Live broadcast monitoring system

Abstract

The invention discloses a live broadcast monitoring system, which adopts a modular architecture and comprises a data query module, a data import module, a billboard module and an account management module. According to the system, a cloud mobile phone technology is used for collecting live broadcast pictures, the identity of an anchor is recognized through a face recognition algorithm, a big data processing framework is used for processing and processing data, and the functions of live broadcast room state monitoring, anchor management, data visualization, abnormity traceability and the like are achieved. All the modules work cooperatively, and data query, import editing and account permission setting are supported. In the claims, the functional details and the technical means application range of each module are limited. In practical application, the system can effectively improve live broadcast management efficiency, guarantee compliance operation and promote industry development, such as e-commerce live broadcast sales increase, game live broadcast fault rapid processing and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of live broadcast operation management, and in particular to a live broadcast monitoring system. Background Art

[0002] In today's digital age, the live broadcast industry has seen explosive growth, with live broadcast activities becoming more frequent and diverse. From e-commerce live broadcasts to boost product sales, to entertainment live broadcasts to enrich people's lives, to knowledge popularization live broadcasts to spread professional knowledge, live broadcasts have penetrated into all areas of social and economic life. However, with the rapid expansion of the live broadcast industry, the drawbacks of traditional live broadcast management methods have become increasingly prominent.

[0003] Traditional live broadcast management relies heavily on manual regular inspections of the live broadcast room status, which is not only inefficient but also prone to human errors. Due to limited human energy, it is difficult to conduct all-round and uninterrupted monitoring of many live broadcast rooms, making it difficult to accurately grasp the start and end time of broadcasts. Once the start of broadcast is delayed, the traffic peak will be missed, and the audience will not be fully attracted, reducing the influence and commercial value of the live broadcast; and if the broadcast is not ended in time, it will trigger negative feedback from the audience, damaging the image and user experience of the live broadcast platform.

[0004] In terms of anchor management, when faced with complex scenarios with a multi-person operation team or frequent anchor changes, the traditional human management method can hardly ensure the authenticity of the anchor's identity, the standardization of appearances, and the accurate statistics of appearance time. This makes it easy for anchors to use other people's identities to broadcast live, destroying the fair competition environment of the live broadcast industry; at the same time, statistical bias will also affect the accuracy of anchor performance evaluation, and thus affect the anchor's enthusiasm and the scientific nature of live broadcast operations.

[0005] In addition, traditional live broadcast management lacks the integration and visualization of multi-dimensional data of live broadcast rooms and anchors. Faced with massive live broadcast data, it is difficult for operators to quickly identify abnormal situations, let alone conduct effective traceability analysis, which makes it difficult to solve problems in a timely manner, further hindering the improvement of live broadcast operation efficiency and management level.

[0006] To sum up, the existing live broadcast management methods can no longer meet the needs of the rapid development of the industry. There is an urgent need for an innovative and intelligent live broadcast monitoring system to improve the efficiency and quality of live broadcast management and promote the healthy and sustainable development of the live broadcast industry. Summary of the invention

[0007] The purpose of the present invention is to provide a live broadcast monitoring system, which aims to solve many problems existing in current live broadcast management, specifically including:

[0008] Improve the efficiency of live broadcast monitoring: In view of the problems that traditional manual regular inspection of the live broadcast room status is inefficient and error-prone, and it is difficult to accurately capture the start and end nodes of the broadcast, the real-time and accurate monitoring of the live broadcast room status is realized to ensure that the live broadcast process is complete and smooth, and no key traffic nodes are missed.

[0009] Strengthen anchor management: Solve the problems of anchor identity identification, standardized management of appearances, and difficulty in counting appearance time in scenarios with multiple operation teams or frequent changes of anchors, prevent impersonation, accurately count the anchor's appearance time, and provide solid data support for scientific and reasonable anchor performance evaluation.

[0010] Realize deep data integration and visualization: Change the current situation of lack of integration and visualization of multi-dimensional data in traditional live broadcast management, and create an integrated visualization data platform that can not only integrate various data of the live broadcast room and the anchor, but also provide abnormal picture tracing function, greatly improving the efficiency and management level of live broadcast operations.

[0011] The present invention is achieved through the following technical solutions:

[0012] A live broadcast monitoring system includes a data query module, a data import module, a data dashboard module and an account management module, and each module cooperates to realize the collection, processing, display and management of live broadcast related data: the data query module supports real-time data query and historical data query. Users can enter keywords of the live broadcast room or anchor through the search box for fuzzy matching, and filter according to business unit, category and platform dimensions. The query results are displayed in a specific table form, and users can export no more than 10,000 data according to the filtering conditions;

[0013] The data import module supports the import and editing of live broadcast room information, schedules and anchor data, and provides drag-and-drop upload of Excel tables or local path selection import methods. After importing, users are allowed to edit or delete related records;

[0014] The data dashboard module displays key indicators of the live broadcast room and anchors through dynamic cards and tables, including the duration of suspension, the number of times the face is not shown, and the duration of the face is not shown. It has a search box and filter items, which can quickly locate abnormal data and is linked to the monitoring screen. Clicking on abnormal data can display the corresponding monitoring screen and analysis information. The analysis information includes whether the broadcast is on, the anchor is identified, and whether the anchor shows his face;

[0015] The account management module is used for user authority management of the system. Administrators can use the search box to find users, view account lists, and edit or delete accounts. When creating a new account, they need to fill in the business unit, category, name, and mobile phone number. The default login password is 123abc.

[0016] As a further improvement of the technical solution of the present invention, cloud phone technology is used for data collection and analysis: each cloud phone can monitor 4 live broadcast rooms, the switching time between live broadcast rooms is 15 seconds, and each live broadcast room updates the screen information every 1 minute; the monitoring platform priority is set, among which Douyin, JD.com, Tmall, and Pinduoduo are the first priority, and Vipshop, Suning, Kuaishou, and Xiaohongshu are the second priority; the collected data includes data collection time, live broadcast room name, whether the broadcast is started, anchor identity, and whether the anchor shows his face. The anchor identity is obtained by matching face recognition with the anchor database.

[0017] As a further improvement to the technical solution of the present invention, in terms of anchor identity and appearance management: using face recognition technology, the collected anchor images are matched with the anchor database to confirm the legitimacy of the anchor identity; the anchor's live broadcast screen is tracked throughout the entire process, the anchor's appearance time is accurately counted, and detailed reports are generated on a daily and monthly basis.

[0018] As a further improvement of the technical solution of the present invention, the data processing and processing module processes the collected raw data to generate the following key indicators: live broadcast room conditions, including three situations: stop broadcasting, showing face when starting broadcasting, and not showing face when starting broadcasting. If the host's face is not detected for 3 minutes during the live broadcast, it is determined to be the start of the broadcast without showing face; the actual start time, that is, the time of the earliest picture of the live broadcast; the actual end time, that is, the time of the last picture of the live broadcast; the stop duration, that is, the cumulative duration of the picture being stopped during the live broadcast; the number of times the host does not show his face, during the live broadcast, if the host does not show his face for more than 3 minutes, it is counted as one time; the duration of the host's face not showing, that is, the total cumulative duration of the host not showing his face.

[0019] As a further improvement of the technical solution of the present invention, the system is implemented based on the following technical means: cloud phone technology, used for real-time acquisition of live broadcast images; face recognition algorithm, used for anchor identity identification and matching; database, used for storing collected and processed data; Web interface, used to realize data query, display and management functions.

[0020] As a further improvement of the technical solution of the present invention, the account management module sets permissions according to business units and categories, and users with different permissions can only access and operate corresponding data and functional modules.

[0021] As a further improvement of the technical solution of the present invention, the key indicators displayed by the data dashboard module also include comparative data of the actual broadcast time, the actual broadcast time, the scheduled broadcast time, and the scheduled end time.

[0022] As a further improvement of the technical solution of the present invention, in the data import module, the format of the imported live broadcast room information, schedule and anchor information is checked, and if the format does not meet the requirements, the user is prompted to re-import.

[0023] As a further improvement of the technical solution of the present invention, when performing data query, the data query module can store the query records in the order of query time, so as to facilitate users to quickly view historical query content.

[0024] As a further improvement of the technical solution of the present invention, when an abnormality occurs in the key indicators displayed in the data dashboard module, the system automatically sends an abnormal reminder message to the administrator, and the reminder method includes text messages and in-site messages.

[0025] It should be noted that:

[0026] In the live monitoring system architecture design:

[0027] System modular construction: The system is composed of four core functional modules: data query, data import, data dashboard and account management. Each module has a clear division of labor and close cooperation. The data query module is responsible for providing flexible data query services, the data import module realizes efficient data entry, the data dashboard module intuitively displays key data, and the account management module ensures secure access and permission control of the system, and together completes the entire process of data collection, processing, display and management.

[0028] Data collection and analysis: Use advanced cloud phone technology to collect live broadcast images in real time, and use intelligent algorithms to accurately analyze key data such as whether the live broadcast room is on, the identity of the anchor, and whether the anchor shows his face. At the same time, reasonably set the priority of the monitoring platform to ensure that the live broadcast images of important platforms such as Douyin, JD.com, Tmall, and Pinduoduo are updated in time, providing a real-time and accurate data basis for subsequent data analysis.

[0029] Data processing and handling: The collected original live broadcast screen data is deeply processed, and professional data algorithms are used to generate a series of key indicators such as the live broadcast room status (including three states: off broadcast, with face shown on broadcast, and without face shown on broadcast), actual broadcast start time, actual broadcast end time, off broadcast duration, number of times the host does not show his face, and duration the host does not show his face, to provide strong data support for live broadcast operation analysis.

[0030] Anchor identity and appearance management

[0031] Application of face recognition technology: Using cutting-edge face recognition technology, the real-time images of the anchor collected are quickly and accurately matched with the anchor database, instantly confirming the legitimacy of the anchor's identity, effectively eliminating impersonation, and ensuring the standardization and fairness of the live broadcast.

[0032] Accurate statistics of appearance duration: By continuously tracking the live broadcast screen of the anchor, the intelligent time statistics algorithm is used to accurately count the anchor's appearance duration, and detailed detailed reports are automatically generated according to user-defined cycles such as day and month, providing a reliable data basis for objective and fair anchor performance evaluation.

[0033] Data visualization and anomaly tracing

[0034] Dynamic data dashboard design: Carefully construct a dynamic data dashboard to display key indicators of the live broadcast room and anchors in an intuitive and easy-to-understand way, such as the duration of suspension, the number of times the face is not shown, the duration of the face not shown, etc. At the same time, it is equipped with powerful filtering conditions and search functions to facilitate operators to quickly locate abnormal data and improve the efficiency of problem troubleshooting.

[0035] Rapid tracing of abnormal images: Realize seamless linkage between the data dashboard and the monitoring screen. When the operator clicks on abnormal data on the data dashboard, the system can immediately display the corresponding monitoring screen and mark the analysis information in detail, such as whether the broadcast has started, identification of the anchor, whether the anchor shows his face, etc., to help operators quickly trace the source and deeply analyze the root cause of the problem.

[0036] Account management and permission settings

[0037] Complete account management functions: Administrators can easily create accounts, comprehensively view account information, and flexibly configure account permissions through the account management module to achieve refined management of system users and ensure the safe and stable operation of the system.

[0038] Hierarchical authority management: Scientifically set authorities based on business units, categories, and other dimensions to ensure that users of different roles can only access and operate data and functional modules related to their responsibilities, effectively preventing data leakage and misoperation, and improving system security and management efficiency.

[0039] In summary, the present invention has the following beneficial effects:

[0040] Real-time and efficient monitoring: Through the intelligent identification system, the system can realize uninterrupted monitoring of the live broadcast room 24 hours a day, and the data update frequency is accurate to about 1 minute. This means that the operation team can grasp the dynamics of the live broadcast room in real time and deal with various emergencies in a timely manner. Compared with the traditional manual inspection method, the monitoring efficiency is improved several times, which greatly enhances the timeliness and accuracy of live broadcast management.

[0041] Accurate identity recognition: Using advanced facial recognition technology, the system can accurately recognize the identity of the anchor with an accuracy rate of over 99%. It can instantly match the collected real-time images of the anchor with the anchor database, effectively eliminating the phenomenon of impersonation, ensuring the compliance of the live broadcast subject, and maintaining fair order in the live broadcast industry.

[0042] Detailed data statistics: With the help of intelligent time statistics algorithm, the system can accurately track the host's appearance time throughout the whole process, control the error within a very small range, and automatically generate detailed detailed reports according to user-defined cycles such as day and month, providing extremely reliable data guarantee for the host's performance evaluation, making the evaluation results more scientific and fair.

[0043] Economic Effect

[0044] Reduced labor costs: After the introduction of this live broadcast monitoring system, the workload of manual inspections of the live broadcast room has been greatly reduced. Taking a medium-sized live broadcast operation team as an example, it can save about 50% of the manpower investment, significantly reducing labor cost expenditure, allowing enterprises to invest more resources in the development of core businesses.

[0045] Improved operational efficiency: The system provides accurate data statistics and visual displays, which help the operation team quickly locate problems during the live broadcast process, such as low traffic periods in the live broadcast room and poor performance of the anchor, and then optimize the live broadcast process in a targeted manner. Practice has proved that after using this system, the overall efficiency of live broadcast operations has increased by more than 30% on average, effectively increasing commercial value.

[0046] Social Effects

[0047] Promotion of industry norms: This system provides strong technical support for the compliance operations of the live broadcast industry, encourages various platforms and companies in the industry to strengthen anchor management and live broadcast content supervision, promotes the improvement of the management level of the entire industry, and creates a healthy and orderly live broadcast ecological environment.

[0048] User experience upgrade: By reducing the occurrence of abnormal live broadcast situations, such as anchors being late and live broadcast interruptions, the audience's viewing experience has been significantly improved. According to user feedback data, audience satisfaction with the live broadcast platform has increased by more than 20%, effectively enhancing users' trust and loyalty to the live broadcast platform, and laying a solid user foundation for the sustainable development of the live broadcast industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0050] Figure 1 This is one of the structural diagrams of a live broadcast monitoring system according to an embodiment of the present invention;

[0051] Figure 2 This is a second structural diagram of a live broadcast monitoring system according to an embodiment of the present invention;

[0052] Figure 3 This is a third structural diagram of a live broadcast monitoring system according to an embodiment of the present invention;

[0053] Figure 4 A schematic diagram of the system architecture of a live broadcast monitoring system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, features, and advantages of this application more obvious and easy to understand, the technical solutions in the embodiments of this application are clearly and completely described. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] Reference Figures 1 to 4 , a live broadcast monitoring system, its system architecture design

[0056] This live broadcast monitoring system is built with a modular architecture. The four functional modules of data query, data import, data dashboard and account management are both independent and work together. Each module exchanges data through a standardized data interface to ensure the efficiency and stability of the system operation. For example, after the data acquisition module obtains the live screen data, it transmits the data to the data processing module through the interface for processing, and the processed results are then transmitted to the data dashboard module for display.

[0057] The system uses cloud phone technology to achieve real-time collection of live broadcast images, and uses high-performance server clusters to host cloud phones to ensure the stability and smoothness of data collection. The face recognition algorithm is deployed on a dedicated algorithm server, and through docking with the data collected by the cloud phone, the anchor's identity can be quickly and accurately identified. The database uses a distributed relational database, which can be dynamically expanded according to the growth of data volume to ensure the efficiency of data storage and reading.

[0058] Data collection and analysis

[0059] The data collection module uses the multi-opening function of the cloud phone, and each cloud phone can monitor 4 live broadcast rooms at the same time. Through the self-developed intelligent scheduling algorithm, the live broadcast room screen can be quickly switched, and the switching time is about 15 seconds, ensuring that the screen information of each live broadcast room can be updated every 1 minute.

[0060] The priority setting of the monitoring platform is realized through the configuration file, and the operator can flexibly adjust it according to the actual business needs. For the Tik Tok, JD.com, Tmall, Pinduoduo and other platforms with the first priority, the system uses a dedicated network access to ensure the rapid collection and transmission of live broadcast images and the timeliness of data.

[0061] After preliminary cleaning, the collected data is sent to the data processing module in real time through the data transmission protocol. During the data analysis process, the deep learning algorithm is used to analyze the live broadcast screen to accurately identify the name of the live broadcast room, whether the broadcast is on, the identity of the anchor, and whether the anchor shows his face. For example, for anchor identity recognition, the facial feature points in the picture are first extracted, and then compared with the feature vectors in the anchor database. If the similarity exceeds the set threshold, it is determined to be a successful match.

[0062] Data processing and processing

[0063] After receiving the collected raw data, the data processing module uses the big data processing framework for distributed computing. To judge the situation in the live broadcast room, a time series analysis algorithm is used to continuously monitor the host's face in the live broadcast screen. If the host's face is not detected for three consecutive minutes, it is judged as "the host's face is not shown during the broadcast."

[0064] The actual start time and the actual end time are determined by analyzing the timestamps of the live images and selecting the earliest and latest valid image times as the start time and the end time. The statistics of the suspension time are calculated by accumulating the time intervals of the live image interruptions.

[0065] The statistics of the number of times and duration of the host's face not showing are based on an event-driven programming model. When an event of the host not showing their face is detected, a timer is started. If it exceeds 3 minutes, it is counted as a face-not showing event and the duration of the face-not showing is accumulated.

[0066] Data query and display

[0067] The data query module is implemented based on search engine technology and supports real-time data query and historical data query. After the user enters a keyword in the search box, the system parses the keyword through a word segmentation algorithm and quickly locates relevant data in combination with the index. At the same time, using the multi-dimensional screening function, users can filter data according to multiple dimensions such as business units, categories, platforms, etc. to achieve accurate query.

[0068] The query results are displayed in a paginated table format, with 20 data items displayed per page, which is convenient for users to browse. Users can also export data according to their needs. The exported data format is Excel. The system will limit the amount of exported data to no more than 10,000 items to ensure export efficiency.

[0069] Data dashboard and exception tracing

[0070] The data dashboard module is built with a visual component library, and dynamic cards and tables display key indicators in an intuitive way. After the user enters a keyword in the search box or selects a specific condition through the filter item, the system updates the data dashboard in real time and displays the data that meets the conditions.

[0071] The linkage between the data dashboard and the monitoring screen is achieved through front-end event binding. When the user clicks on a live broadcast data in the table, the front-end page sends a request to the back-end. The back-end obtains the corresponding monitoring screen according to the request parameters, and returns the screen and analysis information to the front-end for display. The annotated information includes whether the broadcast has started, identification of the anchor, whether the anchor shows his face, etc., to help operators quickly understand the live broadcast situation.

[0072] Data import and management

[0073] The data import module supports drag-and-drop upload of Excel tables and import by selecting a local path. Before importing, the system will perform format verification on the Excel table to ensure that the data format meets the requirements. For example, for the live broadcast room information import form, it is required to include required fields such as the live broadcast room name, platform, and anchor ID. If the format is incorrect or a field is missing, the system will pop up a prompt box to guide the user to modify it.

[0074] The imported data is stored in the database, and users can edit and delete it in the data management interface. The editing operation is implemented through a pop-up modal box. After the user modifies the data, click Save, and the system will synchronize the new data to the database; the deletion operation prompts the user through a confirmation pop-up window to prevent accidental deletion.

[0075] Account Management

[0076] The account management module uses the RBAC (role-based access control) model to implement permission management. When creating an account, the administrator needs to select the business unit and category to which the account belongs, and assign corresponding roles, such as operator, administrator, data analyst, etc. Different roles correspond to different operating permissions.

[0077] Administrators can search for users by entering usernames or phone numbers in the search box, and view the basic information, roles, and permissions of an account in the account list. When editing an account, you can modify the basic information and permissions of the account; when deleting an account, the system will first check whether the account has any unfinished operations, and then delete it after confirming that everything is correct.

[0078] User operation process

[0079] Data query: After the user logs in to the system, enter the keywords of the live broadcast room or anchor in the search box on the data query page, select the filtering conditions such as business unit, category, platform, etc., click the query button, and the system will display the query results. The user can click the export button as needed to export the data into an Excel file.

[0080] Data dashboard: On the data dashboard page, users can locate abnormal data through the search box and filter items. By clicking the table row corresponding to the abnormal data, users can view the monitoring screen and analysis information and conduct problem tracing and analysis.

[0081] Data import: Enter the data import page, the user selects the import entry corresponding to the live broadcast room information, schedule or anchor information, and imports the Excel table by dragging and dropping or selecting the local path. The data import is completed after the system verification passes, and the user can perform subsequent operations in the data management interface.

[0082] Account management: On the account management page, administrators can create new accounts, fill in basic account information, and assign roles and permissions; they can also edit and delete existing accounts to ensure the standardization and security of system account management.

[0083] Technical Implementation

[0084] Cloud phone technology: Select a mature cloud phone service provider, implement remote control and data collection of cloud phones through API interfaces, use the virtual environment of cloud phones to simulate real phone operations, and ensure the stability and compatibility of live screen collection.

[0085] Face recognition algorithm: Based on deep learning frameworks such as TensorFlow or PyTorch, train face recognition models. Through training with a large amount of face image data, continuously optimize model parameters and improve recognition accuracy. In practical applications, use GPU to accelerate computing to ensure real-time face recognition.

[0086] Data processing and storage: Use big data processing frameworks such as Hadoop and Spark for data processing, use the distributed file system HDFS to store massive data, and use technologies such as MapReduce and Spark Streaming to achieve distributed computing and real-time processing of data.

[0087] Front-end display: The front-end page is developed with the Vue.js framework, combined with the Element UI component library to implement data query, display and management functions. The front-end and back-end data interaction is realized through the Axios library to ensure the real-time update of page data and the smoothness of user operations.

[0088] Practical Example:

[0089] System architecture design: A medium-sized e-commerce live broadcast company adopted a modular architecture when building a live broadcast monitoring system. During a large-scale promotional event, the data collection module collected nearly 10,000 live broadcast screen data every hour, which was transmitted to the data processing module with the help of an efficient data interface. The processed data was displayed on the data dashboard. The operation staff adjusted the live broadcast strategy in a timely manner based on this data, and the sales during the event increased by 30% compared with the past.

[0090] Data collection and analysis: An MCN agency specializing in beauty live broadcasting, whose anchors broadcast live on multiple platforms. During the prime time of live broadcasting on the Douyin platform, the system quickly collects and updates the images of each live broadcast room through a dedicated network and intelligent scheduling algorithm. In one live broadcast, an anchor who used someone else's identity to broadcast live was successfully identified, and the live broadcast was blocked and handled in time to avoid potential risks.

[0091] Data processing and processing: A comprehensive live broadcast platform has thousands of live broadcasts every day. When analyzing the 3-hour live broadcast data of a well-known anchor, the time series analysis algorithm found that he did not show his face twice for more than 3 minutes, with a total of 8 minutes without showing his face. Based on this, the platform communicated with the anchor, and the anchor's face exposure during subsequent live broadcasts improved, and the audience interaction rate increased.

[0092] Data query and display: When a live broadcast operation team was preparing for a new product live broadcast, they entered product keywords in the data query module, selected filter conditions such as "beauty category" and "Tmall platform", and quickly obtained detailed data of related live broadcasts in the past month, including broadcast time, number of viewers, sales, etc. Based on these data, a live broadcast strategy was formulated, and the sales of the new product live broadcast exceeded expectations by 20%.

[0093] Data dashboard and abnormal source tracing: An operator of a game live broadcast platform found an abnormal increase in the length of time a popular game live broadcast room was suspended on the data dashboard. After clicking on the data, the system displayed the corresponding monitoring screen, showing that the live broadcast was interrupted due to a technical failure. The operator quickly notified the technical team to handle it and resumed the live broadcast within 15 minutes, avoiding a large number of viewers from losing.

[0094] Data import and management: When a newly established live broadcast guild entered the information of its anchors, it uploaded an Excel spreadsheet through the data import module. Due to errors in the format of some fields, the system popped up a prompt box. The staff successfully imported the data after modification. When adjusting the anchor schedule later, the anchor information was edited through the data management interface, and the system updated the database in time to ensure the smooth arrangement of the live broadcast.

[0095] Account management: When a large live broadcast group launched a new business, the administrator created an account for the new operator, selected the business unit and the "operator" role, and assigned corresponding permissions. Later, due to business adjustments, the administrator found the operator's account through the search box and modified its permissions so that it could manage the new live broadcast room and help the business to proceed smoothly.

[0096] User operation process: An operations supervisor in charge of multiple live broadcast rooms found an abnormal decrease in traffic in a live broadcast room on the data dashboard after logging into the system at work. He located the data through the search box and clicked to view the monitoring screen and found that the anchor was in poor condition. The supervisor exported the data of the past week in the data query module for analysis, and notified the anchor to adjust the status in the account management module. After that, the traffic of the live broadcast room gradually recovered.

[0097] Technical implementation: A new live broadcast platform uses cloud phone technology to cooperate with Alibaba Cloud to stably collect live broadcast images through API interfaces. Based on the TensorFlow framework, the face recognition model is trained, and the GPU server is used to accelerate the calculation. In a live broadcast event, the accuracy rate of identity recognition of thousands of anchors reached 99.5%. Hadoop and Spark are used for data processing and storage of massive live broadcast data. The front-end interface developed based on Vue.js operates smoothly, with good user feedback and continuous improvement of platform user activity.

[0098] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0099] Technical effect: With the help of intelligent recognition system, the live broadcast room can be monitored 24 hours a day with high efficiency, and the data update frequency can reach about 1 minute, which greatly improves the monitoring efficiency. The advanced face recognition technology is used to accurately identify the identity of the anchor, effectively prevent impersonation, and ensure the compliance of the live broadcast subject. By accurately tracking the appearance time of the anchor throughout the whole process and accurately counting the relevant data, reliable data guarantee is provided for the anchor performance evaluation.

[0100] Economic effect: Significantly reduce the workload of manual inspection of live broadcast rooms and greatly reduce labor costs; through accurate data statistics and visual display, help the operation team quickly identify problems and optimize the live broadcast process in a timely manner, comprehensively improve the overall efficiency of live broadcast operations, and increase commercial value.

[0101] Social effect: Provide strong technical support for the compliance operation of the live broadcast industry, and effectively promote the improvement of the management level of the entire industry; by reducing the occurrence of abnormal situations in live broadcasts, significantly improve the audience's viewing experience, enhance users' trust in live broadcast platforms, and promote the healthy and sustainable development of the live broadcast industry.

[0102] The technical solutions provided by the embodiments of the present invention are introduced in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the embodiments of the present invention. The description of the above embodiments is only applicable to help understand the principles of the embodiments of the present invention. At the same time, for those skilled in the art, according to the embodiments of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A live broadcast monitoring system, characterized in that: It includes data query module, data import module, data dashboard module and account management module. Each module collaborates to collect, process, display and manage live broadcast related data: The data query module supports real-time data query and historical data query. Users can enter the keywords of the live broadcast room or anchor in the search box for fuzzy matching, and filter by business unit, category, and platform dimensions. The query results are displayed in a specific table, and users can export no more than 10,000 data according to the filtering conditions; The data import module supports the import and editing of live broadcast room information, schedules and anchor data, and provides the Excel spreadsheet drag and drop upload or local path selection import method. After importing, users are allowed to edit or delete related records; The data dashboard module displays key indicators of the live broadcast room and anchors through dynamic cards and tables, including the duration of suspension, the number of times the face is not shown, and the duration of the face is not shown. It has a search box and filter items, which can quickly locate abnormal data and is linked to the monitoring screen. Clicking on abnormal data can display the corresponding monitoring screen and analysis information. The analysis information includes whether the broadcast is on, the anchor is identified, and whether the anchor shows his face; The account management module is used for user authority management of the system. Administrators can use the search box to find users, view account lists, and edit or delete accounts. When creating a new account, they need to fill in the business unit, category, name, and mobile phone number. The default login password is 123abc.

2. A live broadcast monitoring system according to claim 1, characterized in that: Cloud phone technology is used for data collection and analysis: each cloud phone can monitor 4 live broadcast rooms, the switching time between live broadcast rooms is 15 seconds, and each live broadcast room updates the screen information every 1 minute; the monitoring platform priority is set, among which Douyin, JD.com, Tmall, and Pinduoduo are priority I, and Vipshop, Suning, Kuaishou, and Xiaohongshu are priority II; the collected data includes data collection time, live broadcast room name, whether the broadcast is started, anchor identity, and whether the anchor shows his face. The anchor identity is obtained by matching face recognition with the anchor database.

3. A live broadcast monitoring system according to claim 1, characterized in that: In terms of anchor identity and appearance management: using facial recognition technology, the collected anchor images are matched with the anchor database to confirm the legitimacy of the anchor's identity; the anchor's live broadcast screen is tracked throughout the entire process, the anchor's appearance time is accurately counted, and detailed reports are generated on a daily and monthly basis.

4. A live broadcast monitoring system according to claim 1, characterized in that: The data processing and processing module processes the collected raw data to generate the following key indicators: live broadcast room status, including three situations: stop broadcasting, showing face when broadcasting, and not showing face when broadcasting. If the host's face is not detected for 3 minutes during the live broadcast, it is judged as not showing face when broadcasting; actual broadcast start time, that is, the time of the earliest screen of the live broadcast; actual broadcast end time, that is, the time of the last screen of the live broadcast; stop broadcast duration, that is, the cumulative duration of the screen stop during the live broadcast; the number of times the host does not show his face. During the live broadcast, if the host does not show his face for more than 3 minutes, it is counted as one time; the duration of the host's face not showing, that is, the total duration of the host's face not showing.

5. A live broadcast monitoring system according to claim 1, characterized in that: The system is implemented based on the following technical means: cloud mobile phone technology, used to collect live broadcast images in real time; Face recognition algorithm, used for anchor identity recognition and matching; database, used to store collected and processed data; Web interface, used to realize data query, display and management functions.

6. A live broadcast monitoring system according to claim 1, characterized in that: The account management module sets permissions based on business units and categories. Users with different permissions can only access and operate corresponding data and functional modules.

7. A live broadcast monitoring system according to claim 1, characterized in that: The key indicators displayed in the data dashboard module also include comparison data between the actual broadcast time, the actual broadcast time, and the scheduled broadcast time and the scheduled end time.

8. A live broadcast monitoring system according to claim 1, characterized in that: In the data import module, the format of the imported live broadcast room information, schedule and anchor information is checked. If the format does not meet the requirements, the user is prompted to re-import.

9. A live broadcast monitoring system according to claim 1, characterized in that: When performing data query, the data query module can store the query records in chronological order of query time, making it convenient for users to quickly view historical query content.

10. A live broadcast monitoring system according to claim 1, characterized in that: When an abnormality occurs in the key indicators displayed in the data dashboard module, the system automatically sends an abnormal reminder message to the administrator, including SMS and in-site messages.

Images