Personnel scheduling method and device, computer device and storage medium
By identifying the first and second personnel in the surveillance images of the target area and generating scheduling instructions based on the personnel trajectories, the problem of missing personnel caused by false detections by surveillance cameras is solved, and more efficient scheduling and resource utilization are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INDUSTRIAL AND COMMERCIAL BANK OF CHINA
- Filing Date
- 2023-02-10
- Publication Date
- 2026-08-04
AI Technical Summary
In existing technologies, using surveillance cameras to detect missing staff and fill in the gaps can easily lead to misjudgments, resulting in wasted human resources and a decline in service quality.
By acquiring multiple surveillance images of the target area, a personnel detection model is used to identify the first and second personnel, obtain their trajectories, and generate scheduling instructions based on the trajectories, thereby improving scheduling accuracy.
This reduced unnecessary staff deployment, avoided wasting human resources, and improved service quality and the accuracy of scheduling.
Smart Images
Figure CN115995060B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and in particular to a personnel scheduling method, apparatus, computer equipment, and storage medium. Background Technology
[0002] In some interactive scenarios, it is often necessary to handle user business. In order to enable users to complete their business, staff members are usually stationed in different areas of the interactive area to help users complete their business.
[0003] Currently, although each service area is staffed with personnel to assist with transactions, these staff may need to temporarily leave the area for various reasons. If a customer needs to conduct business in that area at this time, there will be a situation where no one is available, negatively impacting customer feedback on the service quality of the interaction area. To address this, staff replacements are typically used to avoid long customer wait times. For example, surveillance cameras can be used to monitor each service area; if an area is found to be without staff, staff from other service areas can be dispatched to fill the gap.
[0004] However, this method of using surveillance cameras to film the service areas of the target area can easily lead to misjudgments of staff shortages when it detects that a service area is missing staff. This results in low accuracy in determining staff shortages. Summary of the Invention
[0005] Therefore, it is necessary to provide a personnel scheduling method, device, computer equipment, and storage medium that can reduce the waste of human resources in response to the above-mentioned technical problems.
[0006] Firstly, this application provides a personnel scheduling method, which includes:
[0007] Acquire monitoring images of multiple areas corresponding to the target area; the target area contains multiple sub-areas, and the monitoring image of each area is the monitoring image of each sub-area;
[0008] From multiple area surveillance images, acquire the target area surveillance image; the target area surveillance image contains a first person but does not contain a second person; the first person is the person conducting the business, and the second person is the person assisting the first person in conducting the business.
[0009] Acquire the target sub-region corresponding to the surveillance image of the target area, and determine the target second person corresponding to the target sub-region;
[0010] Based on surveillance images from multiple areas, the trajectory of the target second person is obtained;
[0011] Based on personnel trajectories, obtain dispatch instructions for the target sub-area; the dispatch instructions are used to direct personnel dispatch in the target sub-area.
[0012] In one embodiment, acquiring a target area surveillance image from multiple area surveillance images includes:
[0013] Multiple area surveillance images are input into a pre-trained personnel detection model. The personnel detection model then extracts images of areas with personnel from the area surveillance images.
[0014] From the personnel area image, obtain the first personnel area image containing the first person and the second personnel area image containing the second person;
[0015] The surveillance image of the area containing the first personnel area image but not the second personnel area image is used as the target area surveillance image.
[0016] In one embodiment, obtaining a first person area image containing a first person and a second person area image containing a second person from the person area image includes:
[0017] Obtain personnel feature information from the personnel area image;
[0018] From the personnel region image, based on the similarity between personnel feature information and standard feature information, a first personnel region image and a second personnel region image are obtained; the standard feature information is the personnel feature information of the second personnel.
[0019] In one embodiment, obtaining a first person region image and a second person region image from the person region image based on the similarity between person feature information and standard feature information includes:
[0020] The similarity level is compared with a preset similarity threshold to obtain the evaluation result of the image of the human area;
[0021] If the evaluation results meet the preset conditions, the personnel area image will be used as the second personnel area image;
[0022] If the evaluation results do not meet the preset conditions, the personnel area image will be used as the first personnel area image.
[0023] In one embodiment, a personnel detection model is used to obtain images of areas with people captured from area surveillance images, including:
[0024] Human characteristics are obtained through a personnel detection model;
[0025] Based on human characteristics, identify human-area images from multiple surveillance images;
[0026] The area monitoring image containing the personnel area image is cropped to obtain the personnel area image.
[0027] In one embodiment, the trajectory of the target second person is obtained based on multiple area surveillance images, including:
[0028] The multi-frame regional surveillance images containing the target second person are input into a pre-trained personnel tracking processing model, which then outputs the trajectory of the target second person.
[0029] In one embodiment, based on personnel trajectories, dispatch instruction information for a target sub-region is obtained, including:
[0030] Based on the personnel trajectory, determine the target time when the second target personnel arrives at the target sub-area;
[0031] If the target time is greater than the preset time threshold, then the scheduling instruction information for the target sub-region is obtained.
[0032] Secondly, this application also provides a personnel dispatching device, which includes:
[0033] The monitoring image acquisition module is used to acquire monitoring images of multiple areas corresponding to the target area; the target area contains multiple sub-areas, and the monitoring image of each area is the monitoring image of each sub-area;
[0034] The target image acquisition module is used to acquire a target area monitoring image from multiple area monitoring images; the target area monitoring image contains a first person but does not contain a second person; the first person is a person who is conducting business, and the second person is a person who is assisting the first person in conducting business.
[0035] The target personnel identification module is used to acquire the target sub-region corresponding to the monitoring image of the target area, and to identify the second target personnel corresponding to the target sub-region.
[0036] The personnel trajectory determination module is used to obtain the personnel trajectory of a target second person based on multiple area monitoring images;
[0037] The dispatch information acquisition module is used to acquire dispatch instruction information for a target sub-area based on personnel trajectories; the dispatch instruction information is used to instruct personnel dispatch in the target sub-area.
[0038] Thirdly, this application also provides a computer device. This computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described above.
[0039] Fourthly, this application also provides a computer-readable storage medium. This computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.
[0040] Fifthly, this application also provides a computer program product. This computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method.
[0041] The aforementioned personnel scheduling method, apparatus, computer equipment, and storage medium acquire multiple regional monitoring images corresponding to a target area; the target area includes multiple sub-areas, and each regional monitoring image is a monitoring image of its respective sub-area; and acquires a target area monitoring image from the multiple regional monitoring images; the target area monitoring image includes a first person but does not include a second person; the first person is a person handling business, and the second person is a person assisting the first person in handling business; acquires the target sub-area corresponding to the target area monitoring image and determines the target second person corresponding to the target sub-area; acquires the personnel trajectory of the target second person based on the multiple regional monitoring images; and acquires scheduling instruction information for the target sub-area based on the personnel trajectory; the scheduling instruction information is used to instruct personnel scheduling in the target sub-area. Thus, it is possible to acquire the personnel trajectory of the target second person based on multiple regional monitoring images; and to acquire scheduling instruction information for the target sub-area based on the personnel trajectory; and to determine the scheduling instruction information based on the personnel trajectory, thereby improving the accuracy of personnel scheduling in the target sub-area and avoiding personnel scheduling when it is not necessary, thus preventing waste of human resources. Attached Figure Description
[0042] Figure 1 This is a flowchart illustrating a personnel scheduling method in one embodiment;
[0043] Figure 2 This is a flowchart illustrating the steps of acquiring a monitoring image of a target area in one embodiment;
[0044] Figure 3 This is a flowchart illustrating the steps of acquiring an image of a human-occupied area in one embodiment;
[0045] Figure 4 This is a structural block diagram of a personnel dispatching device in one embodiment;
[0046] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0047] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0048] In some interactive scenarios, it is often necessary to handle user business. In order to enable users to complete their business, staff members are usually stationed in different areas of the interactive area to help users complete their business.
[0049] Currently, although each service area is staffed with personnel to assist with transactions, these staff may need to temporarily leave the area for various reasons. If a customer needs to conduct business in that area at this time, there will be a situation where no one is available, negatively impacting customer feedback on the service quality of the interaction area. To address this, staff replacements are typically used to avoid long customer wait times. For example, surveillance cameras can be used to monitor each service area; if an area is found to be without staff, staff from other service areas can be dispatched to fill the gap.
[0050] However, when surveillance cameras are used to film the service areas of the target area, if a service area is found to be without staff, staff from other service areas are dispatched to fill the gap. However, it is possible that the staff may only be temporarily away from their work area and will return after a certain period of time. In this case, there is actually no need to dispatch staff. Therefore, dispatching staff in the above way may result in a waste of human resources.
[0051] In one embodiment, such as Figure 1 As shown, a personnel scheduling method is provided. This embodiment illustrates the method applied to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:
[0052] S102, acquire multiple area monitoring images corresponding to the target area; the target area contains multiple sub-areas, and the monitoring image of each area is the monitoring image of each sub-area.
[0053] The target area can be an area related to behavioral interaction, such as a behavioral interaction point area. Sub-areas can be areas divided from the target area according to preset rules, and the target area can contain multiple sub-areas. Area monitoring images can be monitoring images of each sub-area, which can be obtained from cameras in each sub-area. Each sub-area can correspond to multiple monitoring images, and the same sub-area can have multiple area monitoring images.
[0054] For example, the behavioral interaction point is equipped with multiple surveillance cameras for monitoring the behavioral interaction point. At the same time, different surveillance cameras can be used to monitor different service areas of the behavioral interaction point. By collecting the surveillance images of the behavioral interaction point captured by different surveillance cameras, surveillance images of different service areas of the behavioral interaction point can be obtained.
[0055] Optionally, based on the 3D design plan of the target area, the target service area to be filled by personnel is delineated, and the monitoring coverage area corresponding to the above monitoring images is marked on the network point plan. One target service area may correspond to multiple monitoring images simultaneously, avoiding incomplete monitoring coverage.
[0056] For example, the target area can be divided into 6 sub-areas, each with 3 cameras. Then, each sub-area can have 3 monitoring images at the same time, and the target area can have 18 monitoring images at the same time.
[0057] S104, acquire a target area monitoring image from multiple area monitoring images; the target area monitoring image contains a first person but does not contain a second person; the first person is a person conducting business, and the second person is a person assisting the first person in conducting business.
[0058] The target area surveillance image can be an area where a first person is captured, but no second person is captured. The first person can be someone conducting business, such as a customer. The second person can be someone in the target area assisting the first person in conducting business, such as a branch employee.
[0059] For example, from multiple area monitoring images, a target area monitoring image is obtained that captures the first person but not the second person. Further, based on the target area monitoring image, it can be determined that the sub-area corresponding to the target area monitoring image contains a missing second person.
[0060] For example, multiple area surveillance images can be processed to identify staff. If a user conducting business is captured in an area surveillance image, but no staff are included in the image, then that area surveillance image can be used as the target area surveillance image. Furthermore, based on the target area surveillance image, the staff shortage situation in the corresponding sub-area can be determined, and staff in that sub-area can be reasonably dispatched.
[0061] S106, acquire the target sub-region corresponding to the target area monitoring image, and determine the target second person corresponding to the target sub-region.
[0062] Here, the target sub-area refers to the sub-area corresponding to the surveillance image of the target area, which may be the sub-area where the camera that captured the surveillance image of the target area is located. The target second person refers to the second person responsible for assisting the first person in handling business within the target sub-area.
[0063] For example, the sub-area where the camera that captured the surveillance image of the target area is located is designated as the target sub-area. Based on the correspondence between the sub-area and the second person, the second person responsible for assisting the first person in handling business in the target sub-area can be identified.
[0064] For example, if the sub-area where the camera in the target area monitoring image is located is area A, then area A is taken as the target sub-area. Based on the correspondence between each sub-area and the second person, it can be determined that the second person corresponding to area A is the second person a. That is, the second person a is responsible for assisting the first person in handling business in area A.
[0065] Optionally, after obtaining the type of the personnel area images contained in each area's monitoring images, i.e., whether they belong to customer area images or employee area images, if the personnel area images contained in the monitoring images corresponding to a certain service area are all of the type of customer area images, it indicates that there are customers in the service area, but no corresponding staff members. In this case, the service area captured by the monitoring images of that area will be regarded as the target service area that may need to be filled by personnel.
[0066] S108, based on multiple area monitoring images, obtains the trajectory of the target second person.
[0067] Among them, the personnel trajectory can be the behavioral trajectory of the second target person within the target area.
[0068] For example, surveillance images of an area containing a target second person can be acquired, and a tracking processing algorithm can be used to perform human behavior tracking processing on multiple surveillance images of an area containing a target second person to obtain the behavioral trajectory of the target second person within the target area.
[0069] For example, in multiple area monitoring images, 100 frames of area monitoring images capture the target second person. That is, 100 frames of area monitoring images contain the area image of the second person. These 100 frames of area monitoring images can be used to perform behavior tracking processing of the target second person (the staff member corresponding to the target sub-area) and obtain the behavior trajectory of the target second person.
[0070] Optionally, the target second person can be the target staff member. The target staff member refers to the staff member assigned to the target service area in advance. After the target service area (target sub-area) is determined, the staff member responsible for the target service area, i.e. the target staff member, can be determined based on the correspondence between the service area and the staff member. Then, the personnel trajectory of the target staff member can be obtained by using the captured area monitoring images and personnel trajectory route algorithm.
[0071] S110: Based on personnel trajectories, obtain dispatch instruction information for the target sub-area; the dispatch instruction information is used to instruct personnel dispatch in the target sub-area.
[0072] The scheduling instruction information can be information on whether the target sub-region needs to be scheduled.
[0073] For example, the arrival time of the second target person in the target sub-area can be determined based on the personnel trajectory. If the arrival time of the second target person in the target sub-area is within the preset time range, personnel scheduling in the target sub-area should not be carried out. This can avoid having too many second target persons in the target sub-area and avoid the waste of human resources caused by inaccurate personnel scheduling.
[0074] Optionally, the time when the target staff member returns to the target service area can be determined based on the target staff member's trajectory, that is, the time when the target service area is in a state of no staff member. If the time is greater than 5 minutes, the branch supervisor needs to be notified to dispatch and fill the gap.
[0075] In this embodiment, multiple regional monitoring images corresponding to a target area are acquired; the target area includes multiple sub-areas, and each regional monitoring image is a monitoring image of its respective sub-area; the target area monitoring image is acquired from the multiple regional monitoring images; the target area monitoring image includes a first person but does not include a second person; the first person is a person handling business, and the second person is a person assisting the first person in handling business; the target sub-area corresponding to the target area monitoring image is acquired, and the target second person corresponding to the target sub-area is determined; the trajectory of the target second person is acquired based on the multiple regional monitoring images; and scheduling instruction information for the target sub-area is acquired based on the trajectory; the scheduling instruction information is used to instruct personnel scheduling for the target sub-area. Thus, the trajectory of the target second person can be acquired based on multiple regional monitoring images; and scheduling instruction information for the target sub-area can be acquired based on the trajectory; determining the scheduling instruction information based on the trajectory improves the accuracy of personnel scheduling for the target sub-area, avoids unnecessary personnel scheduling, and thus avoids waste of human resources.
[0076] In one embodiment, such as Figure 2 As shown, the target area monitoring image is obtained from multiple area monitoring images, including:
[0077] S202, input multiple area monitoring images into a pre-trained personnel detection model, and use the personnel detection model to obtain images of areas with personnel captured from the area monitoring images;
[0078] S204, from the personnel area image, obtain a first personnel area image containing a first person and a second personnel area image containing a second person;
[0079] S206, the area monitoring image that contains the first personnel area image but does not contain the second personnel area image is used as the target area monitoring image.
[0080] The personnel detection model can be a model for detecting personnel in area surveillance images. The personnel area image can be a bounding box containing a person's image within the area surveillance image; it can be obtained by cropping the bounding box of the personnel area image in the area surveillance image. The first personnel area image can be an image of a personnel area containing a first person, who may be someone conducting business in the target area. The second personnel area image can be an image of a personnel area containing a second person, who may be a staff member in the target area.
[0081] For example, multiple area surveillance images are input into a pre-trained personnel detection model. The target detection model extracts human features and outputs bounding boxes of human image regions contained in the area surveillance images. The bounding boxes of the human image regions in the surveillance images are then cropped to obtain human region images. These human region images are then classified, and a first human region image containing a first person and a second human region image containing a second person are obtained. The area surveillance image containing the first human region image but not the second human region image is then used as the target area surveillance image.
[0082] For example, a personnel area image might be a first personnel area image containing a first person, or a second personnel area image containing a second person. If the area monitoring image contains both the first and second personnel area images, then the sub-area corresponding to the area monitoring image can be determined to contain personnel. If the area monitoring image contains the second personnel area image but not the first personnel area image, then the sub-area corresponding to the area monitoring image can also be determined to contain personnel. If the area monitoring image contains the first personnel area image but not the second personnel area image, then the sub-area corresponding to the area monitoring image can be determined to not contain personnel. In this case, the personnel trajectory of the personnel corresponding to that sub-area can be further determined, and personnel dispatch instructions can be further determined.
[0083] Optionally, since the target area network includes not only staff members but also customers, after obtaining the personnel area images, it is necessary to filter out the employee area images that depict staff members. Specifically, this is done using a personnel image similarity algorithm. This algorithm extracts features from the clothing of individuals photographed in the employee area images to obtain the clothing characteristics of each person. Then, based on these clothing characteristics, the algorithm compares them to the standard clothing characteristics of staff members in the target area network, such as ties, shirts, name tags, trousers, and shoes. By comparing the similarity between the clothing characteristics and the standard clothing characteristics, the images of the people photographed in the obtained personnel area images are scored. If the score meets a certain scoring criterion, the personnel area image is considered an employee area image containing staff members of the target area; otherwise, if the score does not meet a certain scoring criterion, the personnel area image is considered a customer area image containing customers. After obtaining the type of each personnel area image, i.e. whether it belongs to a customer area image or an employee area image, if the personnel area images contained in the monitoring image corresponding to a certain service area are all of the type of customer area images, it indicates that there are customers in the service area, but no corresponding staff. In this case, the service area captured by the monitoring image will be regarded as the target service area that may need to be filled by personnel.
[0084] In this embodiment, acquiring personnel area images through a personnel detection model can improve the accuracy of personnel area image acquisition, thereby improving the accuracy of personnel identification. Furthermore, using a monitoring image of a region containing the first personnel area image but not the second personnel area image as the target area monitoring image can improve the accuracy of determining whether the second person is missing from the target sub-region.
[0085] In one embodiment, obtaining a first person area image containing a first person and a second person area image containing a second person from a person area image includes:
[0086] Obtain personnel feature information from the personnel area image;
[0087] From the personnel region image, based on the similarity between personnel feature information and standard feature information, a first personnel region image and a second personnel region image are obtained; the standard feature information is the personnel feature information of the second personnel.
[0088] Among these, personnel characteristic information can be feature information used to characterize a person's image, such as clothing features. Standard characteristic information can be feature information specific to a second person, such as the standard clothing features of a second person, for example, the standard clothing features of staff. Similarity refers to the degree of similarity between characteristic information.
[0089] For example, feature information of the people captured in the personnel area image can be extracted to obtain feature information of each person. Based on the above feature information, standard feature information can be compared. By comparing the similarity between the personnel feature information and the standard feature information, if the similarity meets the preset condition, the personnel area image is considered to belong to the second personnel area image containing the second person; if the similarity does not meet the preset condition, the personnel area image is considered to belong to the first personnel area image containing the first person.
[0090] For example, a similarity threshold of 90% can be set. If the similarity between personnel feature information and standard feature information is less than 90%, the personnel region image corresponding to that personnel feature information is determined as the first personnel region image. If the similarity between personnel feature information and standard feature information is greater than or equal to 90%, the personnel region image corresponding to that personnel feature information is determined as the second personnel region image.
[0091] In this embodiment, by obtaining a first person region image and a second person region image based on the similarity between the person's characteristic information and the standard characteristic information, the recognition accuracy of the second person region image containing the second person can be improved, thereby improving the accuracy of the judgment on whether the second person is missing in the target sub-region.
[0092] In one embodiment, obtaining a first person region image and a second person region image from a person region image based on the similarity between person feature information and standard feature information includes:
[0093] The similarity level is compared with a preset similarity threshold to obtain the evaluation result of the image of the human area;
[0094] If the evaluation results meet the preset conditions, the personnel area image will be used as the second personnel area image;
[0095] If the evaluation results do not meet the preset conditions, the personnel area image will be used as the first personnel area image.
[0096] The evaluation result can be an evaluation value for the personnel area image, such as a score value like 70 or 80. The preset condition can be a preset evaluation threshold, such as a score threshold of 90.
[0097] For example, the similarity level is compared with a preset similarity threshold. Based on the similarity level and the preset similarity threshold, the person area image is scored. If the score meets a certain scoring criterion, the person area image is considered to belong to a second person area image containing a second person. If the score does not meet a certain scoring criterion, the person area image is considered to belong to a first person area image containing a first person.
[0098] For example, by comparing the similarity with a preset similarity threshold, the score of the person area image is 95. If the scoring standard (scoring threshold) is 90, then because 95 is greater than 90, the score of the person area image meets the scoring standard, and it can be considered that the person area image belongs to the second person area image with a second person.
[0099] Alternatively, clothing features can be used to compare the standard clothing features of staff in the target area with those of employees in the target area, such as ties, shirts, name tags, trousers, and leather shoes. By comparing the similarity between the clothing features and the standard clothing features, the images of the people captured in the obtained personnel area images can be scored. If the score meets a certain scoring standard, the personnel area image is considered to belong to the employee area image where staff in the target area are captured. If the score does not meet a certain scoring standard, the personnel area image is considered to belong to the customer area image where customers are captured.
[0100] In this embodiment, an evaluation result is obtained by comparing the similarity with a similarity threshold. If the evaluation result meets the preset conditions, the personnel area image is used as the second personnel area image. If the evaluation result does not meet the preset conditions, the personnel area image is used as the first personnel area image. This can improve the accuracy of classifying the target personnel area image, and thus improve the accuracy of judging whether the second person is missing in the target sub-region.
[0101] In one embodiment, such as Figure 3 As shown, the personnel detection model is used to obtain images of areas with people captured from regional surveillance images, including:
[0102] S302, using a personnel detection model to obtain human characteristics;
[0103] S304, based on human characteristics, identifies personnel-area images from multiple area surveillance images;
[0104] S306, crop the area monitoring image containing personnel area images to obtain personnel area images.
[0105] The personnel detection model can be a model that detects people in a monitored area image. Person features can be features used to characterize people, features that can distinguish people, such as clothing features and body shape features.
[0106] For example, the area surveillance image is input into a pre-trained people detection model, the people detection model extracts human features, the people detection model outputs the rectangular boxes of the people image regions contained in the surveillance image, and the rectangular boxes of the people image regions in the surveillance image are cropped to obtain the people region image.
[0107] For example, multiple area surveillance images are input into a pre-trained personnel detection model. The personnel detection model identifies the clothing and body features of personnel, and divides the personnel image region rectangles from the area surveillance images based on the clothing and body features of personnel.
[0108] In this embodiment, a person detection model is used to obtain person features; based on these features, a person image region bounding box is identified from the area surveillance image; the person image region bounding boxes contained in the area surveillance image are cropped to obtain the person region image. Thus, obtaining the person region image through a person detection model improves the accuracy of obtaining the person region image, thereby further improving the accuracy of the second person behavior recognition.
[0109] In one embodiment, the trajectory of the target second person is obtained based on multiple area surveillance images, including:
[0110] The multi-frame regional surveillance images containing the target second person are input into a pre-trained personnel tracking processing model, which then outputs the trajectory of the target second person.
[0111] Among them, the people tracking processing model can be a convolutional neural network model capable of pedestrian detection or trajectory tracking.
[0112] For example, a multi-frame area surveillance image containing the target second person is input into a pre-trained personnel tracking processing model. The personnel tracking processing model is used to perform behavior detection processing on the multi-frame area surveillance image with time information to obtain the personnel trajectory of the target second person.
[0113] For example, if there are 100 frames of surveillance images of an area containing the target second person, then behavior detection processing can be performed based on these 100 frames of surveillance images to obtain the trajectory of the target second person.
[0114] In this embodiment, by using a personnel tracking processing model for behavior detection processing, the accuracy of personnel trajectories can be improved, thereby improving the accuracy of scheduling instructions and avoiding the waste of human resources.
[0115] In one embodiment, obtaining dispatch instruction information for a target sub-region based on personnel trajectories includes:
[0116] Based on the personnel trajectory, determine the target time when the second target personnel arrives at the target sub-area;
[0117] If the target time is greater than the preset time threshold, then the scheduling instruction information for the target sub-region is obtained.
[0118] Here, "personnel trajectory" can be the behavioral trajectory of the second target person within the target area. "Target time" can be the time it takes for the second target person to return to the target sub-area. "Time threshold" can be the waiting time threshold required for users to process business transactions. "Schedule instruction information" can be the instruction information for personnel scheduling based on the second target person arriving at the target sub-area within the time corresponding to the time threshold.
[0119] For example, the time when staff return to the target service area can be determined based on personnel trajectories, that is, the time when the target service area is in a state of no staff. If the time is greater than 5 minutes, the branch manager needs to be notified to dispatch replacements.
[0120] For example, consider a target sub-region, sub-region A, and the second target person is person A. If, based on person A's trajectory, it's determined that person A will return to sub-region A in 10 minutes, and the time threshold is set to 5 minutes (because 10 is greater than 5), then person A cannot return to sub-region A within 5 minutes. In this case, a notification can be sent to the branch manager corresponding to sub-region A for dispatching replacement personnel, such as person B, to assist the user with their business. If, based on person A's trajectory, it's determined that person A will return to sub-region A in 3 minutes, and the time threshold is set to 5 minutes (because 3 is less than 5), then person A can return to sub-region A within 5 minutes. In this case, no dispatching replacement is needed, and person A will assist the user with their business after 3 minutes.
[0121] In this embodiment, the target time for the second target person to arrive at the target sub-area is determined based on the personnel trajectory. If the target time is greater than a preset time threshold, scheduling instruction information for the target sub-area is obtained. Thus, determining the target time based on the personnel trajectory and further determining the scheduling instruction information improves the accuracy of personnel scheduling in the target sub-area, avoids unnecessary personnel scheduling, and prevents waste of human resources.
[0122] In one embodiment, a personnel scheduling method is provided, comprising the following steps:
[0123] The system collects surveillance images of different service areas within the target area network. Multiple surveillance cameras are installed in the target area network to monitor different service areas. By collecting surveillance images from these cameras, the system obtains surveillance images of different service areas within the target area network.
[0124] Based on the 3D design plan of the network points, the target service areas requiring personnel replacement are delineated, and the monitoring coverage areas corresponding to the aforementioned monitoring images are marked on the network point plan. One target service area may correspond to multiple monitoring images simultaneously, avoiding incomplete monitoring coverage.
[0125] The object detection algorithm identifies the human regions in each surveillance image. The obtained surveillance images are then input into a pre-trained object detection model, which extracts human features and outputs bounding boxes of the human regions within the surveillance images. These bounding boxes are then cropped to obtain the human region images.
[0126] Using a personnel image similarity algorithm, employee area images and customer area images are selected from the personnel area images. Since the target area network includes not only staff but also customers, after obtaining the personnel area images, it is necessary to further select employee area images that depict staff from the target area. Specifically, this is achieved through a personnel image similarity algorithm. This algorithm extracts features from the clothing of individuals in the employee area images to obtain the clothing characteristics of each person. These clothing characteristics are then compared to the standard clothing characteristics of staff in the target area network, such as ties, shirts, name tags, trousers, and shoes. By comparing the similarity between these clothing characteristics and the standard clothing characteristics, the personnel images in the obtained personnel area images are scored. If the score meets a certain scoring criterion, the personnel area image is considered an employee area image containing staff from the target area; otherwise, it is considered a customer area image containing customers.
[0127] If a surveillance image contains only customer area images and no employee area images, then the service area corresponding to that surveillance image is designated as the target service area. After determining the type of each personnel area image (i.e., whether it belongs to a customer area image or an employee area image), if the surveillance image corresponding to a service area contains only customer area images, it indicates that there are customers in that service area, but no corresponding staff. In this case, the service area captured by that surveillance image is designated as the target service area where personnel may be needed.
[0128] The system obtains the target staff corresponding to the target service area, as well as their personnel trajectories. Target staff refers to the staff pre-assigned to the target service area. After determining the target service area, the staff responsible for the target service area can be identified based on the correspondence between the service area and the staff in the target area. Then, the personnel trajectories of the target staff can be obtained using captured surveillance images and personnel trajectory algorithms.
[0129] Based on personnel trajectories, the return time of target staff to the target service area is predicted. If the time meets a set threshold, the branch's supervisor is notified to redeploy staff to the target service area. Finally, based on personnel trajectories, the return time of staff to the target service area can be determined, i.e., the time during which the target service area is unmanned. If this time exceeds 5 minutes, the branch supervisor needs to be notified to dispatch replacement staff.
[0130] In this embodiment, surveillance images of each service area are captured by a camera. A target detection algorithm identifies individuals within these images, and a personnel image similarity algorithm further identifies target service area staff and customers. If only customers are present in the service area, the path predicts the return time of the target staff. A personnel trajectory algorithm is used to obtain the staff's path from the surveillance images, predicting their return time. If this time exceeds a pre-set threshold, the branch manager is notified to fill the gap. Compared to existing technologies that immediately fill gaps when no staff are present, this application further utilizes the staff's trajectory to determine user waiting time, only filling gaps when the waiting time exceeds a set threshold. This improves the accuracy of gap filling and reduces wasted human resources.
[0131] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the embodiments described above may include steps or stages, which are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.
[0132] Based on the same inventive concept, this application also provides a personnel scheduling device for implementing the personnel scheduling method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one of the personnel scheduling device embodiments provided below can be found in the limitations of the personnel scheduling method described above, and will not be repeated here.
[0133] In one embodiment, such as Figure 4 As shown, a personnel dispatching device 400 is provided, including: a monitoring image acquisition module 410, a target image acquisition module 420, a target personnel determination module 430, a personnel trajectory determination module 440, and a dispatching information acquisition module 450, wherein:
[0134] The monitoring image acquisition module 410 is used to acquire multiple monitoring images of a target area; the target area contains multiple sub-areas, and the monitoring image of each area is the monitoring image of each sub-area.
[0135] The target image acquisition module 420 is used to acquire a target area monitoring image from multiple area monitoring images; the target area monitoring image contains a first person but does not contain a second person; the first person is a person who is conducting business, and the second person is a person who is assisting the first person in conducting business.
[0136] The target personnel identification module 430 is used to acquire the target sub-region corresponding to the target area monitoring image and identify the target second person corresponding to the target sub-region;
[0137] The personnel trajectory determination module 440 is used to obtain the personnel trajectory of the target second person based on multiple area monitoring images;
[0138] The dispatch information acquisition module 450 is used to acquire dispatch instruction information for a target sub-area based on personnel trajectories; the dispatch instruction information is used to instruct personnel dispatch in the target sub-area.
[0139] In one embodiment, the target image acquisition module includes a personnel region image unit, a personnel image classification unit, and a target image determination unit.
[0140] The personnel area image unit is used to input multiple area monitoring images into a pre-trained personnel detection model. Through the personnel detection model, personnel area images containing personnel are obtained from the area monitoring images. The personnel image classification unit is used to obtain a first personnel area image containing a first person and a second personnel area image containing a second person from the personnel area images. The target image determination unit is used to take the area monitoring image containing the first personnel area image but not the second personnel area image as the target area monitoring image.
[0141] In one embodiment, the personnel image classification unit includes a personnel feature information acquisition unit and a similarity unit.
[0142] The personnel feature information acquisition unit is used to acquire personnel feature information from the personnel region image; the similarity degree unit is used to acquire a first personnel region image and a second personnel region image from the personnel region image based on the similarity degree between the personnel feature information and the standard feature information; the standard feature information is the personnel feature information of the second person.
[0143] In one embodiment, the similarity unit is used for the comparison unit, the second person region image determination unit, and the first person region image determination unit.
[0144] The comparison unit is used to compare the similarity with a preset similarity threshold to obtain the evaluation result of the personnel region image; the second personnel region image determination unit is used to use the personnel region image as the second personnel region image when the evaluation result meets the preset conditions; the first personnel region image determination unit is used to use the personnel region image as the first personnel region image when the evaluation result does not meet the preset conditions.
[0145] In one embodiment, the personnel area image unit includes a personnel feature acquisition unit, a personnel area image recognition unit, and a cropping unit.
[0146] The personnel feature acquisition unit is used to acquire personnel features through a personnel detection model; the personnel region image recognition unit is used to identify personnel region images from multiple regional monitoring images based on personnel features; and the cropping unit is used to crop the regional monitoring images containing personnel region images to obtain personnel region images.
[0147] In one embodiment, the personnel trajectory determination module is used to input a multi-frame area monitoring image containing the target second person into a pre-trained personnel tracking processing model, and output the personnel trajectory of the target second person through the personnel tracking processing model.
[0148] In one embodiment, the scheduling information acquisition module includes a target time determination unit and a time judgment unit.
[0149] The target time determination unit is used to determine the target time for the second person to arrive at the target sub-area based on the personnel trajectory; the time judgment unit is used to obtain the scheduling instruction information for the target sub-area if the target time is greater than the preset time threshold.
[0150] Each module in the aforementioned personnel dispatching device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.
[0151] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores area surveillance image data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a personnel scheduling method.
[0152] The interaction objects in this field are understandable. 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.
[0153] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.
[0154] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.
[0155] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.
[0156] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0157] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0158] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those with ordinary technical expertise in the field can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A personnel scheduling method, characterized in that, The method includes: Acquire multiple regional monitoring images corresponding to a target area; the target area includes multiple sub-regions, and each regional monitoring image is a monitoring image of its respective sub-region. A target area monitoring image is obtained from the multiple area monitoring images; the target area monitoring image contains a first person but does not contain a second person; the first person is a person who conducts business processing, and the second person is a person who assists the first person in conducting business processing; Obtain the target sub-region corresponding to the monitoring image of the target area, and determine the target second person corresponding to the target sub-region. The target second person refers to the second person responsible for assisting the first person in handling business in the target sub-region. Based on the surveillance images of the multiple areas, the trajectory of the target second person is obtained through a convolutional neural network personnel tracking processing model. Based on the personnel trajectory, determine the target time when the second target personnel arrives at the target sub-area; If the target time is greater than a preset time threshold, then scheduling instruction information for the target sub-region is obtained; the scheduling instruction information is used to instruct personnel scheduling for the target sub-region.
2. The method according to claim 1, characterized in that, The step of obtaining the target area monitoring image from the plurality of area monitoring images includes: The multiple area surveillance images are input into a pre-trained personnel detection model, and the personnel detection model is used to obtain images of areas with personnel captured from the area surveillance images. From the personnel area image, obtain a first personnel area image containing the first person and a second personnel area image containing the second person; The area monitoring image that contains the first personnel area image but does not contain the second personnel area image is used as the target area monitoring image.
3. The method according to claim 2, characterized in that, The step of obtaining a first personnel area image containing the first person and a second personnel area image containing the second person from the personnel area image includes: From the personnel area image, obtain personnel feature information; From the personnel region image, based on the similarity between the personnel feature information and the standard feature information, a first personnel region image and a second personnel region image are obtained; the standard feature information is the personnel feature information of the second personnel.
4. The method according to claim 3, characterized in that, The step of obtaining the first person region image and the second person region image from the person region image based on the similarity between the person feature information and standard feature information includes: The similarity is compared with a preset similarity threshold to obtain an evaluation result for the personnel region image; If the evaluation result meets the preset conditions, the personnel area image will be used as the second personnel area image. If the evaluation result does not meet the preset conditions, the personnel area image is used as the first personnel area image.
5. The method according to claim 2, characterized in that, The step of obtaining images of areas with people captured from the area monitoring images using the personnel detection model includes: The personnel detection model is used to obtain human characteristics. Based on the aforementioned personal characteristics, the personnel area image is identified from the multiple area surveillance images; The personnel area image is obtained by cropping the area monitoring image containing the personnel area image.
6. The method according to claim 2, characterized in that, The step of obtaining the trajectory of the target second person based on the multiple area surveillance images includes: The multi-frame regional surveillance images containing the target second person are input into a pre-trained personnel tracking processing model, and the personnel trajectory of the target second person is output through the personnel tracking processing model.
7. A personnel dispatching device, characterized in that, The device includes: The monitoring image acquisition module is used to acquire multiple monitoring images of a target area; the target area includes multiple sub-areas, and each monitoring image of a sub-area is a monitoring image of that sub-area. The target image acquisition module is used to acquire a target area monitoring image from the plurality of area monitoring images; the target area monitoring image contains a first person but does not contain a second person; the first person is a person who performs business processing, and the second person is a person who assists the first person in performing business processing; The target personnel determination module is used to acquire the target sub-region corresponding to the monitoring image of the target area, and determine the target second personnel corresponding to the target sub-region. The target second personnel refers to the second personnel responsible for assisting the first personnel in handling business in the target sub-region. The personnel trajectory determination module is used to obtain the personnel trajectory of the target second person based on the multiple area monitoring images and through a convolutional neural network personnel tracking processing model; The scheduling information acquisition module is used to determine the target time for the second target person to arrive at the target sub-area based on the personnel trajectory; if the target time is greater than a preset time threshold, scheduling instruction information for the target sub-area is acquired; the scheduling instruction information is used to instruct the personnel scheduling of the target sub-area.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.