Video content generation method and device, and storage medium

Abstract

The embodiment of the invention provides a video content generation method and device and a storage medium. In the embodiment of the invention, a video script is introduced as a guide for video content generation. And after the uploading operation of the video demander is responded, obtaining the video material and description information of the target object. The method comprises the steps of obtaining a plurality of to-be-processed video clips with semantic tags based on semantic segmentation of a video material; and determining the industry attribute of the target object based on the description information. And matching one or more target video scripts in combination with the semantic tags and the industry attributes, and combining and decorating the to-be-processed video clips according to the combination modes and the content decoration information of the sample video clips included in the target video scripts to generate a plurality of target video contents. In the whole generation process, the video content demander can generate the video content related to the target object in batches based on the video script only by completing the uploading operation, so that the video production threshold is reduced, and the video production efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a method, device and storage medium for generating video content. Background Art

[0002] With the rise of various short video content platforms, the traffic of the entire network is increasingly concentrated in the field of short video content. Some e-commerce platforms have also begun to actively create their own short video content, attracting user traffic and increasing transaction volume through short video content around products.

[0003] At present, short videos in the e-commerce field are produced by merchants around products. In the existing short video production solutions, merchants are required to manually participate in video editing and splicing. For a large number of merchants, it is obvious that the current way of producing short videos is not only time-consuming and labor-intensive, but also inefficient. Summary of the invention

[0004] Multiple aspects of the present application provide a video content generation method, apparatus, device and storage medium to lower the threshold of video production and improve video production efficiency.

[0005] An embodiment of the present application provides a video content generation method, comprising: responding to an upload operation of a video content demander, obtaining a video material containing a target object and description information of the target object; semantically segmenting the video material to obtain a plurality of to-be-processed video clips with semantic labels, and determining the industry attributes of the target object according to the description information of the target object; selecting at least one target video script from existing video scripts according to the semantic labels of the plurality of to-be-processed video clips and the industry attributes of the target object, each video script containing a plurality of sample video clips and a combination method and content decoration information of the plurality of sample video clips; combining and decorating the plurality of to-be-processed video clips according to the plurality of sample video clips contained in the at least one target video script and the combination method and content decoration information of the plurality of sample video clips, so as to obtain a plurality of target video contents.

[0006] The embodiment of the present application also provides a video content generation device, including: an acquisition module, which is used to respond to the upload operation of the video content demander, and obtain the video material containing the target object and the description information of the target object. A semantic segmentation module, which is used to semantically segment the video material to obtain multiple video clips to be processed with semantic tags. A determination module, which is used to determine the industry attributes of the target object according to the description information of the target object. A selection module, which is used to select at least one target video script from existing video scripts according to the semantic tags of the multiple video clips to be processed and the industry attributes of the target object, each video script containing multiple sample video clips and the combination method and content decoration information of the multiple sample video clips. A generation module, which is used to combine and decorate the multiple video clips to be processed according to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips, so as to obtain multiple target video contents.

[0007] An embodiment of the present application also provides a video content generating device, including: a memory and a processor; the memory is used to store a computer program, and the processor is coupled to the memory and is used to execute the computer program to implement the steps in the above method.

[0008] An embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor is enabled to implement the steps in the above-described method.

[0009] In an embodiment of the present application, after responding to the upload operation, the video material and description information of the target object are obtained. Based on the semantic segmentation of the video material, a plurality of video clips to be processed with semantic tags are obtained. Based on the description information, the industry attributes of the target object are determined. One or more target video scripts are matched in combination with the semantic tags and industry attributes, and the video clips to be processed are combined and decorated according to the combination method and content decoration information of the sample video clips included in the target video script to generate a plurality of target video contents. During the entire generation process, the video content demander only needs to complete the upload operation of the relevant information, and then the video content related to the target object can be automatically generated in batches based on the video script, thereby lowering the threshold for video production and improving the efficiency of video production. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0011] Figure 1 A flowchart of a method for generating video content provided by an exemplary embodiment of the present application;

[0012] Figure 2 This is a flowchart of a method for semantically segmenting video material based on a video segmentation model, another exemplary embodiment of the present application;

[0013] Figure 3 A structural schematic diagram of a video segmentation model is provided for another exemplary embodiment of the present application;

[0014] Figure 4 A schematic diagram of an architecture of a segmentation method including a video industry prediction model provided for another exemplary embodiment of the present application;

[0015] Figure 5 A schematic structural diagram of a video content generating device provided by another exemplary embodiment of the present application;

[0016] Figure 6 A schematic structural diagram of a video content generating device provided as yet another exemplary embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0018] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0019] In view of the technical problem of low efficiency in the existing short video production, in the embodiment of the present application, a "video script" is introduced as a guide for video content generation. In the embodiment of the present application, the concept of video script draws on the concept of script in the film and television industry. In the film and television industry, the script describes the storyline, character dialogue, shot description and transition, sound effects, etc. of the movie, providing a structured framework for film production. Accordingly, in the present embodiment, the "video script" describes how the video should be generated. In order to achieve this purpose, the video script includes sample video clips, and the video script can also present the combination method between the sample video clips. Further, the video script also includes content decoration information for the sample video clips. This embodiment does not make too many restrictions on the content decoration information, for example, it can include but is not limited to: Lottie (a lightweight JSON-based animation format), pictures, audio, transitions, videos, subtitles, stickers, flower characters, motion effects, etc.

[0020] Based on the video script, the technical solution provided by this embodiment includes: after responding to the upload operation, the video material and description information of the target object are obtained. Based on the semantic segmentation of the video material, a plurality of video clips to be processed with semantic tags are obtained. Based on the description information, the industry attributes of the target object are determined. One or more target video scripts are matched in combination with the semantic tags and industry attributes, and the video clips to be processed are combined and decorated according to the combination method and content decoration information of the sample video clips included in the target video script to generate multiple target video contents. During the entire generation process, the video content demander only needs to complete the upload operation of the relevant information, and then the video content related to the target object can be automatically generated in batches based on the video script, thereby lowering the threshold for video production and improving the efficiency of video production.

[0021] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0022] Figure 1 A schematic diagram of a video content generation method provided in an embodiment of the present application. Figure 1 This method is described exemplarily.

[0023] S1: responding to an upload operation of a video content demander, obtaining video material containing a target object and description information of the target object;

[0024] S2: semantically segmenting the video material to obtain a plurality of video clips to be processed with semantic labels, and determining the industry attribute of the target object according to the description information of the target object;

[0025] S3: selecting at least one target video script from existing video scripts according to the semantic labels of the multiple video clips to be processed and the industry attributes of the target object, each video script comprising multiple sample video clips and a combination method and content decoration information of the multiple sample video clips;

[0026] S4: According to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips, the multiple video clips to be processed are combined and decorated to obtain multiple target video contents.

[0027] In the embodiment of the present application, the video content demander can be any individual, enterprise or organization that needs to produce video content, such as various merchants in the field of e-commerce, anchors in short video platforms, content creators of various video platforms, etc. Correspondingly, the target object is the object that the video content demander wants the audience to understand, which can be a specific concept, event, product or service, such as a mobile phone, and the video can explain the basic introduction, design features, technical specifications, functional features, etc. of the mobile phone. It can also be a certain person, company or organization. Taking the target object as a certain task as an example, the video can explain the background information, career or major achievements, influence or contribution of the person. It can also be a certain place or region. Taking the target object as a certain city, the video can introduce the geographical location, history, main attractions, food characteristics, local customs, etc. of the city. Of course, it can also be other objects that can be introduced through videos, and the embodiment of the present application does not limit this.

[0028] In the embodiment of the present application, in order to help the video content demander improve the efficiency of video production. A relevant video mixing tool can be provided to the video demander, and any embodiment or any combination of embodiments provided in the embodiment of the present application can be implemented by running the tool through an electronic device. Optionally, the tool can exist in the form of an application (an independent application, or a functional module, applet, etc. in other applications). Or it can also exist in the form of a page such as the Web, HTML5 (HyperTextMarkup Language 5, a language description method for building Web content).

[0029] Among them, by running the video remix tool, a function of uploading materials required for video production can be provided to video content demanders. For example, a page can be provided with an upload control. The video content demander clicks the upload control, and an upload window pops up. Based on the upload window, the video material corresponding to the target object and the description information of the target object are uploaded. When performing the upload operation, the video content demander can upload the video material and description information of the target object. In this embodiment, the video material can be one or more segments, as long as the video material contains a common target object.

[0030] The description information of the target object at least includes the relevant introduction of the target object, such as the name, ID or details link of the target object, etc. The details link is the link address pointing to the details page of the target object; at least the industry attribute of the target object can be confirmed according to the description information. Optionally, the industry attribute can be represented by the category or industry name to which the target object belongs, and the category to which the target object belongs can include a primary category, a secondary category, etc. Taking the target object as a mobile phone as an example, the primary industry attribute to which the target object belongs can be a primary category, such as electronic products, and the secondary industry attribute can be a secondary category, such as a business mobile phone. In another example, the industry attribute can be represented by a specific field of the target object. Taking the target object as a software engineer as an example, the primary industry attribute to which the target object belongs can be a primary field, such as a technology practitioner, and the secondary industry attribute can be a secondary field, such as an Internet technology practitioner. Among them, the finer the category level used by the industry attribute to which the target object belongs, the more conducive it is to matching a more relevant video script for it, while improving the matching efficiency.

[0031] There are multiple ways to obtain the description information of the target object. Optionally, the video demander can directly upload text content or audio content related to the target object, and obtain the description information of the target object based on the text content or audio content. Optionally, some options can also be provided to the video demander, and these options can be displayed in the form of a graphical user interface, which at least includes the target object's logo, name, and detailed link.

[0032] In this embodiment, after obtaining the video material of the target object, the video material can be semantically segmented, that is, the video is divided into different segments according to its inherent semantic meaning, and multiple video segments to be processed with semantic tags are obtained. Each semantic tag corresponds to at least one video segment to be processed. In the case where a user uploads multiple video materials, each video material can be semantically segmented separately, or multiple video materials can be spliced, and then the spliced ​​video materials are semantically segmented, and there is no limitation on this. In an example, the video demander is a merchant, and the target object is uploaded as a video material of its main product, such as a mobile phone. The semantic tags that may be obtained include mobile phone overview, appearance design, hardware specifications, software functions, actual experience, price and purchase information, etc. Among them, the "mobile phone overview" label may correspond to the basic information of the mobile phone, including video clips such as brand, model, release date, etc., the "appearance design" label may correspond to the appearance design of the mobile phone, such as size, weight, color options, etc. video clips, and the "hardware specifications" label may correspond to a detailed introduction to the hardware configuration of the mobile phone, such as processor, RAM (Random Access Memory), storage space, battery capacity, etc. video clips. Other semantic tags may correspond to multiple video clips, or may correspond to only one video clip. For example, the "software function" tag may correspond to only a video clip introducing the face recognition function.

[0033] Further, at least one video script is selected from the existing videos according to the semantic tags of the multiple video clips to be processed and the industry attributes of the target object. In an optional embodiment, the video scripts are first filtered according to the industry attributes of the target object, so that the search scope can be narrowed, and then the scripts are filtered according to the semantic tags. After two screenings, at least one target video script is obtained. It should be noted that the execution order of the above-mentioned screening operations is not limited by this application.

[0034] In an embodiment of the present application, each video script includes multiple sample video clips, as well as a combination method and content decoration information of the multiple sample video clips. In this embodiment, based on a sample video clip in the target video script, at least one to-be-processed video clip with the same or similar semantic tag can be matched, and the multiple to-be-processed video clips matched with each sample video clip are combined and decorated according to the combination method and content decoration information of the multiple sample clips provided by the target video script to obtain multiple target video contents.

[0035] Optionally, the video script can be based on the JSON format. JSON (JavaScript Object Notation) is a lightweight data exchange format.

[0036] An embodiment of the present application provides a video script about a mobile phone introduction. The video script is exemplarily described below. It should be noted that the video script is only an example and does not constitute a limitation to the present application.

[0037] Optionally, the video script includes a title, which is used to introduce the main content of the video script. Secondly, the video script also includes three video segments.

[0038] The first segment A0, starting from 0 seconds ("start_time":0) and lasting to 20 seconds ("end_time":20), contains the semantic tag of Exterior Design. The video mainly shows the exterior design of the phone. In this part, the video will also add the "Fade In" animation to leave a deep first impression on the audience.

[0039] Optionally, video_path is the storage path of each sample video clip in a Linux system (a free and open source UNIX-like operating system). The storage path representation method may be different for different operating systems. Optionally, the sample video clips can also be stored in the cloud by replacing the local storage path with a cloud link.

[0040] The second segment B0, starting at 20 seconds ("start_time":20) and ending at 40 seconds ("start_time":40), contains the semantic tag of Hardware Specifications. This part of the video introduces the hardware specifications of the iPhone. In order to better show the details of the hardware, a dynamic effect of "Zoom" will be added to allow the audience to have a deeper understanding of the hardware structure and design.

[0041] The third segment C0, starting at 40 seconds ("start_time":20) and lasting to 70 seconds ("end_time":20), contains semantic tags for hardware specifications (Software Features). This part of the video introduces the software features of the phone. In order to highlight specific software functions or features, the dynamic effect of "Highlight" will be used.

[0042] In the above example, the video script includes three sample video clips, which are respectively recorded as sample video clips A0, B0 and C0. In one example, the to-be-processed clips matched by sample video clip A0 include A1 and A2, the to-be-processed clips matched by sample video clip B0 include B1 and B2, and the to-be-processed clips matched by sample video clip C0 include C1. Optionally, the combination of them can be presented in the form of (A1, B1, C1), (A2, B1, C1) (A1, B2, C1) (A2, B2, C1), any of which can form a target video content.

[0043] In an optional embodiment, the video script includes a plurality of "slots", which are respectively used to insert different sample video clips or access links of the sample video clips. The video script may also include the position, duration and order of the slots in the entire video script, which together form a combination of the plurality of sample video clips. The sample video clips are matched with the videos to be processed, and each sample video clip matches at least one video clip to be processed. Based on the matching results, the video clip to be processed is inserted into the "slot" of the corresponding sample video clip. After the insertion, the plurality of video clips to be processed are decorated based on the content decoration information of the plurality of sample video clips to generate a plurality of target video contents.

[0044] In this embodiment, a "video script" is introduced as a guide for video content generation. After responding to the upload operation, the video material and description information of the target object are obtained. Based on the semantic segmentation of the video material, multiple video clips to be processed with semantic tags are obtained. Based on the description information, the industry attributes of the target object are determined. One or more target video scripts are matched in combination with the semantic tags and industry attributes, and the video clips to be processed are combined and decorated according to the combination method and content decoration information of the sample video clips included in the target video script to generate multiple target video contents. During the entire generation process, the video content demander only needs to complete the upload operation to batch generate video content related to the target object based on the video script, thereby lowering the threshold for video production and improving video production efficiency.

[0045] Artificial Intelligence (AI) is a technology that uses digital computers to simulate human perception of the environment, acquire knowledge, and use knowledge. This technology can enable machines to have functions similar to human perception, reasoning, and decision-making. Artificial intelligence technology mainly includes computer vision technology, speech processing technology, natural language processing technology, machine learning, deep learning, and other major directions.

[0046] Machine learning is to predict unknown information by learning from existing data. Based on machine learning, the present application embodiment provides a method for semantically segmenting video material based on a video segmentation model. Figure 2 The flowchart of the method is shown in FIG. 1 and FIG. 2 is a flowchart of the method. The following is an exemplary description of how the video segmentation model performs semantic segmentation in conjunction with the accompanying drawings. The segmentation method of the video material can be referred to the following example, which is not limited thereto.

[0047] S21: inputting the video material into a pre-trained video segmentation model, wherein the video material is separated into speech and picture by using a pre-processing layer to obtain speech information and picture data with time information;

[0048] S22: inputting the voice information into a first semantic analysis layer, converting the voice information into text information, and performing semantic segmentation on the text information to obtain a plurality of first semantic labels and their corresponding relationships with time information;

[0049] S23: inputting the picture data into a second semantic analysis layer, dividing the picture data into a plurality of picture segments according to feature similarities between adjacent frames in the picture data, converting the plurality of picture segments into text information and performing semantic extraction on the text information to obtain a plurality of second semantic labels and their corresponding relationships with time information;

[0050] S24: Segment the video material according to the plurality of first semantic tags and their correspondence with time information and the plurality of second semantic tags and their correspondence with time information to obtain a plurality of to-be-processed video segments with semantic tags.

[0051] In the embodiments of the present application, Figure 3 As shown, it is a structural schematic diagram of a video segmentation model provided in an embodiment of the present application. In this embodiment, the pre-trained video segmentation model includes a preprocessing layer, a first semantic analysis layer, and a second semantic analysis layer. Among them, when pre-training the video segmentation model, in order to improve the performance of the model, it is optional to train by industry, classify the training data according to the industry, and train a separate model for each industry.

[0052] If the industry attribute of the video material is unknown, optionally, the video material is first classified using a pre-trained video industry prediction model, and then segmented using a corresponding video segmentation model. Figure 4 The schematic diagram of the architecture of a segmentation method including a video industry prediction model provided in an embodiment of the present application. Figure 4As shown, it includes multiple video segmentation models, such as video segmentation model 1, video segmentation model 2, ... video segmentation model n, each video segmentation model corresponds to a different industry attribute. In this embodiment, the video material is first classified using a pre-trained video industry prediction model, and then segmented using the corresponding video segmentation model. The steps of segmenting using the corresponding segmentation model are the same as those in the above embodiment and will not be repeated here. It should be noted that Figure 4 The middle dotted line indicates that each video segmentation model divided by industry attributes has the possibility of being input with video materials, but only one is selected each time.

[0053] In this embodiment, after responding to the upload operation of the video demander, the obtained video material is input into the pre-processing layer of the pre-trained video segmentation model, and the voice and picture of the video material are separated to obtain the voice information and picture data with time information, such as Figure 3 As shown. Time information is usually a timestamp in a media content (audio or video). A timestamp is a mark used to indicate the relative position of a specific audio sample or video frame in a media stream. In other words, the timestamp provides the positioning information of the media content on the timeline, and in this embodiment, the voice information and picture data separated from the video material have the same timeline. The main tasks of the preprocessing layer include: decomposing composite media files containing multiple data streams (such as audio streams, video streams, subtitle streams, etc.) such as MP4 (MPEG-4 Part 14, a standard digital multimedia container format), AVI (Audio Video Interleaved, a multimedia container format), etc. into individual data streams, and inputting the decomposed data streams into the semantic analysis layer. The data streams here mainly include two data streams: voice information and picture data.

[0054] On the one hand, the voice information is input into the first semantic analysis layer, in which the voice information is converted into text information, and the text information is semantically segmented to obtain a plurality of first semantic labels and their corresponding relationships with time information, such as Figure 3 As shown. In this embodiment, the first semantic tag is associated with the content of the voice information, and multiple first semantic tags together constitute a semantic description of the voice information. In an optional embodiment, the first semantic tag may be a keyword extracted from the voice information, and the keyword may be the most frequently appearing word, which can accurately capture the key points that the voice information wants to express. In another optional embodiment, the first semantic tag may also be a topic extracted from the voice information, and the topic is composed of key sentences extracted from the content of the voice information. In some other embodiments, "keywords" and "topics" can also be generated by AI to generate new keywords and new topics to summarize the text. Among them, each first semantic tag has time information corresponding to the voice information.

[0055] On the other hand, the picture data is input into the second semantic analysis layer, and in the second semantic analysis layer, the picture data is divided into a plurality of picture segments according to the feature similarity between adjacent frames in the picture data, such as Figure 3 In this embodiment, the picture data is first pre-processed, that is, the picture data is divided into frame sequences.

[0056] In an optional embodiment, an implementation method of dividing the picture data into multiple picture segments includes: calculating the similarity of adjacent frames contained in the picture data, dividing two or more consecutive frames whose similarity of adjacent frames is greater than a set threshold into the same picture segment, and then obtaining multiple picture segments. It should be noted that the threshold can be set according to the specific application.

[0057] Further, the multiple screen segments are respectively converted into text information and semantic extraction is performed on the text information to obtain multiple second semantic tags and their corresponding relationship with time information. In an optional embodiment, visual feature extraction is performed on each frame, and optionally, a pre-trained convolutional neural network (CNN) is used to extract visual features of each frame. The extracted features are input into the next sequence generation model, which is trained to generate text describing the content of the frame. Further, the generated text is semantically analyzed, optionally through natural language processing (NLP), and the semantic changes between continuous texts are detected. When a significant semantic change is detected, such as a change in the theme, it can be considered that this is a text segmentation point, and the text information is then segmented into multiple text segments with specific semantics. The text segments are segmented into independent words or phrases and represented by feature vectors. Based on the word frequency method, the most frequently appearing words are selected as labels, or the theme of each text information is identified and the theme is used as a label, or a label is generated based on an AI model. Among them, each second semantic label has time information corresponding to the screen data.

[0058] Further, according to the plurality of first semantic tags and their corresponding relationships with time information and the plurality of second semantic tags and their corresponding relationships with time information, the video material is segmented to obtain a plurality of to-be-processed video segments with semantic tags, such as Figure 3 In an optional embodiment, a plurality of semantic tags corresponding to the video material and their corresponding relationships with time information are generated according to the plurality of first semantic tags and their corresponding relationships with time information and the plurality of second semantic tags and their corresponding relationships with time information.

[0059] From the above, it can be seen that the first semantic tag corresponds to the time information of the voice information, and the second semantic tag corresponds to the time information of the picture data. When determining multiple semantic tags of the video material, if there is voice information, then the time information corresponding to the first semantic tag corresponding to the voice information and the time information corresponding to the second semantic tag corresponding to a certain picture data should be roughly synchronized. If the time information is not synchronized, it may cause the voice information and the picture data content to not match.

[0060] Therefore, further optionally, based on the degree of overlap between the time information corresponding to the multiple first semantic tags and the time information corresponding to the multiple second semantic tags, the first type of time information corresponding to both the first semantic tag and the second semantic tag and the second type of time information corresponding only to the second semantic tag are determined.

[0061] In an optional embodiment, the degree of overlap between the time information of the first semantic tag and the time information of the second semantic tag can be the length of the overlapping area on the time axis between the time information of the voice information corresponding to the first semantic tag and the picture data corresponding to the second semantic tag. Generally speaking, a video clip contains both voice and picture, so the first type of time information corresponds to this situation. The second type of time information corresponds to the situation where only the picture exists in the video clip.

[0062] For the first type of time information, a semantic tag corresponding to the first type of time information is generated according to the first semantic tag and the second semantic tag corresponding to the first type of time information. In an optional embodiment, the first semantic tag and the second semantic tag are the same or similar, and the two are merged to obtain a final semantic tag. In another optional embodiment, if the difference between the two is large, the first semantic tag can be selected as the final semantic tag. In some other embodiments, if the difference between the two is large, two semantic tags can be output for the video content demander to manually select the final semantic tag. Optionally, when the video demander uploads the prompt information, the prompt word option can be displayed to the video demander so that the video demander can select the relative importance of the voice and picture in the video material. If the video demander believes that the information contained in the video is more important, the second semantic tag can be used as the final semantic tag.

[0063] As mentioned above, the second type of time information only corresponds to the second semantic tag. Therefore, for the second type of time information, the second semantic tag corresponding to the second type of time information is used as the semantic tag corresponding to the second type of time information. In this embodiment, the second semantic tag corresponding to the second type of time information is used as the final semantic tag.

[0064] Furthermore, after obtaining multiple semantic tags corresponding to the video material, the video material is segmented according to the multiple semantic tags corresponding to the video material and their correspondence with time information to obtain multiple video segments to be processed, and the corresponding semantic tag is marked for each video segment to be processed.

[0065] In each embodiment of the present application, in response to the upload operation of the video demander, the description information of the target object is obtained at the same time. According to the description information of the target object, at least the industry attribute of the target object can be determined. Regarding the implementation method of determining the industry attribute of the target object according to the description information of the target object, please refer to the following example, which is not limited to this.

[0066] In an optional embodiment, the industry attribute of the target object is determined according to the description information of the target object, including: the description information of the target object is the identification information of the target object, and according to the identification information of the target object, the object information management system to which the target object belongs is queried to obtain the industry attribute of the target object. Optionally, the industry attribute includes at least one primary structure, for example, according to the identification information of the target object, it is determined that the mobile phone is an industry attribute, and further, it can be determined that the mobile phone belongs to the industry attribute of a business mobile phone according to the identification information.

[0067] Alternatively, the description information of the target object is an access link of the target object, and the details page of the target object is accessed according to the access link of the target object; the industry attribute of the target object is obtained from the details page of the target object. Generally speaking, the details page of the target object usually contains information such as the name, picture, description, price, and comments of the product. In an optional embodiment, web crawling is used to obtain detailed information of the target object from a web page. In an example, the target object is a product. When obtaining the industry attribute according to the details page of the product, most e-commerce websites will classify the products they sell. In an optional embodiment, the industry attribute of the product can be obtained through the classification information of the product. Optionally, keywords related to the industry attribute can also be obtained through the name and description of the product, and the industry attribute can be determined according to the keywords. Optionally, the URL (Uniform Resource Locator) of the picture can also be crawled from the web page, the picture can be downloaded, and the industry attribute of the product can be obtained using image recognition technology.

[0068] In each embodiment of the present application, after obtaining the semantic labels of multiple video clips to be processed and the industry attributes of the target object, at least one target video script is selected from the existing video scripts based on this. In an optional embodiment, the industry attributes corresponding to the existing video script are determined according to the multiple sample video clips contained in the existing video script. Optionally, each sample video clip has a semantic label corresponding to it, and the industry attributes corresponding to the existing video script can be determined based on these semantic labels. In an optional embodiment, a pre-trained video industry prediction model is used to predict the industry attributes of the existing script based on the semantic labels of multiple sample video clips. Optionally, the industry attributes of the existing script include at least one level, and may also include higher levels such as the second level. More levels of subdivision are to ensure that the industry attributes of the target object can match it more closely.

[0069] In an optional embodiment, the existing video script contains semantic tags corresponding to the video clips of the multiple samples, and the method of obtaining the semantic tags can be the same as the semantic tags corresponding to the multiple video clips to be processed from the video material segmented above. In another optional embodiment, the existing video script can also be labeled by manual annotation, which has high accuracy.

[0070] Further, at least one candidate video script is selected from the existing video scripts according to the industry attributes of the target object and the industry attributes corresponding to the existing video scripts. Optionally, the highest level of the industry attributes of the target object are used for matching first. If a candidate script of the first threshold is matched, the match is considered successful. Otherwise, the industry attributes of the lower level are matched. If a candidate script of the second threshold is matched, the match is considered successful. Otherwise, the industry attributes of the lower level are matched. Each level of industry attributes has its corresponding minimum threshold for matching results. If the number of candidate scripts matched is too low, the industry attributes of the lower level continue to be matched. The minimum threshold of each level can be set as needed. It should be noted that although the number of results of higher-level matching may be small and the number cannot meet the requirements, it is necessary to consider using the matching results of the lower level. In this case, the matching results of the higher level and the matching results of the lower level are combined to form candidate videos that meet the quantity requirements.

[0071] After obtaining at least one candidate script, further matching the target video script from the candidate script. Optionally, at least one target video script is selected from the at least one candidate video script based on the semantic tags of the multiple video clips to be processed and the semantic tags of the multiple sample video clips included in the at least one candidate video script.

[0072] In an optional embodiment, before matching, the semantic tags of multiple video clips to be processed are optimized to improve their matching degree. Optionally, multiple video tags to be processed are preprocessed to remove duplicate, erroneous, irrelevant or abnormal tags. Further, the tags are standardized, and some necessary conversions are performed on the tags to eliminate the inconsistency of the tags and improve the matching accuracy. Further optionally, representative tags are screened according to the industry attributes. For different industry attributes, they all have specific and relevant keywords or tags, which can accurately describe the characteristics and content of the target objects involved in the video material. For example, the industry attribute of a certain video material is mobile phone, and the semantic tags corresponding to the multiple video clips to be processed that are cut out include: mobile phone overview, appearance design, hardware specifications, software functions, coffee tasting. It is easy to see that the semantic tag "coffee tasting" does not belong to the mobile phone industry, so it can be eliminated, which can improve the matching rate of the sample video clips.

[0073] In an embodiment of the present application, when the semantic tags corresponding to a plurality of video clips to be processed filter the target video script, the higher the match degree with the semantic tags of the video script, the better, and the more matches, the better. Optionally, when the semantic tags of a plurality of video clips to be processed filter the target video script, it is not necessary to completely match the semantic tags of the video script. Due to the diversity and complexity of semantic tags, complete matching is not realistic in many cases. However, if most of the semantic tags are in the current video script, but individual semantic tags do not appear, the match can also be considered successful. Optionally, at this time, the user can also be prompted with the video clip corresponding to the missing semantic tag. In order to obtain better results, more video materials can be uploaded to complete the corresponding video clip.

[0074] In an optional embodiment, fuzzy matching can also be used to screen video scripts. The process generally includes: label data preprocessing → semantic label vectorization corresponding to each video to be processed → fusion of label vectors to obtain the overall video vector → label vectorization of video scripts and fusion to obtain video script vector → calculation of the similarity between the overall video vector and the video script vector → setting a similarity threshold and judging whether there is a match.

[0075] In another optional embodiment, weighted fuzzy matching may also be used. This method assigns different weights to different semantic tags. For example, higher weights may be assigned to tags that are more important to the content of the video material. This helps to match video scripts with higher relevance.

[0076] In an optional embodiment, after obtaining at least one target video script, the multiple video clips to be processed are combined and decorated according to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips to obtain multiple target video contents, including: for each target video script, according to the semantic labels of the multiple sample video clips contained in the target video script, at least one video clip to be processed with the same or similar semantics as each sample video clip is selected from the multiple video clips to be processed. In this embodiment, the semantic labels of the multiple sample video clips contained in the target video script are matched with the semantic labels of the multiple video clips to be processed. Among them, each sample video clip matches at least one video clip to be processed.

[0077] In an optional embodiment, a semantic label corresponds to only one video clip to be processed. In this case, when at least one video clip to be processed having the same or similar semantics as each sample video clip is cross-combined according to the combination mode of multiple sample video clips included in the target video script to obtain at least one initial video content, when the label successfully matches the label of the sample video clip, it is only necessary to insert the video clip to be processed into the "slot" of the sample video. In another optional embodiment, there are multiple video clips to be processed that correspond to one semantic label at the same time. In this case, when at least one video clip to be processed having the same or similar semantics as each sample video clip is cross-combined according to the combination mode of multiple sample video clips included in the target video script to obtain at least one initial video content, when the label successfully matches the label of the sample video clip, it is only necessary to select any video to be processed from the label and insert it into the "slot" of the sample video clip until all video clips to be processed under the label are traversed. At the same time, the video material may have multiple semantic tags after segmentation, and the same semantic tag may also correspond to multiple video segments to be processed. Based on this, when the video segments to be processed under different semantic tags are cross-combined, the initial video content can be generated in batches.

[0078] Optionally, the content decoration information includes sample decoration content and the decoration position and decoration method of the sample decoration content in the multiple sample video clips. Further optionally, according to the content decoration information of the multiple sample video clips contained in the target video script, the at least one initial video content is content decorated to obtain at least one target video content, including: generating target decoration content according to the sample decoration content; adding the target decoration content to a corresponding position in the at least one initial video content according to the decoration position and decoration method of the sample decoration content in the multiple sample video clips to obtain at least one target video content.

[0079] In an optional embodiment, generating target decorative content according to the sample decorative content includes: directly using the sample decorative content as the target decorative content. Optionally, the sample decorative content can be at least one of a picture, a sticker, a flower character, a voice, a transition, an animation, and background music. Optionally, the animation can also be Lottie. Optionally, the video script is based on the JSON format. In an example, its content is as follows: Figure 4 As shown, the video script introduces new summer clothes. It includes semantic tags such as "opening announcement", "selling point introduction", and "usage process", as well as the start and end time, special effects, and storage path of the video clip corresponding to each tag. The video script is only an example and does not constitute a limitation to this application.

[0080] In another optional embodiment, target decorative content is generated based on the sample decorative content, including: obtaining at least one multimedia data corresponding to the target object based on the description information of the target object; and replacing the sample decorative content of the same type in the sample decorative content with the at least one multimedia data to obtain the target decorative content.

[0081] Optionally, the description information may include one or more prompt words, which may preset the decorative content of the generated video content to meet the special needs of different demand parties. For example, the description information may indicate that the video content displays custom content at a specified location, such as displaying personal contact information, product links, etc. at the end of the video, or indicate the video resolution, background music, language of generated subtitles, presence or absence of special effects and transitions, and duration, etc. At least one decorative content of the target object is obtained according to the prompt words included in the description information.

[0082] In an optional embodiment, the video content demander is allowed to make certain restrictions or constraints on the video content to be generated. Based on this, in the process of uploading video materials and description information, the video content demander is allowed to upload constraint information on the video content at the same time, for example, prompt information can be uploaded, wherein the prompt information may include one or more prompt words. Optionally, the prompt words can preset the generated video content to meet the special needs of different demanders, for example, the prompt information can indicate that the video content displays custom content at a specified location, such as displaying personal contact information, product links, etc. at the end of the video, or indicating the resolution of the video, background music, language of generated subtitles, special effects, and the presence or absence of transitions. A video theme can also be given directly, which can not only set the tone for generating video content, but also select a video script related to a given theme when matching video scripts.

[0083] In one example, the prompt words included in the prompt information may be "subtitles in English, background music M". Of course, the prompt words can also be expressed in English or other languages, and these grammatical rules can be pre-set as needed. Further optionally, multiple multimedia data corresponding to the target object can be obtained according to the prompt information, such as the aforementioned content, and English subtitles and specified background music M can be obtained according to the prompt words in the prompt information. For example, the subtitles of the sample video clip are replaced with English subtitles, and the background music of the sample video clip is replaced with the specified background music M, and finally the target decorative content is obtained. In this embodiment, there is no limitation on the method of obtaining English subtitles. It can be that the voice information is first recognized as text and then translated into English by machine, or if the subtitles have been burned into the video material, OCR (Optical Character Recognition) technology can be used to recognize and extract the subtitles, or the video material has a matching subtitle file that can be read directly, and the subtitles can be translated into English by machine translation.

[0084] In an optional embodiment, the prompt words included in the prompt information can be directly used to screen video scripts, and only when they meet the conditions preset by the prompt words will they be selected as target video scripts.

[0085] In another optional embodiment, the prompt words included in the prompt information will be incorporated into the selected target video script, and the target video content will be generated based on the incorporated video script.

[0086] After obtaining the target decorative content, the target decorative content needs to be used to decorate the initial video content. Optionally, according to the decorative position and decoration method of the sample decorative content in the multiple sample video clips, the target decorative content is added to the corresponding position in the at least one initial video content to obtain at least one target video content. In this embodiment, the decorative position of the sample decorative content in the sample video clip can be a time position, which is the most common position definition method. The decorative content can be added in a specific time period of the video, such as adding subtitles from the 10th second to the 20th second of the video. Optionally, the decorative position of the sample decorative content in the sample video clip can also refer to adding decorative content in a specific spatial position of the video frame. For example, a map can be added to the upper right corner or center of the video. In some other embodiments, the decorative position can also be specified based on the position of the content, which usually requires the help of AI, such as target detection, face recognition and other technologies, which can allow the decorative content to move with specific targets, such as objects, characters, etc. Further, according to the aforementioned possible decorative positions and decoration methods, the target decorative content is added to the corresponding position in at least one video content to obtain at least one target video content.

[0087] In an embodiment of the present application, after the target video content is generated, the plurality of target video contents are optionally displayed to the video content demander. Optionally, the target video contents are displayed to the user according to the matching degree of the video scripts that generate the target video contents. Alternatively, based on the frequency of use of the video scripts, the target video contents generated by the video scripts with high frequency are preferentially displayed to the user. Alternatively, based on the rating of the video scripts, the target video contents are displayed to the user from high to low according to the rating, wherein the rating of the video scripts comes from the rating given by users who have used the scripts based on their satisfaction.

[0088] Optionally, a rating option may be displayed to the video content demander so that the demander can rate the generated video content; in response to the selection operation of the video content demander, the selected target video content is determined; and the selected target video content is published to the details page of the target object or a third-party platform.

[0089] Figure 5 The structure diagram of a video content generating device provided by the exemplary embodiment of the present application is shown in FIG. Figure 5 As shown, the device 500 includes: an acquisition module 51, a semantic segmentation module 52, a determination module 53, a selection module 54 and a generation module 55, wherein:

[0090] The acquisition module 51 is used to respond to the upload operation of the video content demander and acquire the video material containing the target object and the description information of the target object.

[0091] The semantic segmentation module 52 is used to perform semantic segmentation on the video material to obtain a plurality of to-be-processed video segments with semantic labels.

[0092] The determination module 53 is used to determine the industry attribute of the target object according to the description information of the target object.

[0093] The selection module 54 is used to select at least one target video script from the existing video scripts according to the semantic tags of the multiple video clips to be processed and the industry attributes of the target object, each video script comprising multiple sample video clips and the combination method and content decoration information of the multiple sample video clips.

[0094] The generating module 55 is used to combine and decorate the multiple video clips to be processed according to the multiple sample video clips contained in the at least one target video script and the combination mode and content decoration information of the multiple sample video clips to obtain multiple target video contents.

[0095] Further optionally, when the semantic segmentation module 52 performs semantic segmentation on the video material to obtain a plurality of video segments to be processed with semantic labels, it is also used to: input the video material into a pre-trained video segmentation model, in which the preprocessing layer is used to separate the voice and picture of the video material to obtain voice information and picture data with time information; input the voice information into the first semantic analysis layer, convert the voice information into text information, perform semantic segmentation on the text information to obtain a plurality of first semantic labels and their correspondence with time information; input the picture data into the second semantic analysis layer, divide the picture data into a plurality of picture segments according to the feature similarity between adjacent frames in the picture data, convert the plurality of picture segments into text information and perform semantic extraction on the text information to obtain a plurality of second semantic labels and their correspondence with time information; and segment the video material according to the plurality of first semantic labels and their correspondence with time information and the plurality of second semantic labels and their correspondence with time information to obtain a plurality of video segments to be processed with semantic labels.

[0096] Further optionally, when the semantic segmentation module 52 segments the video material according to the multiple first semantic tags and their correspondence with time information and the multiple second semantic tags and their correspondence with time information to obtain a plurality of video segments to be processed with semantic tags, it is specifically used to: generate a plurality of semantic tags corresponding to the video material and their correspondence with time information according to the multiple first semantic tags and their correspondence with time information and the multiple second semantic tags and their correspondence with time information; segment the video material according to the multiple semantic tags corresponding to the video material and their correspondence with time information to obtain a plurality of video segments to be processed, and mark each video segment to be processed with a corresponding semantic tag.

[0097] Further optionally, the semantic segmentation module 52 generates multiple semantic tags corresponding to the video material and their correspondence with the time information based on the multiple first semantic tags and their correspondence with the time information and the multiple second semantic tags and their correspondence with the time information, and is specifically used to: determine the first type of time information corresponding to both the first semantic tag and the second semantic tag and the second type of time information corresponding only to the second semantic tag based on the degree of overlap between the time information corresponding to the multiple first semantic tags and the time information corresponding to the multiple second semantic tags; for the first type of time information, generate the semantic tag corresponding to the first type of time information based on the first semantic tag and the second semantic tag corresponding to the first type of time information; for the second type of time information, use the second semantic tag corresponding to the second type of time information as the semantic tag corresponding to the second type of time information.

[0098] Further optionally, when determining the industry attributes of the target object based on the description information of the target object, the determination module 53 is also used to: the description information of the target object is the identification information of the target object, and according to the identification information of the target object, query the object information management system to which the target object belongs to obtain the industry attributes of the target object; or, the description information of the target object is the access link of the target object, and according to the access link of the target object, access the details page of the target object; and obtain the industry attributes of the target object from the details page of the target object.

[0099] Further optionally, when selecting at least one target video script from existing video scripts based on the semantic labels of the multiple video clips to be processed and the industry attributes of the target object, the selection module 54 is specifically used to: determine the industry attributes corresponding to the existing video scripts based on the multiple sample video clips contained in the existing video scripts; select at least one candidate video script from the existing video scripts based on the industry attributes of the target object and the industry attributes corresponding to the existing video scripts; select at least one target video script from the at least one candidate video script based on the semantic labels of the multiple video clips to be processed and the semantic labels of the multiple sample video clips contained in the at least one candidate video script.

[0100] Further optionally, when the generation module 55 combines and decorates the multiple video clips to be processed to obtain multiple target video contents according to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips, the generation module 55 is used to: for each target video script, select at least one video clip to be processed with the same or similar semantics as each sample video clip from the multiple video clips to be processed according to the semantic labels of the multiple sample video clips contained in the target video script; cross-combine at least one video clip to be processed with the same or similar semantics as each sample video clip according to the combination method of the multiple sample video clips contained in the target video script to obtain at least one initial video content; and perform content decoration on the at least one initial video content according to the content decoration information of the multiple sample video clips contained in the target video script to obtain at least one target video content.

[0101] Further optionally, the content decoration information includes sample decoration content and the decoration position and decoration method of the sample decoration content in the multiple sample video clips. When the generation module 55 performs content decoration on the at least one initial video content according to the content decoration information of the multiple sample video clips contained in the target video script to obtain at least one target video content, it is specifically used to: generate target decoration content according to the sample decoration content; add the target decoration content to the corresponding position in the at least one initial video content according to the decoration position and decoration method of the sample decoration content in the multiple sample video clips to obtain at least one target video content.

[0102] Further optionally, when generating target decorative content based on the sample decorative content, the generation module 55 is specifically used to: directly use the sample decorative content as the target decorative content; or, based on the description information of the target object, obtain at least one multimedia data corresponding to the target object; and use the at least one multimedia data to replace the sample decorative content of the same type in the sample decorative content to obtain the target decorative content.

[0103] Further optionally, the selection module 54 is also used to: display the multiple target video contents to the video content demander; and determine the selected target video content in response to the selection operation of the video content demander; and publish the selected target video content to the details page of the target object or a third-party platform.

[0104] Figure 6 FIG. 1 is a schematic diagram of a video content generation device provided by another exemplary embodiment of the present application. Figure 5 As shown, the device includes: a processor 61 and a memory 62, wherein the processor includes one or more and the memory also includes one or more.

[0105] The memory 62 is used to store computer programs and can be configured to store various other data to support operations on the computing platform. Examples of such data include instructions for any application or method operating on the computing platform, contact data, phone book data, messages, pictures, videos, etc.

[0106] The memory 62 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.

[0107] Among them, the processor 61 is coupled to the memory 62 and is used to execute the computer program stored in the memory 62, so as to: respond to the upload operation of the video content demander, obtain the video material containing the target object and the description information of the target object; semantically segment the video material to obtain multiple video clips to be processed with semantic labels, and determine the industry attributes of the target object according to the description information of the target object; select at least one target video script from the existing video scripts according to the semantic labels of the multiple video clips to be processed and the industry attributes of the target object, each video script contains multiple sample video clips and the combination method and content decoration information of the multiple sample video clips; combine and decorate the multiple video clips to be processed according to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips to obtain multiple target video contents.

[0108] In an optional embodiment, when the processor 61 performs semantic segmentation on the video material to obtain a plurality of video segments to be processed with semantic labels, the processor 61 is specifically used to: input the video material into a pre-trained video segmentation model, in which the pre-processing layer is used to separate the voice and picture of the video material to obtain voice information and picture data with time information; input the voice information into a first semantic analysis layer, convert the voice information into text information, perform semantic segmentation on the text information, and obtain a plurality of first semantic labels and their correspondence with time information; input the picture data into a second semantic analysis layer, divide the picture data into a plurality of picture segments according to the feature similarity between adjacent frames in the picture data, convert the plurality of picture segments into text information and perform semantic extraction on the text information to obtain a plurality of second semantic labels and their correspondence with time information; and segment the video material according to the plurality of first semantic labels and their correspondence with time information and the plurality of second semantic labels and their correspondence with time information to obtain a plurality of video segments to be processed with semantic labels.

[0109] In an optional embodiment, when the processor 61 segments the video material according to the multiple first semantic tags and their correspondence with time information and the multiple second semantic tags and their correspondence with time information to obtain a plurality of video segments to be processed with semantic tags, the processor 61 is specifically used to: generate a plurality of semantic tags corresponding to the video material and their correspondence with time information according to the multiple first semantic tags and their correspondence with time information and the multiple second semantic tags and their correspondence with time information; segment the video material according to the multiple semantic tags corresponding to the video material and their correspondence with time information to obtain a plurality of video segments to be processed, and mark each video segment to be processed with a corresponding semantic tag.

[0110] In an optional embodiment, when the processor 61 generates multiple semantic tags corresponding to the video material and their correspondence with time information based on the multiple first semantic tags and their correspondence with time information and the multiple second semantic tags and their correspondence with time information, the processor 61 is specifically used to: determine the first type of time information corresponding to both the first semantic tag and the second semantic tag and the second type of time information corresponding only to the second semantic tag based on the degree of overlap between the time information corresponding to the multiple first semantic tags and the time information corresponding to the multiple second semantic tags; for the first type of time information, generate the semantic tag corresponding to the first type of time information based on the first semantic tag and the second semantic tag corresponding to the first type of time information; for the second type of time information, use the second semantic tag corresponding to the second type of time information as the semantic tag corresponding to the second type of time information.

[0111] In an optional embodiment, when the processor 61 determines the industry attributes of the target object based on the description information of the target object, it is specifically used to: the description information of the target object is the identification information of the target object, and according to the identification information of the target object, query the object information management system to which the target object belongs to obtain the industry attributes of the target object; or, the description information of the target object is the access link of the target object, and according to the access link of the target object, access the details page of the target object; and obtain the industry attributes of the target object from the details page of the target object.

[0112] In an optional embodiment, when the processor 61 selects at least one target video script from existing video scripts based on the semantic tags of the multiple video clips to be processed and the industry attributes of the target object, it is specifically used to: determine the industry attributes corresponding to the existing video scripts based on the multiple sample video clips contained in the existing video scripts; select at least one candidate video script from the existing video scripts based on the industry attributes of the target object and the industry attributes corresponding to the existing video scripts; select at least one target video script from the at least one candidate video script based on the semantic tags of the multiple video clips to be processed and the semantic tags of the multiple sample video clips contained in the at least one candidate video script.

[0113] In an optional embodiment, when the processor 61 combines and decorates the multiple video clips to be processed to obtain multiple target video contents according to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips, the processor 61 is used to: for each target video script, select at least one video clip to be processed with the same or similar semantics as each sample video clip from the multiple video clips to be processed according to the semantic labels of the multiple sample video clips contained in the target video script; cross-combine at least one video clip to be processed with the same or similar semantics as each sample video clip according to the combination method of the multiple sample video clips contained in the target video script to obtain at least one initial video content; and perform content decoration on the at least one initial video content according to the content decoration information of the multiple sample video clips contained in the target video script to obtain at least one target video content.

[0114] In an optional embodiment, the content decoration information includes sample decoration content and the decoration position and decoration method of the sample decoration content in the multiple sample video clips. When the processor 61 performs content decoration on the at least one initial video content according to the content decoration information of the multiple sample video clips included in the target video script to obtain at least one target video content, it is specifically used to: generate target decoration content according to the sample decoration content; add the target decoration content to the corresponding position in the at least one initial video content according to the decoration position and decoration method of the sample decoration content in the multiple sample video clips to obtain at least one target video content.

[0115] In an optional embodiment, when the processor 61 generates the target decorative content based on the sample decorative content, it is specifically used to: directly use the sample decorative content as the target decorative content; or, based on the description information of the target object, obtain at least one multimedia data corresponding to the target object; and use the at least one multimedia data to replace the sample decorative content of the same type in the sample decorative content to obtain the target decorative content.

[0116] In an optional embodiment, the processor 61 is also used to: display the multiple target video contents to the video content demander; and determine the selected target video content in response to the selection operation of the video content demander; and publish the selected target video content to the details page of the target object or a third-party platform.

[0117] Further, if Figure 6 As shown, the device also includes: a communication component 63, a display 64, a power component 65, an audio component 66 and other components. Figure 6 Only some components are shown schematically, which does not mean that the device only includes Figure 6 In addition, Figure 6 The components in the dashed box are optional components, not mandatory components, and the specific components depend on the product form of the video content generation device. The video content generation device of this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, or an IOT device, or a server device such as a conventional server, a cloud server, or a server array. If the video content generation device of this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, etc., it can include Figure 6 If the video content generation device of this embodiment is implemented as a server device such as a conventional server, a cloud server or a server array, it may not include Figure 6 Components within the dashed box.

[0118] The detailed implementation and beneficial effects of each step in the method of this embodiment have been described in detail in the aforementioned embodiments and will not be elaborated here.

[0119] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement the steps that can be executed by the computing platform in the above method embodiment.

[0120] The above-mentioned memory 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.

[0121] The above-mentioned communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology and other technologies.

[0122] The above-mentioned display includes a screen, and 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 touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0123] The power supply assembly provides power to various components of the device where the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device where the power supply assembly is located.

[0124] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (Microphone, MIC), and when the device where the audio component is located is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting an audio signal.

[0125] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-readable storage media (including but not limited to disk storage, compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), optical storage, etc.) containing computer-usable program code.

[0126] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0129] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interface, network interface and memory.

[0130] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0131] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0132] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0133] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for generating video content, characterized in that: include: In response to an upload operation of a video content demander, obtaining video material containing a target object and description information of the target object; Performing semantic segmentation on the video material to obtain a plurality of video clips to be processed with semantic tags, and determining the industry attribute of the target object according to the description information of the target object; According to the semantic labels of the multiple video clips to be processed and the industry attributes of the target object, at least one target video script is selected from the existing video scripts, each video script comprising multiple sample video clips and a combination method and content decoration information of the multiple sample video clips; According to the multiple sample video clips contained in the at least one target video script and the combination method and content decoration information of the multiple sample video clips, the multiple video clips to be processed are combined and decorated to obtain multiple target video contents.

2. The method according to claim 1, characterized in that The video material is semantically segmented to obtain a plurality of video clips to be processed with semantic labels, including: Inputting the video material into a pre-trained video segmentation model, wherein the video material is separated into speech and picture by using a pre-processing layer to obtain speech information and picture data with time information; Inputting the voice information into a first semantic analysis layer, converting the voice information into text information, and performing semantic segmentation on the text information to obtain a plurality of first semantic tags and their corresponding relationships with time information; Inputting the picture data into a second semantic analysis layer, dividing the picture data into a plurality of picture segments according to feature similarities between adjacent frames in the picture data, converting the plurality of picture segments into text information and performing semantic extraction on the text information to obtain a plurality of second semantic labels and their corresponding relationships with time information; According to the plurality of first semantic tags and their corresponding relationships with time information and the plurality of second semantic tags and their corresponding relationships with time information, the video material is segmented to obtain a plurality of to-be-processed video segments with semantic tags.

3. The method according to claim 2, characterized in that According to the plurality of first semantic tags and the corresponding relationship between the tags and the time information and the plurality of second semantic tags and the corresponding relationship between the tags and the time information, the video material is segmented to obtain a plurality of to-be-processed video segments with semantic tags, including: Generate a plurality of semantic tags corresponding to the video material and their corresponding relationships with time information according to the plurality of first semantic tags and their corresponding relationships with time information and the plurality of second semantic tags and their corresponding relationships with time information; According to the multiple semantic tags corresponding to the video material and the corresponding relationship between the semantic tags and the time information, the video material is segmented to obtain multiple video segments to be processed, and the corresponding semantic tag is marked for each video segment to be processed.

4. The method according to claim 3, characterized in that Generating a plurality of semantic tags corresponding to the video material and their corresponding relationships with the time information according to the plurality of first semantic tags and their corresponding relationships with the time information and the plurality of second semantic tags and their corresponding relationships with the time information, including: Determine, according to the degree of overlap between the time information corresponding to the multiple first semantic tags and the time information corresponding to the multiple second semantic tags, the first type of time information corresponding to both the first semantic tags and the second type of time information corresponding only to the second semantic tags; For the first type of time information, generating a semantic tag corresponding to the first type of time information according to the first semantic tag and the second semantic tag corresponding to the first type of time information; For the second type of time information, the second semantic label corresponding to the second type of time information is used as the semantic label corresponding to the second type of time information.

5. The method according to claim 1, characterized in that Determining the industry attribute of the target object according to the description information of the target object includes: The description information of the target object is the identification information of the target object, and according to the identification information of the target object, the object information management system to which the target object belongs is queried to obtain the industry attribute of the target object; or The description information of the target object is an access link of the target object. According to the access link of the target object, the detail page of the target object is accessed; and the industry attribute of the target object is obtained from the detail page of the target object.

6. The method according to claim 1, characterized in that According to the semantic labels of the plurality of video clips to be processed and the industry attributes of the target object, at least one target video script is selected from the existing video scripts, including: Determine the industry attributes corresponding to the existing video script based on multiple sample video clips contained in the existing video script; Selecting at least one candidate video script from the existing video scripts according to the industry attributes of the target object and the industry attributes corresponding to the existing video scripts; At least one target video script is selected from the at least one candidate video script according to the semantic tags of the multiple video clips to be processed and the semantic tags of the multiple sample video clips included in the at least one candidate video script.

7. The method according to claim 1, characterized in that According to the plurality of sample video clips included in the at least one target video script and the combination mode and content decoration information of the plurality of sample video clips, the plurality of video clips to be processed are combined and decorated to obtain a plurality of target video contents, including: For each target video script, according to the semantic labels of the multiple sample video clips contained in the target video script, at least one to-be-processed video clip having the same or similar semantics as each sample video clip is selected from the multiple to-be-processed video clips; According to the combination mode of the plurality of sample video segments contained in the target video script, cross-combining at least one to-be-processed video segment having the same or similar semantics as each sample video segment, to obtain at least one initial video content; According to the content decoration information of the plurality of sample video clips contained in the target video script, the at least one initial video content is content decorated to obtain at least one target video content.

8. The method according to claim 7, characterized in that The content decoration information includes sample decoration content and the decoration position and decoration method of the sample decoration content in the plurality of sample video clips; According to the content decoration information of the plurality of sample video clips included in the target video script, the at least one initial video content is content decorated to obtain at least one target video content, including: generating target decoration content according to the sample decoration content; According to the decoration position and decoration method of the sample decoration content in the multiple sample video clips, the target decoration content is added to the corresponding position in the at least one initial video content to obtain at least one target video content.

9. The method according to claim 8, characterized in that Generating target decoration content according to the sample decoration content includes: Using the sample decoration content directly as the target decoration content; or, Acquire at least one multimedia data corresponding to the target object according to the description information of the target object; The at least one multimedia data is used to replace the sample decoration content of the same type in the sample decoration content to obtain the target decoration content.

10. The method according to any one of claims 1 to 9, characterized in that: Also includes: Displaying the plurality of target video contents to the video content demander; as well as In response to a selection operation by the video content demander, determining a selected target video content; The selected target video content is published to the detail page of the target object or a third-party platform.

11. A video content generating device, characterized in that: include: Memory and processor; The memory is used to store a computer program, and the processor is coupled to the memory and is used to execute the computer program in the memory to implement the steps in the method according to any one of claims 1 to 10.

12. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is enabled to implement the steps in the method according to any one of claims 1 to 10.

Images