The query unit is configured to query the target video identifier corresponding to the multimodal features of the target video segment based on the correspondence relationship;

[0173] The determining unit is configured to determine the target video corresponding to the target video identifier as the next consecutive video of the current video.

[0174] Optionally, the device further includes:

[0175] The timing module is configured to determine the next consecutive video of each video in each of the plurality of video clusters, and to record the timing relationship of multiple video pairs, each video pair including a video and its next consecutive video;

[0176] The header video determination module is configured to determine the header video based on the temporal relationship of the multiple video pairs, wherein the header video is not the next consecutive video of any video;

[0177] The concatenation module is configured to concatenate the video identifier of the header video and the identifiers of multiple subsequent videos of the header video in sequence according to the temporal relationship of the multiple video pairs, thereby generating a video chain corresponding to the header video.

[0178] Optionally, the device further includes:

[0179] The quality determination module is configured to determine the quality of each of the multiple video chains corresponding to the header video, wherein the quality includes at least one of the following: video clarity, multimodal feature similarity, similarity of video description information, video chain duration, and manual review score;

[0180] The filtering module is configured to filter the multiple video chains corresponding to the header video based on their respective quality, and obtain the filtered video chains.

[0181] Optionally, the device further includes:

[0182] The detection module is configured to detect the video currently playing on the user terminal;

[0183] The video chain acquisition module is configured to acquire a video chain containing the currently playing video, and then acquire the next consecutive video of the currently playing video;

[0184] The push module is configured to push the next consecutive video of the currently playing video to the user terminal when it is detected that the currently playing video has finished playing.

[0185] Optionally, the device further includes:

[0186] The time determination module is configured to determine the start playback time of the target video segment as the start playback time of the next consecutive video of the currently playing video;

[0187] The push module includes:

[0188] The instruction generation unit is configured to generate a playback instruction for the next consecutive video of the currently playing video when it is detected that the currently playing video has finished playing, so as to instruct the user terminal to play the next consecutive video of the currently playing video from the start playback time;

[0189] The push unit is configured to push the next consecutive video of the currently playing video and the playback instruction to the user terminal.

[0190] Optionally, the comparison determination module 32 includes:

[0191] The similarity comparison unit is configured to compare the multimodal features of the tail video segment with the multimodal features of each of the multiple candidate videos;

[0192] The video determination unit is configured to determine the target video whose similarity to the multimodal features of the tail video segment exceeds a preset threshold as the next consecutive video of the current video.

[0193] The apparatus for determining the next consecutive video provided in this exemplary embodiment compares the multimodal features of the end video segment of the current video with the multimodal features of multiple videos for similarity. The target video whose similarity to the multimodal features of the end video segment exceeds a preset threshold is determined as the next consecutive video of the current video. The comparison of multimodal feature similarity allows for characterization and comparison of videos from multiple video feature dimensions, resulting in high accuracy. If the similarity between the multimodal features of the target video and the multimodal features of the end video segment of the current video exceeds the preset threshold, then the target video is similar to the end video segment of the current video in multiple video feature dimensions. Therefore, the target video can be considered the video immediately adjacent to the end video segment of the current video, and thus can be determined as the next consecutive video of the current video. This solves the problem of needing to view multiple videos to determine the next consecutive video of the current video.

[0194] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0195] Figure 4 This is a block diagram illustrating an apparatus 400 for determining the next video according to an exemplary embodiment. For example, apparatus 400 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0196] Reference Figure 4 The electronic device 400 may include one or more of the following components: processing component 402, memory 404, power component 406, multimedia component 408, audio component 410, input / output (I / O) interface 412, sensor component 414, and communication component 416.

[0197] Processing component 402 typically controls the overall operation of device 400, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 402 may include one or more processors 420 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 402 may include one or more modules to facilitate interaction between processing component 402 and other components. For example, processing component 402 may include a multimedia module to facilitate interaction between multimedia component 408 and processing component 402.

[0198] Memory 404 is configured to store various types of data to support the operation of device 400. Examples of this data include instructions for any application or method operating on device 400, contact data, phonebook data, messages, pictures, videos, etc. Memory 404 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0199] Power supply component 406 provides power to various components of device 400. Power supply component 406 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 400.

[0200] Multimedia component 408 includes a screen that provides an output interface between the device 400 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 408 includes a front-facing camera and / or a rear-facing camera. When the device 400 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0201] Audio component 410 is configured to output and / or input audio signals. For example, audio component 410 includes a microphone (MIC) configured to receive external audio signals when device 400 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 404 or transmitted via communication component 416. In some embodiments, audio component 410 also includes a speaker for outputting audio signals.

[0202] I / O interface 412 provides an interface between processing component 402 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0203] Sensor assembly 414 includes one or more sensors for providing status assessments of various aspects of device 400. For example, sensor assembly 414 may detect the on / off state of device 400, the relative positioning of components such as the display and keypad of device 400, changes in the position of device 400 or a component of device 400, the presence or absence of user contact with device 400, the orientation or acceleration / deceleration of device 400, and temperature changes of device 400. Sensor assembly 414 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 414 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 414 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0204] Communication component 416 is configured to facilitate wired or wireless communication between device 400 and other devices. Device 400 can access wireless networks based on communication standards, such as WiFi, carrier networks (such as 2G, 3G, 4G, or 5G), or combinations thereof. In one exemplary embodiment, communication component 416 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 416 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0205] In an exemplary embodiment, the electronic device 400 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the method described above for determining the next continuous video.

[0206] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 404 including instructions, which can be executed by a processor 420 of the device 400 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0207] In an exemplary embodiment, a computer program product is also provided, comprising readable program code executable by a processor 420 of an electronic device 400 to perform the method for determining the next video described above. Optionally, the program code may be stored in a storage medium of the electronic device 400, which may be a non-transitory computer-readable storage medium, such as a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0208] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0209] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for determining the next video, characterized in that, The method comprises: obtaining multi-modal features of a tail video segment of a current video, the multi-modal features being extracted from a plurality of video feature dimensions, the plurality of video feature dimensions comprising at least two of the following: a visual feature dimension, a text feature dimension, and an audio feature dimension; determining a next continuous video of the current video according to a comparison result of the multi-modal features of the tail video segment and multi-modal features of a plurality of candidate videos; The method further comprises: determining a next continuous video of each video included in each video cluster of a plurality of video clusters, and recording a time sequence relationship of a plurality of video pairs, each video pair comprising a video and its next continuous video, the plurality of video clusters being obtained by clustering the plurality of candidate videos using description information of the videos; determining a head video according to the time sequence relationship of the plurality of video pairs, the head video not being a next continuous video of any video; concatenating, in sequence, a video identifier of the head video and identifiers of a plurality of subsequent videos of the head video according to the time sequence relationship of the plurality of video pairs, to generate a video chain corresponding to the head video; in a case where the video chain corresponding to the head video is a plurality of video chains, the method further comprises: determining a quality of each of the plurality of video chains corresponding to the head video, the quality comprising at least one of the following: video definition, multi-modal feature similarity, video description information similarity, video chain length, and artificial review score; screening the plurality of video chains corresponding to the head video according to the quality of each of the plurality of video chains corresponding to the head video, to obtain a screened video chain.

2. The method of claim 1, wherein, The method further comprises: obtaining description information of each of the plurality of candidate videos; clustering the plurality of candidate videos according to the description information of each of the plurality of candidate videos, to obtain the plurality of video clusters; obtaining multi-modal features of each of the candidate videos included in each of the plurality of video clusters, and storing the multi-modal features of each of the candidate videos included in each of the plurality of video clusters in a video feature library of each of the plurality of video clusters; determining a next continuous video of the current video according to a comparison result of the multi-modal features of the tail video segment and multi-modal features of a plurality of candidate videos, comprises: obtaining a video cluster to which the current video belongs; determining a next continuous video of the current video according to a comparison result of the multi-modal features of the tail video segment and the description information of each of the plurality of candidate videos in a video feature library of the video cluster to which the current video belongs.

3. The method of claim 2, wherein, obtaining multi-modal features of each of the candidate videos included in each of the plurality of video clusters, and storing the multi-modal features of each of the candidate videos included in each of the plurality of video clusters in a video feature library of each of the plurality of video clusters, comprises: cutting each of the candidate videos included in each of the plurality of video clusters into a plurality of video segments of a preset length; extracting a plurality of video frames from each of the video segments; obtaining multi-modal features of each of the video frames; splicing the multi-modal features of a plurality of video frames belonging to a same video segment to obtain multi-modal features of the video segment; The multimodal features of the video segments of each candidate video included in each of the plurality of video clusters are stored in a video feature library of each of the plurality of video clusters.

4. The method of claim 3, wherein, After the plurality of video segments of each candidate video included in each of the plurality of video clusters are cut into video segments of a preset time length, the method further includes: obtaining a video identifier of each of the plurality of candidate videos; storing a correspondence between the multimodal features of each video segment and a video identifier of a video to which the video segment belongs; determining a next continuous video of the current video according to a comparison result of the multimodal features of the tail video segment and the multimodal features of the plurality of candidate videos, including: determining the multimodal features of the target video segment according to a comparison result of the multimodal features of the tail video segment and the multimodal features of the plurality of video segments in the video feature library; querying a target video identifier corresponding to the multimodal features of the target video segment according to the correspondence; determining a target video corresponding to the target video identifier as the next continuous video of the current video.

5. The method of claim 4, wherein, The method further includes: detecting a currently played video of a user terminal; obtaining a video chain containing the currently played video, and further obtaining a next continuous video of the currently played video; when it is detected that the currently played video is played to the end, pushing the next continuous video of the currently played video to the user terminal.

6. The method of claim 5, wherein, The method further includes: determining a start playing time of the target video segment as a start playing time of the next continuous video of the currently played video; when it is detected that the currently played video is played to the end, pushing the next continuous video of the currently played video to the user terminal, including: when it is detected that the currently played video is played to the end, generating a playing instruction of the next continuous video of the currently played video to instruct the user terminal to play the next continuous video of the currently played video from the start playing time; pushing the next continuous video of the currently played video and the playing instruction to the user terminal.

7. The method of claim 1, wherein, determining the next continuous video of the current video according to a comparison result of the multimodal features of the tail video segment and the multimodal features of the plurality of candidate videos, including: performing similarity comparison between the multimodal features of the tail video segment and the multimodal features of the plurality of candidate videos respectively; determining a target video with a similarity to the multimodal features of the tail video segment exceeding a preset threshold as the next continuous video of the current video.

8. An apparatus for determining a next video, the apparatus comprising: The apparatus includes: a feature acquisition module configured to obtain multimodal features of a tail video segment of a current video, the multimodal features being extracted from a plurality of video feature dimensions, the plurality of video feature dimensions including at least two of the following: a visual feature dimension, a text feature dimension, and an audio feature dimension; a comparison and determination module configured to determine a next continuous video of the current video according to a comparison result of the multimodal features of the tail video segment and multimodal features of a plurality of candidate videos; The apparatus further includes: a timing module configured to determine a next continuous video of each video included in each of the plurality of video clusters, and record a timing relationship of a plurality of video pairs, each of the plurality of video pairs including a video and its next continuous video; a head video determination module configured to determine a head video according to the timing relationship of the plurality of video pairs, the head video not being the next continuous video of any video; a concatenation module configured to concatenate, in sequence, a video identifier of the head video and identifiers of a plurality of subsequent videos of the head video according to the timing relationship of the plurality of video pairs, to generate a video chain corresponding to the head video; a quality determination module configured to determine a quality of each of a plurality of video chains corresponding to the head video, the quality including at least one of the following: video definition, multi-modal feature similarity, video description information similarity, video chain length, and artificial review score; a screening module configured to screen the plurality of video chains corresponding to the head video according to the quality of each of the plurality of video chains, to obtain screened video chains.

9. The apparatus of claim 8, wherein, The apparatus further includes: a description information acquisition module configured to acquire description information of each of a plurality of candidate videos; a clustering module configured to cluster the plurality of candidate videos according to the description information of each of the plurality of candidate videos, to obtain a plurality of video clusters; a storage module configured to acquire multi-modal features of each of the plurality of candidate videos included in each of the plurality of video clusters, and store the multi-modal features of each of the plurality of candidate videos included in each of the plurality of video clusters in a video feature library of each of the plurality of video clusters; The comparison determination module includes: a video cluster unit configured to acquire a video cluster to which the current video belongs; a comparison determination unit configured to determine a next continuous video of the current video according to a comparison result of the multi-modal features of the tail video segment and multi-modal features of each of a plurality of candidate videos in a video feature library of the video cluster to which the current video belongs.

10. The apparatus of claim 9, wherein, The storage module includes: a cutting unit configured to cut each of the plurality of candidate videos included in each of the plurality of video clusters into a plurality of video segments of a preset length; an extraction unit configured to extract a plurality of video frames from each of the video segments; a video frame unit configured to acquire multi-modal features of each of the video frames; a splicing unit configured to splice the multi-modal features of a plurality of video frames belonging to a same video segment, to obtain multi-modal features of the video segment; a video segment feature storage unit configured to store the multi-modal features of a plurality of video segments of each of the plurality of candidate videos included in each of the plurality of video clusters in a video feature library of each of the plurality of video clusters.

11. The apparatus of claim 10, wherein, The apparatus further includes: a video identifier acquisition module configured to acquire a video identifier of each of the plurality of candidate videos; a storage correspondence module configured to store a correspondence between the multi-modal features of each of the video segments and a video identifier of a video to which each of the video segments belongs; The comparison determination module includes: The target video segment determination unit is configured to determine the multimodal feature of the target video segment according to a comparison result of the multimodal feature of the tail video segment and multimodal features of a plurality of video segments in a video feature library. The query unit is configured to query a target video identifier corresponding to the multimodal feature of the target video segment according to the correspondence. The determination unit is configured to determine a target video corresponding to the target video identifier as a next continuous video of the current video.

12. The apparatus of claim 11, wherein, The apparatus further includes: The detection module is configured to detect a currently played video of a user terminal. The video chain acquisition module is configured to acquire a video chain containing the currently played video, and further acquire a next continuous video of the currently played video. The push module is configured to push the next continuous video of the currently played video to the user terminal when it is detected that the currently played video is played to the end.

13. The apparatus of claim 12, wherein, The apparatus further includes: The time determination module is configured to determine a start playing time of the target video segment as a start playing time of the next continuous video of the currently played video. The push module includes: The indication generation unit is configured to generate a playing indication of the next continuous video of the currently played video to instruct the user terminal to play the next continuous video of the currently played video from the start playing time when it is detected that the currently played video is played to the end. The push unit is configured to push the next continuous video of the currently played video and the playing indication to the user terminal.

14. The apparatus of claim 8, wherein, The comparison determination module includes: The similarity comparison unit is configured to perform similarity comparison between the multimodal feature of the tail video segment and respective multimodal features of a plurality of candidate videos. The video determination unit is configured to determine a target video with a similarity to the multimodal feature of the tail video segment exceeding a preset threshold as the next continuous video of the current video.

15. An electronic device, comprising: It includes: A processor; A memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method for determining a next video according to any one of claims 1 to 7.

16. A computer readable storage medium, when instructions in the computer readable storage medium are executed by a processor of an electronic device, enable the electronic device to perform the method for determining a next video according to any one of claims 1 to 7.

17. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the method for determining a next video according to any one of claims 1 to 7.

Images