A multimedia interface generation method, device, equipment and medium

Abstract

The present disclosure provides a multimedia interface generation method, device, equipment and medium, which realizes one-key generation of a multimedia interface based on a multimedia list screenshot. The method comprises: obtaining a multimedia list screenshot and performing multimedia information identification to obtain related information of a multimedia in the multimedia list screenshot; determining similar multimedia interfaces of the multimedia list screenshot from a multimedia database according to the obtained related information of the multimedia, and obtaining cover information and multimedia description information of each similar multimedia interface; and generating a target multimedia interface according to the cover information and the multimedia description information of the similar multimedia interfaces.

Description

Technical Field

[0001] This disclosure relates to the field of multimedia technology, and in particular to a method, apparatus, device and medium for generating a multimedia interface. Background Technology

[0002] With the continuous development of various multimedia platforms, multimedia playback via multimedia playlists has become increasingly popular among users. As a result, various multimedia platforms have launched multimedia playlist modes. However, since multimedia playlists are not universal across different multimedia platforms, users need to recreate their multimedia playlists after switching platforms. Therefore, various multimedia platforms have introduced the function of importing external multimedia playlists.

[0003] Currently, the most common method for importing multimedia playlists is through screenshots of the playlist. During the import process, users can import screenshots of the multimedia playlist into their current multimedia software, which then identifies the multimedia content based on the screenshot. However, this method still requires users to manually create the multimedia playlist and manually input its information, significantly increasing the user's workload. Summary of the Invention

[0004] This disclosure provides a multimedia interface generation method, apparatus, device, and medium, enabling multimedia list-based screenshot generation. Figure 1 Create multimedia interfaces with a single key.

[0005] According to a first aspect of the present disclosure, a multimedia interface generation method is provided, the method comprising:

[0006] Obtain a screenshot of the multimedia list and perform multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot;

[0007] Based on the relevant information of the multimedia obtained, similar multimedia interfaces to the screenshot of the multimedia list are determined from the multimedia database, and the cover information and multimedia description information of each similar multimedia interface are obtained.

[0008] The target multimedia interface is generated based on the cover information and multimedia description information of the similar multimedia interface.

[0009] The method described above extracts relevant information about the multimedia from a screenshot of the multimedia playlist, filters similar multimedia interfaces from the multimedia database, and generates a new multimedia interface based on the cover art and description information of the similar multimedia interfaces. When importing a multimedia playlist from a screenshot, users do not need to manually create a playlist and enter relevant information. Instead, the system automatically filters similar multimedia based on the information in the playlist and generates a multimedia interface based on the information of the similar multimedia. This achieves one-click generation of multimedia interfaces based on screenshots of the multimedia playlist, improving the efficiency of users importing playlists, enhancing the user experience, and reducing the user's usage cost.

[0010] One optional implementation involves acquiring a screenshot of the multimedia list and performing multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot, including:

[0011] Obtain a screenshot of the multimedia list and perform multimedia information area detection; then perform text content recognition on the detected multimedia information areas.

[0012] Based on the identified text content, match the target keywords in the multimedia keyword database;

[0013] Based on the target keywords, obtain relevant information about the multimedia in the multimedia list screenshot.

[0014] The above method extracts text from multimedia list screenshots by performing region detection and text recognition, and further extracts relevant information from the extracted text by performing keyword matching in a multimedia keyword database, thus ensuring the accuracy and effectiveness of the extracted information.

[0015] An optional implementation involves acquiring a screenshot of the multimedia list and detecting multimedia information regions, then performing text content recognition on the detected multimedia information regions, including:

[0016] The multimedia list screenshot is divided into text regions by line to obtain multi-line text regions of the multimedia list screenshot;

[0017] Remove redundant information from each line of text in the multimedia list screenshot and determine the envelope region of each line of text in the multimedia screenshot;

[0018] By using a deep learning neural network to perform character recognition on the envelope region of each line of text, the text content corresponding to the envelope region of each line of text can be obtained.

[0019] The above method segments and recognizes text regions by row, making full use of the characteristic that text is distributed by row in multimedia list screenshots, thereby improving the efficiency and accuracy of text region segmentation and text recognition.

[0020] An optional implementation involves determining similar multimedia interfaces to the multimedia list screenshot from a multimedia database based on the obtained multimedia-related information, including:

[0021] The relevant information of each multimedia extracted from the multimedia list screenshot is encoded and integrated to obtain the first encoded vector corresponding to the multimedia list screenshot;

[0022] Obtain the second encoding vector of multimedia-related information from different multimedia interfaces in the multimedia database, and calculate the similarity between the second encoding vector and the first encoding vector;

[0023] Based on the calculated similarity, similar multimedia interfaces to the multimedia list screenshots are identified from the multimedia database.

[0024] The above method calculates the similarity between the first code corresponding to the relevant information of the multimedia information and the second code corresponding to each multimedia interface in the multimedia database, and determines the similar multimedia interfaces based on the similarity, thereby accurately identifying similar multimedia interfaces from the multimedia database.

[0025] An optional implementation involves generating a target multimedia interface based on cover information and multimedia description information of similar multimedia interfaces, including:

[0026] Identify the cover styles of similar multimedia interfaces, and determine the cover information corresponding to the multimedia list screenshot based on the cover styles of similar multimedia interfaces;

[0027] Identify words in the multimedia description information of similar multimedia interfaces, and determine the multimedia description information corresponding to the multimedia list screenshot based on the identified words;

[0028] Based on the multimedia information in the multimedia list screenshot, as well as the corresponding cover information and multimedia description information, generate the target multimedia interface.

[0029] The above method determines the cover information of the multimedia list screenshot based on the cover style of similar multimedia content, and determines the multimedia description information of the multimedia list screenshot based on the description information of similar multimedia content, thus ensuring the style consistency between the final generated multimedia interface and the interface corresponding to the multimedia list screenshot.

[0030] One optional implementation involves identifying the cover styles of similar multimedia interfaces, including:

[0031] Feature extraction is performed on the cover information of similar multimedia interfaces;

[0032] Based on the extracted cover style feature vectors, the probability of different cover style classifications is predicted, and the cover style classification with the highest probability is determined as the cover style of the corresponding similar multimedia interface.

[0033] The above method extracts features from the cover information and predicts the probability of cover style classification based on the extracted features. The cover style with the highest predicted probability is determined as the cover style of similar multimedia interfaces, thus ensuring the accuracy of the cover style of each similar multimedia interface.

[0034] One optional implementation involves determining the cover information corresponding to the multimedia list screenshot based on the cover style of similar multimedia interfaces, including:

[0035] Calculate the frequency of occurrence / user preference for each cover style among similar multimedia interface cover styles;

[0036] Determine the target cover style corresponding to the multimedia list screenshot based on the frequency of occurrence / user preference;

[0037] Select the cover information that corresponds to the multimedia list screenshot from the similar multimedia interface that matches the target cover style.

[0038] The above method determines the target cover style corresponding to the multimedia list screenshot based on the frequency of occurrence of each cover style and user preference, thus ensuring the accuracy of the determined target cover style.

[0039] An optional implementation involves identifying words in the multimedia description information of each similar multimedia interface, and generating multimedia description information corresponding to the multimedia list screenshot based on the identified words, including:

[0040] Identify words in the multimedia description information of similar multimedia interfaces and count the number of times each word appears;

[0041] The system filters out a preset number of target words based on their frequency of occurrence, and then determines the multimedia description information corresponding to the multimedia list screenshot based on the target words.

[0042] The above method determines target words based on the frequency of word occurrences in the multimedia description information of similar multimedia interfaces, and then determines the multimedia description information corresponding to the multimedia list screenshot based on the target words, thereby increasing the probability that the multimedia description information is similar to the description information in the multimedia list screenshot.

[0043] An optional implementation involves filtering a preset number of target words based on their frequency of occurrence, and determining multimedia description information corresponding to the multimedia list screenshot based on the target words, including:

[0044] Based on the frequency of word occurrence, a preset number of target words are selected in descending order of frequency.

[0045] Based on the preset number of target words selected, predict the corresponding semantic expression, and select target multimedia description information with the same semantic expression from the multimedia description information database. Then, determine the target multimedia description information as the multimedia description information corresponding to the multimedia list screenshot.

[0046] The above method predicts the corresponding semantic expression based on the target words, and then filters target multimedia description information with the same semantic expression from the multimedia description information database based on the semantic expression, thus ensuring the accuracy of the determined target multimedia description information.

[0047] According to a second aspect of the present disclosure, a multimedia interface generation apparatus is provided, comprising:

[0048] The acquisition module is used to acquire a screenshot of the multimedia list and perform multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot;

[0049] The determination module is used to determine similar multimedia interfaces of the multimedia list screenshot from the multimedia database based on the relevant information of the obtained multimedia, and to obtain the cover information and multimedia description information of each similar multimedia interface.

[0050] The production module is used to generate a target multimedia interface based on the cover information and multimedia description information of the similar multimedia interfaces.

[0051] According to a third aspect of the present disclosure, an electronic device is provided, comprising:

[0052] processor;

[0053] Memory used to store the aforementioned processor-executable instructions;

[0054] The processor is configured to execute the instructions to implement the steps in any of the methods in the first aspect.

[0055] According to a fourth aspect of the present disclosure, a computer-readable non-volatile storage medium is provided, including computer-readable instructions that, when read and executed by a computer, cause the computer to perform the steps in any of the methods in the first aspect described above. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a schematic diagram illustrating an application scenario provided by an embodiment of this disclosure;

[0058] Figure 2 A system framework diagram of a multimedia interface generation method provided in this disclosure embodiment;

[0059] Figure 3 A flowchart illustrating a multimedia interface generation method provided in this embodiment of the disclosure;

[0060] Figure 4 A schematic diagram illustrating text region segmentation provided in an embodiment of this disclosure;

[0061] Figure 5 A schematic diagram of an envelope region provided in an embodiment of this disclosure;

[0062] Figure 6 A flowchart illustrating a cover style classification model training method provided in this embodiment of the disclosure;

[0063] Figure 7 A flowchart illustrating a multimedia description generation model training method provided in this embodiment of the disclosure;

[0064] Figure 8 A schematic diagram of a multimedia interface generation device provided in an embodiment of this disclosure;

[0065] Figure 9 A schematic diagram of an electronic device provided according to an embodiment of this disclosure;

[0066] Figure 10 This is a schematic diagram of a program product provided for an embodiment of this disclosure. Detailed Implementation

[0067] The principles and spirit of this disclosure will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement this disclosure, and are not intended to limit the scope of this disclosure in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0068] Those skilled in the art will recognize that embodiments of this disclosure can implement a method, apparatus, device, and medium for transmitting data via bandwidth estimation. Therefore, this disclosure can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0069] According to embodiments of this disclosure, a method, apparatus, device, and medium for generating a multimedia interface are proposed.

[0070] It is important to understand in this article that any number of elements in the accompanying figures is for illustrative purposes and not for limitation, and that any naming is for distinction only and has no limiting meaning.

[0071] The principles and spirit of this disclosure will be explained in detail below with reference to several representative embodiments. Invention Overview

[0073] The author has discovered that with the continuous development of various multimedia platforms, multimedia playback via multimedia playlists is becoming increasingly popular among users. Various multimedia platforms have launched multimedia playlist modes. However, since multimedia playlists are not universal across different multimedia platforms, users need to recreate their multimedia playlists after switching platforms. Therefore, various multimedia platforms have introduced external multimedia playlist import functions.

[0074] Currently, the most common method for importing multimedia playlists is through screenshots. During the import process, users can take screenshots of their multimedia playlists and import them into their current multimedia software. The software then identifies the multimedia elements within the screenshot. However, this method still requires users to manually create the playlist and manually input its information. For example, when importing a playlist screenshot, the software first identifies the song and artist information, then the user manually creates the playlist, enters the playlist name, and imports the songs, significantly increasing the user's workload.

[0075] In view of this, the present disclosure provides a method for generating a multimedia interface.

[0076] This method acquires a screenshot of a multimedia list and performs multimedia information recognition to obtain relevant information about the multimedia in the screenshot. Based on the obtained multimedia information, it filters similar multimedia interfaces from the multimedia database and obtains the cover information and multimedia description information of each similar multimedia interface. Based on the cover information and multimedia description information of the similar multimedia interfaces, it generates a new multimedia interface to achieve multimedia list-based multimedia interface recognition. Figure 1 Create multimedia interfaces with a single key.

[0077] Application Scenarios Overview

[0078] First refer to Figure 1 This is a schematic diagram illustrating an application scenario of an embodiment of this disclosure. User 10 logs into network server 12 through a client installed on user device 11. The client can be a web browser or an application client installed on a mobile user device, such as a mobile phone or tablet computer.

[0079] User equipment 11 and network server 12 communicate via a network, which can be a local area network (LAN), a wide area network (WAN), or the like. User equipment 11 can be a portable device (e.g., a mobile phone, tablet, laptop computer) or a personal computer (PC). Network server 12 can be any device capable of providing internet services.

[0080] One possible form of communication between user device 11 and network server 12 is that the user can log in to the corresponding multimedia platform and upload screenshots of the multimedia playlist. Network server 12 receives the multimedia playlist screenshots uploaded by user device 11, such as screenshots of playlists and video playlists.

[0081] In this embodiment of the disclosure, after receiving a screenshot of a multimedia list uploaded by user 10 after logging into the corresponding multimedia platform, the network server 12 performs multimedia information recognition on the multimedia list screenshot to obtain relevant information about the multimedia in the multimedia list screenshot; based on the obtained relevant information about the multimedia, it determines similar multimedia interfaces of the multimedia list screenshot from the multimedia database and obtains the cover information and multimedia description information of each similar multimedia interface; and generates a target multimedia interface based on the cover information and multimedia description information of the similar multimedia interfaces.

[0082] Exemplary methods

[0083] For ease of understanding, the terminology appearing in the embodiments of this disclosure is uniformly explained below:

[0084] Cross-entropy loss function: A loss function in deep learning used to measure the difference between the predicted distribution and the true labeled distribution. Assuming M represents the number of classes in the sample... The indicator variable is the true label; it is 1 if the predicted class matches the sample class, and 0 otherwise. To predict the probability that an observed sample belongs to a certain category, the corresponding cross-entropy loss function is:

[0085] .

[0086] CNN: Also known as Convolutional Neural Network, it is a type of feedforward neural network where artificial neurons can respond to surrounding units. Convolutional neural networks include convolutional layers and pooling layers.

[0087] RNN: Also known as Recurrent Neural Network, the internal state of the network can exhibit dynamic temporal behavior and can use internal memory to process variable-length input sequences.

[0088] CRNN: Convolutional Recurrent Neural Network, also known as CNN+RNN, is used to solve image-based sequence recognition problems, especially scene text recognition problems. The CRNN network implements variable-length verification. It combines CNN and RNN network structures, uses a bidirectional LSTM recurrent network for temporal training, and finally introduces the CTC loss function to achieve end-to-end variable-length sequence recognition.

[0089] CTC: Short for Connectionist Temporal Classification, it's an algorithm commonly used in speech recognition, text recognition, and other fields to solve problems where input and output sequences have different lengths and cannot be aligned. In CRNN, it's actually the loss function corresponding to the model.

[0090] The following is combined Figure 1 Application scenarios, refer to Figure 2 This document describes a multimedia interface generation method according to exemplary embodiments of the present disclosure. It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of the present disclosure, and the embodiments of the present disclosure are not limited in any way. Rather, the embodiments of the present disclosure can be applied to any scenario in which they are used.

[0091] Figure 2 A system framework diagram of a multimedia interface generation method provided in this disclosure embodiment is shown below. Figure 2 As shown, the system is mainly divided into four parts: multimedia detection module, information extraction module, multimedia description matching module, and cover matching module. The functions of each module are as follows:

[0092] (1) Multimedia detection module, used to detect the position of cover information and multimedia description information (including multimedia list title and multimedia list introduction) in the multimedia list screenshot, and filter out useless background areas.

[0093] (2) Information extraction module, used to identify information in the local area output by the multimedia detection module of the song, that is, to extract relevant multimedia information in the text using the weak error correction capabilities of text recognition and natural language processing.

[0094] Based on relevant information about the multimedia, filter similar multimedia interfaces that match the screenshots of the multimedia list from the multimedia database.

[0095] (3) Multimedia description matching module, which is used to generate multimedia description information based on the output of the information extraction module. That is, based on the selected similar multimedia interfaces, its multimedia description information is extracted, and these text information are used as input to generate multimedia description information using a multimedia description generation model (such as the GPT-2 algorithm model).

[0096] (4) Cover matching module, which is used to identify the cover style of similar multimedia interfaces by using a cover style classification model based on the cover information of similar multimedia interfaces selected, and to determine the corresponding cover information based on the identified cover style.

[0097] After obtaining the relevant information, cover information, and multimedia description information corresponding to the multimedia list screenshot using the above model, a brand-new multimedia interface is generated in the multimedia application.

[0098] By adopting the above methods, the user cost can be effectively reduced and a more convenient experience can be provided.

[0099] Figure 3 This is a flowchart illustrating a multimedia interface generation method provided in an embodiment of the present disclosure, such as... Figure 3 As shown in the figure, this disclosure provides a multimedia interface generation method, including the following steps:

[0100] Step 301: Obtain a screenshot of the multimedia list and perform multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot;

[0101] Step 302: Based on the obtained multimedia information, determine the similar multimedia interfaces of the multimedia list screenshot from the multimedia database, and obtain the cover information and multimedia description information of each similar multimedia interface.

[0102] Step 303: Generate the target multimedia interface based on the cover information and multimedia description information of similar multimedia interfaces.

[0103] The above method, based on the relevant information of the multimedia extracted from the multimedia list screenshot, determines similar multimedia interfaces from the multimedia database; and generates a new target multimedia interface based on the cover information and multimedia description information of the similar multimedia interfaces. When importing the multimedia list based on the screenshot, the user does not need to manually create the list and enter relevant information, but the multimedia interface is automatically generated, realizing one-click generation of multimedia interfaces based on multimedia list screenshots.

[0104] In step 301 above, the multimedia list screenshot includes multimedia information, multimedia cover, multimedia description information, and background. Taking a playlist as an example, a playlist screenshot usually includes song information (such as song name, artist name, and song introduction), playlist information (such as playlist title and playlist introduction), and playlist cover.

[0105] The aforementioned information includes the names, authors, and descriptions of each multimedia element in the multimedia list screenshot. For example, a playlist includes the song title, artist name, and playlist title.

[0106] As an optional implementation, a screenshot of the multimedia list is acquired and multimedia information is identified to obtain relevant information about the multimedia in the multimedia list screenshot, including:

[0107] Obtain a screenshot of the multimedia list and perform multimedia information area detection; then perform text content recognition on the detected multimedia information areas.

[0108] Based on the identified text content, match the target keywords in the multimedia keyword database;

[0109] Based on the target keywords, obtain relevant information about the multimedia in the multimedia list screenshot.

[0110] Specifically, the process involves obtaining a screenshot of the multimedia list and detecting multimedia information regions, then performing text content recognition on the detected multimedia information regions. This includes the following steps:

[0111] Step 1: Divide the multimedia list screenshot into text regions by line to obtain multi-line text regions of the multimedia list screenshot;

[0112] In some embodiments, an adaptive binary algorithm is used to segment the text region line by line, and the resulting screenshot of the multimedia list is shown below. Figure 4 As shown, an adaptive binary algorithm is used to distinguish regions containing content from regions without content using different values. Figure 4 The black areas represent areas with content, while the white areas represent areas without content.

[0113] It should be noted that the aforementioned areas containing content refer not only to areas containing text, but also to areas containing images and icons.

[0114] Then, a row segmentation algorithm is used to segment the processed multimedia list screenshot. The regions containing content, obtained through the adaptive binary algorithm, are segmented in a row-like manner. Figure 4 The black area in the image is divided into rows to obtain the multi-line text area of ​​the multimedia list screenshot.

[0115] The main purpose of this step is to remove redundant information from the multimedia list screenshot, such as background information of the multimedia interface.

[0116] Step 2: Remove redundant information from each line of text in the multimedia list screenshot and determine the envelope area of ​​each line of text in the multimedia screenshot.

[0117] In some embodiments, the envelope region is a rectangular region that includes all the text.

[0118] Specifically, the deep learning CRNN algorithm is used to detect text regions in the multi-line text regions of the segmented multimedia list screenshot, determining the envelope region of each line of text, such as... Figure 5 As shown in the figure, the gray area is the detected envelope area. In some embodiments, the position information corresponding to the envelope area is represented by the coordinates of the envelope area combined with the size. For example, it can be represented by (x, y, w, h), where (x, y) is the coordinate point of the upper left corner of each gray area (or other positions, such as the center point, lower right corner, etc.), and (w, h) is the length and width of the entire detection area (i.e., the gray area).

[0119] It should be noted that this step determines the envelope region of each line of text based on the segmented multi-line text region, in the form of lines.

[0120] Before using the CRNN algorithm described above to detect text regions in the multi-line text regions of the segmented multimedia list screenshot, the algorithm needs to be trained. The specific training process is as follows:

[0121] The first step is to collect multiple screenshots of multimedia lists as samples, and to label the envelope region of each line of text in the samples using the labeling method (x, y, w, h).

[0122] The second step is to select a text detection CNN algorithm (such as CRNN algorithm) and perform regression training on the envelope region. The method used in this embodiment is, for example, YOLOv5.

[0123] The third step is to determine the end of training when the corresponding envelope region label is output after the input sample multimedia list screenshot is determined.

[0124] Step 3: Use a deep learning neural network to perform text recognition on the envelope area of each line of text, and obtain the text content corresponding to the envelope area of each line of text;

[0125] Specifically, first, imitate the text format in the multimedia list, randomly generate pictures with text, and record the text information in the pictures as the training set. Then, generate a text dictionary using a common Chinese character library, where each character corresponds to a digital label, such as "你" corresponding to the digital label 1.

[0126] Then, adopt the CRNN+CTC method to perform recognition training on the text in the pictures. When it is determined that the output text content matches the corresponding pre-recorded text information, it is determined that the training is completed, and thus a trained text classification one-hot model can be obtained.

[0127] During application, input the envelope area of each line of text (in the form of a picture) into the trained text classification one-hot model, and the text content corresponding to the output envelope area can be obtained.

[0128] As an optional implementation manner, the above-mentioned matching of the target keyword in the multimedia keyword library according to the recognized text content, and obtaining the relevant information of the multimedia in the multimedia list screenshot are implemented by the above-mentioned information extraction module. The process of the information extraction module extracting relevant information will be elaborated in detail below.

[0129] In the text content corresponding to the envelope area of each line of text recognized by the above-mentioned deep learning neural network, there is some redundant information interference, and the final relevant information of the multimedia cannot be directly extracted from it. Therefore, it is necessary to use the natural language processing (NLP, Natural Language Processing) algorithm to identify the relevant information of the multimedia in the recognized text content according to the multimedia keyword library.

[0130] In the embodiments of the present disclosure, the NLP algorithm adopts the TF-IDF algorithm, which is a commonly used weighting technique for information retrieval and text mining. In addition to the above TF-IDF algorithm, other NLP algorithms can also be used to implement the above steps, and the specific algorithms used in the embodiments of the present disclosure are not limited.

[0131] In the embodiments of the present disclosure, use the TF-IDF algorithm to match and retrieve the text content corresponding to the obtained envelope area and the information in the multimedia keyword library, retrieve the most matching content, that is, the target keyword, and determine the relevant information of the multimedia in the multimedia screenshot according to the target keyword.

[0132] The following example, using a playlist screenshot, illustrates the process by which the TF-IDF algorithm is used to determine relevant information about multimedia.

[0133] Suppose the text content of the envelope region identified using a deep learning neural network is as follows:

[0134] [['VIP Paid'],[],['Song Name 1','1','D','L+','Artist Name 1'],['Song Name 2','2','D','Artist Name 2,Selected'],['Song Name 3','Y','D','L+','Artist Name 3,Mandarin..'],['Song Name 4 (Old Japanese Version)','4','D','L+','Artist Name 4 Original Song...']。

[0135] The TF-IDF algorithm is used to match the above text content with the text content in the song library, which pre-stores song titles and artist names. The matched songs and artist names from the library are output, and the output song information includes each song title and artist name in the playlist screenshot:

[0136] [['Song Name 1', 'Artist Name 1'], ['Song Name 2', 'Artist Name 2'], ['Song Name 3', 'Artist Name 3'], ['Song Name 4', 'Artist Name 4']].

[0137] Based on the target keywords mentioned above, relevant information about the multimedia in the multimedia list screenshot was obtained.

[0138] In step 302 above, in this embodiment of the disclosure, based on the extracted multimedia information, a clustering algorithm is used from the multimedia database to determine similar multimedia interfaces of the multimedia list screenshot. That is, using the multimedia information as the cluster center, similar multimedia interfaces of the multimedia list screenshot are determined from the multimedia database according to a preset matching principle.

[0139] In step 302 above, based on the obtained multimedia information, similar multimedia interfaces to the multimedia list screenshot are determined from the multimedia database, specifically including:

[0140] The relevant information of each multimedia extracted from the multimedia list screenshot is encoded and integrated to obtain the first encoded vector corresponding to the multimedia list screenshot;

[0141] Obtain the second encoding vector of multimedia-related information from different multimedia interfaces in the multimedia database, and calculate the similarity between the second encoding vector and the first encoding vector;

[0142] Based on the calculated similarity, similar multimedia interfaces to the multimedia list screenshots are identified from the multimedia database.

[0143] After obtaining the similarity between the multimedia screenshot and different multimedia interfaces in the multimedia database, the multimedia interfaces in the multimedia database are sorted according to the similarity, and a preset number of multimedia interfaces are determined as similar multimedia interfaces for the multimedia list screenshot based on the sorting order.

[0144] In step 303 above, generating the target multimedia interface based on the cover information and multimedia description information of similar multimedia interfaces includes the following three sub-steps:

[0145] Identify the cover styles of similar multimedia interfaces, and determine the cover information corresponding to the multimedia list screenshot based on the cover styles of similar multimedia interfaces;

[0146] Identify words in the multimedia description information of similar multimedia interfaces, and determine the multimedia description information corresponding to the multimedia list screenshot based on the identified words;

[0147] Based on the multimedia information in the multimedia list screenshot, as well as the corresponding cover information and multimedia description information, generate the target multimedia interface.

[0148] The aforementioned identification of the cover styles of similar multimedia interfaces includes:

[0149] Feature extraction is performed on the cover information of similar multimedia interfaces;

[0150] Based on the extracted cover style feature vectors, the probability of different cover style classifications is predicted, and the cover style classification with the highest probability is determined as the cover style of the corresponding similar multimedia interface.

[0151] The above-mentioned identification of the cover style of each similar multimedia interface can be achieved using a cover style classification model. Before using the cover style classification model to identify the cover style of each similar multimedia interface, the cover style classification model needs to be trained first.

[0152] Figure 6 This is a flowchart illustrating a cover style classification model training method provided in an embodiment of this disclosure, as shown below. Figure 6 As shown, the training method for the above cover style classification model includes:

[0153] Step 601: Obtain cover information of multiple sample multimedia interfaces with different cover styles, and determine the cover style tags of each multimedia interface.

[0154] In some embodiments, a preset number of sample multimedia interface cover information (usually cover images) can be manually collected and labeled with style and color tags. The style tags include, but are not limited to, tech style, sweet style, European and American rock style, minimalist style, retro style, etc., and black and white color, warm color, cool color, etc.

[0155] Step 602: Use the cover style classification model to extract features from the cover information of each sample multimedia interface, and predict the probability of different cover style classifications based on the extracted cover style feature vectors to determine the cover style corresponding to each sample multimedia interface.

[0156] Step 603: Determine the first loss function value based on the determined cover style of each sample multimedia interface and the corresponding cover style label. Train the cover style classification model based on the first loss function value. After training, the trained cover style classification model is obtained.

[0157] In some embodiments, the cross-entropy loss function between the corresponding cover style label and the cover style output by the model is calculated and optimized using the gradient descent algorithm.

[0158] The cover information of the sample multimedia interface is input into the cover style classification model. The model training ends when the cover style output by the model is consistent with the labeled cover style.

[0159] After training, the cover information of each similar multimedia interface is input into the trained cover style classification model to obtain the cover style of each similar multimedia interface.

[0160] As an optional implementation, based on the cover style of similar multimedia interfaces, the cover information corresponding to the multimedia list screenshot is determined, including:

[0161] Calculate the frequency of occurrence / user preference for each cover style among similar multimedia interface cover styles;

[0162] Determine the target cover style corresponding to the multimedia list screenshot based on the frequency of occurrence / user preference;

[0163] Select the cover information that corresponds to the multimedia list screenshot from the similar multimedia interface that matches the target cover style.

[0164] In some embodiments, cover information can be randomly selected from the cover information corresponding to the cover style that appears most frequently among similar multimedia interface cover styles, or the cover information corresponding to the cover style with the highest user preference, as the target cover style corresponding to the multimedia list screenshot.

[0165] User preference can be calculated based on user behavior data such as following, collecting, and liking of logged-in accounts, or basic user information such as age, gender, and region.

[0166] The above-mentioned identification of words in the multimedia description information of similar multimedia interfaces, and the generation of multimedia description information corresponding to the multimedia list screenshot based on the identified words, includes:

[0167] Identify words in the multimedia description information of similar multimedia interfaces and count the number of times each word appears;

[0168] The system filters out a preset number of target words based on their frequency of occurrence, and then determines the multimedia description information corresponding to the multimedia list screenshot based on the target words.

[0169] The above process can be achieved through a multimedia description generation model. Before using the multimedia description generation model to identify words in the multimedia description information of similar multimedia interfaces, the multimedia description generation model needs to be trained first.

[0170] Figure 7 This is a flowchart illustrating a multimedia description generation model training method provided in an embodiment of this disclosure, as shown below. Figure 7 As shown in the embodiments of this disclosure, a method for training a multimedia description generation model is also provided, including:

[0171] Step 701: Obtain multiple similar sample multimedia interfaces and sample multimedia description information for each sample multimedia interface.

[0172] Step 702: Use the multimedia description generation model to identify words in the description information of the multimedia interface of each sample;

[0173] Step 703: Count the number of times words appear, filter out a preset number of target words based on the number of times words appear, and predict and output multimedia description information based on the target words;

[0174] Step 704: Determine the second loss function value based on the output multimedia description information and the corresponding sample multimedia description information, and train the multimedia description generation model based on the second loss function value. After training, the trained multimedia description generation model is obtained.

[0175] As an optional implementation, a preset number of target words are selected based on the frequency of their occurrence, and multimedia description information is predicted based on these target words, including:

[0176] Based on the frequency of word occurrence, a preset number of target words are selected in descending order of frequency.

[0177] Based on the preset number of target words selected, predict the corresponding semantic expressions, and select multimedia description information with the same semantic expressions from the multimedia description information database.

[0178] In some embodiments, multimedia description information includes multimedia title information and multimedia introduction information. Since multimedia title information and multimedia introduction information differ in length, content and other aspects, when training a multimedia description generation model, it is necessary to predict the output multimedia title information and multimedia introduction information based on different databases.

[0179] Specifically, the multimedia description generation model is used to count the frequency of words and filter out a preset number of target words. Based on the target words and the corresponding title database, the corresponding multimedia title information is predicted and output. At the same time, based on the words and the corresponding introduction database, the corresponding multimedia introduction information is predicted and output.

[0180] Using the predicted multimedia title information and multimedia description information, the multimedia description generation model is trained. The training ends when the model's output is determined to be the predicted multimedia title information and multimedia description information.

[0181] After training, the trained multimedia description generation model can identify words in the multimedia description information of similar multimedia interfaces and generate multimedia description information corresponding to the multimedia list screenshots based on the identified words.

[0182] Exemplary device

[0183] Having described the exemplary embodiments of this disclosure, the following references are made. Figure 8 This disclosure describes a multimedia interface generation apparatus based on exemplary embodiments. The multimedia interface generation apparatus provided in this disclosure and the multimedia interface generation method provided by the exemplary method of this invention belong to the same disclosure concept. Various embodiments of the multimedia interface generation method provided by the exemplary method can be applied to the multimedia interface generation apparatus in this embodiment for implementation, and will not be repeated here. Figure 8 As shown in the embodiments of this disclosure, a multimedia interface generation apparatus is also provided, including:

[0184] The acquisition unit 801 is used to acquire a screenshot of the multimedia list and perform multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot;

[0185] The determining unit 802 is used to determine similar multimedia interfaces of the multimedia list screenshot from the multimedia database based on the obtained multimedia information, and to obtain the cover information and multimedia description information of each similar multimedia interface.

[0186] The generation unit 803 is used to generate a target multimedia interface based on the cover information and multimedia description information of similar multimedia interfaces.

[0187] Optionally, the acquisition unit 801 is used to acquire a screenshot of the multimedia list and perform multimedia information recognition to obtain relevant information about the multimedia in the multimedia list screenshot, including:

[0188] Obtain a screenshot of the multimedia list and perform multimedia information area detection; then perform text content recognition on the detected multimedia information areas.

[0189] Based on the identified text content, match the target keywords in the multimedia keyword database;

[0190] Based on the target keywords, obtain relevant information about the multimedia in the multimedia list screenshot.

[0191] Optionally, the acquisition unit 801 is used to acquire a screenshot of the multimedia list and perform multimedia information region detection, and to perform text content recognition on the detected multimedia information region, including:

[0192] The multimedia list screenshot is divided into text regions by line to obtain multi-line text regions of the multimedia list screenshot;

[0193] Remove redundant information from each line of text in the multimedia list screenshot and determine the envelope region of each line of text in the multimedia screenshot;

[0194] By using a deep learning neural network to perform character recognition on the envelope region of each line of text, the text content corresponding to the envelope region of each line of text can be obtained.

[0195] Optionally, the determining unit 802 is used to determine similar multimedia interfaces to the multimedia list screenshot from the multimedia database based on the obtained multimedia-related information, including:

[0196] The relevant information of each multimedia extracted from the multimedia list screenshot is encoded and integrated to obtain the first encoded vector corresponding to the multimedia list screenshot;

[0197] Obtain the second encoding vector of multimedia-related information from different multimedia interfaces in the multimedia database, and calculate the similarity between the second encoding vector and the first encoding vector;

[0198] Based on the calculated similarity, similar multimedia interfaces to the multimedia list screenshots are identified from the multimedia database.

[0199] Optionally, the generation unit 803 is used to generate a target multimedia interface based on the cover information and multimedia description information of similar multimedia interfaces, including:

[0200] Identify the cover styles of similar multimedia interfaces, and determine the cover information corresponding to the multimedia list screenshot based on the cover styles of similar multimedia interfaces;

[0201] Identify words in the multimedia description information of similar multimedia interfaces, and determine the multimedia description information corresponding to the multimedia list screenshot based on the identified words;

[0202] Based on the multimedia information in the multimedia list screenshot, as well as the corresponding cover information and multimedia description information, generate the target multimedia interface.

[0203] Optionally, the generation unit 803 is used to identify the cover styles of various similar multimedia interfaces, including:

[0204] Feature extraction is performed on the cover information of similar multimedia interfaces;

[0205] Based on the extracted cover style feature vectors, the probability of different cover style classifications is predicted, and the cover style classification with the highest probability is determined as the cover style of the corresponding similar multimedia interface.

[0206] Optionally, the generation unit 803 is used to determine the cover information corresponding to the multimedia list screenshot based on the cover style of similar multimedia interfaces, including:

[0207] Calculate the frequency of occurrence / user preference for each cover style among similar multimedia interface cover styles;

[0208] Determine the target cover style corresponding to the multimedia list screenshot based on the frequency of occurrence / user preference;

[0209] Select the cover information that corresponds to the multimedia list screenshot from the similar multimedia interface that matches the target cover style.

[0210] Optionally, the generation unit 803 is used to identify words in the multimedia description information of each similar multimedia interface, and generate multimedia description information corresponding to the multimedia list screenshot based on the identified words, including:

[0211] Identify words in the multimedia description information of similar multimedia interfaces and count the number of times each word appears;

[0212] The system filters out a preset number of target words based on their frequency of occurrence, and then determines the multimedia description information corresponding to the multimedia list screenshot based on the target words.

[0213] Optionally, the generation unit 803 is used to filter out a preset number of target words based on the frequency of their occurrence, and to determine the multimedia description information corresponding to the multimedia list screenshot based on the target words, including:

[0214] Based on the frequency of word occurrence, a preset number of target words are selected in descending order of frequency.

[0215] Based on the preset number of target words selected, predict the corresponding semantic expression, and select target multimedia description information with the same semantic expression from the multimedia description information database. Then, determine the target multimedia description information as the multimedia description information corresponding to the multimedia list screenshot.

[0216] The following reference Figure 9 This describes a multimedia interface generation device 900 according to such an embodiment of the present disclosure. Figure 9 The multimedia interface generation device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments disclosed herein.

[0217] like Figure 9 As shown, the multimedia interface generation device 900 can be manifested in the form of a general-purpose computing device, such as a terminal device. The components of the multimedia interface generation device 900 may include, but are not limited to: at least one processor 901, at least one memory 902, and a bus 903 connecting different system components (including memory 902 and processor 901).

[0218] Bus 903 represents one or more of several bus architectures, including a memory bus or memory controller, peripheral bus, processor, or local bus using any of the various bus architectures.

[0219] The memory 902 may include a readable medium in the form of volatile memory, such as random access memory (RAM) 9021 and / or cache memory 9022, and may further include read-only memory (ROM) 9023.

[0220] The memory 902 may also include a program / utility 9025 having a set (at least one) of program modules 9024, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0221] The multimedia interface generation device 900 can also communicate with one or more external devices 904 (e.g., keyboard, pointing device, etc.), one or more devices that enable users to interact with the multimedia interface generation device 900, and / or any device that enables the multimedia interface generation device 900 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 905. Furthermore, the multimedia interface generation device 900 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 906. As shown, network adapter 906 communicates with other modules of the multimedia interface generation device 900 via bus 903. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with the multimedia interface generation device 900, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0222] The aforementioned memory 902 is used to store program instructions; the processor 901 is used to call the program instructions stored in the memory and execute the multimedia interface generation method provided in the above embodiment according to the obtained program instructions.

[0223] Exemplary program product

[0224] In some possible implementations, various aspects of this disclosure may also be implemented as a program product comprising program code that, when run on a terminal device, causes the terminal device to perform the steps of the modules in the multimedia interface generation apparatus according to various exemplary embodiments of this disclosure as described in the "Exemplary Methods" section of this specification.

[0225] For example, a terminal device can perform actions such as acquiring a screenshot of a multimedia list and recognizing multimedia information to obtain relevant information about the multimedia in the screenshot; based on the obtained multimedia information, it can determine similar multimedia interfaces from the multimedia database and obtain the cover information and multimedia description information of each similar multimedia interface; and generate a target multimedia interface based on the cover information and multimedia description information of the similar multimedia interfaces.

[0226] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0227] like Figure 10 As shown, a program product 1000 according to an embodiment of the present disclosure is described, which may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, the readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0228] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take many forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0229] The program code contained on the readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wired, fiber optic, RF, etc., or any suitable combination thereof.

[0230] Program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing devices can be connected to the user's computing device via any type of network—including local area networks (LANs) or wide area networks (WANs)—or can be connected to external computing devices (e.g., via the Internet using an Internet service provider).

[0231] It should be noted that although several modules or sub-modules of the system have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided and embodied by multiple modules.

[0232] Furthermore, although the operation of the modules of the system disclosed herein is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain operations may be omitted, multiple operations may be combined into one operation, and / or one operation may be decomposed into multiple operations.

[0233] While the spirit and principles of this disclosure have been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for convenience of expression. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A method of generating a multimedia interface, characterized by, The method comprises: obtaining the multimedia list screenshot and performing multimedia information recognition to obtain the related information of the multimedia in the multimedia list screenshot; the related information comprises the name, author and multimedia description information of each multimedia in the corresponding multimedia list screenshot; encoding and integrating the related information of each multimedia extracted from the multimedia list screenshot to obtain a first encoding vector corresponding to the multimedia list screenshot; obtaining a second encoding vector of the related information of the multimedia of different multimedia interfaces in the multimedia database, and calculating the similarity between the second encoding vector and the first encoding vector; determining the similar multimedia interface of the multimedia list screenshot from the multimedia database according to the calculated similarity, and obtaining the cover information and multimedia description information of each similar multimedia interface; generating a target multimedia interface according to the cover information and multimedia description information of the similar multimedia interface.

2. The method of claim 1, wherein, Obtaining the multimedia list screenshot and performing multimedia information recognition to obtain the related information of the multimedia in the multimedia list screenshot comprises: obtaining the multimedia list screenshot and performing multimedia information area detection, and performing text content recognition on the detected multimedia information area; matching a target keyword in a multimedia keyword library according to the recognized text content; obtaining the related information of the multimedia in the multimedia list screenshot according to the target keyword.

3. The method of claim 2, wherein, Obtaining the multimedia list screenshot and performing multimedia information area detection, and performing text content recognition on the detected multimedia information area comprises: performing text area segmentation on the multimedia list screenshot by row to obtain multiple rows of text areas of the multimedia list screenshot; removing the redundant information of each row of text areas of the multimedia list screenshot, and determining the envelope area of each row of text in the multimedia screenshot; performing text recognition on the envelope area of each row of text by using a deep learning neural network to obtain the text content corresponding to the envelope area of each row of text.

4. The method of claim 1, wherein, Generating a target multimedia interface according to the cover information and multimedia description information of the similar multimedia interface comprises: recognizing the cover style of each similar multimedia interface, and determining the cover information corresponding to the multimedia list screenshot according to the cover style of the similar multimedia interface; recognizing the words in the multimedia description information of each similar multimedia interface, and determining the multimedia description information corresponding to the multimedia list screenshot according to the recognized words; generating a target multimedia interface according to the related information of the multimedia in the multimedia list screenshot, and the corresponding cover information and multimedia description information.

5. The method of claim 4, wherein, Recognizing the cover style of each similar multimedia interface comprises: performing feature extraction on the cover information of each similar multimedia interface; predicting the probability of different cover style categories according to the extracted cover style feature vector, and determining the cover style category with the maximum probability as the cover style of the corresponding similar multimedia interface.

6. The method of claim 4, wherein, Determining the cover information corresponding to the multimedia list screenshot according to the cover style of the similar multimedia interface comprises: calculating the occurrence frequency / user preference degree of each cover style in the cover style of the similar multimedia interface; According to the occurrence frequency / user preference degree, a target cover style corresponding to the multimedia list screenshot is determined; From a similar multimedia interface corresponding to the target cover style, cover information corresponding to the multimedia list screenshot is selected.

7. The method of claim 4, wherein, The words in the multimedia description information of each similar multimedia interface are identified, and multimedia description information corresponding to the multimedia list screenshot is generated according to the identified words, including: The words in the multimedia description information of each similar multimedia interface are identified, and the occurrence frequency of the words is counted; According to the occurrence frequency of the words, a preset number of target words are screened out, and multimedia description information corresponding to the multimedia list screenshot is determined according to the target words.

8. The method of claim 7, wherein, According to the occurrence frequency of the words, a preset number of target words are screened out, and multimedia description information corresponding to the multimedia list screenshot is determined according to the target words, including: According to the occurrence frequency of the words, a preset number of target words are screened out in descending order of frequency; According to the screened out preset number of target words, the corresponding semantic expression is predicted, and target multimedia description information with the same semantic expression is screened out from a multimedia description information library, and the target multimedia description information is determined as the multimedia description information corresponding to the multimedia list screenshot.

9. A multimedia interface generating apparatus, characterized by comprising: Including: The acquisition unit is configured to acquire a multimedia list screenshot and perform multimedia information identification to obtain related information of the multimedia in the multimedia list screenshot; The related information includes the name, author and multimedia description information of each multimedia in the corresponding multimedia list screenshot; The determination unit is configured to encode and integrate the related information of each multimedia extracted from the multimedia list screenshot to obtain a first encoding vector corresponding to the multimedia list screenshot, acquire a second encoding vector of the related information of the multimedia of different multimedia interfaces in a multimedia database, and calculate the similarity between the second encoding vector and the first encoding vector; according to the calculated similarity, similar multimedia interfaces of the multimedia list screenshot are determined from the multimedia database, and cover information and multimedia description information of each similar multimedia interface are acquired; The generation unit is configured to generate a target multimedia interface according to the cover information and multimedia description information of the similar multimedia interfaces.

10. The apparatus of claim 9, wherein, The acquisition unit is configured to acquire a multimedia list screenshot and perform multimedia information identification to obtain related information of the multimedia in the multimedia list screenshot, including: The acquisition unit is configured to acquire a multimedia list screenshot and perform multimedia information identification to obtain related information of the multimedia in the multimedia list screenshot, including: Text content recognition is performed on the detected multimedia information region; According to the recognized text content, a target keyword is matched in a multimedia keyword library; 11. The apparatus of claim 10, wherein, According to the target keyword, the related information of the multimedia in the multimedia list screenshot is obtained. The acquisition unit is configured to acquire a multimedia list screenshot and perform multimedia information identification to obtain related information of the multimedia in the multimedia list screenshot, including: The multimedia list screenshot is segmented into multiple lines of text regions according to rows, and the multiple lines of text regions of the multimedia list screenshot are obtained; remove redundant information of each text area of the multimedia list screenshot, and determine an envelope area of each line of text in the multimedia list screenshot; perform text recognition on the envelope area of each line of text by using a deep learning neural network to obtain text content corresponding to the envelope area of each line of text.

12. The apparatus of claim 9, wherein, The generation unit is configured to generate a target multimedia interface according to cover information and multimedia description information of the similar multimedia interfaces, including: identifying a cover style of each similar multimedia interface, and determining cover information corresponding to the multimedia list screenshot according to the cover style of the similar multimedia interface; identifying a word in multimedia description information of each similar multimedia interface, and determining multimedia description information corresponding to the multimedia list screenshot according to the identified word; generating a target multimedia interface according to related information of multimedia in the multimedia list screenshot, and the corresponding cover information and multimedia description information.

13. The apparatus of claim 12, wherein, The generation unit is configured to identify a cover style of each similar multimedia interface, including: performing feature extraction on cover information of each similar multimedia interface; predicting a probability of different cover style categories according to the extracted cover style feature vector, and determining a cover style category with the maximum probability as the cover style of the corresponding similar multimedia interface.

14. The apparatus of claim 12, wherein, The generation unit is configured to determine cover information corresponding to the multimedia list screenshot according to the cover style of the similar multimedia interface, including: calculating a number of occurrences / user preference degrees corresponding to each cover style in the cover style of the similar multimedia interface; determining a target cover style corresponding to the multimedia list screenshot according to the number of occurrences / user preference degrees; selecting cover information corresponding to the multimedia list screenshot from the similar multimedia interface corresponding to the target cover style.

15. The apparatus of claim 12, wherein, The generation unit is configured to identify a word in multimedia description information of each similar multimedia interface, and determine multimedia description information corresponding to the multimedia list screenshot according to the identified word, including: identifying a word in multimedia description information of each similar multimedia interface, and counting a number of occurrences of the word; screening a preset number of target words according to the number of occurrences of the word, and determining multimedia description information corresponding to the multimedia list screenshot according to the target words.

16. The apparatus of claim 15, wherein, The generation unit is configured to screen a preset number of target words according to the number of occurrences of the word, and determine multimedia description information corresponding to the multimedia list screenshot according to the target words, including: screening a preset number of target words in a descending order of the number of occurrences according to the number of occurrences of the word; predicting a corresponding semantic expression according to the screened preset number of target words, and screening target multimedia description information with the same semantic expression from a multimedia description information database, and determining the target multimedia description information as the multimedia description information corresponding to the multimedia list screenshot.

17. An electronic device, comprising: The device includes: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the method of any one of claims 1 to 8.

18. A computer-readable non-transitory storage medium, comprising: comprising computer readable instructions which, when read and executed by a computer, cause the computer to perform the method of any one of claims 1 to 8.

Images