Risk identification method and device based on target detection, computer equipment and medium

Abstract

The invention provides a risk identification method and device based on target detection, computer equipment and a medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: acquiring monitoring video data of a employee performance place; performing target detection on the monitoring video data through a pre-trained target detection model to obtain a risk target in the monitoring video data; the risk target represents that a target employee in the employee performance place records secret-related data content; continuously tracking the risk target to obtain the duration of recording the confidential data content by the target employee; and when the duration is greater than or equal to a preset duration, performing early warning for the risk target. According to the invention, intelligent detection identification and early warning can be carried out on illegal behaviors of leakage of secret-related data of employees, so that the cost for checking and inspecting the illegal behaviors of leakage of secret-related data is reduced.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a risk identification method, device, computer equipment and medium based on target detection. Background Art

[0002] At present, the investigation and inspection of the violation of confidential data leakage by employees in the workplace is mainly carried out through on-site visits and review of historical monitoring data after the data leakage incident occurs. Since this method often requires a complicated process, it is very time-consuming and labor-intensive, and in order to meet the monitoring needs, the corresponding economic cost has to be increased for long-term storage of historical monitoring data. This means that the overall cost of investigating and inspecting the violation of confidential data leakage is currently high.

[0003] Therefore, how to reduce the cost of investigating and examining violations that lead to the leakage of confidential data has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose a risk identification method, device, computer equipment and medium based on target detection, aiming to intelligently detect, identify and warn of violations of confidential data by employees in the workplace, thereby reducing the cost of investigating and examining violations of confidential data.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes a risk identification method based on target detection, the method comprising:

[0006] Obtain surveillance video data of the workplace where employees perform their duties;

[0007] Performing target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content;

[0008] Continuously track the risk target to obtain the duration for which the target employee records confidential data;

[0009] When the duration is greater than or equal to the preset duration, an early warning is issued for the risk target.

[0010] In some embodiments, before performing target detection on the surveillance video data using a pre-trained target detection model to obtain risk targets in the surveillance video data, the method further includes:

[0011] Acquire surveillance video sample data; the surveillance video sample data includes positive samples and negative samples, the positive samples represent that the employee records confidential data content, the negative samples represent that the employee does not record confidential data content, and the number of positive samples is greater than the number of negative samples;

[0012] The target detection model is obtained by performing model training on the positive samples detected by a preset target detection algorithm.

[0013] In some embodiments, obtaining surveillance video sample data includes:

[0014] In the surveillance video data of the workplace where the employee performs his / her duties, a first target image data frame containing confidential data content and employee violation data content is detected, and a second target image data frame not containing the employee violation data content is detected; wherein the employee violation data content is image data content representing the confidential data content recorded by the employee;

[0015] The first target image data frame in the monitoring video data is marked as a positive sample, and a part of the second target image data frame in the monitoring video data is marked as a negative sample to obtain monitoring video sample data; the number of the positive samples is higher than the number of the negative samples.

[0016] In some embodiments, the confidential data content includes: image data content showing the identity information of the insurance customer, and / or image data content showing the insurance policy information of the insurance customer.

[0017] In some embodiments, the early warning for the risk target includes at least one of the following:

[0018] Broadcasting a preset voice warning prompt at the employee's workplace;

[0019] Output a preset voice warning prompt to the target employee who records confidential data in the workplace where the employee performs his / her duties;

[0020] The target employees who record confidential data content in the workplaces where the employees perform their duties are marked as employees with high risk of leaking confidential data.

[0021] In some embodiments, after acquiring the surveillance video data of the workplace where the employee performs his / her duties, the method further includes:

[0022] Performing target detection on the surveillance video data using a pre-trained target detection model to obtain confidential data content in the surveillance video data;

[0023] Continuously tracking the confidential data content to obtain the duration of the target employee recording the confidential data content at the employee's workplace;

[0024] When the duration is greater than or equal to the preset duration, an early warning is issued for recording the confidential data content for the target employee.

[0025] In some embodiments, the continuously tracking the confidential data content to obtain the duration of the target employee recording the confidential data content in the employee's workplace includes:

[0026] The confidential data content is continuously tracked based on a preset multi-target tracking algorithm to identify illegal behaviors in the surveillance video data; the illegal behaviors include: the target employee in the employee's workplace records the confidential data content;

[0027] When the violation is identified, the duration for which the target employee records the confidential data content is recorded.

[0028] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application proposes a risk identification device based on target detection, the device comprising:

[0029] An acquisition module is used to obtain surveillance video data of the workplace where employees perform their duties;

[0030] A target detection module is used to perform target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content;

[0031] A target tracking module is used to continuously track the risk target and obtain the duration of the target employee recording confidential data content;

[0032] The risk warning module is used to issue a warning for the risk target when the duration is greater than or equal to a preset duration.

[0033] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes a computer device, which includes a memory and a processor, the memory stores a computer program, and the processor implements the method described in the first aspect when executing the computer program.

[0034] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0035] To achieve the above-mentioned purpose, the fifth aspect of the embodiments of the present application proposes a computer program product, which stores a computer program. When the computer program is executed by a processor, the method described in the first aspect is implemented.

[0036] The target detection-based risk identification method, device, computer equipment, computer-readable storage medium and computer program product proposed in the present application obtain surveillance video data of employees' workplaces; perform target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employees' workplaces recording confidential data content; the risk targets are continuously tracked to obtain the duration of the target employees recording confidential data content; and when the duration is greater than or equal to a preset duration, an early warning is issued for the risk targets.

[0037] The embodiment of the present application pre-trains a target detection model, and uses the target detection model to perform target detection on the surveillance video data of the employee's job performance scene, obtains the risk target in the surveillance video data that represents the target employee recording confidential data content, and then combines it with continuous tracking of the risk target to issue an early warning for the risk target when it is found that the target employee records confidential data content for a duration greater than the preset duration. In this way, compared with the method of investigating and inspecting violations through on-site visits and viewing historical monitoring data, the embodiment of the present application uses the target detection model to intelligently detect, identify and warn of workplace employees' violations of confidential data leakage, which can shift security risks to the left. When the employee only records confidential data content but has not actually caused confidential data leakage, the employee's illegal behavior that may cause confidential data leakage can be stopped in time, thereby greatly reducing the cost of investigating and inspecting violations of confidential data leakage in the later stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of the steps of the target detection-based risk identification method provided in some embodiments of the present application;

[0039] Figure 2 A schematic diagram of a model training step flow involved in a target detection-based risk identification method provided in an embodiment of the present application;

[0040] Figure 3 for Figure 2 Schematic diagram of detailed process flow of step S201;

[0041] Figure 4 A schematic diagram of the steps of the target detection-based risk identification method provided in the embodiment of the present application in other embodiments;

[0042] Figure 5 A logical flow diagram of a complete embodiment of a target detection risk identification method provided in an embodiment of the present application;

[0043] Figure 6 is a schematic diagram of the structure of a risk identification device based on target detection provided in an embodiment of the present application;

[0044] Figure 7 It is a schematic diagram of the hardware structure of the computer device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0050] First, some nouns involved in this application are analyzed:

[0051] Artificial intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0052] Natural language processing (NLP): NLP uses computers to process, understand and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is often used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0053] Information Extraction: A text processing technology that extracts specified types of entity, relationship, event and other factual information from natural language text and forms structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data are all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.

[0054] Image caption generates a natural language description for an image and uses the generated description to help applications understand the semantics expressed in the image visual scene. For example, image captioning can convert image retrieval into text retrieval, which is used to classify images and improve image retrieval results. People usually only need to take a quick glance to describe the details of the image visual scene, while automatically adding descriptions to images is a comprehensive and arduous computer vision task that requires converting the complex information contained in the image into a natural language description. Compared with ordinary computer vision tasks, image captioning not only requires identifying objects from images, but also requires associating the identified objects with natural semantics and describing them in natural language. Therefore, image description requires people to extract deep features of images, associate them with semantic features, and convert them for generating descriptions.

[0055] Next, the overall concept of the embodiments of the present application is described.

[0056] With the continuous development of the company's business, the company's employees and business systems are frequently released and upgraded. In particular, insurance business personnel often come into contact with users' insurance information. The security control of these data assets within the company is becoming more and more important. Internal data leakage will not only cause serious economic losses to the company, but also cause reputational losses. Although the company has legal or internal system documents such as the "Confidentiality Agreement", "Terminal Security Management Measures", and "Employee Code of Conduct" to manage employees, it is impossible to avoid data leakage caused by employees unintentionally. What's more, some company employees may also maliciously leak company data, for example, by taking screenshots of the company's internal application (Application, APP) software or taking photos of confidential data at the workplace, and then deliberately leaking the obtained company confidential information to the outside, thereby causing a bad reputation for the company. In response to this, it is an indispensable part for the company to conduct internal information security inspections and inspections, and targeted security promotion is also essential, but it is particularly important to take timely measures to predict and stop the risk of leaking company content information in advance.

[0057] At present, the investigation and inspection of the violation of confidential data leaked by employees in the workplace is mainly carried out through on-site visits and reviewing historical monitoring data after the data leakage incident occurs. For example, after the company's information security department detects an emergency response incident, it will notify a specific security administrator to conduct a detailed on-site investigation of the incident without informing the parties in advance to confirm how the parties leaked the information. In order to further verify, it is necessary to retrieve the monitoring and view the historical time records. Since this method often requires a complicated process, it is very time-consuming and labor-intensive, and in order to meet the monitoring needs, the corresponding economic cost has to be increased for long-term storage of historical monitoring data. This means that the overall cost of investigating and inspecting the violation of confidential data leakage is currently high.

[0058] Therefore, how to reduce the cost of investigating and examining violations that lead to the leakage of confidential data has become a technical problem that needs to be solved urgently.

[0059] Based on this, the embodiments of the present application provide a risk identification method, apparatus, computer equipment, computer-readable storage medium and computer program product based on target detection, which obtains surveillance video data of employees' workplaces; performs target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employees' workplaces recording confidential data content; the risk targets are continuously tracked to obtain the duration of the target employees recording confidential data content; and when the duration is greater than or equal to a preset duration, an early warning is issued for the risk targets.

[0060] Compared to the method of investigating and examining violations through on-site visits and checking historical monitoring data, the embodiment of the present application pre-trains a target detection model, and uses the target detection model to perform target detection on the monitoring video data of the employee's job performance scenario, to obtain the risk target in the monitoring video data that represents that the target employee is recording confidential data content, and then continuously tracks the risk target, so as to issue an early warning for the risk target when it is found that the target employee has been recording confidential data content for a duration greater than the preset duration.

[0061] In this way, the embodiment of the present application uses a target detection model to intelligently detect, identify and warn of violations by employees in the workplace who leak confidential data, which can shift security risks to the left. When employees are only recording confidential data content but have not actually caused confidential data leakage, the employees' violations that may cause confidential data leakage can be stopped in time, thereby greatly reducing the cost of later investigation and inspection of violations that leak confidential data.

[0062] Based on the overall concept of the above-mentioned embodiments of the present application, specific embodiments of the risk identification method based on target detection, device, computer equipment, computer readable storage medium and computer program product provided in the embodiments of the present application are proposed. First, each specific embodiment of the risk identification method based on target detection in the embodiments of the present application is described in detail.

[0063] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0064] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0065] It should be noted that in each specific implementation of the embodiments of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiments of the present application need to obtain sensitive personal information of the user, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiments of the present application will be obtained.

[0066] In addition, the target detection-based risk identification method provided in the embodiment of the present application can be applied to the terminal, can also be applied to the server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the risk identification method based on target detection, etc., but is not limited to the above forms.

[0067] In addition, the present application embodiment can also be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer computer devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present 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. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0068] For ease of understanding and explanation, the following text will take the terminal device application provided by the embodiment of the present application as an example for detailed explanation. The implementation of the risk identification method based on target detection provided by the embodiment of the present application in any of the above-mentioned forms of the subject application can refer to the process of the terminal device application based on target detection risk identification method described later.

[0069] Please refer to Figure 1 , Figure 1 The following is a flow chart of the steps of the target detection-based risk identification method in some embodiments of the present application. Figure 1 The execution order of some method steps is shown in the figure, but based on different design requirements of actual applications, the target detection-based risk identification method provided in the embodiment of the present application can certainly adopt an execution order different from the method steps shown in the figure. That is, Figure 1 The order of the steps of the method shown does not constitute a limitation on the execution logic order of the risk identification method based on target detection provided in the embodiment of the present application. Figure 1 Reasonable changes in the order of the steps of the method shown should be included in the scope of protection of the target detection-based risk identification method provided in the embodiments of the present application.

[0070] like Figure 1 As shown, in some embodiments, the target detection-based risk identification method provided in the embodiments of the present application may include but is not limited to steps S101 to S104.

[0071] Step S101, obtaining surveillance video data of the workplace where the employee performs his / her duties.

[0072] When the terminal device identifies and warns employees of risky behaviors that may leak confidential data, it first obtains surveillance video data of the employees' workplace.

[0073] In some embodiments, the terminal device can apply for daily surveillance video data of the employee's workplace through the workplace employee reporting system.

[0074] Step S102, performing target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content.

[0075] After obtaining the surveillance video data, the terminal device immediately uses the surveillance video data as the model input of a pre-trained target detection model, thereby performing target detection on the surveillance video data through the target detection model to obtain risk targets in the surveillance video data that are used to characterize one or more target employees in the employee's workplace when recording confidential data content.

[0076] It should be noted that the risk target in the surveillance video data may be an image data frame, which shows the illegal behavior of the target employee in recording confidential data content.

[0077] In addition, the confidential data content includes: image data content showing the identity information of the insurance customer, and / or image data content showing the insurance customer's policy information.

[0078] Based on this, in some embodiments, the terminal device performs target detection on the monitoring video data through the target detection model to obtain the risk target. The target detection model can be used to first detect and identify whether there is image data content (image data content 1) showing the identity information of the insurance customer, and / or image data content (image data content 2) showing the insurance customer's policy information in the monitoring video data. Afterwards, when it is detected that the monitoring video data has image data content 1 and / or image data content 2, the target detection model is used to further detect and identify whether there is a violation of the target employee recording image data content 1 and / or image data content 2 in the image data frame where image data content 1 and / or image data content 2 exists. Therefore, when it is further identified that there is a violation in the image data frame, the terminal device marks the image data frame as a risk target in the monitoring video data.

[0079] Step S103, continuously tracking the risk target to obtain the duration of the target employee recording confidential data content.

[0080] After the terminal device performs target detection on the surveillance video data through the target detection model to obtain the risk targets in the surveillance video data, it further continuously tracks the risk targets to count the duration of the target employees recording confidential data content.

[0081] In some embodiments, the terminal device can continuously track risk targets in the video sequence through a multi-target tracking algorithm. For example, after the terminal device performs target detection on the surveillance video data through the target detection model to obtain the risk target in the surveillance video data, it further inputs the risk target as the target detection result into the multi-target tracking algorithm for training, thereby realizing continuous data analysis of the confidential data content recorded by the target employee in the surveillance video data, and then statistically obtaining the duration of the target employee recording the confidential data content.

[0082] It should be noted that the multi-target tracking algorithm can be a deepsort algorithm. The deepsort algorithm uses a Kalman filter to predict the target trajectory, so that risk targets in different frames of surveillance video data can be continuously tracked. It should be understood that based on the different design needs of actual applications, in other feasible implementations, the terminal device can of course also adopt other types of multi-target tracking algorithms that are different from those here. The risk identification method based on target detection provided in the embodiment of the present application is not limited to the specific type of multi-target tracking algorithm.

[0083] In some embodiments, the terminal device can perform an image description of the risk targets in the surveillance video data detected by the target detection model to output all risk targets in the surveillance video data as text data expressing the content of the confidential data recorded by the target employee. Then, the terminal device can obtain the duration of the target employee recording the confidential data content by performing natural language processing on these text data. For example, the terminal device can obtain text data such as "14:25:38, employee 1 took a photo of the insurance customer's insurance policy information", "14:25:39, employee 1 took a photo of the insurance customer's insurance policy information", ..., "14:26:05, employee 1 took a photo of the insurance customer's insurance policy information" by performing an image description on the risk targets in the surveillance video data. In this way, the terminal device extracts the time information by performing natural language processing on these text data, and then calculates the time information, and can obtain the duration of the violation of "employee 1 taking a photo of the insurance customer's insurance policy information" as 27 seconds.

[0084] Step S104: when the duration is greater than or equal to a preset duration, a warning is issued for the risk target.

[0085] After the terminal device obtains the duration of the target employee recording confidential data, it will compare the duration obtained each time with the preset duration (such as 0.5 seconds, 1 second, 3 seconds or 5 seconds, etc.). If the comparison finds that the duration is greater than or equal to the preset duration, it is determined that the target employee records confidential data and there is a risk that the confidential data will be leaked by the target employee. The risk target in the monitoring video data is immediately marked as a warning content to issue an early warning for the risk target. This is to stop the target employee's illegal behavior of recording confidential data in a timely manner, thereby avoiding the leakage of confidential data.

[0086] In addition to marking the risk target as warning content in the monitoring video data to issue an early warning for the risk target, the terminal device can also directly issue an early warning for violations of confidential data content recorded by the target employee represented by the risk target.

[0087] In some embodiments, in the above step S104, the early warning for the risk target may include at least one of the following:

[0088] Broadcasting a preset voice warning prompt at the employee's workplace;

[0089] Output a preset voice warning prompt to the target employee who records confidential data in the workplace where the employee performs his / her duties;

[0090] The target employees who record confidential data content in the workplaces where the employees perform their duties are marked as employees with high risk of leaking confidential data.

[0091] It should be noted that the preset voice warning prompt can be a voice prompt to remind the target employee to stop recording confidential data content. It should be understood that based on the different design needs of actual applications, in different feasible implementations, the terminal device can of course specifically set voice warning prompts with different voice contents to remind the target employee to stop recording confidential data content. The target detection-based risk identification method provided in the embodiment of the present application is not limited to the specific voice content of the voice warning prompt.

[0092] When the terminal device has determined that there is a risk of confidential data being leaked by the target employee, it can also automatically play a preset voice warning prompt in the form of a broadcast at the employee's workplace through an audio output device installed at the employee's workplace.

[0093] In addition to broadcasting voice warning prompts in the employee's work scene, the terminal device can also output a preset voice warning prompt separately to the target employee who is recording confidential data content in the employee's work site, thereby reminding the target employee to stop recording confidential data content.

[0094] While outputting voice warning prompts, the terminal device can also mark the target employees who record confidential data content at the workplace as high-risk confidential employees in the background, thereby facilitating subsequent management and control of the target employees to prevent them from leaking confidential data content.

[0095] In an embodiment of the present application, when identifying and warning the risk behavior of employees that may leak confidential data content through the terminal device, the monitoring video data of the employee's workplace is first obtained. Afterwards, the terminal device uses the monitoring video data as the model input of the pre-trained target detection model, so as to perform target detection on the monitoring video data through the target detection model to obtain the risk target in the monitoring video data used to characterize one or more target employees in the employee's workplace recording confidential data content. After that, the risk target is further tracked by the terminal device to count the duration of the target employee recording confidential data content. Finally, when the duration is greater than or equal to the preset duration, it is determined that the target employee records the confidential data content and there is a risk that the confidential data content is leaked by the target employee, so that the terminal device immediately issues an early warning for the risk target. In order to promptly stop the illegal behavior of the target employee recording confidential data content, thereby avoiding the leakage of confidential data content.

[0096] Compared to the method of investigating and examining violations through on-site visits and checking historical monitoring data, the embodiment of the present application pre-trains a target detection model, and uses the target detection model to perform target detection on the monitoring video data of the employee's job performance scenario, to obtain the risk target in the monitoring video data that represents that the target employee is recording confidential data content, and then continuously tracks the risk target, so as to issue an early warning for the risk target when it is found that the target employee has been recording confidential data content for a duration greater than the preset duration.

[0097] In this way, the embodiment of the present application uses a target detection model to intelligently detect, identify and warn of violations by employees in the workplace who leak confidential data, which can shift security risks to the left. When employees are only recording confidential data content but have not actually caused confidential data leakage, the employees' violations that may cause confidential data leakage can be stopped in time, thereby greatly reducing the cost of later investigation and inspection of violations that leak confidential data.

[0098] Before the terminal device formally performs target detection on the surveillance video data, it uses a target detection algorithm with pre-configured parameters to perform model training on the constructed surveillance video sample data, thereby obtaining a target detection model that can perform target detection on the surveillance video data of the employee's work site.

[0099] See also Figure 2 , Figure 2A schematic flow chart of a model training step involved in a target detection-based risk identification method provided in an embodiment of the present application.

[0100] like Figure 2 In some embodiments, in the aforementioned step S102, before performing target detection on the monitoring video data through a pre-trained target detection model to obtain risk targets in the monitoring video data, the target detection-based risk identification method provided in the embodiment of the present application may also include steps S201 to S202 as shown below.

[0101] Step S201, obtaining surveillance video sample data; the surveillance video sample data includes positive samples and negative samples, the positive samples represent that the employee records confidential data content, the negative samples represent that the employee does not record confidential data content, and the number of positive samples is higher than the number of negative samples.

[0102] Step S202: Perform model training on the positive samples using a preset target detection algorithm to obtain a target detection model.

[0103] It should be noted that the preset target detection algorithm can be a target detection algorithm based on computer vision (such as yolov5s). It should be understood that based on the different design requirements of actual applications, in other feasible implementations, the terminal device can of course also use other types of target detection algorithms to perform model training to obtain the final target detection model. The target detection-based risk identification method provided in the embodiment of the present application does not limit the specific type of target detection algorithm used to train the target detection model.

[0104] The terminal device obtains the pre-built surveillance video sample data and uses it as the model input. The surveillance video sample data includes positive samples and negative samples. The positive samples are used to represent the confidential data recorded by the employees, while the negative samples are used to represent the confidential data not recorded by the employees. In addition, the number of positive samples is higher than the number of negative samples.

[0105] After configuring the parameters of the preset target detection algorithm, the terminal device inputs the acquired surveillance video sample data into the target detection algorithm, so that the target detection algorithm can perform model training on the positive samples in the surveillance video sample data. After that, after the target detection algorithm converges with the model training for the surveillance video sample data, the terminal device can obtain the target detection model for detecting risk targets in the surveillance video data.

[0106] The terminal device can construct surveillance video sample data for model training based on the surveillance video data of the employees' work sites.

[0107] Please refer to Figure 3 , Figure 3 for Figure 2 Schematic diagram of the detailed step flow of step S201.

[0108] like Figure 3 As shown, in some embodiments, the aforementioned step S201, obtaining surveillance video sample data, may include steps S301 to S302 as shown below.

[0109] Step S301, in the surveillance video data of the employee's workplace, detecting a first target image data frame containing confidential data content and employee violation data content, and detecting a second target image data frame that does not contain the employee violation data content; wherein the employee violation data content is image data content that represents the employee recording the confidential data content.

[0110] When the terminal device trains the target detection model based on the target detection algorithm, for the sample data used to input the target detection algorithm, the terminal device obtains the surveillance video data of the employee's workplace, and then detects the first target image data frame in the surveillance video data that contains both confidential data content and employee violation data content, and detects the second target image data frame in the surveillance video data that only contains confidential data content but does not contain employee violation data content.

[0111] It should be noted that the employee violation behavior data content is the image data content that represents the confidential data content recorded by the employee. For example, in a frame of surveillance video data, an employee takes a mobile phone to take a photo of the insurance customer identity information displayed on the computer screen. In this frame of image, the insurance customer identity information displayed on the computer screen is confidential data content, and the employee takes a mobile phone to take a photo of the insurance customer identity information, which is the employee violation behavior data content. In this way, the terminal device can mark this frame of image as the first target image data frame.

[0112] Step S302: mark the first target image data frame in the surveillance video data as a positive sample, and mark a part of the second target image data frame in the surveillance video data as a negative sample, to obtain surveillance video sample data; the number of the positive samples is higher than the number of the negative samples.

[0113] After detecting the first target image data frame and the second target image data frame in the monitoring video data, the terminal device further marks the first target image data frame as a positive sample through an image data annotation tool, and then marks a part of the second target image data frame as a negative sample according to a preset ratio, thereby obtaining monitoring video sample data.

[0114] It should be noted that the preset ratio is the ratio between the number of positive samples and the number of negative samples, and the preset ratio indicates that the number of positive samples is higher than the number of negative samples. For example, the preset ratio may be 10:1, 5:1, 2:1, 1.1:1, and so on. In addition, the image data annotation tool used by the terminal device may be a target detection annotation tool labelimg, that is, the terminal device labels the first image data frame in the monitoring video data through labelimg to obtain positive samples, and selects some of the second target image data frames labeled as negative samples according to the preset ratio.

[0115] In some embodiments, the terminal device may also use an image data annotation tool to select some image frames in the monitoring video data that contain neither confidential data content nor employee violation data content and annotate them as negative samples.

[0116] In this embodiment, the terminal device obtains the pre-built surveillance video sample data and uses the surveillance video sample data as the model input. Then, after configuring the parameters of the preset target detection algorithm, the terminal device inputs the acquired surveillance video sample data into the target detection algorithm, so that the target detection algorithm performs model training on the positive samples in the surveillance video sample data. Afterwards, when the target detection algorithm converges in the model training for the surveillance video sample data, the terminal device can obtain the target detection model for detecting risk targets in the surveillance video data.

[0117] Thus, compared to the method of checking and inspecting illegal behaviors through on-site visits and checking historical monitoring data, the embodiment of the present application pre-trains a target detection model, and then uses the target detection model to perform target detection on the monitoring video data of the employee's performance of duties, so as to obtain the risk target in the monitoring video data that represents the target employee recording confidential data content. Finally, by continuously tracking the risk target, it is possible to issue an early warning for the risk target when it is found that the target employee records confidential data content for a duration greater than the preset duration. That is, the present embodiment can more simply and quickly achieve accurate speculation by using artificial intelligence methods to conduct supervised analysis and prediction of monitoring video data, thereby focusing on analyzing the predicted employee behavior risk issues such as recording confidential data in advance, thereby minimizing the risk of leaking internal company data.

[0118] See also Figure 4 , Figure 4 A schematic diagram of the step flow of the target detection-based risk identification method provided in the embodiment of the present application in other embodiments.

[0119] like Figure 4As shown, in some embodiments, after obtaining the surveillance video data of the employee's workplace as mentioned above, the target detection-based risk identification method provided in the embodiment of the present application may also include steps S401 to S403 as shown below.

[0120] Step S401: Perform target detection on the surveillance video data using a pre-trained target detection model to obtain confidential data content in the surveillance video data.

[0121] After the terminal device obtains the surveillance video data, in addition to using the target detection model to detect risk targets in the surveillance video data that represent one or more target employees in the employee's workplace recording confidential data content, it can also directly detect the confidential data content in the surveillance video data through a pre-trained target detection model, thereby obtaining the confidential data content in the surveillance video data.

[0122] Exemplarily, the terminal device directly detects and identifies whether there is image data content (image data content 1) showing the identity information of the insurance customer and / or image data content (image data content 2) showing the insurance customer's policy information in the surveillance video data through the target detection model. Afterwards, when it is detected that the surveillance video data contains image data content 1 and / or image data content 2, the terminal device marks the image data content 1 and / or image data content 2 as confidential data content in the surveillance video data.

[0123] Step S402, continuously tracking the confidential data content to obtain the duration of the target employee recording the confidential data content in the employee's workplace.

[0124] After the terminal device performs target detection on the surveillance video data through the target detection model to obtain the confidential data content in the surveillance video data, it further continuously tracks the confidential data content, and when it further identifies that one or more target employees record the confidential data content in the employee's workplace, it counts the duration for which the target employees record the confidential data content.

[0125] It should be noted that the terminal device can also track confidential data content through a multi-target tracking algorithm, and at the same time predict the behavior trajectories of other targets (such as employees) in the image frame containing confidential data content based on the multi-target tracking algorithm, thereby identifying the situation in which the target employee records the confidential data content, and when such a situation is identified, counting the duration of the target employee recording the confidential data content.

[0126] In some embodiments, step S402 may include:

[0127] The confidential data content is continuously tracked based on a preset multi-target tracking algorithm to identify illegal behaviors in the surveillance video data; the illegal behaviors include: the target employee in the employee's workplace records the confidential data content;

[0128] When the violation is identified, the duration for which the target employee records the confidential data content is recorded.

[0129] After the terminal device detects the target of the surveillance video data through the target detection model and obtains the confidential data content in the surveillance video data, it continuously tracks the detected data content based on the preset multi-target tracking algorithm (such as the deepsort algorithm), thereby identifying the illegal behavior of the target employee in the employee's workplace recording the confidential data content in the image frame containing the confidential data content. Afterwards, when the target employee is identified as violating the regulations by recording the confidential data content, the duration of the target employee recording the confidential data content is counted.

[0130] In some embodiments, the terminal device can describe the violations identified by the multi-target tracking algorithm in an image to output the violations of the target employee recording confidential data in the surveillance video data as text data. Then, the terminal device performs natural language processing on these text data to count the duration of the target employee recording confidential data. For example, the terminal device can obtain text data such as "14:25:38, employee 1 took a photo of the insurance customer's insurance policy information", "14:25:39, employee 1 took a photo of the insurance customer's insurance policy information", ..., "14:26:05, employee 1 took a photo of the insurance customer's insurance policy information" by describing the violations in the surveillance video data in an image. In this way, the terminal device extracts the time information by performing natural language processing on these text data, and then calculates the time information, and can statistically obtain that the duration of the violation "employee 1 took a photo of the insurance customer's insurance policy information" is 27 seconds.

[0131] Step S403: When the duration is greater than or equal to a preset duration, an alert is issued to the target employee to record the confidential data content.

[0132] Basically the same as the aforementioned terminal device issuing early warnings for risky targets, after the terminal device obtains the duration of the target employee recording confidential data content through statistics, it will synchronously compare the duration obtained for each statistics with the preset duration (such as 0.5 seconds, 1 second, 3 seconds or 5 seconds, etc.). When the comparison finds that the duration is greater than or equal to the preset duration, it is determined that there is a risk of the target employee recording confidential data content and the confidential data content is leaked by the target employee. Therefore, the violation of the target employee recording confidential data in the surveillance video data is immediately marked as a warning content, so as to issue an early warning for the target employee recording confidential data content.

[0133] In this embodiment, the monitoring video data is obtained through the terminal device, and the confidential data content in the monitoring video data is directly detected through the pre-trained target detection model, so as to obtain the confidential data content in the monitoring video data. After that, the confidential data content is continuously tracked through the terminal device, so that when one or more target employees are identified in the employee's workplace and record the confidential data content, the duration of the target employee recording the confidential data content is counted. When the duration is greater than or equal to the preset duration, the violation of the target employee recording confidential data in the monitoring video data is marked as a warning content, so as to issue an early warning for the target employee recording confidential data content.

[0134] Compared with the method of investigating and examining violations through on-site visits and checking historical monitoring data, this embodiment uses a target detection model to intelligently detect, identify and warn of violations by employees who leak confidential data, which can shift security risks to the left. In cases where employees are only recording confidential data content but have not actually caused confidential data leakage, employees' violations that may cause confidential data leakage can be stopped in time, thereby greatly reducing the cost of investigating and examining violations that leak confidential data in the later stage.

[0135] Next, a complete embodiment of the target detection risk identification method provided in the embodiments of the present application is proposed.

[0136] Please refer to Figure 5 , Figure 5 A logical flow chart of a complete embodiment of a method for risk identification based on target detection provided in an embodiment of the present application.

[0137] like Figure 5 As shown, the terminal device uses the target detection algorithm yolov5s and the target tracking algorithm deepsort method to identify and process the illegal behavior of target employees taking photos and recording confidential data in the workplace as a risk. The specific process is as follows:

[0138] 1. The terminal device applies for the daily surveillance video data of the employee's workplace through the reporting system. The surveillance video data includes image frames related to data content such as insurance policies.

[0139] 2. The terminal device uses the image data labeling tool labelimg to mark the image data frames in the surveillance video data in which the target employee takes illegal photos of confidential data as positive samples, and selects some image data frames that do not contain confidential data as negative samples according to the preset ratio (5:1).

[0140] 3. The terminal device uses computer vision yolov5s to integrate the positive and negative samples marked above. For example, the positive and negative sample images are feature extracted through convolutional neural networks, and the position and category of objects are predicted using anchor boxes, so as to detect real-time risk targets in surveillance video data (target representation of target employees recording confidential data in the workplace).

[0141] 4. The terminal device further uses the multi-target tracking algorithm deepsort to continuously track the risk targets in the surveillance video data. That is, the terminal device inputs the risk targets as target detection results into the deepsort algorithm for training, so that the deepsort algorithm can continuously analyze the behavior of the target employee in the surveillance video data taking a photo of the confidential data displayed on the computer screen with a mobile phone. Among them, the deepsort algorithm uses the Mann filter to predict the trajectory of the risk target, so as to continuously track the risk target in different image frames of the surveillance video data.

[0142] 5. The terminal device continuously tracks the risk target and issues an early warning when it counts the duration of the target employee recording confidential data content for a preset duration (such as 0.5 seconds). That is, if the terminal device detects that an employee has taken risky photos for a certain period of time and the content of the photos is the customer's insurance policy information, it will set a voice warning prompt and define the employee as a high-risk employee in the background, so that the company's security staff can take effective measures in time to avoid data leakage risks in advance.

[0143] In this embodiment, the DeepSort algorithm is combined with the target detection algorithm Yolov5 to continuously track and locate the specific targets of employees recording confidential data, so as to monitor and analyze the violations of insurance business employees at their workplaces, especially the accurate monitoring of the photography behavior that is expressly prohibited by the company. It can realize early warning analysis and prevention of employees' violations of confidential data recording, thereby avoiding the conventional manual screening of surveillance videos after the event (after the confidential data leakage incident occurs) which takes a lot of time. In addition, the artificial intelligence method is used to conduct supervised analysis and prediction of the surveillance video data of the employee's job performance scenes, which can more simply and quickly realize accurate speculation, so as to focus on the analysis of the predicted behavioral risk issues such as employees' illegal photography in advance, and try to avoid the occurrence of internal data leakage risk problems.

[0144] See also Figure 6 The embodiment of the present application further provides a risk identification device based on target detection, which can implement the above-mentioned risk identification method based on target detection, and the device includes:

[0145] An acquisition module is used to obtain surveillance video data of the workplace where employees perform their duties;

[0146] A target detection module is used to perform target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content;

[0147] A target tracking module is used to continuously track the risk target and obtain the duration of the target employee recording confidential data content;

[0148] The risk warning module is used to issue a warning for the risk target when the duration is greater than or equal to a preset duration.

[0149] In some embodiments, the target detection-based risk identification device provided in the embodiments of the present application may further include:

[0150] A model training module is used to obtain surveillance video sample data; the surveillance video sample data includes positive samples and negative samples, the positive samples represent that employees record confidential data content, and the negative samples represent that employees do not record confidential data content, and the number of positive samples is higher than the number of negative samples; and a model training is performed on the positive samples through a preset target detection algorithm to obtain a target detection model.

[0151] In some embodiments, the model training module is also used to detect a first target image data frame containing confidential data content and employee violation data content in the surveillance video data of the employee's workplace, and to detect a second target image data frame that does not contain the employee violation data content; wherein the employee violation data content is image data content that characterizes the employee's recording of the confidential data content; the first target image data frame in the surveillance video data is marked as a positive sample, and a part of the second target image data frame in the surveillance video data is marked as a negative sample to obtain surveillance video sample data; the number of positive samples is higher than the number of negative samples.

[0152] In some embodiments, the confidential data content includes: image data content showing the identity information of the insurance customer, and / or image data content showing the insurance policy information of the insurance customer.

[0153] In some embodiments, the risk warning module is also used to broadcast a preset voice warning prompt at the employee's workplace; output a preset voice warning prompt to the target employee who records confidential data content at the employee's workplace; and mark the target employee who records confidential data content at the employee's workplace as an employee at high risk of leakage.

[0154] In some embodiments, the target detection module is further used to perform target detection on the monitoring video data using a pre-trained target detection model to obtain confidential data content in the monitoring video data;

[0155] The target tracking module is further used to continuously track the confidential data content to obtain the duration of the target employee recording the confidential data content in the employee's work site;

[0156] The risk warning module is also used to issue a warning to the target employee when the duration is greater than or equal to a preset duration and the target employee records the confidential data content.

[0157] In some embodiments, the target tracking module is also used to continuously track the confidential data content based on a preset multi-target tracking algorithm to identify violations in the surveillance video data; the violations include: target employees in the employee's workplace recording the confidential data content; and, when the violation is identified, recording the duration of the target employee recording the confidential data content.

[0158] It should be noted that the specific implementation of the target detection-based risk identification device provided in the embodiment of the present application is basically the same as the specific implementation of the target detection-based risk identification method described above, and will not be repeated here.

[0159] The embodiment of the present application also provides a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned risk identification method based on target detection when executing the computer program. The computer device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0160] See also Figure 7 , Figure 7 The hardware structure of a computer device according to another embodiment is shown, and the computer device includes:

[0161] The processor 701 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (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 the present application;

[0162] The memory 702 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 702 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 702, and the processor 701 calls and executes the risk identification method based on target detection in the embodiment of this application;

[0163] Input / output interface 703, used to implement information input and output;

[0164] Communication interface 704, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0165] A bus 705 that transmits information between the various components of the device (e.g., the processor 701, the memory 702, the input / output interface 703, and the communication interface 704);

[0166] The processor 701 , the memory 702 , the input / output interface 703 and the communication interface 704 are connected to each other in communication within the device via a bus 705 .

[0167] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned risk identification method based on target detection is implemented.

[0168] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0169] An embodiment of the present application also provides a computer program product, which stores a computer program. When the computer program is executed by a processor, the above-mentioned risk identification method based on target detection is implemented.

[0170] The target detection-based risk identification method, target detection-based risk identification device, computer equipment and storage medium provided in the embodiment of the present application pre-train a target detection model, and use the target detection model to perform target detection on the monitoring video data of the employee's job performance scene, obtain the risk target in the monitoring video data that represents the target employee recording the confidential data content, and then combine the continuous tracking of the risk target to warn the risk target when it is found that the target employee records the confidential data content for a duration greater than the preset duration. In this way, compared with the method of investigating and inspecting violations through on-site visits and viewing historical monitoring data, the embodiment of the present application uses the target detection model to intelligently detect, identify and warn the violation of workplace employees leaking confidential data, which can shift the security risk to the left. When the employee only records the confidential data content but has not actually caused the confidential data to be leaked, the employee's violation that may cause the confidential data to be leaked can be stopped in time, thereby greatly reducing the cost of investigating and inspecting the violation of leaking confidential data in the later stage.

[0171] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems. In addition, the non-Company software tools or components (if any) that appear in the embodiments of the present application are merely examples and do not represent actual use.

[0172] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0173] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0174] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0175] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0176] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0177] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0178] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0179] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0180] If the integrated unit is implemented in the form of 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 the present application, 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, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0181] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A risk identification method based on target detection, characterized in that: The method comprises: Obtain surveillance video data of the workplace where employees perform their duties; Performing target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content; Continuously track the risk target to obtain the duration for which the target employee records confidential data; When the duration is greater than or equal to the preset duration, an early warning is issued for the risk target.

2. The method according to claim 1, characterized in that Before performing target detection on the surveillance video data by using a pre-trained target detection model to obtain risk targets in the surveillance video data, the method further includes: Acquire surveillance video sample data; the surveillance video sample data includes positive samples and negative samples, the positive samples represent that the employee records confidential data content, the negative samples represent that the employee does not record confidential data content, and the number of positive samples is greater than the number of negative samples; The target detection model is obtained by performing model training on the positive samples detected by a preset target detection algorithm.

3. The method according to claim 2, characterized in that The obtaining of surveillance video sample data includes: In the surveillance video data of the workplace where the employee performs his / her duties, a first target image data frame containing confidential data content and employee violation data content is detected, and a second target image data frame not containing the employee violation data content is detected; wherein the employee violation data content is image data content representing the confidential data content recorded by the employee; The first target image data frame in the monitoring video data is marked as a positive sample, and a part of the second target image data frame in the monitoring video data is marked as a negative sample to obtain monitoring video sample data; the number of the positive samples is higher than the number of the negative samples.

4. The method according to any one of claims 1 to 3, characterized in that The confidential data content includes: image data content showing the identity information of the insurance customer, and / or image data content showing the insurance customer's policy information.

5. The method according to claim 1, characterized in that: The early warning for the risk target includes at least one of the following: Broadcasting a preset voice warning prompt at the employee's workplace; Output a preset voice warning prompt to the target employee who records confidential data in the workplace where the employee performs his / her duties; The target employees who record confidential data content in the workplaces where the employees perform their duties are marked as employees with high risk of leaking confidential data.

6. The method according to claim 1, characterized in that After acquiring the surveillance video data of the workplace where the employee performs his / her duties, the method further includes: Performing target detection on the surveillance video data using a pre-trained target detection model to obtain confidential data content in the surveillance video data; Continuously tracking the confidential data content to obtain the duration of the target employee recording the confidential data content at the employee's workplace; When the duration is greater than or equal to the preset duration, an early warning is issued for recording the confidential data content for the target employee.

7. The method according to claim 6, characterized in that The continuously tracking the confidential data content to obtain the duration of the target employee recording the confidential data content at the employee's workplace includes: The confidential data content is continuously tracked based on a preset multi-target tracking algorithm to identify illegal behaviors in the surveillance video data; the illegal behaviors include: the target employee in the employee's work site records the confidential data content; When the violation is identified, the duration for which the target employee records the confidential data content is recorded.

8. A risk identification device based on target detection, characterized in that: The device comprises: An acquisition module is used to obtain surveillance video data of the workplace where employees perform their duties; A target detection module is used to perform target detection on the surveillance video data through a pre-trained target detection model to obtain risk targets in the surveillance video data; the risk targets represent the target employees in the employee's workplace recording confidential data content; A target tracking module is used to continuously track the risk target and obtain the duration of the target employee recording confidential data content; The risk warning module is used to issue a warning for the risk target when the duration is greater than or equal to a preset duration.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the target detection-based risk identification method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the target detection-based risk identification method according to any one of claims 1 to 7 is implemented.

Images