Employee checking method for entertainment venue, server, medium and product

Abstract

The invention discloses an employee checking method for an entertainment venue, a server, a medium and a product. The method comprises the following steps: identifying real-time behavior characteristic information of employees; determining a work starting position and a work finishing position of the finished work content of the employee; identifying an actual moving route of the employee from the working starting position to the working finishing position; acquiring a common moving route of the employee from the working starting position to the working finishing position; when the actual moving route is not matched with the common moving route and the common moving route can pass normally, acquiring target video data, and identifying the actual shielding condition of the employee in the target monitoring area; determining the abnormal frequency of abnormal shielding of the employee in the target monitoring area; and when the abnormal frequency of the continuous monitoring area exceeds a preset frequency threshold, determining that the employee is a suspicious person. By implementing the technical scheme, the accuracy of checking the suspicious behaviors of the employees in the entertainment venue is improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, server, medium, and product for screening employees in entertainment venues. Background Art

[0002] As a special setting with dense crowds and high mobility, security management in entertainment venues has always been a focus in the field of public security.

[0003] Currently, there are two main methods for screening suspicious individuals in entertainment venues. One is identity verification based on facial recognition, which identifies suspicious individuals by comparing them against a blacklist or abnormal personnel database. The other is anomaly detection based on user behavior analysis. This uses deep learning models to analyze characteristics such as a person's movements, trajectory, and duration of stay to identify suspicious behaviors and identify them.

[0004] However, since employees in entertainment venues have worked in this environment for a long time and are familiar with the location of cameras and image collection areas in the venues, if employees deliberately avoid cameras in the image collection area and prevent the cameras from collecting image data of suspicious behavior, it will be impossible to promptly detect suspicious behavior of employees by identifying image data, affecting the daily safe operation of entertainment venues. Summary of the Invention

[0005] The present application provides a method, server, medium and product for screening employees in entertainment venues, which can improve the accuracy of screening for suspicious behavior of employees in entertainment venues.

[0006] In a first aspect, the present application provides a method for screening employees in an entertainment venue, the method comprising: identifying the real-time behavioral characteristic information of the employees based on real-time video data of various areas in the entertainment venue obtained in advance; determining the starting work position and the finishing work position corresponding to the completed work content of the employee based on the real-time behavioral characteristic information and the behavioral characteristic information corresponding to each preset work content; identifying the actual movement route of the employee from the starting work position to the finishing work position based on the real-time video data; obtaining the commonly used movement route from the starting work position to the finishing work position in the historical work record of the employee, the commonly used movement route being the route with the highest movement frequency in the historical movement routes; when the actual movement route does not match the commonly used movement route, and the commonly used movement route is passable normally, obtaining Target video data of the practitioner in the target monitoring area, where the target monitoring area is one of all the monitoring areas through which the actual moving route passes; based on the target video data, identifying the actual occlusion situation of the practitioner when moving in the target monitoring area, the actual occlusion situation includes the occlusion range and occlusion duration of various parts of the practitioner's body; according to the actual occlusion situation, determining the abnormal number of abnormal occlusions of the practitioner in the target monitoring area, where the abnormal occlusion is that the occlusion range of various parts of the practitioner's body exceeds the preset normal occlusion range, and the occlusion duration exceeds the preset normal occlusion duration; when the abnormal number of times in a continuous monitoring area exceeds the preset number threshold, determining that the practitioner is a suspicious person, where the continuous monitoring area is a plurality of monitoring areas that the practitioner passes through in sequence according to the actual moving route.

[0007] Using the above technical solution, by determining whether an employee's actual work route matches their usual work route, it is determined whether the employee is intentionally avoiding the usual route. If the employee's actual work route does not match their usual work route, and the usual route is accessible, it indicates that the employee may have intentionally changed their route to carry out suspicious behavior. In this case, the acquired video image information is used to identify whether the employee is intentionally avoiding the image capture device to prevent the device from capturing image data of the employee performing suspicious actions. If the employee repeatedly and intentionally obscures certain body parts within the image capture device's capture area, preventing the device from capturing image data of these parts (i.e., if the employee repeatedly displays abnormal obscuration in video data from multiple consecutive monitoring areas), it indicates that the employee is intentionally avoiding the image capture device to prevent the device from capturing image data of the employee performing suspicious actions, and the employee is determined to be a suspicious person. Using the above method to identify employees as suspicious persons improves the accuracy of screening for suspicious behavior among entertainment venue employees and avoids missed detections due to employees intentionally avoiding the image capture device to capture image data of suspicious actions.

[0008] In combination with some embodiments of the first aspect, in some embodiments, after the step of determining the abnormal number of times the practitioner is abnormally occluded in the target monitoring area based on the actual occlusion situation, the method also includes: when the abnormal number of times there is no continuous monitoring area exceeds a preset number threshold, obtaining the actual movement time of the practitioner in the target unmonitored area, and the target unmonitored area is one of all areas where no image data of the practitioner is captured during the movement; obtaining the starting position and end position of the practitioner's movement in the target unmonitored area; planning an unmonitored route with the longest moving distance from the starting position to the end position; calculating the predicted movement time of the practitioner based on the unmonitored route and the real-time moving speed of the practitioner; when the actual movement time exceeds the predicted movement time, determining that the practitioner is a suspicious person.

[0009] The above technical solution further improves the suspicious behavior identification system by adding analysis of movement in unmonitored areas in addition to abnormal obstruction screening. When conventional monitoring area screening cannot determine whether a worker is suspicious, the worker's movement duration and path in the unmonitored area are analyzed. By comparing the predicted movement duration with the actual movement duration, workers who deliberately extend their stay in unmonitored blind spots to engage in suspicious behavior can be identified. This not only fills the screening loopholes caused by insufficient monitoring coverage, but also improves the accuracy of the screening, providing more reliable protection for safe operations within the venue.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the predicted movement duration of the practitioner is calculated based on the unmonitored route and the real-time movement speed of the practitioner, specifically including: calculating the initial movement duration based on the unmonitored route and the real-time movement speed of the practitioner; identifying the actual monitoring area set in the entertainment venue based on the real-time video data; determining the abnormal unmonitored area in the actual monitoring area set based on the preset normal monitoring area set; when the area of the abnormal unmonitored area is larger than the preset area, obtaining the preset increased duration corresponding to the number of users in the abnormal unmonitored area; adding the initial movement duration to the increased duration to obtain the predicted movement duration of the practitioner.

[0011] Using the above technical solution, when the area of the abnormally unmonitored area caused by normal factors such as the image acquisition device's acquisition angle offset is larger than the preset area, it indicates that there are users engaged in entertainment activities in the abnormally unmonitored area. The movement time of employees may be longer due to the needs of users in the abnormally unmonitored area. Directly judging whether employees have engaged in suspicious behavior based on the predicted movement time calculated from their walking speed will increase the false positive rate. In this case, considering the impact of user needs in the abnormally unmonitored area on the movement time of employees, increasing the predicted movement time according to certain rules, and judging whether employees have engaged in suspicious behavior based on the increased predicted movement time can reduce the false positive rate and improve the accuracy of the investigation.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after the step of determining that the employee is a suspicious person when the number of abnormalities in the continuous monitoring area exceeds a preset number threshold, the method also includes: sending a preset out-of-town work content to the employee's mobile terminal; when the employee performs the out-of-town work content, obtaining the security inspection result of the employee going out through the preset security inspection equipment; when the security inspection result is normal, based on the actual moving route and the target moving route from the employee receiving the out-of-town work content to going out, determining a first screening area set, a second screening area set and a screening person, the first screening area set is a non-monitored area to be checked, and the second screening area set is a monitored area to be checked; sending the first screening area set, the second screening area set and the location information of the screening person to the mobile terminal of the security personnel.

[0013] Using this technical solution, once a worker is identified as suspicious, the worker's security check results are obtained by assigning them to out-of-town work, further confirming whether they are carrying dangerous goods. After the worker leaves the office, security personnel are notified to investigate suspicious areas they have passed through and suspicious individuals they have come into contact with. This enables targeted investigations, avoids the manpower and time consumption of comprehensive investigations, and improves investigation efficiency. Furthermore, by conducting area-based investigations, key areas where suspicious individuals may be active can be effectively controlled, ensuring that no potential safety hazards are overlooked and safeguarding the safety of both consumers and businesses.

[0014] In combination with some embodiments of the first aspect, in some embodiments, when the security check result is normal, based on the actual moving route and the target moving route of the employee from receiving the out-of-town work content to going out, the first screening area set, the second screening area set and the screening personnel are determined, specifically including: when the security check result is normal, obtaining the target moving route of the employee from receiving the out-of-town work content to going out; obtaining the unmonitored area passed by the actual moving route and the target moving route as the first screening area; based on the video data of the monitoring area passed by the actual moving route and the target moving route, selecting the monitoring area where the number of abnormalities exceeds the preset number threshold as the second screening area; obtaining the first user who entered the first screening area within a preset time period and the second user who is within the preset range of the employee when the employee is in the second screening area as the screening personnel.

[0015] Using this technical solution, assuming normal security check results, through in-depth analysis of actual and target movement routes, we can identify unmonitored and abnormally monitored areas as key areas for investigation. This avoids blind investigations, saves manpower and time, and improves investigation efficiency. The first user who appears in the first investigation area within a preset time period, and the second user who is within the preset range of employees in the second investigation area, are identified as persons for investigation, making the investigation more targeted.

[0016] In combination with some embodiments of the first aspect, in some embodiments, the first user who enters the first screening area within a preset time period and the second user who is within the preset range of the practitioner in the second screening area are obtained as screening personnel, specifically including: obtaining the first user who enters the first screening area within a preset time period; obtaining video data of the practitioner when moving in the second screening area; identifying the second user who blocks the target body area of the practitioner in the video data, the target body area being the location area of the body part where items can be transferred and the location area where items can be placed; and using the first user and the second user as screening personnel.

[0017] The above technical solution ensures coverage of potential related persons in unmonitored areas by capturing the first user entering the primary screening area. Simultaneously, by analyzing video data from the secondary screening area, the system can identify secondary users who obstruct target body areas of employees. These areas are often key locations for the transfer or placement of items. This enhances the targeted nature of the screening, avoids the distraction caused by broad screening, and allows for focused investigations on those who may actually be involved in suspicious behavior, improving the accuracy and efficiency of the screening.

[0018] In combination with some embodiments of the first aspect, in some embodiments, after the step of determining that the practitioner is a suspicious person when the number of abnormalities in the continuous monitoring area exceeds a preset number threshold, the method also includes: if the practitioner is marked as a suspicious person for a preset number of consecutive times, adjusting the current work content of the practitioner to the preset work content, and sending the preset work content to the mobile terminal of the practitioner; when the practitioner works according to the preset work content, if the practitioner is still marked as a suspicious person, pushing the personal information of the practitioner to the management personnel.

[0019] Using this technical solution, if an employee is repeatedly flagged as suspicious, but no unusual issues occur within the entertainment venue, the employee's work content will be adjusted to that of a person who has always worked within the monitored area, to further determine whether the employee has engaged in suspicious behavior that does not affect the normal operation of the entertainment venue. If the employee is still flagged as suspicious while performing the adjusted work content, it indicates that the employee may be engaging in suspicious behavior that does not affect the normal operation of the entertainment venue. The employee's personal information will be forwarded to management personnel, allowing them to conduct a focused review of the individual or take further management measures to prevent any impact on the safe operation of the entertainment venue.

[0020] In second aspect, an embodiment of the present application provides a practitioner screening server, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the practitioner screening server to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions. When the instructions are executed on a practitioner screening server, the practitioner screening server executes the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, the present application provides a computer program product, which, when running on a practitioner screening server, enables the practitioner screening server to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] It is understandable that the practitioner screening server provided in the second aspect, the storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the method provided in this application. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects of the corresponding methods and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This application determines whether an employee's actual work route is inconsistent with their usual work route to determine whether the employee has intentionally avoided the conventional route. When the employee's actual work route is inconsistent with their usual work route and the usual route is accessible, the application uses the acquired video image information to identify whether the employee has intentionally circumvented the image capture device to determine whether the employee has engaged in suspicious behavior. This improves the accuracy of screening for suspicious behavior by entertainment venue employees and avoids missed detections due to employees intentionally circumventing the image capture device.

[0025] 2. This application arranges outbound work tasks to obtain security check results for employees and further determine whether they are carrying dangerous goods. After the employees go out, security personnel are notified to check for suspicious areas they have passed through and suspicious persons they have come into contact with. This enables targeted deployment of screening work, avoids the manpower and time consumption of comprehensive screening, and improves screening efficiency.

[0026] 3. This application further determines whether employees have engaged in suspicious behavior that does not affect the normal operation of the entertainment venue by adjusting their job description to that of always working within the monitored area. If an employee is still flagged as a suspicious person while performing their duties according to the adjusted job description, this indicates that the employee may be engaging in suspicious behavior that does not affect the normal operation of the entertainment venue. The employee's personal information will be forwarded to management personnel, allowing them to conduct a focused review of the individual or take further management measures to prevent any impact on the safe operation of the entertainment venue. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a structural diagram of a system architecture applicable to the method for screening employees in entertainment venues according to an embodiment of the present application; Figure 2 This is a flow chart of a method for screening employees in entertainment venues according to an embodiment of the present application; Figure 3 This is another flowchart of the method for screening employees in entertainment venues according to an embodiment of the present application; Figure 4 It is an exemplary hardware structure diagram of the practitioner investigation server in the embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of this application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0030] Figure 1 It is a structural diagram of a system architecture that can be applied to the method for screening employees in entertainment venues in the embodiment of the present application.

[0031] See also Figure 1 ,The employee screening system includes an image acquisition device, a mobile terminal and an employee ,screening server.

[0032] The employee screening server, the core component of the system, analyzes and processes video data captured by image acquisition devices and transmits work information to mobile terminals of employees or security personnel. Image acquisition devices capture video data from various areas within the entertainment venue and transmit it to the employee screening server. Mobile terminals receive work information sent from the employee screening server to the corresponding employees or security personnel. Terminal devices include mobile phones, tablets, and computers, which are used to receive work information.

[0033] Through the above system architecture, the employee screening system can monitor employees in entertainment venues in real time, determine whether employees are suspicious based on the identified actions or behaviors of employees, and track and identify suspicious persons when they are detected to prevent suspicious persons from affecting the daily safe operation of entertainment venues.

[0034] In related technologies, there are two main methods for screening suspicious individuals in entertainment venues. One is identity verification based on facial recognition, which identifies suspicious individuals by comparing them against a blacklist or abnormal person database. The other is anomaly detection based on user behavior analysis, which uses deep learning models to analyze characteristics such as a person's movements, trajectory, and dwell time to identify various suspicious behaviors and identify suspicious individuals. However, since entertainment venue employees have long worked in these environments and are familiar with the location of cameras and image collection areas within the venue, if employees deliberately avoid cameras within the image collection area, preventing the cameras from capturing image data of suspicious behavior, suspicious behavior cannot be detected promptly through image recognition, affecting the daily safe operation of the entertainment venue.

[0035] The method for screening employees in entertainment venues, as described in the embodiments of the present application, determines whether an employee's actual work route is inconsistent with their commonly used work route to determine whether the employee has intentionally avoided the conventional route. When an employee's actual work route is inconsistent with their commonly used work route, and the commonly used route is accessible, the method uses the acquired video image information to identify whether the employee has intentionally circumvented the image capture device to determine whether the employee has engaged in suspicious behavior. This improves the accuracy of screening for suspicious behavior among entertainment venue employees and avoids missed detections due to employees intentionally circumventing the image capture device.

[0036] The following combination Figure 2 To illustrate the method of the embodiment of the present application.

[0037] See also Figure 2 , which is a flow chart of a method for screening employees in entertainment venues in an embodiment of the present application.

[0038] S201. Based on previously acquired real-time video data of various areas in the entertainment venue, identify real-time behavioral feature information of employees.

[0039] Specifically, first, real-time video data collected by image acquisition devices deployed in various areas of the entertainment venue is obtained through a network connection.

[0040] Next, image recognition technology is used to identify and label employees in real-time video data. The received video data is first parsed into frames. Object detection algorithms, such as YOLO (You Only Look Once) and Faster R-CNN, are then used to detect all human targets within the frames. These algorithms can quickly and accurately locate each person in the image and generate corresponding bounding boxes. After detecting the human targets, the system analyzes features such as clothing, work identification, and work area to classify them as either employees or users. Once a human target is identified as a human, the system determines the human's identity by reading information such as their work ID number and name, or by using facial recognition technology to obtain facial features and compare them with a pre-stored database of human information. Using a pre-set annotation method, the human target and their identity are annotated in the image.

[0041] Next, a pre-trained human action recognition model is used to extract features from the practitioners in each frame. This human action recognition model is built based on deep learning algorithms, such as convolutional neural networks (CNNs) and long short-term memory networks (LSTMs). During the construction phase, the model's network structure is first determined. The convolutional neural network (CNN) serves as the front-end, extracting features from each frame and capturing spatial features within the image. Next, the long short-term memory network (LSTM) is connected to the CNN as the back-end. LSTMs are capable of processing data with time series characteristics. The frame-by-frame features extracted by the CNN are input into the LSTM in chronological order, thereby capturing temporal variations and dependencies in the practitioners' behavior, such as the sequence and duration of actions.

[0042] During the training phase, a large amount of image data collected in advance and labeled with different behaviors of entertainment venue employees is input into the constructed model as training data. The backpropagation algorithm is used to calculate the error between the model output and the labeled result, and the model parameters (such as the convolution kernel weights in CNN, the gating parameters in LSTM, etc.) are adjusted according to the error, so that the model can accurately identify the behavioral characteristics of the employees in the image.

[0043] The frames obtained from parsing the video data are sequentially fed into a trained human action recognition model. The model first identifies the identities of the practitioners annotated in the video frames. For each practitioner with different identities, the model performs feature extraction on the corresponding image sequences. During feature extraction, the CNN, following a pre-defined network structure, performs convolution and pooling operations on each frame, deeply exploring the spatial features contained in the image, such as body posture and the outline of the movements. These frame-by-frame spatial features are sequentially fed into the LSTM. Using its inherent gating mechanism, the LSTM processes the features in the time series, accurately capturing the temporal variations and dependencies of the behaviors of practitioners with different identities. For example, the sequence of actions taken by a particular practitioner over a period of time and the duration of each action are captured. During this processing, the model creates a separate set of feature vectors for each identified practitioner. These feature vectors comprehensively reflect the practitioner's behavior in the video frame sequence. Ultimately, the model will integrate different identity information with the corresponding real-time behavioral feature information and output it in a structured form, such as in the form of key-value pairs, where the key is the identity information of the practitioner and the value is a detailed description or feature vector containing their behavioral characteristics (such as action type, action frequency, action duration, sequence of actions, etc.), to achieve accurate identification and output of real-time behavioral feature information of practitioners with different identity information.

[0044] S202: Determine the starting work position and the finishing work position corresponding to the completed work content of the employee based on the real-time behavior characteristic information and the preset behavior characteristic information corresponding to each work content.

[0045] Specifically, the server pre-stores a database of behavioral characteristic information corresponding to each preset work content. The database records in detail the specific manifestations of different work contents in behavioral characteristics, such as the typical range or pattern of characteristic parameters such as action type, action frequency, action duration, and action sequence.

[0046] The real-time behavioral characteristic information of the practitioners is compared and analyzed with the preset behavioral characteristic information in the database one by one. The similarity between the real-time behavioral characteristic information and the preset behavioral characteristic information is calculated through the set matching algorithm. When the similarity with the target work content reaches the preset threshold, the current work content of the practitioner is determined to be the target work content.

[0047] After identifying matching work content, the system searches for the earliest video frame that identifies behavioral characteristics matching the current work content. In this earliest video frame, the worker is located using image recognition technology. Using spatial positioning algorithms such as triangulation, the camera's field of view, installation location, and calibration parameters are combined to calculate the worker's physical coordinates within the entertainment venue, thereby determining the worker's starting work position. The worker's real-time behavioral characteristics are continuously monitored throughout the work process. When behavioral characteristics change and the new characteristics have a higher degree of similarity with other work content in the database, or when the new characteristics have a similarity with the target work content below a set threshold, the target work content is determined to be completed. At this point, the server uses the same image recognition and spatial positioning methods based on the current real-time video frame to calculate the worker's physical coordinates and determine the completed work position. Furthermore, the worker's work start and completion times are determined based on the acquisition times of the earliest and current real-time video frames.

[0048] S203: Based on the real-time video data, identify the actual movement route of the employee from the starting work position to the finishing work position.

[0049] Specifically, the start time and the completion time of the employees are first obtained, and then the target video data of each area in the entertainment venue from the start time to the completion time in the real-time video data is obtained.

[0050] Based on the target video data, the worker's actual movement path from the start position to the completion position is identified. The target video data is first parsed frame by frame. An object detection algorithm (such as YOLO or Faster R-CNN) is then used to locate the worker in each frame and obtain the worker's pixel coordinates. Based on the camera's field of view, installation location, and calibration parameters, spatial positioning algorithms such as triangulation are used to convert the pixel coordinates into physical coordinates within the entertainment venue. The worker's physical coordinates are then arranged in chronological order and connected sequentially to generate a preliminary movement trajectory. Because discontinuities and noise may exist, the trajectory is smoothed using a smoothing filter algorithm (such as a Kalman filter or a moving average filter), and data interpolation algorithms (such as linear interpolation or spline interpolation) are used to fill in large time intervals. The optimized trajectory is then compared with pre-set entertainment venue map information to determine whether it crosses an inaccessible area. If so, the trajectory is corrected by combining information from adjacent frames and the entertainment venue map information to obtain the worker's actual movement path from the start position to the completion position.

[0051] In some embodiments, by communicating with a device with a positioning function carried by the employee, the real-time position of the employee at different time points between the start time and the completion time of the work can be obtained, and the actual movement route of the employee from the start position to the completion position of the work can be obtained.

[0052] S204: Obtain the commonly used movement routes from the starting work position to the finishing work position in the historical work records of the practitioners.

[0053] The frequently used moving route is the route with the highest moving frequency among all historical moving routes from the starting work location to the finishing work location.

[0054] Specifically, the historical work records of the practitioners are first obtained, which record the movement route information of the practitioners from various starting work locations to the completion work locations in different historical time periods.

[0055] Then, a set of movement routes whose starting and ending positions are the same as or close to the starting and finishing positions of the work is filtered out from the historical records.

[0056] Next, a clustering algorithm is used to analyze the routes in the route set. Algorithms such as K-Means and DBSCAN can be used. Clustering is based on route similarity, which takes into account factors such as route shape, direction, and key nodes. For example, two routes are considered similar if their coordinate deviation across most sections is less than a set value, or if their primary direction aligns. Through cluster analysis, similar routes are grouped into the same category.

[0057] Finally, the number of routes in each cluster is counted. The route corresponding to the cluster with the largest number of routes is the common route used by employees to move from their current work start location to their work completion location.

[0058] S205: When the actual moving route does not match the commonly used moving route and the commonly used moving route is passable, target video data of the employee in the target monitoring area is obtained.

[0059] The target monitoring area is one of all monitoring areas that the actual moving route passes through.

[0060] Specifically, the physical coordinate sequence of the actual movement route is compared point by point with the coordinate sequence of the commonly used movement route, and the distance deviation between the two routes is calculated. Based on a preset distance threshold, if the distance deviation between the two routes exceeds the threshold on most sections, the actual movement route is determined to be mismatched with the commonly used movement route.

[0061] When the actual moving route does not match the commonly used moving route, determine whether the commonly used moving route is passable. Obtain the coordinates of each position that the commonly used moving route passes through. Taking one of the position coordinates as an example, obtain the first monitoring area and the first intercepted image range corresponding to the position coordinate in the preset corresponding table. Obtain the monitoring video data of the first monitoring area between the start time point and the completion time point of the work. Parse the monitoring video data into a series of continuous image frames. Then, according to the first intercepted image range, intercept the image data of the corresponding image range in the image frame to obtain the intercepted image frame. According to the above method, the intercepted image frames of each position coordinate that the commonly used moving route passes through are obtained accordingly. The intercepted image frames are input into the trained image recognition model in sequence. The model analyzes each frame of the image to identify whether there are objects that may hinder passage, such as stacked goods. If there are no objects that may hinder passage, it is determined that the commonly used moving route is passable.

[0062] The model is built based on deep learning algorithms (such as convolutional neural networks). It is trained with a large amount of image data containing various obstacles, so that the model can accurately identify different types of objects that may hinder passage.

[0063] If the commonly used moving route is accessible, obtain the target video data of each monitoring area passed by the actual moving route between the start time point and the completion time point of the work.

[0064] S206: Based on the target video data, identify the actual occlusion situation of the practitioners when they move in the target monitoring area.

[0065] Among them, the actual blocking situation includes the blocking range and blocking duration of various parts of the practitioners' body.

[0066] Specifically, the target video data is first parsed into consecutive image frames. Then, an object detection algorithm is used to locate the worker in each frame, obtaining the worker's pixel coordinates and the corresponding bounding box. The image frames are then cropped based on the bounding box to obtain image data that only contains the worker within the preset range.

[0067] Obtain a pre-trained occlusion recognition model. This model is built on a deep learning algorithm (such as a convolutional neural network) and is trained on a large amount of image data containing different occlusion conditions. It can accurately identify the occlusion of various body parts in the image.

[0068] Taking a convolutional neural network as an example, the captured image data is sequentially input into the occlusion detection model according to acquisition time. The model first preprocesses the input image data. Next, multiple convolution kernels in the convolution layer slide over the image to perform convolution operations, extracting features at different levels, from simple edges and textures to complex shapes, generating a series of feature maps. Subsequently, the pooling layer performs dimensionality reduction on the feature maps, using methods such as max pooling or average pooling, to reduce computational effort and enhance the model's robustness to image position variations. After multiple layers of convolution and pooling, the feature maps are flattened into one-dimensional vectors and input into the fully connected layer. The fully connected layer comprehensively analyzes the extracted features and makes decisions. By operating on the weight matrix and bias term, combined with an appropriate activation function (such as sigmoid for binary classification and linear activation for regression prediction of occlusion coordinates), it outputs the occlusion determination results for each body part and the approximate pixel coordinate range of the occluded area.

[0069] For each frame of image output, record the occlusion range of each body part corresponding to each timestamp. Based on the timestamps, traverse the recorded occlusion ranges of each body part in chronological order, filter out the time points where the occlusion range changes, and calculate the duration between each two time points of change to obtain the occlusion duration corresponding to the occlusion range of different body parts.

[0070] Finally, the occlusion ranges of different body parts and the corresponding occlusion durations of different body parts are sorted and summarized to obtain the actual occlusion conditions of practitioners at different time points during their movement in the target monitoring area.

[0071] S207: Determine the number of abnormal occlusions caused by employees in the target monitoring area based on the actual occlusion situation.

[0072] Among them, abnormal occlusion refers to the situation where the occlusion range of various parts of the practitioner's body exceeds the preset normal occlusion range, and the occlusion duration exceeds the preset normal occlusion duration.

[0073] Specifically, the actual occlusion situations of the employee at different time points during their movement in the target monitoring area are traversed, and the occlusion range and occlusion duration of each body part in the actual occlusion situation are compared with the preset normal occlusion range and preset normal occlusion duration of each body part. When the occlusion range of each body part in the actual occlusion situation exceeds the preset normal occlusion range and the occlusion duration exceeds the preset normal occlusion duration, the abnormal number of abnormal occlusions is increased by one. After traversing all actual occlusion situations, the abnormal number of abnormal occlusions of the employee in the target monitoring area is obtained.

[0074] S208. When the number of abnormalities in the continuous monitoring area exceeds a preset threshold, the employee is determined to be a suspicious person.

[0075] Among them, the continuous monitoring area is a plurality of monitoring areas that employees pass through in sequence according to their actual movement routes.

[0076] Specifically, along the actual movement route of the employee, the data of the monitoring areas passed by the employee are traversed one by one. For each monitoring area, the number of abnormal occlusions in the monitoring area is compared with the preset number threshold. If the number of abnormalities in the monitoring area exceeds the preset number threshold, the consecutive number is increased by one, and the number of abnormal areas is increased by one; if the number of abnormalities in the monitoring area does not exceed the preset number threshold, the consecutive number is set to zero. When the consecutive number exceeds the preset number threshold, the employee is determined to be a suspicious person. At the same time, the ratio of the number of abnormal areas to the total number of all monitoring areas passed by the actual movement route is calculated. If the ratio is greater than the preset ratio threshold, the employee is determined to be a suspicious person.

[0077] In the embodiment of the present application, by judging whether the actual work route of the employee is consistent with the commonly used work route, it is determined whether the employee has intentionally avoided the conventional route. When the actual work route of the employee is inconsistent with the commonly used work route, and the commonly used route is passable normally, it indicates that the employee may have intentionally changed the route to complete the suspicious behavior. At this time, the obtained video image information is used to identify whether the employee has intentionally avoided the image acquisition device and prevented the device from collecting image data of the employee performing suspicious actions to determine whether the employee is a suspicious person. This improves the accuracy of the investigation of suspicious behavior of employees in entertainment venues and avoids missed judgments caused by employees deliberately avoiding the image acquisition device to collect image data of suspicious behavior.

[0078] The following combination Figure 3 To further illustrate the method of the embodiment of the present application.

[0079] See also Figure 3 , is another flow chart of the method for screening employees in entertainment venues in an embodiment of the present application.

[0080] S301. Based on previously acquired real-time video data of various areas in the entertainment venue, identify real-time behavioral feature information of employees.

[0081] S302: Determine the starting work position and the finishing work position corresponding to the completed work content of the employee based on the real-time behavior characteristic information and the preset behavior characteristic information corresponding to each work content.

[0082] S303: Based on the real-time video data, identify the actual movement route of the employee from the starting work position to the finishing work position.

[0083] S304: Obtain the commonly used movement routes from the starting work position to the finishing work position in the historical work records of the practitioners.

[0084] S305: When the actual moving route does not match the commonly used moving route and the commonly used moving route is passable, target video data of the employee in the target monitoring area is obtained.

[0085] S306: Based on the target video data, identify the actual occlusion situation of the practitioners when they move in the target monitoring area.

[0086] S307: Determine the number of abnormal occlusions caused by employees in the target monitoring area based on the actual occlusion situation.

[0087] S308. When the number of abnormalities in the continuous monitoring area exceeds a preset threshold, the employee is determined to be a suspicious person.

[0088] Steps S301-S308 and Figure 2 Steps S201 to S208 in the illustrated embodiment are similar, and the descriptions of steps S201 to S208 may be referred to, and will not be repeated here.

[0089] S309: When the number of abnormal times in which there is no continuous monitoring area exceeds a preset threshold, the actual moving time of the employee in the target non-monitored area is obtained.

[0090] The target unmonitored area is one of all areas where no image data is captured during the movement of the employee.

[0091] Specifically, when the number of abnormal times when there are no consecutive monitoring areas exceeds a preset threshold, the order in which the employees pass through each monitoring area is determined based on the actual movement route.

[0092] First, first video data is acquired from the first surveillance area during the period from the start of work to the completion of work. This acquired first video data is then analyzed frame by frame in chronological order to obtain first video frames. Next, the first video frames are processed using an object detection algorithm to locate the employee's position in each frame and obtain their pixel coordinates. Combining the field of view, installation location, and calibration parameters of the image acquisition device in the first surveillance area, spatial positioning algorithms such as triangulation are used to convert the employee's pixel coordinates into physical coordinates within the entertainment venue. When the employee's physical coordinates are first detected, the acquisition time of that frame is recorded, marking the time the employee first enters the first surveillance area. The employee's physical coordinates at that time are also recorded as the location of entry into the first surveillance area. The first video frames are then analyzed frame by frame in chronological order. When the employee's physical coordinates are not detected for the first time, the acquisition time of that frame is recorded, marking the time the employee leaves the first surveillance area. The employee's physical coordinates at that time are also recorded as the location of exit from the first surveillance area.

[0093] Then, obtain the second video data of the first monitoring area from the time of leaving the first monitoring area to the time of completing the work, and repeat the above steps to obtain the time points and locations of the worker entering and leaving the second monitoring area. Repeat the above steps in the order of passing through the areas until the time points and locations of the worker entering and leaving the last monitoring area are obtained.

[0094] Finally, calculate the duration between the start time point and the time point of entering the first monitoring area to obtain the actual movement duration of the first non-monitored area, use the starting point of the actual movement route as the starting position of the movement in the first non-monitored area, and use the position of entering the first monitoring area as the end position of the movement in the first non-monitored area. Then calculate the duration between the time point of leaving the first monitoring area and the time point of entering the second monitoring area to obtain the actual movement duration of the second non-monitored area, use the position of leaving the first monitoring area as the starting position of the movement in the second non-monitored area, and use the position of entering the second monitoring area as the end position of the movement in the second non-monitored area. Execute the above steps in the order of passing through the areas until the actual movement duration of all non-monitored areas passed by the employee and the starting and end positions of the employee's movement in the non-monitored areas are obtained.

[0095] S310: Obtain the starting position and ending position of the employee moving in the target unmonitored area.

[0096] The starting position and the ending position of the employee moving in the target unmonitored area obtained in step S309 are obtained.

[0097] S311. Planning an unmonitored route with the longest moving distance from the starting position to the end position.

[0098] Specifically, the system first obtains information about a preset entertainment venue map. Then, using a path search algorithm, such as the A* algorithm or the Dijkstra algorithm, it searches the entertainment venue map, starting at the starting location and ending at the ending location. During the search, constraints are set to prioritize unmonitored areas for path construction, generating multiple candidate unmonitored routes from the starting location to the ending location.

[0099] For each candidate route generated, the distances between adjacent nodes in the route are calculated based on the coordinate system and distance calculation method in the map. These distances are then accumulated to obtain the total length of each candidate route.

[0100] Finally, the lengths of all the calculated candidate routes are sorted, and the longest route is selected as the unmonitored route with the longest moving distance from the starting position to the end position.

[0101] S312. Calculate the initial movement duration based on the unmonitored route and the real-time movement speed of the employee.

[0102] Specifically, first obtain the monitoring area corresponding to the starting position of the unmonitored route. Obtain the first time point and the first position when the employee enters the monitoring area, and the second time point and the second position when the employee leaves the monitoring area. Based on the first time point and the second time point, calculate the total time the employee moves in the monitoring area. Based on the coordinate system and the distance calculation method in the map, calculate the distance between each adjacent node from the first position to the second position in the actual moving route, and then add up these distances to obtain the total distance the employee moves in the monitoring area. Based on the total distance and the total time, calculate the real-time moving speed of the employee. Then, obtain the route length of the unmonitored route calculated in step S311. Next, calculate the initial moving time based on the route length and the real-time moving speed.

[0103] S313: Based on the real-time video data, identify a set of actual monitoring areas in the entertainment venue.

[0104] Specifically, real-time video data from various areas in an entertainment venue is obtained. The real-time video data is parsed into continuous video frames. Preset calibration objects in the video frames are identified using image recognition technology. The preset calibration objects are objects that are pre-arranged and fixed in position in the entertainment venue, such as corner markers. The actual position of the preset calibration objects in the video frame is compared with the preset normal acquisition position, and the offset angle of the image acquisition device is calculated through trigonometric functions and geometric relationships. Based on the calculated offset angle, the actual acquisition area range of the image acquisition device (i.e., the actual monitoring area) is recalculated and determined using a spatial geometric model based on the original field of view range parameters, installation location data, and calibration parameters of the image acquisition device.

[0105] S314: Determine abnormal non-monitored areas in the actual monitoring area set according to the preset normal monitoring area set.

[0106] Specifically, the preset monitoring area range and the actual monitoring area range of each image acquisition device are compared, and by calculating the intersection and difference of the two areas, the part of the preset monitoring area that is not covered by the actual monitoring area is located, and the abnormal non-monitoring area where image data should be collected but is not actually collected is obtained.

[0107] S315: When the area of the abnormal non-monitored area is larger than the preset area, according to the number of users in the abnormal non-monitored area, a preset increase time duration corresponding to the number of users is obtained.

[0108] Specifically, the boundary coordinates of the abnormally unmonitored area are substituted into a pre-set area calculation model to determine the area of that area. Based on spatial geometry, the model analyzes the geometric shape of the area's polygons to accurately calculate the area. Once the calculation is complete, the area of the abnormally unmonitored area is compared with the pre-set area.

[0109] When the area of the abnormally unmonitored zone is larger than a preset area, video data collected by image acquisition devices deployed around the abnormally unmonitored zone is acquired. Human targets in the video data are identified and tracked using a multi-target detection and tracking algorithm. By analyzing the video frame by frame, the number of human targets entering or remaining in the abnormally unmonitored zone is counted to determine the number of users in the abnormally unmonitored zone. The preset increase duration corresponding to the number of users is obtained from the preset correspondence table.

[0110] When the area of the abnormal unmonitored region is less than or equal to the preset area, the initial movement duration is used as the predicted movement duration of the practitioner.

[0111] S316. Add the initial moving time to the increased time to obtain the predicted moving time of the practitioner.

[0112] The initial movement duration is added to the increased duration to calculate the predicted movement duration.

[0113] S317. When the actual moving time exceeds the predicted moving time, the employee is determined to be a suspicious person.

[0114] Compare the actual moving time with the predicted moving time. If the actual moving time exceeds the predicted moving time, the employee is determined to be a suspicious person.

[0115] S318: Send the preset out-of-office work content to the mobile terminal of the employee.

[0116] Specifically, based on the employee's historical work information, a set of out-of-home work content performed by the employee is extracted. Based on the out-of-home work content set, a first set of employees currently performing work content in the out-of-home work content set is obtained. The historical work information of each first employee in the first set of employees is obtained, and a second set of employees who have performed the employee's current work is screened out. The target employee with the longest out-of-home work time is screened out from the second set of employees. The target employee's current out-of-home work content is sent to the employee's mobile terminal, and the employee's current work content is sent to the target employee's mobile terminal.

[0117] S319. When the employee is performing work outside the home, the security inspection result of the employee passing through the preset security inspection equipment is obtained.

[0118] Specifically, video data captured by image acquisition devices installed at the entrances and exits of entertainment venues is obtained. The received video data is first parsed into frame-by-frame images, and then a target detection algorithm is used to detect all human targets within the image frames that have passed through the preset security inspection equipment. When the human target is detected as the first employee, the corresponding identity information of the first employee is determined by reading the employee's ID number, name, and other information from their work ID, or by using facial recognition technology to obtain facial features and compare them with a pre-stored database of personnel information. If the identity information matches that of the employee, a communication connection is established with the preset security inspection equipment to obtain the security inspection result corresponding to the time when the employee passed through the preset security inspection equipment (the time when the video frame corresponding to the employee was identified).

[0119] S320: When the security check result is normal, obtain the target moving route of the employee from receiving the out-of-home work content to going out.

[0120] Specifically, the time when the employee receives the out-of-office notification (the time when the out-of-office notification is sent to the employee) and the time when they pass through the preset security check device are first obtained. Then, target video data for each area of the entertainment venue from the time when the out-of-office notification is received to the time when the preset security check device is passed is obtained from the real-time video data.

[0121] According to the target video data, the target movement route of the employee from the time of receiving the outgoing work to the time of passing the preset security inspection equipment is identified. Figure 2 The step of identifying the actual moving route in step S203 is similar, and can be referred to Figure 2 Step S203 in .

[0122] S321. Obtain the unmonitored area passed by the actual moving route and the target moving route as the first inspection area.

[0123] Among them, the first inspection area set is the unmonitored area that needs to be inspected.

[0124] Specifically, based on the actual moving route and the target moving route, the order in which the employees move through the first monitoring areas according to the actual moving route and the order in which they move through the second monitoring areas according to the target moving route are determined.

[0125] Based on the order in which the employee passes through each first monitoring area, the first non-monitored area corresponding to the first monitoring area passed by the employee for the first time and the first monitoring area passed by the employee for the second time is obtained from the preset non-monitored area correspondence table, and the second non-monitored area corresponding to the first monitoring area passed by the employee for the second time and the first monitoring area passed by the employee for the third time is obtained, and so on until the last non-monitored area corresponding to the first monitoring area passed by the employee for the second to last time and the first monitoring area passed by the employee for the last time is obtained. At the same time, based on the order in which the employee passes through each second monitoring area, the non-monitored area in the target movement route is obtained using the same method.

[0126] Finally, all the unmonitored areas obtained above are used as the first screening areas.

[0127] S322. Based on the video data of the monitoring areas passed by the actual moving route and the target moving route, select the monitoring area where the number of abnormalities exceeds a preset number threshold as the second screening area.

[0128] Among them, the second inspection area set is the monitoring area that needs to be inspected.

[0129] Specifically, first video data of the first monitoring area passed by the actual movement route is obtained between the start time and the completion time of the work. At the same time, second video data of the second monitoring area passed by the target movement route is obtained between the time when the target moves out to work and the time when it passes the preset security inspection equipment.

[0130] Based on the first video data and the second video data, a first actual occlusion situation when the employee moves in the first monitoring area and a second actual occlusion situation when the employee moves in the second monitoring area are identified.

[0131] According to the first actual occlusion situation and the second actual occlusion situation, the number of abnormal occlusions by the employee in the first monitoring area and the second monitoring area is determined. When the number of abnormal occlusions in the first monitoring area and the second monitoring area exceeds a preset number threshold, the monitoring area is used as the second screening area.

[0132] The above specific implementation steps are Figure 2 The steps of S206-S208 are similar, you can refer to Figure 2 Steps S206-S208 in .

[0133] S323: Acquire the first user who enters the first screening area within a preset time period.

[0134] Specifically, video data collected by image acquisition devices deployed around the first screening area during a preset time period (which can be the preset working hours of employees) is obtained. The video data is parsed into a sequence of image frames. Using a multi-target detection and tracking algorithm, each frame is analyzed frame by frame to identify and track individuals entering the first screening area. Information about these individuals is recorded to obtain first user information, including the first user's appearance. Simultaneously, based on the first user information and real-time video data from various areas within the entertainment venue, a multi-target detection algorithm is applied to each frame of the real-time video data to detect all individuals in the image. Information such as the bounding box and confidence score for each individual is obtained. For each detected individual, appearance features are extracted, such as facial feature vectors, clothing color, and texture. The extracted features are then compared with the appearance features in the first user information. The similarity between the features (e.g., cosine similarity) is calculated to identify a matching first user. If the similarity exceeds a preset threshold, the detected individual is identified as the corresponding first user. Once the match is successful, based on the installation position, field of view, and internal and external calibration data of the image acquisition device, the center coordinates of the first user's bounding box are converted into physical coordinates within the entertainment venue through coordinate transformation to obtain the first user's real-time location information.

[0135] S324. Obtain video data of employees moving in the second screening area.

[0136] Specifically, first video data of the first monitoring area passed by the actual movement route is obtained between the start time and the completion time of the work. At the same time, second video data of the second monitoring area passed by the target movement route is obtained between the time when the target moves out to work and the time when it passes the preset security inspection equipment.

[0137] S325: Identify a second user who blocks the target body area of the practitioner in the video data.

[0138] Among them, the target body area is the location area of the body parts where items can be transferred and the location area where items can be placed, such as the hands and a certain range around them, as well as the areas where pockets and backpacks are located.

[0139] Specifically, based on the first video data and the second video data, the actual occlusion situation of the practitioner when moving in the first monitoring area and the second monitoring area is identified. Based on the occlusion range of each part of the practitioner's body in the actual occlusion situation, determine whether the target body area is included in the occlusion range of each part of the practitioner's body. If so, obtain the target video frame in which the target body area of the practitioner is identified to be occluded. According to the pixel coordinates of the practitioner obtained previously and the corresponding bounding box range, combined with the preset target range rules (for example, a rectangular area formed by extending a certain pixel distance around the practitioner as the center), determine the target range around the practitioner. Use a multi-target detection algorithm to process the target video frame and identify all users within the target range as the second user. At the same time, the real-time location information of the second user is obtained and recorded by the above method.

[0140] S326. Use the first user and the second user as screening personnel.

[0141] User information of the investigator is generated based on the user information of the first user and the second user.

[0142] S327: Send the first screening area set, the second screening area set, and the location information of the screening personnel to the mobile terminal of the security personnel.

[0143] A communication connection is established with the security personnel's mobile terminal through the built-in communication module, and the first screening area set, the second screening area set and the location information of the screening personnel are sent to the security personnel's mobile terminal.

[0144] S328. If the employee is marked as a suspicious person for a preset number of consecutive times, adjust the current work content of the employee to the preset work content, and send the preset work content to the mobile terminal of the employee.

[0145] Specifically, the system obtains the employee's historical work records. The system then iterates through the employee IDs of each record in chronological order. If the employee is identified as suspicious, the number of marks is incremented by one; if not, the number is reset to zero. When the number of marks exceeds a preset threshold, the system sends the preset work content to the employee's mobile terminal. The preset work content is the work performed under full surveillance.

[0146] S329. When the employee is working according to the preset work content, if the employee is still marked as a suspicious person, the employee's personal information is pushed to the management personnel.

[0147] Specifically, when the employee is working according to the preset work content, steps S301-S317 are continued to be executed. If the employee is still marked as a suspicious person, the personal information of the employee will be sent to the mobile terminal of the manager.

[0148] In an embodiment of the present application, when an employee is determined to be a suspicious person, the employee's security check results are obtained by arranging an out-of-town work assignment to further determine whether the employee is carrying dangerous goods. After the employee leaves the venue, security personnel are notified to investigate suspicious areas the employee has passed through and suspicious persons the employee has come into contact with. This allows for targeted deployment of investigation work, avoids the manpower and time consumption of comprehensive investigations, and improves investigation efficiency. Furthermore, if an employee is repeatedly marked as a suspicious person but no abnormalities occur in the entertainment venue, the employee's work content is adjusted to work within the monitored area to further determine whether the employee has engaged in suspicious behavior that does not affect the normal operation of the entertainment venue. If the employee is still marked as a suspicious person while working according to the adjusted work content, it indicates that the employee may be engaging in some suspicious behavior that does not affect the normal operation of the entertainment venue. The employee's personal information is then pushed to management personnel, allowing management personnel to conduct a focused investigation on the employee or take further management measures to avoid affecting the safe operation of the entertainment venue.

[0149] The above describes the method for screening employees in entertainment venues in the embodiment of the present application. The following describes the employee screening server in the embodiment of the present application in detail in combination with the above-mentioned method for screening employees in entertainment venues.

[0150] See also Figure 4 , which is an exemplary hardware structure diagram of a practitioner investigation server in an embodiment of the present application.

[0151] In some embodiments, the employee screening server 400 includes a computer device, which may be a terminal device. The computer device includes a processor 401, memory 402, a communication module 403, an input device 404, and an output device 405, all connected via a system bus. The processor 401 of the computer device provides computing and control capabilities. The memory 402 of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and computer program in the non-volatile storage medium to run. The database is used to store data. The communication module 403 of the computer device transmits work information to mobile terminals of employees or security personnel, etc. The input device 404 of the computer device receives video data transmitted by an image acquisition device, etc. The output device 405 of the computer device displays information about suspicious individuals, etc. When executed by the processor 401, the computer program implements the method for screening employees in entertainment venues according to the embodiments of the present application.

[0152] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0153] In some embodiments of the present application, a computer-readable storage medium is provided, comprising instructions. When the instructions are executed on the employee screening server 400, the employee screening server 400 can execute the employee screening method for entertainment venues in the embodiments of the present application.

[0154] In some embodiments of the present application, a computer program product is also provided. When the computer program product runs on the employee screening server 400, the employee screening server 400 executes the employee screening method for entertainment venues in the embodiments of the present application.

[0155] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0156] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0157] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disk, hard disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state drive).

[0158] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for screening employees in entertainment venues, characterized in that: include: Based on the real-time video data of various areas in the entertainment venue obtained in advance, the real-time behavioral characteristics of the practitioners are identified; Determine the starting work position and the finishing work position corresponding to the completed work content of the practitioner according to the real-time behavior characteristic information and the preset behavior characteristic information corresponding to each work content; Based on the real-time video data, identifying the actual movement route of the worker from the starting work position to the finishing work position; Obtaining a common movement route from the starting work position to the finishing work position in the historical work records of the employee, wherein the common movement route is the route with the highest movement frequency in the historical movement routes; When the actual movement route does not match the common movement route and the common movement route is passable, obtaining target video data of the employee in a target monitoring area, where the target monitoring area is one of all monitoring areas passed by the actual movement route; Based on the target video data, identifying actual occlusion conditions of the practitioner when moving in the target monitoring area, the actual occlusion conditions including the occlusion range and occlusion duration of various parts of the practitioner's body; Determine, based on the actual occlusion situation, the number of abnormal occlusions that occur in the target monitoring area by the employee, where the abnormal occlusion is when the occlusion range of various parts of the employee's body exceeds a preset normal occlusion range, and the occlusion duration exceeds a preset normal occlusion duration; When the number of abnormalities in the continuous monitoring area exceeds a preset number threshold, the employee is determined to be a suspicious person, and the continuous monitoring area is a plurality of monitoring areas that the employee passes through in sequence according to the actual movement route.

2. The method according to claim 1, characterized in that After the step of determining the number of abnormal occlusions caused by the employee in the target monitoring area based on the actual occlusion situation, the method further includes: When the number of abnormalities in which there is no continuous monitoring area exceeds a preset number threshold, obtaining the actual movement time of the employee in the target unmonitored area, where the target unmonitored area is one of all areas where no image data of the employee is captured during the movement; Obtaining the starting position and ending position of the employee moving within the target unmonitored area; Planning an unmonitored route with the longest travel distance from the starting position to the end position; Calculating a predicted travel time of the employee based on the unmonitored route and the real-time travel speed of the employee; When the actual moving duration exceeds the predicted moving duration, the employee is determined to be a suspicious person.

3. The method according to claim 2, characterized in that The calculating of the predicted travel time of the employee based on the unmonitored route and the real-time travel speed of the employee specifically includes: Calculating an initial movement duration based on the unmonitored route and the real-time movement speed of the employee; Based on the real-time video data, identifying a set of actual monitoring areas in the entertainment venue; Determine, based on a preset normal monitoring area set, an abnormal non-monitored area in the actual monitoring area set; When the area of the abnormal non-monitoring area is larger than a preset area, obtaining a preset increase time length corresponding to the number of users in the abnormal non-monitoring area; The initial moving duration is added to the increased duration to obtain the predicted moving duration of the practitioner.

4. The method according to claim 1, wherein After the step of determining the employee as a suspicious person when the number of abnormalities in the continuous monitoring area exceeds a preset number threshold, the method further includes: Sending preset out-of-office work content to the mobile terminal of the employee; When the employee performs the out-of-home work, obtaining the security inspection result of the employee passing through the preset security inspection equipment; When the security check result is normal, based on the actual movement route and the employee's target movement route from the time the employee receives the out-of-home work content to the time the employee goes out, a first screening area set, a second screening area set, and screening personnel are determined, wherein the first screening area set is a non-monitored area to be checked, and the second screening area set is a monitored area to be checked; The first screening area set, the second screening area set and the location information of the screening personnel are sent to the mobile terminal of the security personnel.

5. The method according to claim 4, characterized in that When the security check result is normal, determining a first screening area set, a second screening area set, and screening personnel based on the actual movement route and the target movement route of the employee from the time the employee receives the out-of-home work content to the time the employee goes out, specifically includes: When the security check result is normal, obtaining the target movement route of the employee from the time when the employee receives the out-of-home work content to the time when the employee goes out; Obtaining the unmonitored area passed by the actual moving route and the target moving route as a first screening area; Based on the video data of the monitoring areas passed by the actual moving route and the target moving route, the monitoring areas where the number of abnormalities exceeds a preset number threshold are selected as the second screening areas; When the first user who entered the first screening area within a preset time period and the employee is in the second screening area, the second user who is within the preset range of the employee is obtained as the screening person.

6. The method according to claim 5, characterized in that The acquiring, when the first user who entered the first screening area within a preset time period and the employee is within the second screening area, a second user who is within a preset range of the employee as the screening person specifically includes: Acquire the first user who enters the first screening area within a preset time period; Acquiring video data of the employee moving within the second screening area; Identifying a second user who obstructs a target body region of the practitioner in the video data, the target body region being a location region where an item can be transferred and a location region where an item can be placed; The first user and the second user are used as investigators.

7. The method according to claim 1, characterized in that After the step of determining the employee as a suspicious person when the number of abnormalities in the continuous monitoring area exceeds a preset number threshold, the method further includes: If the employee is marked as a suspicious person for a preset number of consecutive times, adjusting the current work content of the employee to the preset work content, and sending the preset work content to the mobile terminal of the employee; When the employee is working according to the preset work content, if the employee is still marked as a suspicious person, the personal information of the employee is pushed to the management personnel.

8. A practitioner checking server, characterized in that: include: one or more processors and memory; The memory is coupled to the one or more processors, and the memory is used to store computer program code, wherein the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the practitioner screening server to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing computer instructions, characterized in that: When the computer instructions are executed on the practitioner screening server, the practitioner screening server is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a practitioner screening server, the practitioner screening server is enabled to perform the method according to any one of claims 1 to 7.

Images