Workstation monitoring methods, devices, equipment and storage media

Abstract

This application provides a workstation monitoring method, apparatus, equipment, and storage medium. The method acquires monitoring videos of each workstation by video monitoring; it obtains the image of the workstation under test by frame extraction processing of each monitoring video; and it analyzes the image to obtain the current on-duty status and employee on-duty time of the workstation under test. This allows for the real-time determination of the current on-duty rate of the set of workstations and the employee on-duty time of all workstations under test. Based on the real-time determined current on-duty rate of the set of workstations and the employee on-duty time of all workstations under test, a scheduling scheme for scheduling each workstation can be determined. This can improve the on-duty rate of each workstation under test, thereby improving the stability of real-time business on-duty rate, especially when workstations under test are absent for extended periods or have delayed shift changes, which could lead to unmet business service demands.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to a workstation monitoring method, device, equipment and storage medium. Background Technology

[0002] In daily office work, some positions require long-term, real-time online availability to ensure business continuity, such as scenarios involving human customer service representatives providing timely support to users. Currently, these scenarios rely on manual inspections, which are susceptible to unforeseen circumstances. If staff rotations cannot be scheduled promptly, it will impact on operational availability. Conversely, if operational availability cannot be maintained above a certain level, it will affect the continuity of real-time business operations and reduce throughput. Therefore, addressing the current low stability of real-time operational availability has become a pressing technical challenge. Summary of the Invention

[0003] This application provides a workstation monitoring method, device, equipment, and storage medium, aiming to solve the technical problem of low stability of real-time business on-duty rate.

[0004] Firstly, this application provides a workstation monitoring method, which includes the following steps:

[0005] Obtain monitoring videos of each workstation;

[0006] Based on the workstation number of each workstation, the monitoring videos of each workstation are batch-allocated to obtain at least one batch of workstation sets corresponding to the videos to be tested.

[0007] Based on the first frame extraction rule, the test video corresponding to each test station in the set of workstations is processed sequentially to obtain the test image corresponding to the test station.

[0008] Detect the image to be tested corresponding to the workstation to be tested, and obtain the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations;

[0009] Based on the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations, a scheduling scheme for all workstations to be tested in the set of workstations is determined.

[0010] Secondly, this application also provides a workstation monitoring device, the workstation monitoring device comprising:

[0011] The monitoring video acquisition module is used to acquire monitoring videos of each workstation.

[0012] The monitoring video grouping module is used to batch allocate the monitoring videos of each workstation according to the workstation number of each workstation, so as to obtain at least one batch of workstation sets corresponding to the videos to be tested.

[0013] The image to be tested module is used to process the video to be tested corresponding to each workstation in the set of workstations in turn based on the first frame extraction rule, so as to obtain the image to be tested corresponding to the workstation to be tested.

[0014] The image detection module is used to detect the image to be tested corresponding to the workstation to be tested, and to obtain the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations.

[0015] The scheduling scheme determination module is used to determine the scheduling scheme for all workstations to be tested in the set of workstations based on the current on-duty rate of the set of workstations and the on-duty time of employees in all workstations to be tested in the set of workstations.

[0016] Thirdly, this application also provides a computer device, the computer device including a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the workstation monitoring method described above.

[0017] Fourthly, this application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of the workstation monitoring method described above.

[0018] This application provides a workstation monitoring method, apparatus, equipment, and storage medium. Based on the workstation's request to start work, this application acquires monitoring videos for each workstation; according to the workstation number, the monitoring videos are batch-allocated to obtain at least one set of workstations; based on a first frame extraction rule, frames are sequentially extracted from each video to be tested in each set of workstations to obtain the image to be tested corresponding to the workstation; and based on the image to be tested corresponding to the workstation, the current on-duty status of the workstation and the employee's on-duty time are obtained, thereby... The system determines in real-time the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations under test. Based on this real-time determination, a scheduling scheme for each workstation can be determined. This can improve the on-duty rate of each workstation under test, thereby enhancing the stability of real-time business on-duty rate, especially in situations where workstations under test are absent for extended periods or have delayed job changes, leading to unmet business service demands. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A workstation monitoring system provided for embodiments of this application;

[0021] Figure 2 A flowchart illustrating the first embodiment of the workstation monitoring method provided in this application;

[0022] Figure 3 A flowchart illustrating the process of determining the scheduling scheme provided in the embodiments of this application;

[0023] Figure 4 A flowchart illustrating a second embodiment of the workstation monitoring method provided in this application;

[0024] Figure 5 This is a schematic block diagram of a workstation monitoring device provided in an embodiment of this application;

[0025] Figure 6 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.

[0026] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0029] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0030] The embodiments of this application provide a workstation monitoring method, workstation monitoring device, computer equipment, and storage medium for monitoring the on-duty rate of each workstation by means of facial image recognition, and for timely scheduling of employees based on the monitoring results, thereby ensuring the stability of the real-time on-duty rate.

[0031] like Figure 1 As shown, Figure 1 An embodiment of this application provides a workstation monitoring system, which includes a terminal and a server, the terminal and the server being communicatively connected, and the server being communicatively connected to monitoring devices at each workstation.

[0032] The terminals include electronic devices such as mobile phones, tablets, laptops, desktop computers, personal digital assistants, and wearable devices.

[0033] The server may be a single independent server or a server cluster.

[0034] The following will provide a detailed description of the workstation monitoring method provided in the embodiments of this application based on the workstation monitoring system.

[0035] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the workstation monitoring method provided in this application.

[0036] like Figure 2 As shown, the workstation monitoring method includes steps S101 to S105.

[0037] Step S101: Obtain monitoring videos of each workstation;

[0038] In this embodiment, a camera device can be installed at each workstation. The camera's shooting range can cover a certain area of ​​the employee within the workstation. For example, the camera device can be positioned at the center line of the workstation, with the shooting range being the same on both sides. The height can be adjusted according to the employee's face height while working to ensure complete identification of the employee's facial information. When an employee is working at a workstation, the camera device can monitor the employee's situation.

[0039] In one exemplary embodiment, a trigger device can be installed at each workstation. This trigger device can be connected to a server, a camera device, or all three. When the trigger device is in the off state, the camera device can be in the off state; when the trigger device is in the on state, the camera device can be in recording mode. When an employee needs to start work, they can activate the trigger device at their workstation to activate the camera device, and the video captured by the camera device will be sent to the server, thereby obtaining the monitoring video corresponding to that workstation. The server can monitor workstations where the camera device is activated, while workstations where the camera device is not activated can be excluded from the monitoring scope.

[0040] In one exemplary embodiment, the triggering device may be a button switch; it may also be a card reader, such as triggering job monitoring by reading an employee's ID card; or it may be other devices that can be used to trigger the device switch, such as an infrared scanning device, a human body sensor, etc.

[0041] Step S102: Based on the workstation number of each workstation, the monitoring videos of each workstation are batch-allocated to obtain at least one batch of workstation sets corresponding to the videos to be tested.

[0042] In this embodiment, for each workstation, the server can group the monitoring videos according to the workstation number corresponding to each workstation, so as to process the monitoring videos in batches. This avoids problems such as video processing delays caused by excessive server processing pressure due to too many monitoring videos, thereby improving image processing efficiency. Furthermore, batch processing is more convenient for monitoring and management.

[0043] In one exemplary embodiment, each workstation can be numbered according to its distribution order, and the corresponding camera device for each workstation can also be numbered. Furthermore, the monitoring videos collected by each camera device can be grouped according to the workstation number. For example, if there are 50 workstations, the monitoring videos corresponding to 10 workstations can be grouped into 5 groups, resulting in 5 batches of videos to be tested. Each batch of videos to be tested contains 10 monitoring videos.

[0044] Step S103: Based on the first frame extraction rule, process the test video corresponding to each test station in the set of workstations in sequence to obtain the test image corresponding to the test station.

[0045] In this embodiment, according to the first frame extraction rule, each video to be tested in the same batch of workstation sets is processed by frame extraction. Several images to be tested are selected from each video to be tested as the basis for detection, so as to realize the workstation monitoring of each workstation to be tested according to the correspondence between each video to be tested and the workstation to be tested.

[0046] Understandably, when there are many workstations and a large number of videos captured by the cameras, the processing pressure on the server and image processing model will be significant, potentially leading to reduced efficiency and poor sustained performance. However, by using video grouping and image frame extraction, the real-time processing pressure on the server and image processing model can be reduced, thereby improving image processing efficiency and enhancing the sustainability of image detection.

[0047] Frame extraction, in this context, refers to extracting a number of frames from a video at regular intervals. Frame extraction methods can include extracting keyframes, uniform frame extraction, and extracting frames within a specified time period. A keyframe is a frame containing complete information; other non-keyframes are compressed using the difference between them and the keyframe. Uniform frame extraction refers to extracting a specified number of images from the video at equal intervals (equal time intervals or equal frame number intervals). Extracting frames within a specified time period involves extracting images from a specific time segment, such as extracting images from a 30-35 second segment in a 1-minute video.

[0048] In one exemplary implementation, keyframes of the video can be extracted, for example, by using a complete / clear facial image captured in the video to be tested as the image to be tested.

[0049] In one exemplary implementation, a uniform frame extraction method can be used, for example, extracting a frame as the image to be tested at 5-second intervals.

[0050] In one exemplary implementation, frames can be extracted at a specified time, such as the first 3 seconds of images captured in each minute of video as the images to be tested.

[0051] Optionally, the frame rate of the camera device can be set directly. For example, the camera devices can be divided into several groups. The camera devices in the same group can collect images together every 30 seconds as the images to be tested. The collection time of two adjacent groups of camera devices can be spaced 5 seconds apart, thereby distinguishing the images collected by each group of camera devices. The images to be tested collected by the same group of camera devices can be processed in parallel, reducing the processing pressure on the server and ensuring the continuous operation of the server while ensuring the image processing efficiency of the server.

[0052] Step S104: Detect the image to be tested corresponding to the workstation to be tested, and obtain the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations.

[0053] In this embodiment, facial images can be extracted from the same batch of images to be tested using an image processing model, thereby determining whether there are any absent employees at each workstation. Then, the current on-duty rate of each position in the set of on-duty workstations can be calculated, and the on-duty time of employees at each workstation can be calculated in real time based on the acquisition time of each image to be tested.

[0054] The image processing model can include, but is not limited to, face recognition algorithms, face extraction algorithms, and parallel processing algorithms. The image processing model can process a batch of test images in parallel to improve image processing efficiency.

[0055] In one exemplary implementation, if no face image is extracted from the image to be tested, it can be assumed that there is an employee absenteeism at the workstation corresponding to the image to be tested. Then, based on the total number of workstations currently monitored, the on-duty rate of the currently on-duty workstations can be calculated.

[0056] In one exemplary implementation, the continuous monitoring duration and cumulative monitoring duration of each on-duty workstation are determined based on the employee information of each on-duty employee and the acquisition time of the image to be measured at each on-duty employee's workstation, thereby determining the on-duty time of each on-duty employee.

[0057] Step S105: Based on the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations, determine the scheduling scheme for all workstations to be tested in the set of workstations.

[0058] In this embodiment, the server can promptly schedule employees to perform shift changes and substitute shifts based on the current on-duty rate of each monitored workstation and the on-duty time of employees at each workstation to be tested, so as to ensure that the on-duty rate of the workstation is above the specified requirements.

[0059] Reference Figure 3 , Figure 3 This is a flowchart illustrating the process of determining the scheduling scheme provided in an embodiment of this application. Figure 3 As shown, after obtaining the current on-duty rate and the on-duty time of employees at each workstation to be tested, it is necessary to compare and judge the current on-duty rate and the on-duty time of employees based on the preset on-duty rate and preset working time in order to execute different scheduling schemes.

[0060] In one embodiment, when the current on-duty rate of the set of workstations is greater than the preset on-duty rate and the on-duty time of employees in the set of workstations that are to be tested reaches the first working time, the on-duty workstations in the set of workstations that are not on duty or the on-duty workstations in the set of workstations that are to be tested and the on-duty time of employees in the set of workstations that have reached the first working time are marked as the first target workstations.

[0061] Send job change information for the first target workstation and send the job change information to the employee to be changed, so that the first target workstation can complete the job change within the preset job change time limit.

[0062] In one exemplary implementation, a minimum on-duty rate, such as 90%, can be preset. Based on the nature of the job, continuous working hours or daily cumulative working hours can be set for each position as the preset working duration, i.e., the first working duration. When the continuous or cumulative working hours of the current employee at a certain workstation are about to reach the preset working duration, that workstation is marked as the first target workstation and prioritized for job rotation. Job rotation information is then sent to pre-arrange for other personnel to complete the job rotation within the preset time limit.

[0063] At this time, because there are workstations that require job changes, the real-time on-duty rate may decrease due to untimely job changes. Therefore, for workstations that are within the monitoring range but are not on duty, they are also marked as non-target workstations. By sending return-to-work information, the employees corresponding to the non-on-duty workstations are reminded to return to their posts within the specified time limit to ensure that the real-time on-duty rate remains above the minimum on-duty rate.

[0064] In one exemplary implementation, a maximum continuous working time can be set as the first working time, and a preset reminder time can be set. When the continuous working time at the current workstation is about to reach the first working time, the current employee and the employee taking over the shift will be reminded to change shifts in a timely manner. For example, if the maximum continuous working time is set to 4 hours and the preset reminder time is 15 minutes, the server will generate a reminder message and send it to the current employee and the next employee taking over the shift at the current workstation when the current employee's working time reaches 3 hours and 45 minutes, so as to remind them to change shifts in a timely manner, thereby ensuring the continuous operation of the workstation and ensuring the on-duty rate.

[0065] Continue to refer to Figure 3 In one embodiment, when the current on-duty rate of the set of workstations is less than or equal to the preset on-duty rate and the on-duty time of no employee in the set of workstations who is not to be tested reaches the first working time, the second target workstation in the set of workstations is marked as not being on duty, and a reminder message is sent to the on-duty employee corresponding to the second target workstation to remind the on-duty employee to return to the second target workstation;

[0066] If the on-duty status of the second target workstation is not updated to "on-duty" within the preset time for sending the reminder message, the first scheduling message is sent to the supervisor corresponding to the second target workstation.

[0067] Receive scheduling feedback information sent by the supervisor in response to the first scheduling information;

[0068] Based on the scheduling feedback information, a first job change information is generated for the second target workstation, and the first job change information is sent to the employee to be changed, so that the second target workstation can complete the job change within the preset job change time limit.

[0069] In one exemplary implementation, when the current on-duty rate is detected to be lower than the minimum on-duty rate, the system promptly reminds off-duty personnel to return to their posts or reminds management personnel to promptly relocate employees to off-duty workstations in order to improve the on-duty rate.

[0070] In one exemplary implementation, when the current on-duty rate is less than or equal to the preset on-duty rate, and the on-duty time of employees at each workstation to be tested is less than the first working time, there is no job change phenomenon, which means that there are many positions where employees are absent. The workstations that are not currently on duty among all the workstations to be tested are detected as the second target workstations and marked. After the detection of the test videos in the batch of on-duty workstations is completed, the on-duty employees corresponding to each marked second target workstation are reminded to return to work within a preset time (e.g., 10 minutes) to ensure that the on-duty rate is restored to above the minimum on-duty rate requirement in a timely manner.

[0071] In one exemplary implementation, if the on-duty employee is not detected to have returned to work within a preset time, a first scheduling message is sent to the supervisor to whom the on-duty employee belongs, reminding the supervisor that there is a departure from the post. After understanding the specific situation, the supervisor can return scheduling feedback information to the server to remind the server to promptly schedule the employee to be replaced.

[0072] In one exemplary implementation, if an employee returns to work after a preset time, but the server has already marked the workstation as being in an off-duty state and notified the supervisor, the employee can change the on-duty status of the workstation by sending a status change message to the server through the supervisor.

[0073] In one embodiment, step S105 further includes:

[0074] When the current on-duty rate of the set of workstations is greater than the preset on-duty rate and there are no employees in the set of workstations whose on-duty time has reached the first working time;

[0075] Based on a preset neural network, facial expression features are extracted from the test images corresponding to the test workstations in the set of workstations where the on-duty status is on-duty.

[0076] Based on a pre-trained emotion analysis model, the facial expression features in the test images corresponding to the test positions in the set of workstations where the on-duty status is "on-duty" are analyzed to determine the emotions of the employees in the test positions where the on-duty status is "on-duty" in the set of workstations.

[0077] When the employee's emotion at the test station in the set of workstations is on duty and matches the predetermined emotion label, and the employee's on-duty time at the test station reaches the second working time, the test station is marked as the third target workstation, where the second working time is less than the first working time.

[0078] Send the second job change information to the supervisor corresponding to the third target workstation;

[0079] Receive the job change feedback information sent by the supervisor regarding the second job change information;

[0080] Based on the job change feedback information, a second job change information is generated for the third target workstation, and the second job change information is sent to the employee to be changed, so that the third target workstation can complete the job change within the preset job change time limit.

[0081] In this embodiment, facial recognition algorithms and preset neural networks are combined to collect and analyze the facial expressions of on-duty employees. An emotion analysis model is used to analyze the facial expression features of on-duty employees and analyze their emotional changes. Furthermore, based on predetermined emotion tags, the emotional changes of on-duty employees are compared to determine their work status. The on-duty rate and on-duty working hours are analyzed simultaneously to better design and implement employee scheduling plans.

[0082] Specifically, the preset neural network can be a feedforward neural network (FNN) or a convolutional neural network (CNN), or other models or algorithms that can be used for facial expression feature recognition and extraction, such as facial expression feature extraction methods designed by humans for facial expression recognition.

[0083] In one exemplary implementation, multiple convolutional and pooling layers in a Convolutional Neural Network (CNN) can extract higher-level and multi-layered features from the entire face or local regions, exhibiting good classification performance for facial expression image features. Deep convolutional neural networks can be used to fuse facial expression feature extraction and expression classification into a single end-to-end network; for example, VGG19 and ResNet18 can be used respectively for expression recognition and classification.

[0084] Each block of VGG19 consists of a convolutional layer, a BatchNorm layer, a ReLU layer, and an average pooling layer. ResNet consists of two convolutional layers and two BatchNorm layers. Shortcuts can be set at the input and output of each ResNet module, and a dropout strategy is added before the fully connected layer to increase the model's robustness.

[0085] In a specific embodiment, before performing model analysis, the emotion analysis model is trained using a database containing several images corresponding to different emotions (anger, disgust, happiness, neutrality, etc.). During the training phase, image transformations can be performed on the database images, such as flipping, rotating, and cutting, to perform data augmentation. Then, the database images are randomly cut and randomly mirrored, and then the emotion analysis model is trained to improve the classification accuracy of the emotion analysis model.

[0086] In this embodiment, an emotion analysis model is used to analyze the emotions of on-duty employees. When a poor emotional state of an on-duty employee is detected, such as depression, aversion, or pain, the analysis result may indicate that the employee's work performance is poor. However, if the employee's on-duty time has not reached the first working time limit, i.e., they cannot change shifts normally, then if the employee's on-duty time has reached the second working time limit, the employee's workstation is marked as the third target workstation. The second shift change information is then sent to the supervisor corresponding to that workstation. The supervisor determines whether to change the employee and sends feedback information to the server. If a shift change is required, the second shift change information is sent to the employee awaiting the change, reminding them to complete the shift change within the specified time limit.

[0087] The third target workstation indicates that the employee's work performance is poor, and the server determines that the employee may need to be reassigned. Supervisors can assess the employee's actual condition on-site, such as whether they are sick or have urgent matters to attend to, and then decide whether to reassign the employee.

[0088] This embodiment provides a workstation monitoring method. The method acquires monitoring videos of each workstation based on its workstation entry requests; it batches the monitoring videos according to their workstation numbers to obtain at least one set of workstations; based on a first frame extraction rule, it sequentially extracts frames from each video in each workstation set to obtain the image corresponding to the workstation; and it analyzes the image to obtain the current on-duty status and employee on-duty time of the workstation, thereby enabling real-time monitoring. Determine the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested within the set of workstations. Based on the real-time determination of the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested, determine the current scheduling scheme for each workstation. This can improve the on-duty rate of each workstation to be tested, thereby improving the stability of the real-time business on-duty rate, in situations where the business services provided cannot meet the demand due to long-term absence of employees at workstations or untimely job changes at workstations to be tested.

[0089] Please refer to Figure 4 , Figure 4A flowchart illustrating the second embodiment of the workstation monitoring method provided in this application.

[0090] Based on the above Figure 2 In the illustrated embodiment, step S104 specifically includes:

[0091] Step S1041: Perform personnel identification on the image to be tested corresponding to the workstation to be tested to determine the on-duty status of the workstation to be tested;

[0092] In this embodiment, facial images can be extracted from the image to be tested using a facial recognition algorithm, and the on-duty status of each workstation can be determined based on the extraction results of the facial images.

[0093] Understandably, the workstation status can be on-duty, off-duty, or waiting-to-be-on-duty.

[0094] In one exemplary implementation, when an employee is identified in the image to be tested corresponding to the workstation to be tested, the workstation's on-duty status can be determined to be "on-duty"; when no employee is detected in the images to be tested corresponding to the workstation for a certain period of time or for a certain number of images, the workstation's on-duty status can be determined to be "off-duty"; on-duty employees can also request that the corresponding workstation be put on standby status by application, and the monitoring of the workstation's on-duty status can be suspended for a certain period of time so that on-duty employees can handle emergency matters.

[0095] Specifically, facial recognition refers to technologies that can identify or verify the identity of a subject in an image or video. Facial recognition technology includes traditional methods that combine manually designed features (such as edge and texture descriptors) with machine learning techniques (such as principal component analysis (PCA), linear discriminant analysis (LDA), and support vector machines), as well as methods that use large facial datasets to build and train deep neural networks. The process of facial recognition technology mainly includes a face detection module for extracting faces from images and defining boundaries, a face alignment module for scaling and cropping face images using a set of reference points located at fixed positions in the image, a face representation module for converting the pixel values ​​of face images into compact and discriminative feature vectors or templates, and a face matching module for selecting and calculating the similarity score between two templates to measure the probability that they belong to the same subject.

[0096] In an exemplary embodiment, the algorithm flow for face recognition using Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) is similar, mainly including: reading images and labels from a face image database and performing grayscale processing (histogram equalization can be performed simultaneously); converting the read two-dimensional image data information into a one-dimensional vector, and then combining them column-wise into an original data matrix; normalizing the original matrix, and using PCA or LDA algorithms to perform feature analysis and dimensionality reduction on the original data matrix (during the calculation process, the image obtained after restoring the dimensionality of the one-dimensional mean vector obtained from the original data is the "average face"); reading the image to be recognized, converting it into the same vector representation as in the training set, traversing the training set, and finding the image whose difference from the image to be recognized is less than a threshold (or has the smallest difference), which is the recognition result.

[0097] In one exemplary embodiment, the neural network algorithm mainly includes a convolutional neural network (CNN), which is trained with a large amount of data through deep learning methods, thereby enabling the neural network to learn robust facial representations to changes in the training data.

[0098] In one embodiment, when a face is detected in the image to be tested corresponding to the workstation to be tested, the workstation to be tested corresponding to the image to be tested with a face is determined to be in an on-duty state; when a face is not detected in the image to be tested corresponding to the workstation to be tested, the workstation to be tested corresponding to the image to be tested without a face is marked as a workstation of interest. Based on the second frame extraction rule, the image to be tested is extracted from the video to be tested corresponding to the workstation of interest, and based on the recognition result of whether a face exists in the extracted image to be tested, the workstation to be tested is determined to be in an on-duty state or not in an on-duty state.

[0099] In this embodiment, when a face image is detected as missing in a test image, the corresponding workstation number and the test video can be labeled. Furthermore, by reducing the frame extraction interval (i.e., increasing the frame extraction frequency), targeted detection is performed on the test videos of the workstations of interest. The face is extracted from the images of interest using a facial recognition algorithm. If an employee reappears in the image of interest within a preset time range or a preset number of detected images, the label can be cleared and the on-duty status of the workstation of interest can be adjusted to "on-duty". Otherwise, the on-duty status of the workstation of interest can be determined to be "off-duty".

[0100] Optionally, when a worker is detected working at a workstation of interest again, the face of the worker can be recognized to determine the worker's identity. If the worker's information is recorded, the recording duration can be continued. If the worker is a worker who has changed shifts, the work duration of the worker can be recorded again, and the workstation of interest can be removed from the record.

[0101] In one exemplary implementation, for the video to be tested corresponding to the workstation of interest, the frame extraction interval of the second frame extraction rule can be smaller than that of the first frame extraction rule. For example, assuming that the frame extraction interval of the first frame extraction rule is to extract an image every 30 frames, then the frame extraction interval of the second frame extraction rule can be to extract an image every 5 frames, thereby increasing the detection frequency of the image to be tested at the workstation of interest, so as to be able to determine in a timely manner whether there is a absence or absence at the current workstation. If the time span of the 30-frame interval is 5 minutes, then the time span of the 5-frame interval is 50 seconds. It can be set that if no face image appears after three detections, it is determined that there is an absence or absence at the current workstation. Then the absence detection can be completed in 150 seconds, so as to promptly dispatch employees and ensure the on-duty rate.

[0102] Optionally, the same technical effect as the second frame-skipping rule can be achieved by changing the acquisition frequency of the camera device. For example, the normal acquisition frequency of the camera device can be set to acquire one frame image every 3 minutes, and when a certain workstation is marked as a workstation of interest, the acquisition frequency of the camera device can be set to acquire one frame image every 30 seconds. At the same time, if no face image appears after three detections, it can be determined that there is a situation of absence or absence at the current workstation. The judgment can be completed in 90 seconds, so as to promptly dispatch employees.

[0103] In this embodiment, workstations where no facial images are detected can be designated as pre-departure workstations to avoid misjudgments. Workstations where no facial images are detected are marked as workstations of interest. By adjusting the frame sampling frequency of these workstations, the detection frequency for them can be increased. This avoids misjudgments caused by employees briefly leaving the screen or leaving the screen precisely within the frame corresponding to the first frame sampling rule, thereby improving the accuracy of departure judgment. At the same time, when departure occurs, the detection time can be shortened, allowing for timely scheduling and arrangement of departing workstations.

[0104] Step S1042: Based on the on-duty status of the workstation to be tested and the total number of workstations corresponding to the set of on-duty workstations, calculate the current on-duty rate of the set of on-duty workstations;

[0105] In this embodiment, the current on-duty rate can be calculated in real time based on the total number of workstations corresponding to all on-duty workstation sets and the monitored on-duty status of the workstations.

[0106] In one exemplary implementation, the total number of monitored videos in each set of workstations can be recorded, and the total number of current workstations can be calculated based on the total number of workstations in each set of workstations. Then, the number of currently on-duty workstations and the number of currently off-duty workstations monitored in each set of workstations are added together to obtain the total number of currently on-duty workstations and the total number of currently off-duty workstations. Based on the total number of currently on-duty workstations, the current on-duty rate can be calculated.

[0107] Step S1043: Obtain the shooting time of the video to be tested corresponding to the workstation to be tested, and calculate the on-duty time of employees in all workstations to be tested in the set of workstations based on the shooting time of the video to be tested corresponding to the workstation to be tested.

[0108] In this embodiment, during image processing, the processing time or acquisition time of the current image to be tested is recorded for each image processing step, and is associated with each workstation number to record the continuous working time and cumulative working time of the on-duty employees corresponding to each workstation.

[0109] In one exemplary implementation, a record table can be set up. After the server numbers the jobs, it can record them in the order of the numbers. According to the facial recognition algorithm, the identity information of each employee is identified and recorded after the corresponding workstation number. The time point of the first collection of the facial images of employees at each workstation is also recorded. Then, based on the time point of each facial image recognition, the on-duty time of each employee can be calculated, or the time can be synchronized with the server system time when the first collection time is zero, until the on-duty personnel are detected to have left their posts or changed posts, then the timekeeping stops and the on-duty time is recorded, thereby obtaining a record table that records the working time, employee information and workstation number.

[0110] Please see Figure 5 , Figure 5 This is a schematic block diagram of a workstation monitoring device provided in an embodiment of this application. The workstation monitoring device is used to execute the aforementioned workstation monitoring method. The workstation monitoring device can be configured in a terminal.

[0111] like Figure 5 As shown, the workstation monitoring device 100 includes: a monitoring video acquisition module 101, a video grouping module 102, an image detection module 103, and a scheduling scheme determination module 104.

[0112] The monitoring video acquisition module 101 is used to acquire monitoring videos of each workstation.

[0113] The monitoring video grouping module 102 is used to batch allocate the monitoring videos of each workstation according to the workstation number of each workstation, so as to obtain at least one batch of workstation sets corresponding to the videos to be tested.

[0114] The image acquisition module 103 is used to process the video to be tested corresponding to each workstation in the set of workstations according to the first frame extraction rule, and obtain the image to be tested corresponding to the workstation.

[0115] The image detection module 104 is used to detect the image to be tested corresponding to the workstation to be tested, and to obtain the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations.

[0116] The scheduling scheme determination module 105 is used to determine the scheduling scheme for all workstations to be tested in the set of workstations based on the current on-duty rate of the set of workstations and the on-duty time of employees in all workstations to be tested in the set of workstations.

[0117] In one embodiment, the image detection module 104 is further configured to perform personnel identification on the image to be tested corresponding to the workstation to be tested, determine the on-duty status of the workstation to be tested; calculate the current on-duty rate of the workstation set based on the on-duty status of the workstation to be tested and the total number of workstations corresponding to the set of on-duty workstations; obtain the shooting time of the video to be tested corresponding to the workstation to be tested, and calculate the on-duty time of employees at all workstations to be tested in the set of on-duty workstations based on the shooting time of the video to be tested corresponding to the workstation to be tested.

[0118] In one embodiment, the image detection module 104 is further configured to, when a face is detected in the image corresponding to the workstation under test, determine that the workstation under test with a face is in the "on-duty" status; when a face is not detected in the image under test corresponding to the workstation under test, mark the workstation under test with no face as a "workstation of interest", extract the image under test from the video under test corresponding to the workstation of interest based on the second frame extraction rule, and determine whether the workstation of interest is in the "on-duty" or "off-duty" status based on the recognition result of whether the extracted image under test has a face.

[0119] In one embodiment, the scheduling scheme determination module 105 is further configured to, when the current on-duty rate of the set of workstations is greater than the preset on-duty rate and the on-duty time of employees in the set of workstations with test positions reaches the first working time, mark the test positions in the set of workstations that are not on duty or the test positions in the set of workstations with employees with test positions reaching the first working time as the first target workstations; send job change information for the first target workstations and send the job change information to the employees to be changed, so that the first target workstations complete the job change within the preset job change time limit.

[0120] In one embodiment, the scheduling scheme determination module 105 is further configured to: mark a second target workstation in the set of workstations that is not currently on duty when the current on-duty rate of the set of workstations is less than or equal to a preset on-duty rate and the on-duty time of no employee in the set of workstations that is to be tested reaches a first working time; send a reminder message to the on-duty employee corresponding to the second target workstation to remind the on-duty employee to return to the second target workstation; if the on-duty status of the second target workstation is not updated to on-duty within a preset time for sending the reminder message; send a first scheduling message to the supervisor corresponding to the second target workstation; receive scheduling feedback information sent by the supervisor in response to the first scheduling message; generate a first job change information for the second target workstation based on the scheduling feedback information; and send the first job change information to the employee to be changed so that the second target workstation can complete the job change within a preset job change time limit.

[0121] In one embodiment, the scheduling scheme determination module 105 is further configured to: when the current on-duty rate of the set of workstations is greater than a preset on-duty rate and the on-duty time of employees in the set of workstations who are not in the test workstations reaches a first working time; extract facial expression features from the test images corresponding to the on-duty workstations in the set of workstations whose on-duty status is "on-duty" based on a preset neural network; analyze the facial expression features from the test images corresponding to the on-duty workstations in the set of workstations whose on-duty status is "on-duty" based on a pre-trained emotion analysis model, and determine the emotions of the employees in the test workstations whose on-duty status is "on-duty" in the set of workstations; and in the above... When the employee's emotion at a test workstation in the set of workstations matches a predetermined emotion tag and the employee's on-duty time at that test workstation reaches a second working time, the test workstation is marked as a third target workstation, where the second working time is less than the first working time. Second job change information is sent to the supervisor corresponding to the third target workstation. The supervisor's job change feedback information is received. Based on the job change feedback information, second job change information is generated for the third target workstation, and this second job change information is sent to the employee to be changed, so that the third target workstation completes the job change within a preset job change time limit.

[0122] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the device and each module described above can be referred to the corresponding processes in the aforementioned workstation monitoring method embodiments, and will not be repeated here.

[0123] The apparatus provided in the above embodiments can be implemented as a computer program, which can be used in, for example... Figure 6 It runs on the computer device shown.

[0124] Please see Figure 6 , Figure 6 This is a schematic block diagram illustrating the structure of a computer device according to an embodiment of this application. The computer device may be a terminal.

[0125] See Figure 6 The computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0126] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any workstation monitoring method.

[0127] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0128] Internal memory provides an environment for the execution of computer programs stored in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to perform any workstation monitoring method.

[0129] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0130] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0131] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps:

[0132] Obtain monitoring videos of each workstation;

[0133] Based on the workstation number of each workstation, the monitoring videos of each workstation are batch-allocated to obtain at least one batch of workstation sets corresponding to the videos to be tested.

[0134] Based on the first frame extraction rule, the test video corresponding to each test station in the set of workstations is processed sequentially to obtain the test image corresponding to the test station.

[0135] Detect the image to be tested corresponding to the workstation to be tested, and obtain the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations;

[0136] Based on the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations, a scheduling scheme for all workstations to be tested in the set of workstations is determined.

[0137] In one embodiment, when the processor detects the image to be tested corresponding to the workstation to be tested, obtains the current on-duty rate of the set of workstations and the on-duty time of employees at all workstations to be tested in the set of workstations, it is further configured to:

[0138] Personnel identification is performed on the image to be tested corresponding to the workstation to be tested to determine the on-duty status of the workstation to be tested;

[0139] Based on the on-duty status of the workstation to be tested and the total number of workstations corresponding to the set of on-duty workstations, calculate the current on-duty rate of the set of on-duty workstations.

[0140] Obtain the shooting time of the video to be tested corresponding to the workstation to be tested, and calculate the on-duty time of employees at all workstations to be tested in the set of workstations to be tested based on the shooting time of the video to be tested corresponding to the workstation to be tested.

[0141] In one embodiment, when the processor performs personnel identification on the image corresponding to the workstation under test to determine the on-duty status of the workstation under test, it is also configured to:

[0142] When a face is detected in the image to be tested corresponding to the workstation to be tested, the workstation to be tested corresponding to the image to be tested with a face is determined to be in the on-duty status.

[0143] When no face is detected in the test image corresponding to the test station, the test station corresponding to the test image without a face is marked as a work station of interest. Based on the second frame extraction rule, the test image is extracted from the test video corresponding to the work station of interest. Based on the recognition result of whether the extracted test image has a face, the work station of interest is determined to be either on duty or off duty.

[0144] In one embodiment, when the processor determines the scheduling scheme for all workstations to be tested in the set of workstations based on the current on-duty rate of the set of workstations and the on-duty time of employees in all workstations to be tested in the set of workstations, it is also used to implement:

[0145] When the current on-duty rate of the set of workstations is greater than the preset on-duty rate and the on-duty time of employees in the set of workstations that are to be tested reaches the first working time, the on-duty workstations in the set of workstations that are not on duty or the on-duty workstations in the set of workstations that are to be tested and the on-duty time of employees in the set of workstations that have reached the first working time are marked as the first target workstations.

[0146] Send job change information for the first target workstation and send the job change information to the employee to be changed, so that the first target workstation can complete the job change within the preset job change time limit.

[0147] In one embodiment, when the processor determines the scheduling scheme for all workstations to be tested in the set of workstations based on the current on-duty rate of the set of workstations and the on-duty time of employees in all workstations to be tested in the set of workstations, it is also used to implement:

[0148] When the current on-duty rate of the set of workstations is less than or equal to the preset on-duty rate and the on-duty time of employees in the set of workstations who are not in the test workstations reaches the first working time, mark the second target workstation in the set of workstations that is not on-duty, and send a reminder message to the on-duty employee corresponding to the second target workstation to remind the on-duty employee to return to the second target workstation;

[0149] If the on-duty status of the second target workstation is not updated to "on-duty" within the preset time for sending the reminder message, the first scheduling message is sent to the supervisor corresponding to the second target workstation.

[0150] Receive scheduling feedback information sent by the supervisor in response to the first scheduling information;

[0151] Based on the scheduling feedback information, a first job change information is generated for the second target workstation, and the first job change information is sent to the employee to be changed, so that the second target workstation can complete the job change within the preset job change time limit.

[0152] In one embodiment, when the processor determines the scheduling scheme for all workstations to be tested in the set of workstations based on the current on-duty rate of the set of workstations and the on-duty time of employees in all workstations to be tested in the set of workstations, it is also used to implement:

[0153] When the current on-duty rate of the set of workstations is greater than the preset on-duty rate and there are no employees in the set of workstations whose on-duty time has reached the first working time;

[0154] Based on a preset neural network, facial expression features are extracted from the test images corresponding to the test workstations in the set of workstations where the on-duty status is on-duty.

[0155] Based on a pre-trained emotion analysis model, the facial expression features in the test images corresponding to the test positions in the set of workstations where the on-duty status is "on-duty" are analyzed to determine the emotions of the employees in the test positions where the on-duty status is "on-duty" in the set of workstations.

[0156] When the employee's emotion at the test station in the set of workstations is on duty and matches the predetermined emotion label, and the employee's on-duty time at the test station reaches the second working time, the test station is marked as the third target workstation, where the second working time is less than the first working time.

[0157] Send the second job change information to the supervisor corresponding to the third target workstation;

[0158] Receive the job change feedback information sent by the supervisor regarding the second job change information;

[0159] Based on the job change feedback information, a second job change information is generated for the third target workstation, and the second job change information is sent to the employee to be changed, so that the third target workstation can complete the job change within the preset job change time limit.

[0160] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0161] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A station monitoring method, characterized by, The station monitoring method comprises the following steps: acquiring monitoring videos of each on-duty station; performing batch allocation on the monitoring videos of each on-duty station according to the station numbers of the on-duty stations, to obtain a to-be-tested video corresponding to at least one batch of on-duty station set; processing the to-be-tested video corresponding to each to-be-tested station in the on-duty station set based on a first frame extraction rule, to obtain a to-be-tested image corresponding to the to-be-tested station; detecting the to-be-tested image corresponding to the to-be-tested station, to obtain a current on-duty rate of the on-duty station set and an on-duty duration of employees of all to-be-tested stations in the on-duty station set; determining a scheduling scheme of all to-be-tested stations in the on-duty station set based on the current on-duty rate of the on-duty station set and the on-duty duration of employees of all to-be-tested stations in the on-duty station set; wherein the determination of the scheduling scheme of all to-be-tested stations in the on-duty station set based on the current on-duty rate of the on-duty station set and the on-duty duration of employees of all to-be-tested stations in the on-duty station set comprises: when the current on-duty rate of the on-duty station set is greater than a preset on-duty rate and there is no to-be-tested station in the on-duty station set whose on-duty duration reaches a first working duration; extracting a facial expression feature of a person in the to-be-tested image corresponding to a to-be-tested station in the on-duty station set whose on-duty state is on-duty based on a preset neural network; analyzing the facial expression feature of the person in the to-be-tested image corresponding to the to-be-tested station in the on-duty station set whose on-duty state is on-duty based on a pre-trained emotion analysis model, to determine an employee emotion of the to-be-tested station in the on-duty station set whose on-duty state is on-duty; when the employee emotion of the to-be-tested station in the on-duty station set whose on-duty state is on-duty matches a predetermined emotion label and the on-duty duration of the employee of the to-be-tested station reaches a second working duration, marking the to-be-tested station as a third target station, the second working duration being less than the first working duration; sending second on-duty change information to a supervisor corresponding to the third target station; receiving on-duty change feedback information sent by the supervisor in response to the second on-duty change information; generating second on-duty change information for the third target station based on the on-duty change feedback information, and sending the second on-duty change information to an employee to be changed, so that the third target station completes on-duty change within a preset on-duty change time limit.

2. The station monitoring method according to claim 1, characterized in that, The detection of the to-be-tested image corresponding to the to-be-tested station, to obtain the current on-duty rate of the on-duty station set and the on-duty duration of employees of all to-be-tested stations in the on-duty station set, comprises: performing personnel recognition on the to-be-tested image corresponding to the to-be-tested station, to determine a station on-duty state of the to-be-tested station; calculating the current on-duty rate of the on-duty station set based on the station on-duty state of the to-be-tested station and a total number of stations corresponding to the on-duty station set; acquiring a shooting time of the to-be-tested video corresponding to the to-be-tested station, and calculating the on-duty duration of employees of all to-be-tested stations in the on-duty station set based on the shooting time of the to-be-tested video corresponding to the to-be-tested station.

3. The station monitoring method according to claim 2, wherein The personnel recognition on the to-be-tested image corresponding to the to-be-tested station, to determine the station on-duty state of the to-be-tested station, comprises: In a case where it is identified that the to-be-tested image corresponding to the to-be-tested station exists a face, determining that a station on-duty state of the to-be-tested station corresponding to the to-be-tested image existing the face is on duty; In a case where it is not identified that the to-be-tested image corresponding to the to-be-tested station exists a face, marking the to-be-tested station corresponding to the to-be-tested image not existing the face as a concerned station, extracting a to-be-tested image in the to-be-tested video corresponding to the concerned station based on a second frame extraction rule, and determining that a station on-duty state of the concerned station is on duty or off duty based on an identification result of whether the extracted to-be-tested image exists a face.

4. The station monitoring method of claim 1, wherein, The determining the scheduling scheme of all the to-be-tested stations in the on-duty station set based on the current on-duty rate of the on-duty station set and the on-duty time length of the staff of all the to-be-tested stations in the on-duty station set comprises: In a case where the current on-duty rate of the on-duty station set is greater than a preset on-duty rate and the on-duty time length of the staff of the to-be-tested station in the on-duty station set reaches a first working time length, marking the to-be-tested station in the on-duty station set in off duty or the to-be-tested station in the on-duty station set whose on-duty time length reaches the first working time length as a first target station; sending a shift information for the first target station and sending the shift information to the staff to be shifted to make the first target station complete the shift within a preset shift time limit.

5. The station monitoring method of claim 1, wherein, The determining the scheduling scheme of all the to-be-tested stations in the on-duty station set based on the current on-duty rate of the on-duty station set and the on-duty time length of the staff of all the to-be-tested stations in the on-duty station set comprises: In a case where the current on-duty rate of the on-duty station set is less than or equal to a preset on-duty rate and the on-duty time length of the staff of the to-be-tested station in the on-duty station set reaches a first working time length, marking a second target station in the on-duty station set in off duty and sending a reminder information to the on-duty staff corresponding to the second target station to remind the on-duty staff to return to the second target station; In a case where it is not detected that the on-duty state of the second target station is updated to on duty within a preset time of sending the reminder information, sending a first scheduling information to a supervisor corresponding to the second target station; receiving a scheduling feedback information sent by the supervisor for the first scheduling information; generating a first shift information for the second target station based on the scheduling feedback information and sending the first shift information to the staff to be shifted to make the second target station complete the shift within a preset shift time limit.

6. A station monitoring device, characterized by The station monitoring device comprises: a monitoring video acquisition module configured to acquire monitoring videos of each on-duty station; a monitoring video grouping module configured to perform batch allocation on the monitoring videos of each on-duty station according to station numbers of the on-duty stations to obtain to-be-tested videos corresponding to an on-duty station set of at least one batch; a to-be-tested image acquisition module configured to process to-be-tested videos corresponding to each to-be-tested station in the on-duty station set based on a first frame extraction rule to obtain to-be-tested images corresponding to the to-be-tested stations; and a station on-duty state determination module configured to determine a station on-duty state of each to-be-tested station in the on-duty station set based on whether the to-be-tested images corresponding to the to-be-tested stations exist a face. The image detection module is configured to detect a to-be-tested image corresponding to the to-be-tested station, obtain a current on-duty rate of the on-duty station set and an on-duty duration of an employee of all to-be-tested stations in the on-duty station set; The scheduling scheme determination module is configured to determine a scheduling scheme of all to-be-tested stations in the on-duty station set based on the current on-duty rate of the on-duty station set and the on-duty duration of the employee of all to-be-tested stations in the on-duty station set. The scheduling scheme determination module is further configured to: when the current on-duty rate of the on-duty station set is greater than a preset on-duty rate and there is no to-be-tested station in the on-duty station set whose on-duty duration reaches a first working duration; extract, based on a preset neural network, a facial expression feature in the to-be-tested image corresponding to the to-be-tested station in the on-duty station set in an on-duty state; analyze, based on a pre-trained emotion analysis model, the facial expression feature in the to-be-tested image corresponding to the to-be-tested station in the on-duty station set in the on-duty state, to determine an employee emotion of the to-be-tested station in the on-duty station set in the on-duty state; when the employee emotion of the to-be-tested station in the on-duty station set in the on-duty state matches a predetermined emotion label and the on-duty duration of the employee of the to-be-tested station reaches a second working duration, mark the to-be-tested station as a third target station, the second working duration being less than the first working duration; send second on-duty change information to a supervisor corresponding to the third target station; receive on-duty change feedback information sent by the supervisor in response to the second on-duty change information; generate second on-duty change information for the third target station based on the on-duty change feedback information, and send the second on-duty change information to an employee to be changed, so that the third target station completes on-duty change within a preset on-duty change time limit.

7. The station monitoring device according to claim 6, characterized in that The image detection module is specifically configured to perform personnel identification on the to-be-tested image corresponding to the to-be-tested station, determine a station on-duty state of the to-be-tested station, calculate a current on-duty rate of the on-duty station set based on the station on-duty state of the to-be-tested station and a total number of stations corresponding to the on-duty station set, and obtain a shooting time of a to-be-tested video corresponding to the to-be-tested station and calculate an on-duty duration of an employee of all to-be-tested stations in the on-duty station set based on the shooting time of the to-be-tested video corresponding to the to-be-tested station.

8. A computer device, comprising: The computer device includes a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein the computer program is executed by the processor to implement the steps of the station monitoring method according to any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, wherein the computer program is executed by the processor to implement the steps of the station monitoring method according to any one of claims 1 to 5.

Images