A live broadcast reminder method and related device

Abstract

The present application discloses a live broadcast reminder method and related devices, the method comprising: obtaining additional feature information of each program in the live broadcast; the additional feature information is used to supplement the program list information of the program; whenever a live broadcast screen of the live broadcast is obtained, feature extraction is performed on the live broadcast screen to obtain a target feature, and target additional feature information matching the target feature in the additional feature information of each program is determined, and the program list information corresponding to the target additional feature information is used as the recognition result of the live broadcast screen; target program list information is determined according to multiple recognition results determined by multiple consecutive frames of live broadcast screen; the target program list information represents the program currently being played in the live broadcast; at least based on the target program list information, prompt information is output to a target user; the prompt information is used to remind the target user to watch the live broadcast.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a live broadcast reminder method and related devices. Background Art

[0002] In the current Internet live broadcast industry, the diversity and real-time nature of live broadcast content do provide users with a wealth of viewing options. However, as the length of live broadcasts increases, especially for those long live broadcasts that can easily last for several hours, users' interest and attention in watching tend to gradually weaken. This phenomenon is particularly evident among hot-spot chasers and non-experienced users. Hot-spot chasers are often only interested in certain specific topics or guest appearance times in the live broadcast, while non-experienced users may only want to watch some of the content because they do not have a deep understanding of the live broadcast theme. Therefore, during the live broadcast, when users find that they are not interested in the content being played, they often choose to leave the live broadcast page and browse other content in the application (APP), or even exit the APP directly. This behavior not only damages the user's viewing experience, but also leads to a significant decrease in the live broadcast's activity and interactivity.

[0003] In order to attract and retain users, the traditional approach is to provide a reservation function before the live broadcast begins. Users can use the reservation function to set their attention to a certain live broadcast in advance. When the live broadcast officially starts, the system will send a message reminder to the users who have made the reservation. This mechanism ensures that users can know the start of the live broadcast in time to a certain extent, thereby increasing the chances of users watching the live broadcast.

[0004] However, although the reservation function has improved user engagement to a certain extent, it can only trigger a reminder once when the live broadcast starts. When the live broadcast content does not contain the part they are not interested in, users are still likely to leave the live broadcast page quickly. Summary of the invention

[0005] In view of the above problems, the present application provides a live broadcast reminder method and related devices to achieve the purpose of improving user participation and interactive experience. The specific scheme is as follows:

[0006] The first aspect of the present application provides a live broadcast reminder method, comprising:

[0007] Acquire additional characteristic information of each program in the live broadcast; the additional characteristic information is used to supplement the program list information of the program;

[0008] Whenever a live screen of the live broadcast is obtained, feature extraction is performed on the live screen to obtain a target feature, and target additional feature information matching the target feature in the additional feature information of each program is determined, and program list information corresponding to the target additional feature information is used as a recognition result of the live screen;

[0009] Determine target program list information according to the plurality of recognition results determined by the continuous plurality of frames of live broadcast images; the target program list information represents the program currently being played in the live broadcast;

[0010] At least based on the target program list information, prompt information is output to the target user; the prompt information is used to remind the target user to watch the live broadcast.

[0011] In a possible implementation, obtaining additional feature information of each program in a live broadcast includes:

[0012] Obtain the program name of each program in the live broadcast and the names of the program participants;

[0013] Based on the program name, obtaining detailed text information of the program;

[0014] Based on the names of the personnel, portraits of the program participants are obtained.

[0015] In a possible implementation, feature extraction is performed on the live screen to obtain a target feature, and target additional feature information matching the target feature in the additional feature information of each program is determined, and program list information corresponding to the target additional feature information is used as the recognition result of the live screen, including:

[0016] Extracting text features and image features from the live broadcast screen to obtain target text features and target image features;

[0017] Determining target detailed text information matching the target text feature in the detailed text information of each of the programs;

[0018] Determine a target portrait that matches the target portrait feature among the portraits of the program participants of each of the programs;

[0019] Taking the program list information corresponding to the target detailed text information as the first recognition result of the image frame;

[0020] The program list information corresponding to the target portrait is used as the second recognition result of the image frame.

[0021] In a possible implementation, determining the target program list information according to the multiple recognition results determined from the continuous multiple frames of live broadcast images includes:

[0022] Determine a first frame number of image frames whose first recognition result is target program list information and a second frame number of image frames whose second recognition result is target program list information in a continuous plurality of frames of live broadcast images;

[0023] Based on a first weight factor corresponding to the first recognition result and a second weight factor corresponding to the second recognition result, weighted processing is performed on the first number of frames and the second number of frames to obtain a weighted number of frames; the first weight factor represents the importance of the first recognition result in determining the program currently being played in the live broadcast; the second weight factor represents the importance of the second recognition result in determining the program currently being played in the live broadcast;

[0024] Determining the probability corresponding to the target program list information based on the weighted frame number;

[0025] If the probability is greater than a set probability threshold, it is determined that the target program list information belongs to the program currently being played in the live broadcast.

[0026] In a possible implementation, outputting prompt information to a target user based at least on the target program list information includes:

[0027] Obtaining the popularity parameter and audio content corresponding to the currently playing program;

[0028] Inputting the target program list information, the popularity parameter and the audio content into a large language model to obtain prompt information determined by the large language model;

[0029] Output the prompt information to the target user.

[0030] In a possible implementation, outputting the prompt information to the target user includes:

[0031] Based on the heat parameter, determine the heat value change rate corresponding to the currently playing program.

[0032] The target program list information, the prompt information and the popularity value change rate are sent to the notification service module, so that the notification service module determines the target user and the target push method according to the user feature library, the program feature library and the popularity value, and pushes the prompt information to the target user based on the target push method.

[0033] In another aspect of the present application, a live broadcast reminder device is provided, comprising:

[0034] An acquisition module, used to acquire additional characteristic information of each program in the live broadcast; the additional characteristic information is used to supplement the program list information of the program;

[0035] a feature extraction module, for extracting features from the live screen whenever a live screen of the live broadcast is obtained, obtaining target features, determining target additional feature information that matches the target feature in the additional feature information of each program, and using the program list information corresponding to the target additional feature information as the recognition result of the live screen;

[0036] A determination module, configured to determine target program list information according to a plurality of recognition results determined from a plurality of consecutive frames of live broadcast images; the target program list information represents a program currently being played in the live broadcast;

[0037] A prompt module is used to output prompt information to a target user based at least on the target program list information; the prompt information is used to remind the target user to watch the live broadcast.

[0038] In a third aspect of the present application, a computer program product is provided, comprising computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the live broadcast reminder method as described in any one of the above.

[0039] In a fourth aspect of the present application, an electronic device is provided, including:

[0040] The memory is used to store computer programs;

[0041] The processor is used to execute the computer program so that the electronic device can implement the live broadcast reminder method as described in any one of the above items.

[0042] In the fifth aspect of the present application, a computer storage medium is provided, characterized in that the storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the live broadcast reminder method as described in any one of the above.

[0043] In the present application, additional feature information of each program in the live broadcast is obtained, and the additional feature information is used to supplement the program list information of the program. Whenever a live broadcast screen is obtained, feature extraction is performed on the live broadcast screen to obtain the target feature, and the target additional feature information matching the target feature in the additional feature information of each program is determined, and the program list information corresponding to the target additional feature information is used as the recognition result of the live broadcast screen. According to multiple recognition results determined based on multiple consecutive frames of live broadcast screen, the currently playing program in the live broadcast can be accurately identified, and prompt information is output to the target user based on the target program list information. The prompt information is used to remind the target user to watch the live broadcast, provide users with timely and accurate information feedback, attract users to participate more in live broadcast activities, and improve user participation and interactive experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0045] Figure 1 A flowchart of a live broadcast reminder method provided in Example 1 of the present application;

[0046] Figure 2 A flowchart of a live broadcast reminder method provided in Example 5 of the present application;

[0047] Figure 3 A flowchart of a live broadcast reminder method provided in Example 6 of the present application;

[0048] Figure 4 A schematic diagram of an implementation scenario of a live broadcast reminder method provided in this application;

[0049] Figure 5 A schematic diagram of another implementation scenario of a live broadcast reminder method provided in this application;

[0050] Figure 6 A schematic diagram of another implementation scenario of a live broadcast reminder method provided in this application;

[0051] Figure 7 A schematic diagram of another implementation scenario of a live broadcast reminder method provided in this application;

[0052] Figure 8 A structural schematic diagram of a live broadcast reminder device provided in this application. DETAILED DESCRIPTION

[0053] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0054] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0055] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0056] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0057] Reference Figure 1 , is a flow chart of a live broadcast reminder method provided in Example 1 of the present application, such as Figure 1 As shown, the method may include but is not limited to the following steps:

[0058] Step S101, obtaining additional characteristic information of each program in the live broadcast; the additional characteristic information is used to supplement the program list information of the program.

[0059] In this embodiment, the live broadcast management backend can operate metadata of the live broadcast. For example, as shown in Table 1, the metadata may include a live broadcast ID, a live broadcast name, a live broadcast introduction, a live broadcast stream address, live broadcast program list information (in text or picture form), live broadcast start and end time, etc. In Table 1, "YY1", "YY2", "YY3", "YY4", "YY5", and "YY6" are program names.

[0060] Table 1

[0061]

[0062] Operators can add, modify, view, delete and query live broadcast metadata.

[0063] The additional characteristic information of the program can be obtained based on the program list information in the metadata. The additional characteristic information of the program is used to supplement the program list information of the program and is associated with the program list information. In this embodiment, a correspondence between the program list information and the additional characteristic information can be constructed and stored. The storage structure can be referred to in Table 2. As shown in Table 2, the live broadcast name, program list information (such as the program name and the names of the program participants) and the additional characteristic information can be associated and stored.

[0064] Table 2

[0065]

[0066] Step S102: Whenever a live broadcast screen is obtained, feature extraction is performed on the live broadcast screen to obtain target features, and target additional feature information matching the target features in the additional feature information of each program is determined, and the program list information corresponding to the target additional feature information is used as the recognition result of the live broadcast screen.

[0067] In this embodiment, the live stream address can be obtained from the metadata of the live management background, and the live screen can be read based on the live stream address. For example, the live screen can be read based on the live stream address based on the audio and video codec tool FFmpeg.

[0068] In this embodiment, the additional feature information of each program may be compared with the target feature, and the additional feature information matching the target feature may be determined from the additional feature information of each program, and the additional feature information matching the target feature may be used as the target additional feature information.

[0069] Based on the corresponding relationship stored in step S101, the program list information corresponding to the target additional characteristic information can be determined. For example, if the target additional characteristic information is aa, based on the corresponding relationship, it can be determined that the corresponding program list information includes: "YY1" and Li X.

[0070] Step S103, determining target program list information according to the multiple recognition results determined by the continuous multiple frames of live broadcast images; the target program list information represents the program currently being played in the live broadcast.

[0071] As described in step S102, a recognition result is determined for each frame of the live broadcast picture, and therefore, the plurality of recognition results are recognition results corresponding to each frame of the live broadcast picture in the plurality of consecutive frames of the live broadcast picture.

[0072] In this embodiment, a time window may be set, and target program list information may be determined based on multiple recognition results determined from multiple consecutive frames of live broadcast images within the time window.

[0073] The length of the time window can be adjusted according to the characteristics of the live content and user preferences, and is not limited in this application. For example, if the time window is set to 10 seconds and the live frame rate is 25 frames / second, the target program list information can be determined based on 250 recognition results determined by 10 seconds × 25 frames / second, that is, 250 frames of live pictures.

[0074] Step S104: output prompt information to the target user based at least on the target program list information; the prompt information is used to remind the target user to watch the live broadcast.

[0075] In this embodiment, additional feature information of each program in the live broadcast is obtained, and the additional feature information is used to supplement the program list information of the program. Whenever a live broadcast screen is obtained, feature extraction is performed on the live broadcast screen to obtain the target feature, and the target additional feature information matching the target feature in the additional feature information of each program is determined, and the program list information corresponding to the target additional feature information is used as the recognition result of the live broadcast screen. According to multiple recognition results determined based on multiple consecutive frames of live broadcast screen, the currently playing program in the live broadcast can be accurately identified, and prompt information is output to the target user based on the target program list information. The prompt information is used to remind the target user to watch the live broadcast, provide users with timely and accurate information feedback, attract users to participate more in live broadcast activities, and improve user participation and interactive experience.

[0076] As another optional embodiment of the present application, a live broadcast reminder method is provided in Embodiment 2 of the present application. This embodiment is mainly an implementation method of step S101 in Embodiment 1. Step S11 may include but is not limited to the following steps:

[0077] Step S1011: Obtain the program name of each program in the live broadcast and the names of the program participants.

[0078] As described in Example 1, the program name of each program in the live broadcast and the names of the program participants can be obtained from the metadata of the live broadcast management background.

[0079] Step S1011: Based on the program name, obtain detailed text information of the program.

[0080] In this embodiment, detailed text information of the program can be captured from a third-party website based on the program name. For example, if the program name is the song "YY1", the lyrics information of the song "YY1" can be captured from a third-party website.

[0081] Step S1011: Based on the names of the personnel, obtain portraits of the program participants.

[0082] In this embodiment, the portraits of program participants can be captured from third-party websites based on the names of the participants. For example, if the name of the participant is Li X, the portrait of Li X can be captured from the third-party website.

[0083] The detailed text information of the program and the portraits of the program participants are additional characteristic information of the program.

[0084] In this embodiment, the correspondence between the program list information, the detailed text information of the program, and the portraits of the program participants can be stored. For example, the storage structure can refer to Table 3. As shown in Table 3, the live broadcast name, the program list information (such as the program name and the names of the program participants), the lyrics information, and the portraits of the program participants can be stored in association.

[0085] Table 3

[0086]

[0087] In this embodiment, by obtaining the detailed text information of the program and the portraits of the program participants, whenever a live broadcast screen is obtained, feature extraction is performed on the live broadcast screen to obtain target features, and target additional feature information matching the target features in the detailed text information of each program and the portraits of the program participants is determined, and the program list information corresponding to the target additional feature information is used as the recognition result of the live broadcast screen. According to multiple recognition results determined by continuous multiple frames of live broadcast screen, the currently playing program in the evening party live broadcast (such as New Year's Eve concert, variety show) can be accurately identified, and prompt information is output to the target user based on the target program list information. The prompt information is used to remind the target user to watch the live broadcast, provide timely and accurate information feedback to the user, attract users to participate more in the live broadcast activities, and improve the user's participation and interactive experience.

[0088] As another optional embodiment of the present application, a live broadcast reminder method is provided in Embodiment 3 of the present application. This embodiment is mainly an implementation method of step S102 in Embodiment 2, performing feature extraction on the live broadcast screen to obtain target features, and determining target additional feature information matching the target features in the additional feature information of each program, and using the program list information corresponding to the target additional feature information as the recognition result of the live broadcast screen, which may include but is not limited to the following steps:

[0089] Step S1021: extract text features and image features from the live broadcast image to obtain target text features and target image features.

[0090] In this embodiment, OCR (Optical Character Recognition) technology can be used to convert text information in the live broadcast screen into an editable text format, and regular expressions are used to detect and extract target text features in the text recognized by OCR.

[0091] Feature extraction technology can be used to identify whether target portrait features exist in the live broadcast image.

[0092] Step S1022: Determine the target detailed text information that matches the target text feature in the detailed text information of each of the programs.

[0093] In this embodiment, the detailed text information of each program may be compared with the target text feature, and the detailed text information with the highest similarity to the target text feature may be determined from the detailed text information of each program as the target detailed text information.

[0094] Step S1023: Determine a target portrait that matches the target portrait feature among the portraits of the program participants of each of the programs.

[0095] In this embodiment, the portraits of the program participants of each of the programs may be compared with the target portrait features, and the portrait with the highest similarity to the target portrait features may be determined as the target portrait.

[0096] Step S1024: Use the program list information corresponding to the target detailed text information as the first recognition result of the image frame.

[0097] Based on the corresponding relationship described in the foregoing embodiment, the program list information corresponding to the target detailed text information can be determined.

[0098] Step S1025: Use the program list information corresponding to the target portrait as the second recognition result of the image frame.

[0099] Based on the corresponding relationship described in the foregoing embodiment, the program list information corresponding to the target portrait can be determined.

[0100] Step S103 may include but is not limited to:

[0101] Step S1031, determining target program list information according to a plurality of first recognition results and a plurality of second recognition results determined from a plurality of consecutive frames of live broadcast images; the target program list information represents a program currently being played in the live broadcast.

[0102] For example, if the target detailed text information is the target lyrics information, and the target portrait is the portrait of the program participant, the multiple first recognition results and the multiple second recognition results determined by multiple consecutive frames of live broadcast images can be shown in Table 4. As shown in Table 4, the frame number indicates the frame of the live broadcast image within the time window, the first recognition result is the program name corresponding to the target lyrics information, and the second recognition result is the name of the program participant corresponding to the target portrait.

[0103] Table 4

[0104] Frame rate Current time First recognition result Second recognition result 1 2024-8-7 17:59:00 000 none none 2 2024-8-7 17:59:00 040 《YY1》 Li X 3 2024-8-7 17:59:00 080 《YY1》 none 4 2024-8-7 17:59:00 120 none Li X ......

[0105] In this embodiment, the target program list information is determined based on multiple first recognition results and multiple second recognition results determined based on multiple frames of continuous live broadcast images, which can ensure the accuracy of the target program list information and further improve the accuracy of the currently playing programs in the identified live broadcast. Prompt information is output to the target user based on the target program list information, and the prompt information is used to remind the target user to watch the live broadcast, provide the user with timely and accurate information feedback, attract the user to participate more in the live broadcast activities, and improve the user's participation and interactive experience.

[0106] As another optional embodiment of the present application, a live broadcast reminder method is provided in Embodiment 4 of the present application. This embodiment is mainly an implementation method of step S1031 in Embodiment 3. Step S1031 may include but is not limited to:

[0107] Step S10311, determining a first frame number of image frames whose first recognition result is target program list information and a second frame number of image frames whose second recognition result is target program list information in a continuous plurality of frames of live broadcast images.

[0108] Step S10312: based on a first weight factor corresponding to the first recognition result and a second weight factor corresponding to the second recognition result, weighted processing is performed on the first number of frames and the second number of frames to obtain a weighted number of frames; the first weight factor represents the importance of the first recognition result in determining the program currently being played in the live broadcast; the second weight factor represents the importance of the second recognition result in determining the program currently being played in the live broadcast.

[0109] The first weight factor and the second weight factor can be set by different live broadcasts, which is not limited in this application. For example, the first weight factor can be set to 0.7, and the second weight factor can be set to 0.3.

[0110] Step S10313: Determine the probability corresponding to the target program list information based on the weighted frame number.

[0111] In this embodiment, the probability corresponding to the target program list information can be determined by the following relationship:

[0112] C=(fn1*L1+fn2*S2) / F

[0113] C represents the probability corresponding to the target program list information, fn1 represents the first frame number, L1 represents the first weight factor, fn2 represents the second frame number, S2 represents the second weight factor, and F = the length of the time window × the frame rate of the live broadcast (i.e., F represents the total number of consecutive frames).

[0114] Step S10314: If the probability is greater than the set probability threshold, determine that the target program list information belongs to the program currently being played in the live broadcast.

[0115] The probability threshold may be set as required and is not limited in the present application. For example, the probability threshold may be set to 90%.

[0116] As another optional embodiment of the present application, refer to Figure 2 , is a flowchart of a live broadcast reminder method provided in Example 5 of the present application. This embodiment is mainly an implementation method of step S104 in Example 1. Figure 2 As shown, step S104 may include but is not limited to:

[0117] Step S1041: Obtain the popularity parameter and audio content corresponding to the currently playing program.

[0118] In this embodiment, at least one of the number of online users, the amount of barrage sent, and the value of gifts corresponding to the currently playing program can be obtained, and at least one of the number of online users, the amount of barrage sent, and the value of gifts can be used as a heat parameter.

[0119] In this embodiment, the program introduction and interactive content during the performance of the currently playing program can be obtained and used as audio content.

[0120] Step S1042: input the target program list information, the popularity parameter and the audio content into a large language model to obtain prompt information determined by the large language model.

[0121] In this embodiment, the target program list information, the popularity parameter, the audio content, and the prompt word may be input into a large language model to obtain prompt information determined by the large language model.

[0122] Step S1043: output the prompt information to the target user.

[0123] In this embodiment, by obtaining the popularity parameters of the currently playing program (such as the number of online users, the amount of barrage sent, and the value of gifts), the large language model can understand the popularity of the program and the audience's response in real time. Combined with audio content (such as program introductions and interactive content during performances), the large language model can further understand the specific content of the program and the real-time feedback from the audience. Based on the target program list information, the popularity parameters, and the audio content, the large language model can generate personalized and relevant prompt information for the target user, which helps to improve the overall activity and participation of the program.

[0124] As another optional embodiment of the present application, refer to Figure 3, is a flowchart of a live broadcast reminder method provided in Example 6 of the present application. This embodiment is mainly an implementation method of step S1043 in Example 5. Figure 3 As shown, step S1043 may include but is not limited to:

[0125] Step S10431: Based on the heat parameter, determine the heat value change rate corresponding to the currently playing program.

[0126] In this embodiment, the heat parameter may be converted in proportion using a formula to obtain a heat value corresponding to the currently playing program.

[0127] In this embodiment, for the currently playing program, its real-time popularity value is continuously calculated and updated during the playing process. At the same time, the highest popularity value reached by the program during the playing period is recorded. This highest popularity value represents the peak popularity of the program among the audience.

[0128] In the case of multiple programs, for each program that has been played, its highest heat value is also recorded. Then, the highest heat value of the played program can be compared with the highest heat value of the currently played program to determine the heat value change rate corresponding to the currently played program.

[0129] After obtaining the heat value change rate, the target program list information, the prompt information and the heat value may be stored in association. For example, as shown in Table 5, the program name, the name of the program participant, the start time, the heat value change rate and the prompt information may be stored in association. The start time may be determined in the process of determining the target program list information.

[0130] Table 5

[0131]

[0132] Step S10432: Send the target program list information, the prompt information and the popularity value to the notification service module, so that the notification service module determines the target user and the target push method according to the user feature library, the program feature library and the popularity value, and pushes the prompt information to the target user based on the target push method.

[0133] In this embodiment, the program name, the names of program participants, the start time, the heat value change rate and the prompt information can be sent to the notification service module.

[0134] The notification service module can classify users into groups according to their viewing habits to obtain a user feature library. For example, as shown in Table 6, the user feature library may include: a correspondence between user device ID, interested program types, and interested persons.

[0135] Table 6

[0136]

[0137] The notification service module can classify the program content by characteristics to obtain a program characteristic library. For example, as shown in Table 7, the program characteristic library can include the corresponding relationship between program names and program types.

[0138] Table 7

[0139] Program Name Program Type 《YY1》 National, popular YY2 Folk, Quiet ......

[0140] For example, after receiving the target program list information, the prompt information and the popularity value, the notification service module finds that the popularity value of the currently playing program changes by more than 50%, and considers it as a live broadcast of exciting content. It can send a prompt information to the user who is using the APP but not watching the live broadcast through a long connection. Figure 4 As shown, the prompt information can be pushed to the target user in the form of a pop-up box. After the user clicks the pop-up box, he can enter the live broadcast room and use the live broadcast replay function to locate the start time of the program and watch the live hot programs.

[0141] After receiving the target program list information, the prompt information and the popularity value, the notification service module can recall the program participants in the user feature database by their names, such as Figure 5 As shown, an off-site message is pushed to the user devices that are interested in the program participants. The user can open the off-site push message, click to enter the live broadcast room, and use the live broadcast playback function to locate the start time of the program to watch the live program.

[0142] After receiving the target program list information, the prompt information and the heat value, the notification service module can obtain the program type from the program feature library and recall it in the user feature library, such as Figure 6 As shown, an internal message is sent to the user device that is interested in the program type. The user can open the internal message function at any time, click on the program of interest, enter the live broadcast room, and use the live broadcast playback function to locate the start time of the program to watch the live program.

[0143] In this embodiment, the notification service module builds a user feature library, which helps the notification service module to more accurately identify the user group interested in a specific program. In addition, by building a program feature library, it helps the notification service module understand the characteristics and appeal of the program, so as to match it with the user's interest points. On this basis, the notification service module determines the target user and the target push method based on the user feature library, the program feature library and the heat value, and pushes the prompt information to the target user based on the target push method, which can reduce invalid push and resource waste, and improve the efficiency and effect of push.

[0144] Next, we will use a live broadcast of a singer competition variety show "Who is the Singing God" to illustrate the live broadcast reminder method. For example, assuming that the current time is August 9, 2024, and there is a live broadcast of a singer competition variety show "Who is the Singing God" starting at 20:00 tonight, the company's live broadcast operator configures the live broadcast information in the live broadcast management background as shown in the following table:

[0145]

[0146]

[0147] The live broadcast intelligent analysis module can obtain the names of the singers corresponding to each performance program in the live broadcast from the live broadcast metadata (an implementation method of the names of the program participants), and crawl the singer's portrait (i.e., an implementation method of the portrait of the program participants) and the lyrics information of the song (i.e., an implementation method of the text detailed information of the program) from the third-party website. The corresponding relationship between the program list information, the singer's portrait and the song's lyrics information is structured and stored, and the storage structure is as follows:

[0148]

[0149]

[0150] FFmpeg is used to read the live content from the live stream address and submit it to OCR for recognition. The live stream frame rate is 25, so there is one frame every 40 milliseconds.

[0151] The results are as follows:

[0152] Frame rate Current time Lyrics text recognition results Performer portrait recognition results 1 ...... ...... ....... 2 2024-8-9 20:15:00 000 - Lin XX 3 2024-8-9 20:15:00 080 ZZ1 none 4 2024-8-7 20:15:00 120 ZZ1 Lin XX 5 2024-8-7 20:15:00 160 ZZ1 Lin XX 6 2024-8-7 20:15:00 180 ZZ1 Lin XX 7 ...... ....... ....... 8 2024-8-7 20:20:00 120 ZZ1 Lin XX 9 2024-8-7 20:20:00 160 - Lin XX 10 2024-8-7 20:20:00 180 - - 11 2024-8-7 20:20:00 220 - - 12 ...... ....... ....... 13 2024-8-7 21:30:00 220 - That X 14 2024-8-7 21:30:00 260 ZZ7 - 15 2024-8-7 21:30:00 300 ZZ7 That X 16 2024-8-7 21:30:00 340 ZZ7 That X 17 2024-8-7 21:30:00 380 ZZ7 That X 18 ....... ....... ....... 19 2024-8-7 21:35:00 460 ZZ7 That X 20 2024-8-7 21:35:00 500 - That X 21 2024-8-7 21:35:00 540 - - 22 2024-8-7 21:35:00 580 - -

[0153] Analyze the results of OCR recognition, and calculate the programs and singers performing in each time period through formulas; calculate the popularity value of the program by proportionally converting the number of online users, the amount of barrage sent, and the value of gifts; compare the highest popularity values ​​of each program to get the rate of change of the popularity value; use FFmpeg to read the audio content, combine the relevant information of the live broadcast with the program name and performer information, assemble the prompt words, and submit them to the multimodal large language model to get the prompt information.

[0154] The results are as follows:

[0155]

[0156] like Figure 7 As shown, the push methods that can be selected by the notification service module may include: long connection push, in-station message and external station message.

[0157] The notification service module found that at 20:15:00 080 on 2024-8-9, Lin XX's "ZZ1" caused a 52% increase in popularity, and considered it to be a wonderful content of the live broadcast. It will send notification content to users who are using the APP but not watching the live broadcast through a long connection. After the user clicks the pop-up box, he can enter the live broadcast room and use the live broadcast playback function to locate the start time of "ZZ1" to watch the program.

[0158] The user feature library maintained by the notification service module is as follows:

[0159]

[0160]

[0161] At 20:45:00 220 on 2024-8-7, XX performed "ZZ4". The notification service module recalled the singer XX in the user feature database and pushed an off-site message to the user device 1111 who was interested in the performer. The user opened the off-site push message, clicked to enter the live broadcast room, and used the live broadcast playback function to locate the start time of "ZZ4" to watch the live broadcast program.

[0162] The notification service module includes a program feature library that classifies program content by features.

[0163] Program Name Program Type ZZ1 Popularity ZZ2 love song ZZ3 Rock ZZ4 RAP ZZ5 Skill ZZ6 Ethereal ZZ7 Nostalgia

[0164] At 20:35:00 220 on 2024-8-7, Liang X performed "ZZ3". The notification service recalled the program type "rock" in the user feature library and sent an internal message to the user device 2222 who was interested in rock. The user opened the internal message function, clicked on the program he was interested in, entered the live broadcast room, and used the live broadcast playback function to locate the time point of "ZZ3" to watch the live broadcast program.

[0165] Next, the live broadcast reminder device provided by the present application is introduced. The live broadcast reminder device introduced below and the live broadcast reminder method introduced above can refer to each other.

[0166] Reference Figure 8 The live broadcast reminder device includes: an acquisition module 100, a feature extraction module 200, a determination module 300 and a prompt module 400.

[0167] The acquisition module 100 is used to acquire additional characteristic information of each program in the live broadcast; the additional characteristic information is used to supplement the program list information of the program.

[0168] The feature extraction module 200 is used to extract features from the live broadcast screen whenever the live broadcast screen is obtained, obtain target features, and determine target additional feature information that matches the target feature in the additional feature information of each program, and use the program list information corresponding to the target additional feature information as the recognition result of the live broadcast screen.

[0169] The determination module 300 is used to determine target program list information according to the multiple recognition results determined by the continuous multiple frames of live broadcast pictures; the target program list information represents the program currently being played in the live broadcast.

[0170] The prompt module 400 is used to output prompt information to the target user based on at least the target program list information; the prompt information is used to remind the target user to watch the live broadcast.

[0171] The acquisition module 100 may be specifically used for:

[0172] Obtain the program name of each program in the live broadcast and the names of the program participants;

[0173] Based on the program name, obtaining detailed text information of the program;

[0174] Based on the names of the personnel, portraits of the program participants are obtained.

[0175] The feature extraction module 200 can be specifically used for:

[0176] Extracting text features and image features from the live broadcast screen to obtain target text features and target image features;

[0177] Determining target detailed text information matching the target text feature in the detailed text information of each of the programs;

[0178] Determine a target portrait that matches the target portrait feature among the portraits of the program participants of each of the programs;

[0179] Taking the program list information corresponding to the target detailed text information as the first recognition result of the image frame;

[0180] The program list information corresponding to the target portrait is used as the second recognition result of the image frame.

[0181] The determination module 300 may be specifically used for:

[0182] Determine a first frame number of image frames whose first recognition result is target program list information and a second frame number of image frames whose second recognition result is target program list information in a continuous plurality of frames of live broadcast images;

[0183] Based on a first weight factor corresponding to the first recognition result and a second weight factor corresponding to the second recognition result, weighted processing is performed on the first number of frames and the second number of frames to obtain a weighted number of frames; the first weight factor represents the importance of the first recognition result in determining the program currently being played in the live broadcast; the second weight factor represents the importance of the second recognition result in determining the program currently being played in the live broadcast;

[0184] Determining the probability corresponding to the target program list information based on the weighted frame number;

[0185] If the probability is greater than a set probability threshold, it is determined that the target program list information belongs to the program currently being played in the live broadcast.

[0186] The prompt module 400 may be specifically used for:

[0187] Obtaining the popularity parameter and audio content corresponding to the currently playing program;

[0188] Inputting the target program list information, the popularity parameter and the audio content into a large language model to obtain prompt information determined by the large language model;

[0189] Output the prompt information to the target user.

[0190] The prompt module 400 outputs the prompt information to the target user, which may include:

[0191] Based on the heat parameter, determine the heat value change rate corresponding to the currently playing program.

[0192] The target program list information, the prompt information and the popularity value change rate are sent to the notification service module, so that the notification service module determines the target user and the target push method according to the user feature library, the program feature library and the popularity value, and pushes the prompt information to the target user based on the target push method.

[0193] In another embodiment of the present application, a computer program product is provided, characterized in that it includes computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the live broadcast reminder method introduced in any one of embodiments 1 to 6.

[0194] In another embodiment of the present application, an electronic device is provided, including:

[0195] The memory is used to store computer programs;

[0196] The processor is used to execute the computer program so that the electronic device can implement the live broadcast reminder method introduced in any one of embodiments 1 to 6.

[0197] In another embodiment of the present application, a computer storage medium is provided, wherein the storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement the live broadcast reminder method described in any one of Embodiments 1 to 6.

[0198] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0199] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0200] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0201] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A live broadcast reminder method, characterized in that: include: Get additional feature information of each program in the live broadcast; The additional characteristic information is used to supplement the program list information of the program; Whenever a live screen of the live broadcast is obtained, feature extraction is performed on the live screen to obtain a target feature, and target additional feature information matching the target feature in the additional feature information of each program is determined, and program list information corresponding to the target additional feature information is used as a recognition result of the live screen; Determine the target program list information according to the multiple recognition results determined by the continuous multiple frames of live broadcast images; The target program list information represents the program currently being played in the live broadcast; At least based on the target program list information, prompt information is output to the target user; the prompt information is used to remind the target user to watch the live broadcast.

2. The live broadcast reminder method according to claim 1, characterized in that: Get additional feature information of each program in the live broadcast, including: Obtain the program name of each program in the live broadcast and the names of the program participants; Based on the program name, obtaining detailed text information of the program; Based on the names of the personnel, portraits of the program participants are obtained.

3. The live broadcast reminder method according to claim 2, characterized in that: Extracting features from the live screen to obtain target features, determining target additional feature information matching the target features in the additional feature information of each program, and using program list information corresponding to the target additional feature information as the recognition result of the live screen, including: Extracting text features and image features from the live broadcast screen to obtain target text features and target image features; Determining target detailed text information matching the target text feature in the detailed text information of each of the programs; Determine a target portrait that matches the target portrait feature among the portraits of the program participants of each of the programs; Taking the program list information corresponding to the target detailed text information as the first recognition result of the image frame; The program list information corresponding to the target portrait is used as the second recognition result of the image frame.

4. The live broadcast reminder method according to claim 3, characterized in that: Determining target program list information according to the plurality of recognition results determined by the continuous plurality of frames of live broadcast images includes: Determine a first frame number of image frames whose first recognition result is target program list information and a second frame number of image frames whose second recognition result is target program list information in a continuous plurality of frames of live broadcast images; Based on a first weight factor corresponding to the first recognition result and a second weight factor corresponding to the second recognition result, weighted processing is performed on the first number of frames and the second number of frames to obtain a weighted number of frames; the first weight factor represents the importance of the first recognition result in determining the program currently being played in the live broadcast; the second weight factor represents the importance of the second recognition result in determining the program currently being played in the live broadcast; Determining the probability corresponding to the target program list information based on the weighted frame number; If the probability is greater than a set probability threshold, it is determined that the target program list information belongs to the program currently being played in the live broadcast.

5. The live broadcast reminder method according to claim 1, characterized in that: Outputting prompt information to a target user based at least on the target program list information includes: Obtaining the popularity parameter and audio content corresponding to the currently playing program; Inputting the target program list information, the popularity parameter and the audio content into a large language model to obtain prompt information determined by the large language model; Output the prompt information to the target user.

6. The live broadcast reminder method according to claim 5, characterized in that: Outputting the prompt information to the target user includes: Based on the heat parameter, determine the heat value change rate corresponding to the currently playing program. The target program list information, the prompt information and the popularity value change rate are sent to the notification service module, so that the notification service module determines the target user and the target push method according to the user feature library, the program feature library and the popularity value, and pushes the prompt information to the target user based on the target push method.

7. A live broadcast reminder device, characterized in that: include: The acquisition module is used to obtain additional feature information of each program in the live broadcast; The additional characteristic information is used to supplement the program list information of the program; a feature extraction module, for extracting features from the live screen whenever a live screen of the live broadcast is obtained, obtaining target features, determining target additional feature information that matches the target feature in the additional feature information of each program, and using the program list information corresponding to the target additional feature information as the recognition result of the live screen; A determination module, used to determine target program list information according to a plurality of recognition results determined from a plurality of consecutive live frames; The target program list information represents the program currently being played in the live broadcast; A prompt module is used to output prompt information to a target user based at least on the target program list information; the prompt information is used to remind the target user to watch the live broadcast.

8. A computer program product, characterized in that It includes computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the live broadcast reminder method as described in any one of claims 1 to 6.

9. An electronic device, characterized in that: include: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the live broadcast reminder method as described in any one of claims 1 to 6.

10. A computer storage medium, characterized in that: The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the live broadcast reminder method as described in any one of claims 1 to 6.

Images