Camera gun video resource positioning method and related equipment

Abstract

The invention discloses a camera gun video resource positioning method and related equipment. The method comprises the following steps: acquiring camera gun information of each financial network; obtaining structured risk model data for the daily recovery high-risk model troubleshooting index; determining a target financial network point corresponding to each piece of risk model data in the structured risk model data; according to the structured risk model data, the target financial network point corresponding to each piece of risk model data, the camera gun information of each financial network point and the current popularity value of each camera gun, determining a plurality of current retrieval camera guns; according to the current user operation log, updating the current popularity value corresponding to each camera gun; and according to the updated current popularity value corresponding to each camera gun, determining a plurality of updated current retrieval camera guns. The method can achieve the precise pushing of the video resources needed by the daily redisk of the website, improves the video verification efficiency, and can be widely applied to the technical field of artificial intelligence.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method and related equipment for locating video resources using a camera. Background Technology

[0002] In the field of financial security, daily branch review is a core part of risk control. The process generally involves reviewing the branch's CCTV footage to review the implementation of important business operations such as on-site management, counter services, cash handling, and cash delivery, and to check for potential security risks.

[0003] Currently, daily post-mortem reviews in the industry primarily rely on manual methods. This involves inspectors randomly selecting videos from various locations for review, and tailoring their reviews based on personal experience. However, this model heavily depends on human experience and has significant limitations in practical application. It may result in incomplete reviews of locations, ultimately overlooking some truly risky sites. Additionally, in certain scenarios (such as assisting customers with mobile phone use), the YOLO video dynamic detection algorithm can be used to automatically extract video clips matching the scenario from video data for manual review. However, this video detection solution suffers from high requirements for model training expertise and high deployment costs.

[0004] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0005] The embodiments of this application aim to at least partially solve one of the technical problems in the related art. Therefore, the main objective of the embodiments of this application is to propose a method and related equipment for locating video resources using a camera, which can accurately push the video resources needed for daily branch review and improve video verification efficiency.

[0006] To achieve the above objectives, one aspect of this application proposes a method for locating video resources using a camera, the method comprising the following steps: The raw camera information of several financial outlets is obtained, and the raw camera information of each financial outlet is cleaned to obtain the target camera information of each financial outlet. Obtain the raw risk model data for the high-risk model screening indicators in the daily review, and perform data parsing processing on the raw risk model data to obtain structured risk model data; According to the preset institution matching method, financial outlet matching is performed on each risk model data in the structured risk model data to determine the target financial outlet corresponding to each risk model data in the structured risk model data; Based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera, a number of cameras to be currently accessed are determined. Obtain the current user operation log, and update the current popularity value corresponding to each camera gun according to the current user operation log; Based on the updated current popularity value corresponding to each of the aforementioned cameras, a number of currently accessing cameras are determined; wherein each of the aforementioned cameras has captured corresponding video resources.

[0007] In some embodiments, acquiring raw camera information from several financial outlets and performing data cleaning on the raw camera information from each of the financial outlets to obtain target camera information for each of the financial outlets includes: Obtain the original camera information from several of the aforementioned financial outlets; According to the preset camera naming rules, the original camera information of each of the financial outlets is integrated and processed to obtain the original camera outlet mapping table; The original camera outlet mapping table is cleaned to obtain the target camera outlet mapping table; wherein, the target camera outlet mapping table is used to represent the target camera information corresponding to each of the financial outlets, and the target camera outlet mapping table is a mapping table of the relationship between camera and financial outlet.

[0008] In some embodiments, the step of acquiring raw risk model data for the daily review high-risk model screening indicators and performing data parsing processing on the raw risk model data to obtain structured risk model data includes: Obtain the original risk model data for the high-risk model screening indicators of the daily review, and input the original risk model data into the corresponding risk model; Using regular expression matching and semantic parsing techniques, key fields are extracted from the original risk model data to obtain key field data in the original risk model data. Based on the key field data and the risk dimensions of the risk model, the structured risk model data is constructed and displayed in the video viewing interface; wherein, the structured risk model data includes the key field data and core inspection elements.

[0009] In some embodiments, the step of matching each risk model data in the structured risk model data with financial outlets according to a preset institution matching method to determine the target financial outlet corresponding to each risk model data in the structured risk model data includes: According to the preset institution mapping relationship table, institution information is matched for each piece of risk model data in the structured risk model data; If there is risk model data that fails to match, a multimodal intelligent matching method is used to perform composite matching processing on the risk model data that fails to match, and a composite matching result is obtained; wherein, the composite matching result includes an exact matching result and a fuzzy matching result, and the fuzzy matching result includes a first fuzzy matching result and a second fuzzy matching result; Perform an intersection operation on the first fuzzy matching result and the second fuzzy matching result to generate a double fuzzy matching result; The exact matching result is verified for uniqueness. If the exact matching result passes the uniqueness verification, the target financial outlet corresponding to each risk model data in the structured risk model data is determined based on the exact matching result. If the exact matching result fails the uniqueness verification, then the double fuzzy matching result is subjected to uniqueness verification. If the dual fuzzy matching result passes the uniqueness verification, then the target financial outlet corresponding to each risk model data in the structured risk model data is determined based on the dual fuzzy matching result.

[0010] In some embodiments, determining a number of currently accessed surveillance cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target surveillance camera information of each financial outlet, and the current popularity value of each surveillance camera includes: Based on the preset dual screening mechanism, and combining the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, and the target camera information of each financial outlet, all cameras in the target outlet area are determined. The heat values ​​of each camera within the target network area are sorted to obtain the heat value sorting results; Based on the ranking results of the heat values ​​and the historical judgment results of each risk model data in the structured risk model data, a number of the currently accessing cameras are determined through comprehensive evaluation.

[0011] In some embodiments, obtaining the current user operation log and updating the current popularity value corresponding to each camera based on the current user operation log includes: Obtain the current user operation log; wherein, the current user operation log is the log of the actual access to the camera by the verification personnel; Based on the current user operation log, calculate the model weight value corresponding to each camera; A linear weighted model is used, combining the user operation logs corresponding to each camera and the model weight value, to calculate the updated current popularity value of each camera.

[0012] To achieve the above objectives, another aspect of this application proposes a video resource positioning device for a camera, the device comprising the following modules: The camera information acquisition module is used to acquire the original camera information of several financial outlets, and to perform data cleaning on the original camera information of each financial outlet to obtain the target camera information of each financial outlet. The risk model data parsing module is used to acquire the original risk model data for the high-risk model screening indicators of the daily review, and to perform data parsing processing on the original risk model data to obtain structured risk model data. The financial outlet matching module is used to match financial outlets for each risk model data in the structured risk model data according to a preset institution matching method, and to determine the target financial outlet corresponding to each risk model data in the structured risk model data. The current access camera determination module is used to determine a number of current access cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera. The camera heat value update module is used to obtain the current user operation log and update the current heat value corresponding to each camera according to the current user operation log; The current access camera update module is used to determine a number of updated current access cameras based on the updated current popularity value of each camera; wherein each camera has captured corresponding video resources.

[0013] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0014] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0015] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0016] The embodiments of this application include at least the following beneficial effects: This application provides a method and related equipment for locating video resources from camera outlets. This method obtains raw camera information from several financial outlets, cleans the raw camera information from each financial outlet to obtain target camera information for each financial outlet; obtains raw risk model data for the daily review high-risk model screening indicators, and performs data parsing processing on the raw risk model data to obtain structured risk model data; and matches each risk model data in the structured risk model data with financial outlets according to a preset institution matching method to determine the results. The structured risk model data includes the target financial outlets corresponding to each risk model data point; based on the structured risk model data, the target financial outlets corresponding to each risk model data point, the target camera information for each financial outlet, and the current popularity value of each camera, several currently accessed cameras are determined; the current user operation log is obtained, and the current popularity value corresponding to each camera is updated based on the current user operation log; based on the updated current popularity value corresponding to each camera, several updated currently accessed cameras are determined; each camera has captured corresponding video resources. This application embodiment acquires raw risk model data for the high-risk model screening indicators of the daily review and performs data parsing processing on the raw risk model data. The final result is structured risk model data containing the key points of the daily review's high-risk checks. This allows inspectors to quickly locate the key points of risk data verification. Specifically, inspectors can access video data according to the risk event guidelines, reducing reliance on personnel experience during the daily review process and thus improving review quality and efficiency. By identifying several currently accessed video cameras, video resources can be accurately located, reducing repetitive operations in video retrieval. Furthermore, by collecting video access operation logs generated by inspectors actually accessing video cameras, a heat value for each camera is dynamically generated. Therefore, during the next daily review, the structured risk model data can be combined to accurately locate video data within the time period of the event, achieving precise video resource delivery, reducing the large amount of repetitive operations generated by manual video retrieval, and improving video verification efficiency. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of a video resource localization method using a camera provided in an embodiment of this application. Figure 2 This is a flowchart illustrating a video resource positioning method using a camera provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the generation process of structured risk model data provided in an embodiment of this application; Figure 4 This is a schematic diagram of a process for screening network points based on a multimodal intelligent matching method provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a video resource positioning device for a camera provided in an embodiment of this application; Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0019] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0020] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] In the field of financial security, daily branch review is a core part of risk control. The process generally involves reviewing the branch's CCTV footage to review the implementation of important business operations such as on-site management, counter services, cash handling, and cash delivery, and to check for potential security risks.

[0023] Currently, daily review processes in the industry primarily employ a manual approach. This involves auditors randomly selecting branch videos for review, and developing differentiated review procedures based on their individual experience. For instance, when reviewing branch CCTV footage related to cash deposits and replenishment, bank auditors must manually drag the video progress bar to the estimated time (e.g., the typical cash deposit time is 8 AM) to locate the target operation. If the location is not precise, they must repeatedly drag the progress bar or switch between different cameras until the corresponding video segment is found. Because the timing of cash deposits and replenishment varies across branches, and auditors have different experience levels and operating habits, this model has significant limitations in practical application. The reviews are incomplete, ultimately overlooking some branches with genuine risks.

[0024] In addition, in certain scenarios (such as operating a mobile phone on behalf of a customer), the YOLO video dynamic detection algorithm can be used to automatically extract video segments that match the scenario from video data and submit them for manual review. However, this video detection solution has the following main problems: First, the model training requires a high level of expertise. Since there is no universal pre-trained model, it is necessary to customize the training according to the recognition scenario, collect the corresponding video data, manually annotate, and train according to the video recognition requirements. Second, the deployment cost is high, requiring the addition of a video detection model running environment (such as an edge computing box) at the branch office to extract the video segments to be manually verified.

[0025] Therefore, the current method of checking potential security risks by locating CCTV video resources during daily branch review in the financial security field has the following main shortcomings: (1) High dependence on experience and high learning cost: Newcomers have difficulty quickly mastering the key points of debriefing due to lack of experience, which makes the quality and efficiency of debriefing significantly affected by the ability of the personnel. (2) High repetitiveness of operations and low retrieval efficiency: It is necessary to repeatedly drag the progress bar and switch the camera to locate the target video, resulting in a high proportion of invalid operations; (3) Incomplete coverage of outlets and high risk omission rate: Random sampling or experience-based screening of outlets may lead to the omission of some high-risk outlets; (3) Lack of dynamic optimization mechanism and insufficient push accuracy: It lacks intelligent recommendation capabilities based on historical operation data and cannot target high-risk camera guns.

[0026] In view of this, this application provides a method and related equipment for locating video resources from surveillance cameras. This method obtains raw surveillance camera information from several financial outlets and performs data cleaning on this information to obtain target surveillance camera information for each financial outlet. It also obtains raw risk model data for the high-risk model screening indicators used in daily review and performs data parsing processing on this raw risk model data to obtain structured risk model data. Finally, according to a preset institution matching method, it matches each risk model data point in the structured risk model data with a financial outlet to determine the structured risk. The model data includes the target financial outlets corresponding to each risk model data point; based on the structured risk model data, the target financial outlets corresponding to each risk model data point, the target camera information for each financial outlet, and the current popularity value of each camera, several currently accessed cameras are determined; the current user operation log is obtained, and the current popularity value corresponding to each camera is updated based on the current user operation log; based on the updated current popularity value corresponding to each camera, several updated currently accessed cameras are determined; each camera has captured corresponding video resources. This application embodiment acquires raw risk model data for the high-risk model screening indicators of the daily review and performs data parsing processing on the raw risk model data. The final result is structured risk model data containing the key points of the daily review's high-risk checks. This allows inspectors to quickly locate the key points of risk data verification. Specifically, inspectors can access video data according to the risk event guidelines, reducing reliance on personnel experience during the daily review process and thus improving review quality and efficiency. By identifying several currently accessed video cameras, video resources can be accurately located, reducing repetitive operations in video retrieval. Furthermore, by collecting video access operation logs generated by inspectors actually accessing video cameras, a heat value for each camera is dynamically generated. Therefore, during the next daily review, the structured risk model data can be combined to accurately locate video data within the time period of the event, achieving precise video resource delivery, reducing the large amount of repetitive operations generated by manual video retrieval, and improving video verification efficiency.

[0027] This application provides a method for locating video resources using a camera, relating to the field of artificial intelligence technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited thereto. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing a method for locating video resources using a camera, but is not limited to the above forms.

[0028] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0029] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0030] Please see Figure 1 , Figure 1This is an optional flowchart of a video resource localization method using a camera provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0031] Step S101: Obtain the original camera information of several financial outlets, and perform data cleaning on the original camera information of each financial outlet to obtain the target camera information of each financial outlet. In some embodiments, step S101 may include: acquiring raw camera information of several financial outlets; integrating the raw camera information of each financial outlet according to a preset camera naming rule to obtain a raw camera outlet mapping table; cleaning the raw camera outlet mapping table to obtain a target camera outlet mapping table; wherein, the target camera outlet mapping table is used to characterize the target camera information corresponding to each financial outlet, and the target camera outlet mapping table is a mapping table of the relationship between cameras and financial outlets.

[0032] The original camera information consists of the basic equipment information for all cameras in all locations. This basic equipment information may include, but is not limited to, the unique identifier (SN code), model, installation location, area (cash area / ATM area, etc.), and network connection status.

[0033] The target camera outlet mapping table is used to represent the target camera information corresponding to each financial outlet. The target camera outlet mapping table is a mapping table of the relationship between cameras and financial outlets. The target camera outlet mapping table can be described as a "camera information-outlet" mapping table.

[0034] The default naming rules for cameras can be set according to the actual situation, such as "site code-area code-device serial number".

[0035] In step S101, firstly, by calling the standardized API interface provided by the camera manufacturer, the basic equipment information of all cameras in all branches is collected, that is, the original camera information corresponding to all cameras in all branches is collected; then, according to the preset camera naming rules, the collected camera information is bound to the corresponding branch to obtain the original "camera information-branch" mapping table; next, the original basic equipment information is cleaned by methods such as deduplication (removing duplicate equipment records), verification (filtering network disconnected or faulty cameras), and completion (supplementing missing installation area information) to form a standardized "camera information-branch" mapping table, that is, to obtain the target camera branch mapping table, which is to obtain the target camera information corresponding to each financial branch.

[0036] Step S102: Obtain the original risk model data for the high-risk model screening indicators of the daily review, and perform data parsing processing on the original risk model data to obtain structured risk model data; In some embodiments, step S102 may include: acquiring the original risk model data for the high-risk model screening indicators of the daily review, and inputting the original risk model data into the corresponding risk model; using regular expression matching technology and semantic parsing technology to extract key fields from the original risk model data to obtain key field data in the original risk model data; constructing structured risk model data based on the key field data and the risk dimensions of the risk model, and displaying the structured risk model data in the video viewing interface; wherein, the structured risk model data includes key field data and core inspection elements.

[0037] The daily review of high-risk model screening indicators may include, but are not limited to, proxy operations and business transactions handled by clients requiring close monitoring. Risk models may include, but are not limited to, proxy operation risk models and business transaction risk models for clients requiring close monitoring.

[0038] The original risk model data may include, but is not limited to: data from the agency operation model, and data from the business transactions of customers that require special attention.

[0039] Optionally, key fields in the risk model data may include, but are not limited to, the name of the responsible institution (e.g., branch name), transaction time (e.g., start / end time of cash deposit), teller number, customer number, transaction account name, and video access institution ID. The structured risk model data includes data such as trading institutions (which can be used as the institutions to access the data for matching), teller numbers, and customer numbers (used as the historical basis for risk model data to query whether there are any violations by the customer or teller in the past).

[0040] In step S102, firstly, for the high-risk model screening indicators of the daily review, corresponding risk models are created; then, the risk model data obtained from the external system is uploaded and imported in batches into the corresponding risk models. Before the data import, business personnel can configure the "model-region dictionary" in advance according to the characteristics of the risk model (once configured, it can be dynamically associated later); if not configured, the video viewing areas to be associated are manually selected before uploading; during the data import, the business system will parse the imported risk model data and extract the key field data in the risk model data through regular expression matching and semantic parsing technology; then, based on the key field data obtained from deep parsing, closely focusing on the risk dimensions unique to each risk model, a customized verification point system covering key field data and core inspection elements is automatically constructed (the customized verification point system is the parsed structured risk model data).

[0041] For example, assuming a risk model for transactions on behalf of clients, when importing data into this model, the system first automatically extracts key fields such as transaction time, customer name, customer ID number (anonymized), customer date of birth, and transaction type from the original risk model data. Then, it identifies information such as the customer's age and gender. Finally, it combines the extracted key fields and the identified information to form a system of verification points. The key risk fields include: Transaction Time: 2025-01-01 09:10:05, Customer Name: Zhang San, Gender: Male, Age: 70, Transaction Type: Self-service ITM, Operator: Li Si, Employee ID: 1001. The core inspection elements include: sales personnel purchasing wealth management and agency products on behalf of clients using self-service terminals or other electronic devices. In this embodiment, the parsed structured risk model data includes key field data and core inspection elements.

[0042] Understandably, key field data is extracted from the imported original risk model data according to the verification requirements. For example, the verification requirement for proxy operations is to "focus on checking the operation videos of customers requiring special attention at the self-service terminal to confirm whether sales personnel have engaged in proxy operations." The customers requiring special attention can be those meeting a preset age threshold; this embodiment does not impose such a limitation and can be set according to actual circumstances.

[0043] The structured risk model data will be presented in the video access interface, allowing inspectors to understand the business characteristics of the video to be accessed in advance, which is conducive to quickly reviewing video clips according to the key points of verification.

[0044] Step S103: According to the preset institution matching method, financial outlet matching is performed on each risk model data in the structured risk model data to determine the target financial outlet corresponding to each risk model data in the structured risk model data; In some embodiments, step S103 may include: matching institutional information for each risk model data in the structured risk model data according to a preset institutional mapping table; if there are risk model data that fail to match, then using a multimodal intelligent matching method to perform composite matching processing on the risk model data that fails to match, to obtain a composite matching result; wherein, the composite matching result includes a precise matching result and a fuzzy matching result, and the fuzzy matching result includes a first fuzzy matching result and a second fuzzy matching result; performing an intersection operation on the first fuzzy matching result and the second fuzzy matching result to generate a double fuzzy matching result; performing a uniqueness verification on the precise matching result; if the precise matching result passes the uniqueness verification, then determining the target financial outlet corresponding to each risk model data in the structured risk model data according to the precise matching result; if the precise matching result fails the uniqueness verification, then performing a uniqueness verification on the double fuzzy matching result; if the double fuzzy matching result passes the uniqueness verification, then determining the target financial outlet corresponding to each risk model data in the structured risk model data according to the double fuzzy matching result.

[0045] The preset organization mapping table is a mapping table between the full name of the data organization and the video resource organization.

[0046] In step S103, based on the mapping table between the full name of the data institution and the video resource institution, institution matching (institution is the branch) is performed on each risk model data in the structured risk model data. If the corresponding branch can be directly matched according to the mapping table between the full name of the data institution and the video resource institution, the institution number, institution name, etc. of the risk model data are directly updated, and then the process ends. If the corresponding branch cannot be directly matched according to the mapping table between the full name of the data institution and the video resource institution, i.e., the matching fails, the following matching steps need to be performed: First, using regular expression technology, the core geographical identifiers and feature words in the full name of the institution of each risk model data are dynamically parsed. Taking "AA Bank BB City CC County DD Road Branch Self-Service Area" as an example, this processing can accurately extract key elements such as "BB City", "CC County", and "DD Road". Then, the key elements extracted from the full name of the institution are divided into three combinations for composite query. A dual-track parallel mechanism of "fuzzy matching + precise matching" is used for the search. The double verification ensures that the responsible institution of each risk model data can be accurately mapped to the corresponding branch in the video resource institution table. The specific matching method is as follows: Combination 1 (Full Name Precise Query): Integrates the full name of "city + district / county + outlet" and performs precise matching with the full name field of the organization in the "Data Organization Full Name and Video Resource Organization Mapping Relationship Table" to generate result set A; Combination 2 (City-level Fuzzy Query): Taking "City + Outlet" as an independent item, perform fuzzy matching on the city name and outlet name fields of the "Data Institution Full Name and Video Resource Institution Mapping Relationship Table" to obtain the first fuzzy matching result; Combination 3 (District / County-level Fuzzy Query): Taking "District / County + Outlet" as an independent item, perform fuzzy matching on the "Data Institution Full Name and Video Resource Institution Mapping Relationship Table" fields of the district / county name and outlet name respectively to obtain the second fuzzy matching result; Finally, the intersection operation is performed on the fuzzy query results of combination 2 and combination 3 to generate result set B. Based on priority, the uniqueness of result set A is verified first. If a unique exact match exists in set A, the institution is considered successfully matched. If set A does not match, the uniqueness of set B is verified. If a unique exact match exists in set B, the institution is considered successfully matched; otherwise, it indicates that the keyword is unclear and the institution cannot be accurately matched, i.e., the branch cannot be accurately matched. After verifying the uniqueness of the result set, the mapping table between the full name of the data institution and the video resource institution is updated based on the unique verified result set. The institution number and institution name of the risk model data corresponding to the unique verified result set are also updated, thus determining the target financial branch corresponding to each risk model data in the structured risk model data, and the process ends.

[0047] Step S104: Based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current heat value of each camera, determine a number of currently accessed cameras; In some embodiments, step S104 may include: determining all cameras within the target branch area based on a preset dual screening mechanism, combining structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, and the target camera information of each financial outlet; sorting each camera within the target branch area by its popularity value to obtain a popularity value sorting result; and comprehensively evaluating the popularity value sorting result and the historical judgment result of each risk model data in the structured risk model data to determine a number of cameras currently being accessed.

[0048] In step S104, a dual composite condition of "video access agency (site) + model-selected area" is used to achieve accurate identification and locking of all monitoring devices within a specific area of ​​the target site. Then, based on the access area range selected by the risk model, all cameras within the jurisdiction of the video access agency are sorted according to their own popularity value. By default, the cameras are sorted from high to low popularity value in the target area. Combined with the historical judgment results of the risk model data, a comprehensive evaluation is performed, and finally, a list of cameras to be accessed is pushed. After the cameras to be accessed are determined, the required video resources can be directly located.

[0049] For example, assuming the current risk model is an employee risk transaction model, if a query finds that the employee has a history of violations, the system will prioritize pushing the surveillance cameras associated with the employee's most recent violation record. If there is no relevant historical record or no problems have been found in the past, the system will still sort the surveillance cameras in descending order of popularity value and finally push the video resources of the top N (e.g., N=10) surveillance cameras to cover the basic list of cameras to be accessed.

[0050] In this embodiment, when inspectors access the video viewing interface, the system defaults to displaying the video feed in a four-grid format, automatically playing the videos from the top four most popular surveillance cameras. The video playback start time is precisely set to M seconds (e.g., M=10) before the data transaction occurs. Furthermore, the system simultaneously displays "Inspection Points" prompts, clearly listing the key elements requiring focused inspection in the current risk data and the related user information, assisting inspectors in efficiently carrying out their work and facilitating quick identification of risky users, risky times, and event locations. It should be noted that in specific implementations, the number of surveillance feeds displayed on the visualization page can be set according to actual conditions; this embodiment does not impose any limitations on this.

[0051] Step S105: Obtain the current user operation log, and update the current popularity value corresponding to each camera gun according to the current user operation log; In some embodiments, step S105 may include: obtaining the current user operation log; wherein the current user operation log is the log of the inspector actually accessing the camera; calculating the model weight value corresponding to each camera based on the current user operation log; and calculating the updated current popularity value of each camera by using a linear weighted model and combining the user operation log and model weight value corresponding to each camera.

[0052] In step S105, firstly, the system collects operation logs of the surveillance cameras actually accessed by the inspectors through data collection points. These logs include, but are not limited to, the number of clicks, access duration, and click order. Then, based on the current user's operation logs, a linear weighted model is used to calculate the popularity value of each surveillance camera, providing a basis for subsequent camera recommendations. Specifically, after the surveillance cameras for each risk model data point are matched, inspectors can access the video data point by point. During the video access process, the system records the inspectors' video access operation logs (taking the number of clicks on the surveillance camera, access start and end times, and access order as examples), and uses a linear weighted model to calculate the surveillance camera popularity value, providing a basis for subsequent recommendations.

[0053] Step S106: Based on the updated current popularity value corresponding to each camera gun, determine a number of currently accessing camera guns after the update; wherein each camera gun has captured corresponding video resources.

[0054] Steps S101 to S106 as illustrated in this embodiment involve: acquiring raw camera information from several financial outlets and cleaning the raw camera information from each financial outlet to obtain target camera information for each financial outlet; acquiring raw risk model data for the daily review high-risk model screening indicators and performing data parsing processing on the raw risk model data to obtain structured risk model data; and matching each risk model data in the structured risk model data with financial outlets according to a preset institution matching method to determine each financial outlet in the structured risk model data. The system identifies target financial outlets corresponding to each risk model data point; based on the structured risk model data, the target financial outlets corresponding to each risk model data point, the target camera information for each financial outlet, and the current popularity value of each camera, several cameras are identified for current access; the system obtains current user operation logs and updates the current popularity value for each camera based on these logs; and based on the updated current popularity value for each camera, several updated cameras for current access are identified; each camera captures corresponding video resources. This application embodiment acquires raw risk model data for the high-risk model screening indicators of the daily review and performs data parsing processing on the raw risk model data. The final result is structured risk model data containing the key points of the daily review's high-risk checks. This allows inspectors to quickly locate the key points of risk data verification. Specifically, inspectors can access video data according to the risk event guidelines, reducing reliance on personnel experience during the daily review process and thus improving review quality and efficiency. By identifying several currently accessed video cameras, video resources can be accurately located, reducing repetitive operations in video retrieval. Furthermore, by collecting video access operation logs generated by inspectors actually accessing video cameras, a heat value for each camera is dynamically generated. Therefore, during the next daily review, the structured risk model data can be combined to accurately locate video data within the time period of the event, achieving precise video resource delivery, reducing the large amount of repetitive operations generated by manual video retrieval, and improving video verification efficiency.

[0055] To explain in detail the principles of the technical solution of this application, the overall process of this application will be described below with reference to some specific embodiments. It is easy to understand that the following is an explanation of the technical principles of this application and should not be regarded as a limitation of this application.

[0056] This application provides a method for locating video resources from surveillance cameras. It uses risk model data generated by a custom risk model as the starting point for daily branch review. This data, driven by risky branches, transaction times, teller information, and customer information, extracts and compiles key points for verification based on the risk data. Then, based on a camera heat value algorithm, high-risk cameras are pushed to the verification personnel, thereby improving investigation efficiency. The camera heat value algorithm dynamically generates a dynamic heat value for each camera in the model branch by analyzing past video viewing operation logs of verification personnel (mainly including the number of times cameras were clicked, the start and end times of viewing, and the viewing order). During the next review, the top N (e.g., N=4) most heat-sensitive cameras are selected by default, and combined with the risk model data, the video data of the event's time period is accurately located, achieving precise video resource push. This reduces the large amount of repetitive operations generated by manual video retrieval and improves investigation efficiency.

[0057] It is understood that this application embodiment uses model-driven generation of daily review checkpoints and a dynamic heat value algorithm to determine the order of camera inspections and recommend playback time periods, thus achieving accurate delivery of video resources required for daily review. This application embodiment aims to achieve the following objectives: (1) Use model-driven generation of daily review check points to get rid of dependence on human experience and reduce the threshold for personnel operation; (2) The model-driven generation of daily review data can keep up with the requirements of the network point investigation coverage inspection, so as to realize the dynamic adjustment of network point review priority, inspection scope and resource allocation; (3) Push high-risk camera guns to accurately locate video resources and reduce repetitive operations in video retrieval; (4) A dynamic optimization mechanism based on operational data continuously improves the accuracy of camera push; (5) Customer information (such as customer service scenarios that need to be verified in the business process) comes from risk model data generated by the risk model, which reduces the difficulty of verifying video information from the camera. (6) Achieve deep integration between risk model and video review, and standardize the review process.

[0058] This application provides a method for locating video resources of surveillance cameras, driven by risk model data. By collecting camera information, preprocessing risk data, and constructing a multi-factor heat value calculation model, it can achieve accurate targeting and dynamic optimization of high-risk surveillance cameras. The specific process can be divided into seven core steps, including: First, collecting and cleaning information on network cameras; Second, initializing risk model data; Third, preprocessing video retrieval network data; Fourth, accurately targeting surveillance cameras; Fifth, collecting logs; Sixth, using intelligent analysis algorithms to form dynamic heat values; Seventh, data-driven, dynamically updating weight values.

[0059] Please see Figure 2 , Figure 2 This is a flowchart illustrating a video resource localization method using a camera provided in an embodiment of this application. Figure 2 As shown in the figure, the specific implementation process of the video resource positioning method provided in this application embodiment is as follows: The first step is the collection and cleaning of network camera information (input: camera manufacturer interface data, i.e., raw camera information; output: the cleaned "camera information-network" mapping table, i.e., target camera information). Specifically, the implementation process of the first step is as follows: A1. Information Collection: By calling the standardized API interface provided by the camera manufacturer, the basic device information of all cameras in all outlets is collected, that is, the original camera information corresponding to all cameras in all outlets is collected; the basic device information of the camera may include, but is not limited to, the unique identifier (SN code), device model, installation location, area (cash area / ATM area, etc.), network connection status, etc. B1. Information Attribution: Based on the preset camera naming rules (such as "outlet code-area code-device serial number"), the collected camera information is bound to the corresponding outlet. By binding the collected camera information to the corresponding outlet, when the system identifies abnormal risk segments (such as irregular operations in the cash area or non-standard cash escort procedures), it can directly locate the specific functional area of ​​the specific outlet without the need for manual cross-system verification of camera attribution; C1. Data Cleaning: The original equipment basic information is cleaned by methods such as deduplication (removing duplicate equipment records), verification (filtering network disconnected or faulty cameras), and completion (supplementing missing installation area information) to form a standardized "camera-network point" mapping table, which yields the target camera information.

[0060] The second step is to initialize the risk model data (input: raw risk model data related to the high-risk model screening indicators in the daily review; output: structured risk model data after parsing). Please see Figure 3 , Figure 3 This is a schematic diagram illustrating the generation process of structured risk model data provided in an embodiment of this application, such as... Figure 3 As shown, the specific process for generating structured risk model data is as follows: A2. Model Creation: Create corresponding risk models for the high-risk model screening indicators in the daily review (such as: client operations, business transactions of clients requiring special attention, etc.).

[0061] B2. Data Import: Batch upload and import risk model data obtained from external systems (such as data on agency operations, data on business transactions of customers requiring special attention, etc.) into the corresponding risk models; C2. Region Selection: Before importing data, business personnel can configure the "Model-Region Dictionary" in advance according to the characteristics of the risk model (once configured, it can be dynamically associated later); if no configuration is made, the video viewing regions to be associated should be manually selected before uploading. D2. Field Parsing: During data import, the business system will parse the imported risk model data and extract key field data from the risk model data through regular expression matching and semantic parsing technology. Key field data in the risk model data may include, but are not limited to, the name of the responsible institution (such as the branch name), transaction time (such as the start / end time of cash deposit), teller number, customer number, transaction account name, video access institution ID, etc. E2. Generation of Customized Verification Key Points System: Based on the key field data obtained from in-depth analysis, and closely following the risk dimensions unique to each risk model, a customized verification key points system covering key risk fields and core inspection elements is automatically constructed (the customized verification key points system is the analyzed structured risk model data).

[0062] For example, assuming a client-managed transaction model, during the data import process, key fields such as transaction time, customer name, customer ID number (anonymized), customer date of birth, and transaction type are automatically extracted from the original risk model data. Then, information such as the customer's age and gender is identified. Finally, a verification point system is formed based on the extracted key field data and the identified information. The key risk fields include: Transaction Time: 2025-01-01 09:10:05, Customer Name: Zhang San, Gender: Male, Age: 70, Transaction Type: Self-service ITM, Operator: Li Si, Employee ID: 1001. Core inspection elements include: sales personnel purchasing wealth management and agency products on behalf of clients using self-service terminals or other electronic devices. In this embodiment, the parsed structured risk model data includes key field data and core inspection elements.

[0063] F2. Area Association: The business system automatically selects video retrieval areas (e.g., cash area, cash counter, VIP area, lobby, customer area, ATM, ITM self-service area, front door, back door, etc.) based on the model area mapping relationship configured by the business personnel. The selection of video retrieval areas is used for subsequent filtering of the cameras within that area. As shown in the above example of a customer service operation scenario, based on the core inspection elements, it can be determined that the video of the area belonging to the self-service equipment (ITM) needs to be retrieved, which is usually located in the branch lobby and ITM self-service area.

[0064] The third step is to preprocess the video access branch data; (Input: parsed structured risk model data; Output: a matching "risk model data - branch video resource institution ID" association table, that is, the target financial branch corresponding to each risk model data in the structured risk model data). In the risk model data system, the "Video Access Institution ID" field is assigned the responsible institution ID by default. However, in practical applications, it has been found that this field is difficult to accurately locate in the video resource library in some scenarios (a typical example is that the responsible institution ID in an ATM self-service area lacks a corresponding video access institution ID). For example, the institution ID in the original risk model data comes from the banking system; 24-hour ATMs located next to branches are considered off-site devices in the banking system, and their institution IDs differ from those of on-site devices; and since the video system's cameras all use a single host storage, it cannot distinguish between off-site and on-site cameras, resulting in the inability to query camera ID information by institution ID. Therefore, the "Camera Information - Branch" mapping table can be searched using the institution ID as the primary identifier and the institution name as a secondary identifier, ensuring that the risk data can access all camera ID information. Therefore, in this embodiment, the system needs to perform a refined matching process for the full name of the institution in each risk model data entry that cannot match the institution ID to connect to the institution number in the video resource library, i.e., to determine the target financial branch. Please refer to... Figure 4 , Figure 4 This is a schematic diagram of a process for screening service points based on a multimodal intelligent matching method provided in an embodiment of this application, such as... Figure 4 As shown, the specific implementation process of selecting service points based on the multimodal intelligent matching method is as follows: A3. Data Organization Full Name Matching: Based on the mapping table between the full name of the data organization and the video resource organization, perform organization matching (organization is the network point) on each piece of risk model data in the structured risk model data after the second step of parsing; if the corresponding network point can be directly matched according to the mapping table between the full name of the data organization and the video resource organization, then directly update the organization number, organization name, etc. of the risk model data, and then end the process; if the corresponding network point cannot be directly matched according to the mapping table between the full name of the data organization and the video resource organization, then the following matching steps (B3-D3) need to be performed; B3. Key Element Extraction: Using regular expression technology, the core geographical identifiers and characteristic words in the full name of the institution in each risk model data are dynamically parsed. Taking "AA Bank BB City CC County DD Road Branch Self-Service Area" as an example, this process can accurately extract key elements such as "BB City", "CC County", and "DD Road". C3. Multimodal Intelligent Matching: Key elements extracted from the full name of the organization are divided into three combinations for composite querying. A dual-track parallel mechanism of "fuzzy matching + precise matching" is used for the search, and double verification ensures that the responsible organization of each risk model data can be accurately mapped to the corresponding network in the video resource organization table. The specific matching method is as follows: Combination 1 (Full Name Precise Query): Integrates the full name of "city + district / county + outlet" and performs precise matching with the full name field of the organization in the "Data Organization Full Name and Video Resource Organization Mapping Relationship Table" to generate result set A; Combination 2 (City-level Fuzzy Query): Using "City + Outlet" as an independent item, perform fuzzy matching on the city name and outlet name fields in the "Data Institution Full Name and Video Resource Institution Mapping Relationship Table" respectively; Combination 3 (District / County-level Fuzzy Query): Using "District / County + Outlet" as an independent item, perform fuzzy matching on the "Data Institution Full Name and Video Resource Institution Mapping Relationship Table" fields for the district / county name and outlet name respectively; D3. Result Set Processing: Perform an intersection operation on the fuzzy query results of combination 2 and combination 3 to generate result set B. Based on priority, first verify the uniqueness of result set A. If set A contains a unique exact match, the institution is considered successfully matched. If set A does not match, verify the uniqueness of set B. If set B contains a unique exact match, the institution is considered successfully matched; otherwise, it indicates that the keyword is unclear and the institution cannot be accurately matched, i.e., the branch cannot be accurately matched. After verifying the uniqueness of the result set, update the mapping table between the full name of the data institution and the video resource institution based on the unique verified result set, and update the institution number and institution name in the risk model data corresponding to the unique verified result set, then end the process.

[0065] The fourth step is to accurately push video cameras, including sorting video cameras by popularity value and accurately pushing video resources (input: structured risk model data, "risk model data - network video resource institution ID" association table, "video camera information - network" mapping table (initialization information from the first step); output: list of video cameras to be accessed). The structured risk model data includes data such as trading institutions (which can be used as the institutions to access the data for matching), teller numbers, and customer numbers (used as the historical basis for risk model data to query whether there are any violations by the customer or teller in the past).

[0066] A4. Precise Screening Mechanism: Employing a dual composite condition of "video access agency (site) + model selection area", the system achieves precise identification and locking of all monitoring devices within a specific area of ​​the target site.

[0067] B4. Popularity Ranking: Based on the access area selected by the risk model, all cameras within the jurisdiction of the video access agency are ranked according to their own popularity value.

[0068] C4. Intelligent priority push strategy: By default, the cameras are sorted from high to low based on their popularity value in the target area; a comprehensive evaluation is conducted based on the historical judgment results of the risk model data, and finally a list of cameras to be reviewed is pushed. After the cameras to be reviewed are determined, the required video resources can be located directly.

[0069] For example, assuming the current risk model is an employee risk transaction model, if a query finds that the employee has a history of violations, the system will prioritize pushing the surveillance cameras associated with the employee's most recent violation record. If there is no relevant historical record or no problems have been found in the past, the system will still sort the surveillance cameras in descending order of popularity value and finally push the video resources of the top N (e.g., N=10) surveillance cameras to cover the basic list of cameras to be accessed.

[0070] D4. Initial Configuration for Video Access: In this embodiment, when inspectors enter the video access interface, the system displays the video feed in a four-grid format by default, automatically playing the videos from the top four most popular surveillance cameras. The video playback start time is precisely set to M seconds (e.g., M=10) before the data transaction occurs. In addition, the system simultaneously displays "Inspection Points" prompts, clearly listing the key elements requiring focused inspection in the current risk data and the user information involved, assisting inspectors in carrying out their work efficiently and facilitating quick identification of risky users, risky times, and event locations.

[0071] It should be noted that, in specific implementations, the number of monitoring screens displayed on the visualization page can be set according to the actual situation, and this application embodiment does not impose any restrictions on this.

[0072] Step 5: Log collection; The operation logs of the cameras actually accessed by the verification personnel are collected by embedding data points. The operation logs include, but are not limited to, the number of clicks, the access duration, and the click order.

[0073] Step 6: Based on the dynamic heat value generation algorithm, dynamically update the camera heat value (input: verification personnel operation log, dynamic heat value weight; output: updated camera heat value). By recording the video viewing operation logs of the inspectors (including but not limited to the number of times the camera was clicked, the start and end times of the viewing, and the viewing order), a linear weighted model is used to calculate the camera popularity value, providing a basis for subsequent camera recommendations. Specifically: after the camera data for each risk model is matched, inspectors can view the video of each risk model data item one by one. During the video viewing process, the system records the inspectors' video viewing operation logs (taking the number of times the camera was clicked, the start and end times of the viewing, and the viewing order as an example), and uses a linear weighted model to calculate the camera popularity value, providing a basis for subsequent recommendations. The specific calculation process is as follows: A6. Calculation of scores for each factor: (1) Click count score: Calculated directly based on the actual click frequency. The formula for calculating the click count score is: Click count score = number of clicks * weight w1; For example: If a camera is clicked 5 times, the score for the number of clicks = 5 (number of clicks) × 0.4 (w1) = 2 points; (2) Scoring based on access duration: The score is set according to the access duration range, and then multiplied by the weight w2, as shown in Table 1 below: Table 1

[0074] (3) Score based on the order of access: The score decreases sequentially from the first click to subsequent clicks, and then the scores are accumulated and multiplied by the weight w3, as follows: Specifically, upon entering the viewing interface, the first click on the camera earns 10 points, the second click earns 9.9 points, and so on down to 0.1 points. The formula for calculating the viewing order score is: Viewing order score = cumulative single-click score * weight w3; For example: If a camera is clicked 1st and 3rd in sequence, the cumulative score is 10 + 9.8 = 19.8 points, and the score for the order of access is the cumulative score of 19.8 × 0.3 (w3) = 5.94 points.

[0075] B6. Final Heat Value Calculation: The popularity score is calculated as follows: number of clicks + viewing time + viewing order.

[0076] For example: In the example above, the final heat value of the camera gun = 2 + 0.6 + 5.94 = 8.54 points.

[0077] It should be noted that in practical applications, the weight values ​​and specific scoring rules in the calculation of each factor score can be set according to the actual situation.

[0078] Step 7: Update the weight value dynamically based on the dynamic heat value weighting algorithm (Input: Verification personnel operation log; Output: Updated weight value).

[0079] A7. Set multi-factor weights for different risk models: Each model is set with 3 core influencing factors and their default weights. The weights between each risk model do not affect each other and are only used for the calculation of the camera heat value of the current risk model.

[0080] (1) Weight design principle: In the initial stage, the weights of each risk model are initialized manually based on the data analysis results. Subsequently, the weights will be dynamically adjusted based on the risk model data of each risk model.

[0081] (2) Initial weight allocation: The weights can be adjusted according to the actual review needs. The default values ​​are: w1=0.4, w2=0.3, w3=0.3, w1+w2+w3=1. Among them, the weight w1 of "number of clicks" reflects the frequency of the camera being viewed; the weight w2 of "review duration" reflects the practical value of the camera; and the weight w3 of "review order" reflects the priority of using the camera.

[0082] (3) Weight update rules: 1) Scheduled triggering: For example, every Sunday, the weight of each risk model is calculated and updated according to the rules for the following week.

[0083] 2) Itemized update rules: The weighting update rules for "click count" are shown in Table 2: Table 2

[0084] The weighting update rules for "access duration" are shown in Table 3: Table 3

[0085] The weight update rule for "review order" is as follows: the weight of the review order depends on the weight values ​​of the number of clicks and the review duration, and the update rule is w3=1-w1-w2.

[0086] It should be noted that this embodiment is only a brief illustrative description of the overall process of a video resource positioning method using a camera gun. Detailed descriptions of each step can be found in the relevant content of the foregoing embodiments, and will not be repeated here. It is understood that this application does not impose any limitations on this.

[0087] This application embodiment obtains raw camera information from several financial outlets and cleans the raw camera information of each financial outlet to obtain target camera information for each financial outlet; it obtains raw risk model data for the daily review high-risk model screening indicators and performs data parsing processing on the raw risk model data to obtain structured risk model data; according to a preset institution matching method, it matches each risk model data in the structured risk model data with financial outlets to determine the target financial outlet corresponding to each risk model data in the structured risk model data; based on the structured risk model data, the target financial outlet corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera, it determines several currently accessed cameras; it obtains the current user operation log and updates the current popularity value corresponding to each camera based on the current user operation log; based on the updated current popularity value corresponding to each camera, it determines several updated currently accessed cameras; wherein, each camera has captured corresponding video resources. This application embodiment acquires raw risk model data for the high-risk model screening indicators of the daily review and performs data parsing processing on the raw risk model data. The final result is structured risk model data containing the key points of the daily review's high-risk checks. This allows inspectors to quickly locate the key points of risk data verification. Specifically, inspectors can access video data according to the risk event guidelines, reducing reliance on personnel experience during the daily review process and thus improving review quality and efficiency. By identifying several currently accessed video cameras, video resources can be accurately located, reducing repetitive operations in video retrieval. Furthermore, by collecting video access operation logs generated by inspectors actually accessing video cameras, a heat value for each camera is dynamically generated. Therefore, during the next daily review, the structured risk model data can be combined to accurately locate video data within the time period of the event, achieving precise video resource delivery, reducing the large amount of repetitive operations generated by manual video retrieval, and improving video verification efficiency.

[0088] In summary, the camera video resource location method provided in this application aims to achieve accurate location and push of camera video resources related to daily review by using risk model data as a driving force and employing multiple technologies such as key element extraction algorithms, multimodal intelligent matching algorithms, and camera heat value algorithms. Specifically, the camera video resource location method provided in this application has two advantages: First, it is data-driven, allowing reviewers to access video data according to risk event guidelines, reducing reliance on personnel experience in the daily review process; second, through the application of multiple technologies such as key element extraction, multimodal intelligent matching, and camera heat value algorithms, it achieves recommendations for the order of camera viewing and playback time periods. The camera heat value algorithm, in particular, employs a self-learning mechanism that mimics human operational behavior in its camera heat value and factor weights, exhibiting strong interpretability.

[0089] This application utilizes various technologies, including risk model data-driven approach, key element extraction, multimodal intelligent matching, and camera heat value algorithm, to accurately locate the range of camera video resources and recommend the viewing order, achieving precise delivery of video resources needed for daily review. Inspectors only need to click on the data to view the multiple pushed videos sequentially and conduct risk point investigations, reducing the operational overhead of selecting cameras and dragging and dropping videos, thus improving video inspection efficiency.

[0090] The key points of this application's embodiments are as follows: (1) Video resource pool search method with dual verification of "fuzzy matching + precise matching": The video resources are retrieved from the video resource library of the network site by using algorithms such as key element extraction and multimodal intelligent matching. The dual verification of "fuzzy matching + precise matching" can ensure that each piece of risk data can be accurately and comprehensively mapped to the video resources of the network site.

[0091] (2) Camera heat value algorithm: The heat value of each camera in the model network is dynamically generated by the video viewing operation logs of the past inspectors (including but not limited to the number of times the camera was clicked, the start and end time of the viewing, the viewing order, etc.). So that in the next review, the top N cameras with the highest heat value are selected by default, and the video resource data of the time period when the event occurred is accurately located by combining the risk model data.

[0092] (3) Self-learning algorithm for the weight of heat value factor: Based on the statistical value of the operation behavior of the inspectors, the weight of heat factor is dynamically updated. Furthermore, the heat value algorithm of the camera gun has the self-learning ability to imitate the operation of the personnel, which can greatly improve the efficiency of video inspection and the accuracy of risk identification, while reducing the deployment and maintenance costs, and realizing the intelligent and lightweight upgrade of the daily review of financial security outlets.

[0093] This application embodiment utilizes various technologies, including risk model data-driven analysis, key element extraction algorithms, multimodal intelligent matching algorithms, and camera heat value algorithms, to achieve an intelligent upgrade of daily review video inspection. Specifically, it has the following significant advantages: (1) Lowering the operational threshold and reducing reliance on experience: There is no need for inspectors to rely on experience to screen outlets and locate videos. Inspectors can quickly grasp the key points of the review, which solves the problem of "uneven review quality caused by experience differences". (2) Improve search efficiency and reduce repetitive operations: By using trending topics and precise time positioning, the video search efficiency is greatly improved by avoiding repeatedly dragging the progress bar and switching cameras. (3) Cover more outlets and reduce risk omissions: Based on the risk model data, the outlets are accurately matched, which gets rid of the limitations of "random sampling" and greatly improves the coverage of high-risk outlets. (4) Dynamically optimize push notifications to improve accuracy: The user operation log drives the update of the camera's heat value, and the push accuracy gradually improves with the frequency of use, reducing the amount of invalid camera push notifications; (5) Standardize the review process and strengthen risk control: deeply link risk data with video review to form a closed-loop process of "risk positioning - video verification - data feedback", which significantly enhances the standardization of review and risk control capabilities.

[0094] Please see Figure 5 This application also provides a camera video resource positioning device 500, which can implement the above method. The device includes the following modules: The camera information acquisition module 501 is used to acquire the original camera information of several financial outlets, and to perform data cleaning on the original camera information of each financial outlet to obtain the target camera information of each financial outlet. The risk model data parsing module 502 is used to acquire the original risk model data for the daily review high-risk model screening indicators, and to perform data parsing processing on the original risk model data to obtain structured risk model data. The financial outlet matching module 503 is used to match financial outlets for each risk model data in the structured risk model data according to a preset institution matching method, and to determine the target financial outlet corresponding to each risk model data in the structured risk model data. The current access camera determination module 504 is used to determine a number of current access cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera. The camera heat value update module 505 is used to obtain the current user operation log and update the current heat value corresponding to each camera according to the current user operation log. The current access camera update module 506 is used to determine a number of updated current access cameras based on the updated current popularity value of each camera; wherein each camera has captured corresponding video resources.

[0095] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0096] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0097] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0098] Please see Figure 6 , Figure 6 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 601 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 602 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 602 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 602 and is called and executed by the processor 601 using the methods described in the embodiments of this application. The input / output interface 603 is used to implement information input and output; The communication interface 604 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 605 transmits information between various components of the device (e.g., processor 601, memory 602, input / output interface 603, and communication interface 604); The processor 601, memory 602, input / output interface 603, and communication interface 604 are connected to each other within the device via bus 605.

[0099] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0100] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0101] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0102] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0103] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0104] This application provides a method and related equipment for locating video resources from surveillance cameras. It acquires raw surveillance camera information from several financial outlets, cleans the data of each financial outlet to obtain target surveillance camera information for each outlet; acquires raw risk model data for high-risk model screening indicators in daily review, and performs data parsing processing on the raw risk model data to obtain structured risk model data; and matches each risk model data in the structured risk model data with financial outlets according to a preset institution matching method to determine the number of structured risk models. The system identifies the target financial outlets corresponding to each risk model data point; it also identifies several current access cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data point, the target camera information for each financial outlet, and the current popularity value of each camera; it retrieves the current user operation log and updates the current popularity value for each camera based on the current user operation log; and it identifies several updated current access cameras based on the updated current popularity value for each camera; each camera captures corresponding video resources. This application embodiment acquires raw risk model data for the high-risk model screening indicators of the daily review and performs data parsing processing on the raw risk model data. The final result is structured risk model data containing the key points of the daily review's high-risk checks. This allows inspectors to quickly locate the key points of risk data verification. Specifically, inspectors can access video data according to the risk event guidelines, reducing reliance on personnel experience during the daily review process and thus improving review quality and efficiency. By identifying several currently accessed video cameras, video resources can be accurately located, reducing repetitive operations in video retrieval. Furthermore, by collecting video access operation logs generated by inspectors actually accessing video cameras, a heat value for each camera is dynamically generated. Therefore, during the next daily review, the structured risk model data can be combined to accurately locate video data within the time period of the event, achieving precise video resource delivery, reducing the large amount of repetitive operations generated by manual video retrieval, and improving video verification efficiency.

[0105] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0106] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0107] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0108] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0109] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0110] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0111] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0112] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0113] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0114] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0115] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for locating video resources from a camera, characterized in that, The method includes the following steps: The raw camera information of several financial outlets is obtained, and the raw camera information of each financial outlet is cleaned to obtain the target camera information of each financial outlet. Obtain the raw risk model data for the high-risk model screening indicators in the daily review, and perform data parsing processing on the raw risk model data to obtain structured risk model data; According to the preset institution matching method, financial outlet matching is performed on each risk model data in the structured risk model data to determine the target financial outlet corresponding to each risk model data in the structured risk model data; Based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera, a number of cameras to be currently accessed are determined. Obtain the current user operation log, and update the current popularity value corresponding to each camera gun according to the current user operation log; Based on the updated current popularity value corresponding to each of the aforementioned cameras, a number of currently accessing cameras are determined; wherein each of the aforementioned cameras has captured corresponding video resources.

2. The method according to claim 1, characterized in that, The process of acquiring raw camera information from several financial outlets and cleaning the raw camera information from each of the financial outlets to obtain target camera information for each of the financial outlets includes: Obtain the original camera information from several of the aforementioned financial outlets; According to the preset camera naming rules, the original camera information of each of the financial outlets is integrated and processed to obtain the original camera outlet mapping table; The original camera outlet mapping table is cleaned to obtain the target camera outlet mapping table; wherein, the target camera outlet mapping table is used to represent the target camera information corresponding to each of the financial outlets, and the target camera outlet mapping table is a mapping table of the relationship between camera and financial outlet.

3. The method according to claim 1, characterized in that, The process involves acquiring raw risk model data for the high-risk model screening indicators in daily review, and performing data parsing processing on the raw risk model data to obtain structured risk model data, including: Obtain the original risk model data for the high-risk model screening indicators of the daily review, and input the original risk model data into the corresponding risk model; Using regular expression matching and semantic parsing techniques, key fields are extracted from the original risk model data to obtain key field data in the original risk model data. Based on the key field data and the risk dimensions of the risk model, the structured risk model data is constructed and displayed in the video viewing interface; wherein, the structured risk model data includes the key field data and core inspection elements.

4. The method according to claim 1, characterized in that, The step of matching each risk model data in the structured risk model data with financial outlets according to a preset institution matching method to determine the target financial outlet corresponding to each risk model data in the structured risk model data includes: According to the preset institution mapping relationship table, institution information is matched for each piece of risk model data in the structured risk model data; If there is risk model data that fails to match, a multimodal intelligent matching method is used to perform composite matching processing on the risk model data that fails to match, and a composite matching result is obtained; wherein, the composite matching result includes an exact matching result and a fuzzy matching result, and the fuzzy matching result includes a first fuzzy matching result and a second fuzzy matching result; Perform an intersection operation on the first fuzzy matching result and the second fuzzy matching result to generate a double fuzzy matching result; The exact matching results are verified for uniqueness. If the exact matching results pass the uniqueness verification, the target financial outlets corresponding to each risk model data in the structured risk model data are determined based on the exact matching results. If the exact matching result fails the uniqueness verification, then the double fuzzy matching result is subjected to uniqueness verification. If the dual fuzzy matching result passes the uniqueness verification, then the target financial outlet corresponding to each risk model data in the structured risk model data is determined based on the dual fuzzy matching result.

5. The method according to claim 1, characterized in that, The step of determining several currently accessed surveillance cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data point in the structured risk model data, the target surveillance camera information of each financial outlet, and the current popularity value of each surveillance camera includes: Based on the preset dual screening mechanism, and combining the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, and the target camera information of each financial outlet, all cameras in the target outlet area are determined. The heat values ​​of each camera within the target network area are sorted to obtain the heat value sorting results; Based on the ranking results of the heat values ​​and the historical judgment results of each risk model data in the structured risk model data, a number of the currently accessing cameras are determined through comprehensive evaluation.

6. The method according to claim 1, characterized in that, The step of obtaining the current user operation log and updating the current popularity value corresponding to each camera based on the current user operation log includes: Obtain the current user operation log; wherein, the current user operation log is the log of the actual access to the camera by the verification personnel; Based on the current user operation log, calculate the model weight value corresponding to each camera; A linear weighted model is used, combining the user operation logs corresponding to each camera and the model weight value, to calculate the updated current popularity value of each camera.

7. A video resource positioning device for a camera, characterized in that, The device includes the following modules: The camera information acquisition module is used to acquire the original camera information of several financial outlets, and to perform data cleaning on the original camera information of each financial outlet to obtain the target camera information of each financial outlet. The risk model data parsing module is used to acquire the original risk model data for the high-risk model screening indicators of the daily review, and to perform data parsing processing on the original risk model data to obtain structured risk model data. The financial outlet matching module is used to match financial outlets for each risk model data in the structured risk model data according to a preset institution matching method, and to determine the target financial outlet corresponding to each risk model data in the structured risk model data. The current access camera determination module is used to determine a number of current access cameras based on the structured risk model data, the target financial outlets corresponding to each risk model data in the structured risk model data, the target camera information of each financial outlet, and the current popularity value of each camera. The camera heat value update module is used to obtain the current user operation log and update the current heat value corresponding to each camera according to the current user operation log; The current access camera update module is used to determine a number of updated current access cameras based on the updated current popularity value of each camera; wherein each camera has captured corresponding video resources.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images