An office scheduling system based on behavior recognition

Abstract

The present invention relates to the technical field of resource scheduling, and specifically to an office scheduling system based on behavior recognition. The system includes: a path analysis module, a behavior aggregation module, a scheduling screening module, a resource allocation module, and a preference scheduling module. In the present invention, by combining cross-regional image behavior recognition and a time-series reconstruction mechanism, the connection relationship of behavior trajectories is identified, group linkage behavior is recognized based on the intersection of behavior times and label consistency, and further, the resource overlap situation is judged according to the spatial number and function number. The scheduling priority is determined through the corresponding relationship between the role level and the behavior source, forming a task screening and status adjustment process. The intensity of the task execution intention is evaluated by triple determination of action tags, duty adaptation, and behavior coherence, and the resource locking is driven by the behavior intensity value. In the scheduling matching, the resource call frequency ranking and the employee usage time preference are introduced to construct a priority list, realizing the whole-process linkage from behavior perception to optimal resource allocation in the scheduling path.

Description

Technical Field

[0001] This invention relates to the field of resource scheduling technology, and in particular to an office scheduling system based on behavior recognition. Background Technology

[0002] Resource scheduling technology encompasses the rational allocation and optimization of various limited resources across time, space, and task dimensions. Its core lies in achieving dynamic resource allocation and task collaboration through the perception and evaluation of resource status, ensuring the overall efficiency and continuity of system operation. This includes the comprehensive management of elements such as office equipment, personnel, meeting venues, and time nodes. A resource scheduling framework is constructed based on task modeling, scheduling rule setting, and scheduling instruction execution. In office scenarios, it is applied to meeting scheduling, workstation allocation, employee schedule coordination, and office environment management. Data acquisition methods involve sensor data collection, behavior tracking, and historical scheduling data retrieval. Among these, behavior-based office scheduling systems refer to systems that combine user behavior information... This system enables office resource scheduling, addressing the scheduling needs of identifying personnel behavior and task status in the office environment. It encompasses the acquisition of personnel's actions within the office space, determination of behavior types, matching of scheduling rules, and issuance of resource execution instructions. It combines visual image acquisition devices to acquire personnel behavior characteristics, completes action recognition and behavior classification through an image recognition unit, and inputs the recognition results into a task matching unit. Based on the behavior type, it matches preset office scheduling rules and generates and sends resource scheduling commands through a scheduling control unit, such as automatically reserving meeting rooms, adjusting task order, or changing office locations. The process is based on an image acquisition unit, behavior recognition unit, rule matching unit, and scheduling control unit, forming a complete behavior-driven resource scheduling process.

[0003] In traditional office scheduling technology, there is a disconnect between the recognition of behavior and the scheduling response. Resource allocation is mainly based on task information and historical rule execution, failing to fully capture the impact of behavioral temporal changes on scheduling logic. This makes it difficult to accurately track employee movement between multiple areas, and there is a disconnect in the recognition of behavior start and end. In resource conflict scenarios, there is also a lack of consistency verification of behavioral source events and role level ranking mechanisms, resulting in situations where the same behavior triggers tasks multiple times or low-priority tasks preempt critical resources. For example, in meeting scheduling, it is difficult to determine whether the arrival paths of different employees are continuous behaviors, causing meeting room resources to be incorrectly pre-locked, resulting in scheduling redundancy and priority misalignment. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an office scheduling system based on behavior recognition.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an office scheduling system based on behavior recognition includes:

[0006] The path analysis module calls regional monitoring images to identify the start and end times of employee behavior records in multiple time periods, judges the continuity of behavior time, and judges the cross-regional path connection relationship by comparing the relationship between the trajectory end and the image boundary direction and the angle between the motion vectors, and generates cross-regional behavior recognition records.

[0007] The behavior aggregation module filters behavior records with overlapping time periods and consistent action tags in multiple behavior regions based on the cross-regional behavior recognition records, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path tags, and generates behavior linkage analysis results.

[0008] The scheduling and filtering module uses the behavioral linkage analysis results to determine the combination of spatial number and functional number of task resource code, identify behavioral event identifiers and task triggering sources, adjust the task execution status according to the employee role number level order, and generate resource conflict coordination results.

[0009] Based on the resource conflict coordination results, the resource allocation module identifies the matching relationship between action tags and resource function tags, assesses the responsibility correlation between roles and task categories, judges the continuity of the trajectory and the intensity of the employee's task execution intention, locks the associated resources, and generates resource pre-locking information.

[0010] As a further aspect of the present invention, the cross-regional behavior recognition record includes the start and end time points of the behavior, the direction angle of the trajectory end, and the pointing relationship of the image boundary. The behavior linkage analysis results specifically include the consistency of action labels, the intersection length of time periods, and the correspondence of path labels. The resource conflict coordination results include the matching degree of event identifier numbers, the combination relationship of spatial and functional numbers, and the corresponding order of role levels. The resource pre-locking information specifically includes the matching value of action and functional labels, the trajectory continuity index, and the task intent intensity coefficient.

[0011] As a further aspect of the present invention, the path analysis module includes:

[0012] The trajectory time extraction submodule acquires the frame sequence of regional monitoring images, identifies the timestamps of the first appearance and disappearance frames of each employee in the same area, classifies the timestamps according to the employee identifier, extracts the time interval between adjacent image records and the sequential identifier of the corresponding action number, judges the continuity of the behavior time sequence, and generates a behavior time continuity judgment value.

[0013] The boundary direction determination submodule extracts the continuous position coordinates of the end of the trajectory based on the behavior time continuity determination value, constructs the direction vector of the end segment of the trajectory, obtains the direction vector of the shortest connection from the end point of the trajectory to the image boundary, obtains the average end offset value of the trajectory segment, calculates the direction consistency determination value, and obtains the boundary direction consistency coefficient.

[0014] The path connection verification submodule extracts the coordinate values ​​of the trajectory end point and the trajectory start point in the adjacent area based on the boundary direction consistency coefficient, calculates the distance and direction continuity angle between the two coordinate points, determines the path connection relationship between the trajectory start point and end point in the adjacent area, establishes cross-regional behavior sequence labels, and generates cross-regional behavior recognition records.

[0015] As a further aspect of the present invention, the behavior aggregation module includes:

[0016] The time segment comparison submodule obtains the cross-regional behavior recognition records, extracts the start and end times of the behavior of multiple employees in each behavior region, calculates the length of the intersection interval of the behavior time periods of each group of employees, and performs a ratio conversion between the length of the intersection interval and the length of the original behavior time period to generate the behavior time overlap ratio value.

[0017] The action tag filtering submodule extracts the action tags recorded in each behavior region based on the behavior time overlap ratio value, judges the behavior combinations with the same tags, and calculates the behavior consistency fusion coefficient by combining the time overlap ratio value and the tag consistency.

[0018] The behavior linkage generation submodule determines and identifies behavior combinations with consistent behavior labels in multiple regions based on the behavior consistency fusion coefficient, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path labels, and generates behavior linkage analysis results.

[0019] As a further aspect of the present invention, the scheduling and filtering module includes:

[0020] The resource identifier filtering submodule obtains the behavior linkage analysis results and establishes resource number combinations by identifying and extracting the spatial number and functional number of the associated resource codes in the scheduling task.

[0021] The event association comparison submodule, based on the resource number combination, calls the behavior event identifier in the scheduling task and the event number recorded in the task instruction, compares the matching of the behavior event identifier and the event number recorded in the task instruction, filters the resource scheduling set triggered by the same behavior source, and obtains the event source analysis result.

[0022] The task sorting and adjustment submodule determines the execution priority of tasks and adjusts the task execution status based on the analysis results of the event source and the hierarchical order of employee role numbers, and generates resource conflict coordination results.

[0023] As a further aspect of the present invention, the resource allocation module includes:

[0024] The tag matching and judgment submodule obtains the action tags and target resource function tags in each task based on the resource conflict coordination results, judges the degree of association matching, and generates a tag content association degree value.

[0025] The duty path analysis submodule identifies the role category corresponding to the employee identity label based on the correlation value of the label content, compares the task category information, obtains the role task participation characteristics, and extracts the spatial trajectory sequence of the employee in the task to statistically analyze the path coherence and regional distribution to obtain the behavioral coherence performance index.

[0026] The intent intensity assessment submodule calculates the task execution intent intensity value based on the behavioral coherence performance index, combined with the role task participation characteristics and the degree of correlation of tag content, judges and locks the associated resources, and generates resource pre-locking information.

[0027] As a further aspect of the present invention, the system further includes:

[0028] The preference scheduling module uses the resource pre-locking information to count the frequency of resource calls by employees in multiple time periods, determine the resource sorting index position, construct a priority resource list, adjust the corresponding resource scheduling matching order, and generate a resource preference adjustment record.

[0029] The resource preference adjustment record includes resource frequency statistics, time period priority index, and scheduling order adjustment sequence.

[0030] As a further aspect of the present invention, the preference scheduling module includes:

[0031] The resource frequency statistics submodule obtains the resource pre-locking information, analyzes the resource call sequence of employees in multiple time periods, counts the number of times each resource code is called in multiple time periods, and establishes a resource frequency ranking quantity.

[0032] The priority resource generation submodule, based on the resource frequency ranking, combines priority resource numbers to form a set according to the resource call characteristics of each time period, constructs the priority resource list for employees in the corresponding time period, and establishes the resource priority set data;

[0033] The scheduling order adjustment submodule analyzes the correspondence between the time period of the employee's current scheduled task and the resource number of the priority set based on the resource priority set data, adjusts the resource scheduling matching order, and establishes a resource preference adjustment record.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, by combining cross-regional image behavior recognition and temporal reconstruction mechanisms, the connection relationship of behavior trajectories is identified. Based on the intersection of behavior time and label consistency, group linkage behavior is identified. Further, resource overlap is judged according to spatial number and function number. The scheduling priority is determined by the correspondence between role level and behavior source, forming a task screening and status adjustment process. The intensity of task execution intention is evaluated by a triple judgment of action label, responsibility adaptation and behavior coherence. The resource locking is driven by the behavior intensity value. In the scheduling matching, the resource call frequency ranking and employee usage time preference are introduced to build a priority list, realizing the linkage of the entire process from behavior perception to resource optimization in the scheduling path. Attached Figure Description

[0036] Figure 1 This is a system flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the path analysis module of the present invention;

[0038] Figure 3 This is a flowchart of the behavior aggregation module of the present invention;

[0039] Figure 4 This is a flowchart of the scheduling and filtering module of the present invention;

[0040] Figure 5 This is a flowchart of the resource allocation module of the present invention;

[0041] Figure 6 This is a flowchart of the preference scheduling module of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0044] Please see Figure 1 An office scheduling system based on behavior recognition includes:

[0045] The path analysis module calls regional monitoring images to identify the start and end times of employee behavior records in multiple time periods, judges the continuity of behavior time, and judges the cross-regional path connection relationship by comparing the relationship between the trajectory end and the image boundary direction and the angle between the motion vectors, and generates cross-regional behavior recognition records.

[0046] The behavior aggregation module filters behavior records with overlapping time periods and consistent action tags from multiple behavior regions based on cross-regional behavior recognition records, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path tags, and generates behavior linkage analysis results.

[0047] The scheduling and screening module uses the results of behavioral linkage analysis to determine the combination of spatial number and functional number of task resource code, identify behavioral event identifiers and task triggering sources, adjust the task execution status according to the employee role number level order, and generate resource conflict coordination results.

[0048] Based on the resource conflict coordination results, the resource allocation module identifies the matching relationship between action tags and resource function tags, assesses the responsibility correlation between roles and task categories, judges the continuity of the trajectory and the intensity of the employee's task execution intention, locks the associated resources, and generates resource pre-locking information.

[0049] The preference scheduling module uses resource pre-locking information to count the frequency of resource calls by employees in multiple time periods, determine the resource sorting index position, build a priority resource list, adjust the corresponding resource scheduling matching order, and generate resource preference adjustment records.

[0050] Cross-regional behavior recognition records include the start and end times of the behavior, the direction and angle of the trajectory end, and the pointing relationship of the image boundary. The behavior linkage analysis results specifically include the consistency of action labels, the length of the intersection of time periods, and the correspondence of path labels. The resource conflict coordination results include the matching degree of event identifier numbers, the combination relationship of spatial and functional numbers, and the corresponding order of role levels. The resource pre-locking information specifically includes the matching value of action and functional labels, the trajectory continuity index, and the task intent intensity coefficient. The resource preference adjustment records include resource frequency statistics, time period priority index, and scheduling order adjustment sequence.

[0051] Please see Figure 2 The path analysis module includes:

[0052] The trajectory time extraction submodule acquires the frame sequence of regional monitoring images, identifies the timestamps of the first appearance and disappearance frames of each employee in the same area, classifies the timestamps according to the employee identifier, extracts the time interval between adjacent image records and the sequential identifier of the corresponding action number, judges the continuity of the behavior time sequence, and generates a behavior time continuity judgment value.

[0053] Taking an office area as an example, assuming that the image monitoring frame sequence for office area A between 9:00 and 9:30 is one frame per second, the image record of employee number 1001 is processed by sequentially scanning the pixel differences of each frame and calculating a binary marker indicating whether employee 1001 appears in each frame. The timestamp of the first frame in which the employee appears in the image is determined by identifying the frame number where the employee's identifier pixel first appears. For example, if the employee first appears in frame 180, then the timestamp of the first appearance frame is 9:03:00. The timestamp of the disappearing frames is determined by identifying the continuous disappearance of the employee's identifier pixel. The first frame number for frames exceeding three is determined. For example, if the disappearing frame is frame 900, then the timestamp of the disappearing frame is 9:15:00. The start and end timestamps of employee 1001 are recorded and saved separately, categorized and stored according to employee identifiers. The data for each employee is extracted using the same method. The time interval between adjacent image records is calculated by the frame difference between two consecutive images. If employee 1001 first appears in frame 180 and then appears again in frame 190, the corresponding time interval is 10 seconds, and so on, obtaining the time difference of all consecutive records. Continuous action numbers are assigned to employee 1001, defining the first appearance as action number 1, the second appearance as action number 2, and so on. Based on the continuity of the action number sequence, the nth action number is compared with the (n+1)th action number. The temporal continuity of the behavior is determined by whether the difference between the two numbers is 1. If the action numbers are 3 and 4 respectively, they are confirmed as continuous, and a behavior temporal continuity judgment value is generated and recorded as "continuous"; otherwise, it is recorded as "discontinuous".

[0054] The boundary direction determination submodule extracts the continuous position coordinates of the trajectory end based on the behavior time continuity determination value, constructs the trajectory end segment direction vector, obtains the direction vector of the shortest line connecting the trajectory end point to the image boundary, and obtains the average end offset value of the trajectory segment using the formula:

[0055] ;

[0056] Calculate the directional consistency judgment value to obtain the boundary directional consistency coefficient;

[0057] in, The horizontal component of the motion vector at the end of the trajectory is calculated by extracting the difference in the horizontal coordinates of the first and last points among the last three consecutive position coordinates. The longitudinal component of the motion vector at the end of the trajectory is calculated by extracting the difference in the longitudinal coordinates of the first and last points among the last three consecutive position coordinate points. This is the lateral component of the shortest connection vector from the endpoint of the trajectory to the image boundary, obtained by calculating the lateral distance between the endpoint coordinates and the shortest connection point on the four sides of the boundary. This is the longitudinal component of the shortest connection vector from the endpoint of the trajectory to the image boundary, obtained by calculating the longitudinal distance between the endpoint coordinates and the shortest connection point on the four sides of the boundary. The average coordinate offset of the final segment of the trajectory is obtained by extracting three consecutive coordinate points at the end and averaging the Euclidean distances between adjacent points. This is the trajectory direction stability adjustment coefficient, a dimensionless real constant preset by the system used to adjust the influence of the offset term on the overall judgment value. The maximum permissible final trajectory offset is set for the image region and is obtained by taking the maximum offset tolerance value from the system configuration or scene calibration. The direction consistency judgment value represents the degree of consistency between the current trajectory direction and the direction of approach to the image boundary. It serves as a key judgment value in behavior screening to determine whether the path meets the cross-regional behavior standard.

[0058] Based on the aforementioned "continuous" behavior time continuity judgment value for employee 1001, the continuous position coordinates at the end of the employee's trajectory are extracted. For example, the actual image coordinates of the last three coordinate points are (120, 150), (123, 152), and (127, 154). The horizontal component of the direction vector at the end of the trajectory is then extracted. This is obtained by calculating the difference between the x-coordinates of the first and third points. Longitudinal component Calculated by the difference between the ordinates of the first and third points, i.e. Extract the lateral component of the shortest line vector from the trajectory endpoint (127, 154) to the image boundary. With longitudinal component Taking an image size of 200×200 pixels as an example, if the shortest distance from the endpoint to the boundary is the bottom boundary (y=200), then the horizontal distance is... Longitudinal distance Average coordinate offset at the end of the trajectory The calculation process is as follows: calculate the Euclidean distance between the last three coordinate points respectively: First distance Pixels, second distance Pixels, then average offset Pixel, trajectory direction stability adjustment coefficient The system preset values ​​are set based on scenario calibration experiments, taking the stability of an office monitoring scenario as an example. Maximum allowable trajectory offset Through experimental calibration in the office area, with a maximum offset tolerance of 10 pixels, the following calculation was performed using the given formula:

[0059] ;

[0060] Step-by-step calculation yields:

[0061] First, calculate the vector dot product:

[0062] ;

[0063] Next, calculate the vector magnitudes respectively:

[0064] ;

[0065] ;

[0066] The cosine of the angle between the vectors is then calculated as follows:

[0067] ;

[0068] Add offset item:

[0069] ;

[0070] The value for determining directional consistency is:

[0071] ;

[0072] The directional consistency judgment value is a composite metric used to measure whether a behavioral trajectory faces the image boundary and maintains this directional trend. It comprehensively considers the angular relationship between the trajectory's final segment direction and the boundary connection direction (spatial vector cosine) and the average offset amplitude of the final segment movement (trajectory stability), essentially reflecting whether the current behavioral path possesses integrity and continuity. The closer its value is to the upper threshold (e.g., close to 1), the stronger the stability and the clearer the direction of the trajectory as it moves towards the boundary, making it likely to be identified as a cross-regional continuous behavioral path. When this value is used for behavior filtering, it can serve as a core metric for filtering boundary-approaching behaviors in image behavior data, effectively eliminating directional deviation behaviors, image pause segments, and non-connected trajectories, thus enhancing the continuity and accuracy of path recognition. The results show that the degree of agreement between the trajectory's final segment direction and the image boundary approach direction is 0.5768 (value range 0-1), exceeding the preset benchmark threshold of 0.5, thus the boundary directional consistency coefficient is judged to meet the cross-regional behavior standard.

[0073] The path connection verification submodule extracts the coordinate values ​​of the trajectory end point and the trajectory start point in the adjacent area based on the boundary direction consistency coefficient, calculates the distance and direction continuity angle between the two coordinate points, judges the path connection relationship between the trajectory start point and end point in the adjacent area, establishes cross-regional behavior sequence labels, and generates cross-regional behavior recognition records.

[0074] Based on the boundary direction consistency coefficient of 0.5768, which meets the standard, the coordinates of the end point of employee 1001's trajectory (127, 154) and the starting point coordinates of the adjacent region B's trajectory (135, 160) are extracted. First, the Euclidean distance between the two coordinate points is calculated: Secondly, the directional continuity angle is calculated by taking the dot product of the employee's trajectory vector (7,4) at the end of region A and the inter-region connection vector (135-127,160-154)=(8,6). Then the direction angle Since the distance of 10 pixels is less than the maximum allowable distance for cross-region connection (15 pixels) and the direction continuity angle of 7.2 degrees is less than the maximum allowable angle (15 degrees), the trajectory start and end path connection relationship between region A and region B is determined to be valid. The cross-region behavior sequence label "1001_A to B_Valid" is established, and a cross-region behavior recognition record is generated.

[0075]

[0076] As shown in Table 1, the parameters in the table are obtained from actual monitoring image data. Taking employee 1001 as an example, the distance and direction parameters between the end point of the employee's trajectory and the starting point of the trajectory in the neighboring area are calculated to meet the requirements for effective connection, and finally an effective cross-regional behavior label is established.

[0077] Please see Figure 3 The behavior aggregation module includes:

[0078] The time segment comparison submodule obtains cross-regional behavior recognition records, extracts the start and end times of multiple employees' behaviors in each behavior region, calculates the length of the intersection interval of each group of employees' behavior time periods, and converts the length of the intersection interval to the length of the original behavior time period to generate the behavior time overlap ratio value.

[0079] First, the start and end timestamps of multiple employees' behaviors in the office area are retrieved from the storage. Taking the behavior data of employees 1001 and 1002 in area A and area B as examples, employee 1001's behavior in area A starts at 9:03:00 and ends at 9:15:00, while employee 1002's behavior in area B starts at 9:10:00 and ends at 9:25:00. Based on the above data extraction process, the behavior duration of employee 1001 is defined as follows: (i.e., 12 minutes), the duration of employee 1002's behavior is (i.e., 15 minutes) Calculate the intersection length of each behavior time period. First, compare the start and end timestamps. Take the later start timestamp and the earlier end timestamp of the behavior time period as the start and end points of the intersection interval. For example, if the start timestamp of the intersection interval between employee 1001 and employee 1002 is 9:10:00 and the end timestamp is 9:15:00, then the intersection duration is obtained. (i.e., 5 minutes), the ratio of the intersection duration to the original behavior duration is calculated, and the overlap ratio of the behavior time of employees 1001 and 1002 is calculated respectively. and The above calculation steps are then applied to the cross-regional behavior data of all employees to obtain the total behavior time overlap ratio value, forming a set of behavior time overlap ratio values ​​for use in subsequent steps.

[0080] The action tag filtering submodule extracts action tags recorded in each behavior region based on the behavior time overlap ratio, identifies behavior combinations with the same tags, and calculates the consistency between the time overlap ratio and the tag using the following formula:

[0081] ;

[0082] Calculate the behavioral consistency fusion coefficient;

[0083] in, The behavioral consistency fusion coefficient is a synergy value that comprehensively measures the degree of overlap between behavioral time and the consistency of action labels. For the first The overlap ratio of time segments corresponding to a group of behaviors is obtained by extracting the start and end time periods of the behaviors and calculating the ratio of the intersection time length to the original behavior duration. For the first The action label consistency flag for a group of behaviors; when the labels are the same... Different tags This was obtained by comparing the action types of the same employee's behavior records in different regions. The total number of behavioral combinations within the current analysis scope is obtained by recording the logarithm of employee behaviors that overlap across time periods in different regions. The average time overlap ratio of all behavior combinations is calculated by... Add the values ​​and then divide by get, A fluctuation adjustment item is used for label matching; this is a preset value in the system used to offset evaluation bias caused by fluctuations in behavior over time. The behavior combination index number is used to identify the current behavior participating in the calculation. Group cross-regional behavior combinations;

[0084] Based on the aforementioned set of behavior time overlap ratios, taking the cross-regional behaviors of employees 1001 and 1002 as examples, action labels are extracted and judged. Employee 1001's action label in region A is "discussion," and employee 1002's action label in region B is "discussion." Since the two labels are consistent, the action label consistency flag is denoted as... If the labels are inconsistent, mark it as 0. Perform similar operations on employees 1003 and 1004. Assuming the action labels for 1003 and 1004 are "Rest" and "Work" respectively, if the labels are inconsistent, mark it as 0. Extract the consistency flags of all employees' action tags in this manner. The overlap ratio with the previously calculated behavioral time Joint calculations are performed to ensure that the label matches the fluctuation adjustment term. The rationality of this was demonstrated through preliminary experiments in an office setting, sampling employee label fluctuations at different times within a week, resulting in a reasonable range for the label fluctuation adjustment coefficient. In this embodiment, it is set For the formula:

[0085] ;

[0086] The parameters in the document are explained. The behavioral consistency fusion coefficient measures the consistency and synchronicity of interactive behaviors. This represents the sum of the time overlap ratios among all combinations with identical action labels. This indicates the number of behavioral combinations within the analysis scope. , It is the first The percentage of overlapping time periods for the group's employees has been calculated previously. Represents all combinations Average value, i.e. The specific calculation process is as follows:

[0087] ;

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] The behavioral consistency fusion coefficient is a numerical indicator used to quantify the "collaborative action trend" of multiple employees performing behaviors across different office areas. This coefficient comprehensively considers two key factors: the degree of overlap in the time periods of employee behaviors in different areas, and the consistency of action labels among these overlapping behaviors. If a group of behaviors overlaps in time and the corresponding action labels are completely consistent, and the time distribution of each group of behaviors is stable (with minimal fluctuations), then the coefficient will show a large value. This coefficient is not only used to identify the consistency of individual behaviors but also to reveal whether there is a cross-regional linkage behavior chain, providing a basis for determining whether to use this behavior combination as a synchronous scheduling unit in subsequent scheduling strategies. The larger the value, the more likely the behavior group is to constitute a logically linked behavioral entity, applicable to office scheduling scenarios such as meeting group determination, resource reservation merging, and collaborative path priority scheduling. The result shows that the behavioral consistency fusion coefficient is 0.2665. Based on the preset benchmark range of fusion coefficients for office scenarios [0.2, 0.4], this result is within a reasonable range and can be used to determine whether behavioral linkage is effective.

[0093]

[0094] As shown in Table 2, the data on employee time, overlap time, action tags, and consistency indicators were calculated and marked through monitoring records.

[0095] The behavior linkage generation submodule judges and identifies behavior combinations with consistent behavior labels in multiple regions based on the behavior consistency fusion coefficient, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path labels, and generates behavior linkage analysis results.

[0096] Based on the behavioral consistency fusion coefficient, employee combinations 1001-A and 1002-B with the cross-regional behavioral label "discussion" are extracted. Further judgment and identification of cross-regional time-segment intersection combinations are then performed. First, the fusion coefficient is compared using regional synchronicity features. Values ​​less than 0.2 are considered poorly synchronized and not marked. Values ​​between [0.2, 0.4] are considered moderately synchronized and marked as cross-regional time-segment intersection combinations. Finally, values ​​higher than 0.4 are considered highly synchronized and prioritized for marking. In this embodiment, the coefficient is 0.2665, falling within the [0.2, 0.4] range. Therefore, employee combinations 1001-A and 1002-B with the cross-regional behavioral label "discussion" and a fusion coefficient of 0.2665 are recorded. The cross-regional time-segment intersection combination identifier is established as: "Linked Discussion (1001-A↔1002-B)". Combined with the behavioral path label "1001_A to B_Effective", the linkage relationship between region A and region B is determined, yielding the cross-regional behavioral linkage analysis results.

[0097] Please see Figure 4 The scheduling and filtering module includes:

[0098] The resource identifier filtering submodule obtains the behavioral linkage analysis results and establishes resource number combinations by identifying and extracting the spatial number and functional number of the associated resource codes in the scheduling task.

[0099] Taking the cross-regional behavior combination of employee 1001-A and employee 1002-B as an example, the associated resource information in the scheduling task system is called, and the corresponding space number and function number are extracted one by one. By calling the pre-stored resource configuration information, the space number of the behavior area where employee 1001 is located is A301 and the function number is F12, and the space number of the behavior area where employee 1002 is located is B207 and the function number is F12. The complete resource number combination is established according to the format of "space number-function number", that is, the combination is "A301-F12" and "B207-F12", and stored in an array structure: As input data items for event association comparison, the above process is then repeated to sequentially obtain the resource IDs of all cross-regional behavior combinations, forming a unified set of resource ID combinations.

[0100] The event association comparison submodule, based on resource number combinations, calls the behavior event identifier in the scheduling task and the event number recorded in the task instruction, compares the matching of the behavior event identifier and the event number recorded in the task instruction, filters the resource scheduling set triggered by the same behavior source, and obtains the event source analysis results.

[0101] Using the resource number combinations stored in the above array as input, the system retrieves the behavior event identifiers from the stored task scheduling instruction database and the event numbers from the task instruction records. Each event number is compared sequentially using an event number matching method. First, the resource number combination "A301-F12" is taken. Based on the pre-stored task record, its corresponding behavior event identifier E101 is retrieved and matched character by character with the event numbers stored in the task instruction record. If the event number in the task instruction is E101, it is marked as a successful match; otherwise, it is marked as a mismatch. For the second resource number combination "B207-F12", its behavior event identifier E102 is retrieved, and the same matching operation is performed. Assuming the event number in the task instruction record is E102, it is marked as a successful match. Finally, through the above item-by-item matching method, the resource number combinations marked as successfully matched are filtered and recorded, forming the event source analysis results. The matching results are output in array form. This is used for subsequent task sorting.

[0102] The task sorting and adjustment submodule determines the execution priority of tasks and adjusts their execution status based on the event source analysis results and the hierarchical order of employee role numbers, generating resource conflict coordination results.

[0103] Using the above event source analysis results as input, the pre-stored employee role level database is called to extract the level number and level order of the corresponding employee roles. Assuming that the level order is marked by numbers, the smaller the number, the higher the level. For example, the level order of employee 1001 is 2 and the level order of employee 1002 is 3. The level numbers of the tasks corresponding to employees 1001 and 1002 are compared one by one. The level number 2 of employee 1001's task is compared with the level number 3 of employee 1002's task. Since number 2 < number 3, it is determined that the task of employee 1001 has a higher priority. Then, the current task execution status data is called. Assuming that the initial task status is "pending execution", the status of employee 1001's task is adjusted to "priority execution" and the status of employee 1002's task is marked as "secondary execution". The sorting and adjustment of tasks in all event source analysis results are completed in this way to generate a complete set of resource conflict coordination result data, as shown in Table 3.

[0104]

[0105] As shown in Table 3, by clarifying the order of employee roles and comparing the level numbers item by item, the execution status of the corresponding tasks is adjusted to obtain the final resource conflict coordination result.

[0106] Please see Figure 5 The resource allocation module includes:

[0107] The tag matching and judgment submodule obtains the action tags and target resource function tags in each task based on the resource conflict coordination results, judges the degree of association and matching, and generates a tag content association degree value.

[0108] Based on employee task data from resource conflict coordination results, action tags and target resource function tags are extracted for each task. Specific judgment actions are performed, and the degree of tag association matching is statistically analyzed. Taking employee task 1001 as an example, the action tag for this task is extracted as "discussion," and the target resource function tag is "meeting discussion." A word-by-word tag matching is performed to determine the number of identical keywords in "discussion" and "meeting discussion." The keyword "discussion" is matched once, out of a total of two tags. The number of matched keywords is divided by the total number of tags to calculate the tag content association value, i.e., the tag content association value. Perform the same operation on employee task 1002. Assume the action tag for employee task 1002 is "Write a report," the target resource function tag is "Report Writing," and the matching keywords are "report" and "writing," a total of two tags. The total number of tags is two. What is the tag content relevance value? Applying the above calculation process to all task data generates a set of tag content association values, presented as an array. Stored in a format for later use in subsequent steps.

[0109] The duty path analysis submodule identifies the role category corresponding to the employee's identity label based on the correlation value of the tag content, compares the task category information, obtains the role task participation characteristics, and obtains the behavioral consistency performance index by extracting the spatial trajectory sequence of employees in the task and statistically analyzing the path coherence and regional distribution.

[0110] Based on the set of tag content association values, the role classification data of employee identity tags is called. Taking employee 1001 as an example, the role classification is extracted as supervisor (number R02), and the task category information is management and coordination. The employee role responsibilities and task category responsibilities are compared item by item, with a pre-stored role responsibility matching table as a reference. If the matching degree between supervisor responsibilities and management and coordination tasks is 0.8, this value is directly used as the role task participation feature value. Taking employee 1002 as an example, the role is classified as ordinary employee (ID R03), and the task category is document output. The matching degree between ordinary employee level and document output tasks is 0.9. Therefore, the role-task participation feature value of employee 1002 is... Then, the employee task trajectory data was extracted. Taking the task trajectory data of employee 1001 [(10,20), (11,21), (12,22), (20,30)] as an example, the path continuity was calculated point by point. It was found that the first three points were continuous and uninterrupted, while the latter part had obvious jump intervals. The path had a total of 4 points, with 3 points being continuous segments. The path continuity ratio was calculated by calling the ratio of the number of continuous segments to the total number of points. The path interruption count is 1, and the predefined formula is called. The behavioral consistency performance index was obtained as follows: In this way, data from other employees is calculated one by one to form a complete set of consistent behavioral performance indicators, presented in arrays. express.

[0111] The intent intensity assessment submodule uses a formula based on behavioral coherence performance indicators, combined with role task participation characteristics and the relevance of tag content:

[0112] ;

[0113] Calculate the intensity value of the task execution intent, determine and lock the associated resources, and generate resource pre-locking information;

[0114] in, The relevance value of the tag content is obtained by extracting task action tags and resource function tags, and calculating the ratio of the number of matching tag groups to the total number of tag groups. As the role-task participation feature value, after judging the responsibility mapping results between the role classification and task category corresponding to the employee identity label, the responsibility fit and matching degree value is extracted. As a performance indicator of behavioral consistency, it is calculated by statistically analyzing the proportion of continuous segments and the number of interruptions in the employee's trajectory sequence during task execution, and then normalizing the results. The coherence adjustment parameter is a fixed proportional constant set by the system, used to adjust the weight of behavioral coherence in the expression of intent. The number of time segments for trajectory jumps is obtained by counting the number of time segments corresponding to discontinuous regions in the behavior execution trajectory. The task execution intent strength value is a quantitative indicator used to measure whether employees have a clear execution intent before task scheduling.

[0115] Using the above data as input, set the consistency adjustment parameters. The value is the system's preset value. Actual measurements in an office setting showed a range of 0.1 to 0.3; the median value was taken. Number of time periods for trajectory jumps The number of discontinuous areas in the employee's trajectory is counted; employee 1001 has one skipped area. Call the formula:

[0116] ;

[0117] Based on employee 1001 data Input and calculate the intensity value of task execution intention step by step. :

[0118] First, calculate the average:

[0119] ;

[0120] Next, calculate the adjustment items:

[0121] ;

[0122] Finally, the final result is calculated by combining the two parts:

[0123] ;

[0124] The task execution intent strength value is a combined indicator that measures an employee's willingness and adaptability to the current scheduled task. This value integrates three parameters: label matching degree, responsibility adaptability, and behavioral coherence. A higher value indicates a clear current behavioral path, well-defined resource objectives, and a close connection between the task and responsibility, indicating a high-priority execution intent. Conversely, if behavioral jumps are frequent and the label or responsibility connection is weak, the value will be automatically lowered, indicating a lower scheduling priority or that the task can be cancelled. This indicator supports resource locking decisions, further identifying and judging employees' actual intentions and needs for office resources, and reducing resource waste. The results show that the task execution intent strength of employee 1001 is 0.684, falling within the mid-to-high range of [0,1] (defined as 0.6-0.8), indicating a clear execution intent. Based on this, resource locking actions are performed on resource numbers A301-F12 associated with employee 1001's task, marking them as pre-locked. The above steps are repeated to calculate and judge the task execution intent strength values ​​of other employees, forming complete resource pre-locking information, as shown in Table 4.

[0125]

[0126] As shown in Table 4, the intention intensity value of employee 1002 after being substituted into the calculation is 0.954, which is significantly higher than the lower limit of the high intensity range of 0.8. Therefore, resource pre-locking is also performed on resource numbers B207-F12.

[0127] Please see Figure 6 The preference scheduling module includes:

[0128] The resource frequency statistics submodule obtains resource pre-locking information, analyzes the resource call sequence of employees in multiple time periods, counts the number of times each resource code is called in multiple time periods, and establishes a resource frequency ranking.

[0129] Using the employee ID in the resource pre-locking information as an index, the resource call sequences of employees 1001 and 1002 within the past week are extracted by calling the pre-stored employee resource call records for multiple time periods (e.g., morning 9:00-12:00, afternoon 13:00-17:00). The specific extraction process is as follows: the number of times employee 1001 calls resources A301-F12 in the morning is counted, and the count is made daily. For example, the number of calls on Monday is 3, the number of calls on Tuesday is 2, the number of calls on Wednesday is 4, and so on. The same count is performed on all the resources involved. After the resource call frequency is counted, the number of calls in each time period is added up to the total number of calls for the corresponding resource code. For example, the total number of calls for A301-F12 by employee 1001 in the morning is 16 times and the total number of calls for B207-F12 is 5 times in a week. This forms the resource frequency ranking data, as shown in Table 5.

[0130]

[0131] As shown in Table 5, the frequency ranking of resources called by employees 1001 and 1002 in different time periods has been completed and will be used to generate the priority resource list.

[0132] The priority resource generation submodule sorts resources by frequency and combines priority resource numbers into a set according to the resource call characteristics of each time period, constructs a priority resource list for employees in the corresponding time period, and establishes priority resource set data.

[0133] Based on the resource frequency ranking data in Table 5, the resource IDs ranked highest for each employee in each time period are extracted and sorted in descending order according to the number of times they are called. A selection action is then performed, defining the top-ranked resources as those with the highest number of calls. Taking employee 1001 as an example, in the morning resource call ranking, A301-F12 are called 16 times, ranking first, and B207-F12 are called 5 times, ranking second. Therefore, the set of priority resource IDs for the morning period is defined as: [A301-F1...]. 2. In the afternoon resource call ranking for employee 1002, B207-F12 was called 14 times and ranked 1st, while C105-F05 was called 7 times and ranked 2nd. Therefore, the priority resource number set for the afternoon period is defined as: [B207-F12, C105-F05]. After executing the above process, the priority resource list for all employees in the corresponding time period is completed, forming the final priority resource set data, which is stored in array form as employee 1001: [9:00-12:00 AM → A301-F12, B207-F12], employee 1002: [1:00-5:00 PM → B207-F12, C105-F05], for the next step of resource scheduling order adjustment.

[0134] The scheduling order adjustment submodule analyzes the correspondence between the time period of the employee's current scheduled task and the resource number of the priority set based on the resource priority set data, adjusts the resource scheduling matching order, and establishes a resource preference adjustment record;

[0135] The above resource priority set data is invoked, using the current employee task scheduling time period as input. The matching between the resource IDs involved in each task and the corresponding priority set resource IDs is analyzed one by one. Specifically, the process is as follows: Taking employee 1001's current morning task call resource IDs A301-F12 as an example, each item A301-F12 is compared with the character consistency of resource IDs in the priority resource set [A301-F12, B207-F12]. A resource ID matching judgment action is performed. A301-F12 completely matches the first item A301-F12 in the priority set, therefore, the resource call order does not need to be adjusted and remains at the original order of first. For employee 1002's afternoon task call resource C105-F05, each item is compared with C10... 5-F05 matches the resource number characters in the priority resource set [B207-F12, C105-F05]. C105-F05 fails to match the first item B207-F12 in the priority set, but successfully matches the second item C105-F05. Therefore, the resource calling order is determined to be adjusted from the initial order of 1 to the second order in the priority set. The above comparison judgment and adjustment actions are executed to complete the adjustment of the resource scheduling order of all tasks, and a resource preference adjustment record is generated. The specific record content is as follows: the resource calling order of employee 1001's morning task A301-F12 is 1 (unadjusted), and the resource calling order of employee 1002's afternoon task C105-F05 is adjusted from the initial order of 1 to 2, completing the complete resource preference adjustment record.

[0136] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An office scheduling system based on behavior recognition, characterized in that, The system includes: The path analysis module calls regional monitoring images to identify the start and end times of employee behavior records in multiple time periods, judges the continuity of behavior time, and judges the cross-regional path connection relationship by comparing the relationship between the trajectory end and the image boundary direction and the angle between the motion vectors, and generates cross-regional behavior recognition records. The behavior aggregation module filters behavior records with overlapping time periods and consistent action tags in multiple behavior regions based on the cross-regional behavior recognition records, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path tags, and generates behavior linkage analysis results. The scheduling and filtering module uses the behavioral linkage analysis results to determine the combination of spatial number and functional number of task resource code, identify behavioral event identifiers and task triggering sources, adjust the task execution status according to the employee role number level order, and generate resource conflict coordination results. Based on the resource conflict coordination results, the resource allocation module identifies the matching relationship between action tags and resource function tags, assesses the responsibility correlation between roles and task categories, judges the continuity of the trajectory and the intensity of the employee's task execution intention, locks the associated resources, and generates resource pre-locking information.

2. The office scheduling system based on behavior recognition according to claim 1, characterized in that, The cross-regional behavior recognition record includes the start and end time points of the behavior, the direction angle of the trajectory end, and the pointing relationship of the image boundary. The behavior linkage analysis results specifically include the consistency of action labels, the length of the intersection of time periods, and the correspondence of path labels. The resource conflict coordination results include the matching degree of event identifier numbers, the combination relationship of spatial and functional numbers, and the corresponding order of role levels. The resource pre-locking information specifically includes the matching value of action and functional labels, the trajectory continuity index, and the task intent intensity coefficient.

3. The office scheduling system based on behavior recognition according to claim 1, characterized in that, The path analysis module includes: The trajectory time extraction submodule acquires the frame sequence of regional monitoring images, identifies the timestamps of the first appearance and disappearance frames of each employee in the same area, classifies the timestamps according to the employee identifier, extracts the time interval between adjacent image records and the sequential identifier of the corresponding action number, judges the continuity of the behavior time sequence, and generates a behavior time continuity judgment value. The boundary direction determination submodule extracts the continuous position coordinates of the end of the trajectory based on the behavior time continuity determination value, constructs the direction vector of the end segment of the trajectory, obtains the direction vector of the shortest connection from the end point of the trajectory to the image boundary, obtains the average end offset value of the trajectory segment, calculates the direction consistency determination value, and obtains the boundary direction consistency coefficient. The path connection verification submodule extracts the coordinate values ​​of the trajectory end point and the trajectory start point in the adjacent area based on the boundary direction consistency coefficient, calculates the distance and direction continuity angle between the two coordinate points, determines the path connection relationship between the trajectory start point and end point in the adjacent area, establishes cross-regional behavior sequence labels, and generates cross-regional behavior recognition records.

4. The office scheduling system based on behavior recognition according to claim 3, characterized in that, The behavior aggregation module includes: The time segment comparison submodule obtains the cross-regional behavior recognition records, extracts the start and end times of multiple employees' behavior in each behavior region, calculates the length of the intersection interval of the behavior time periods of each group of employees, and performs a ratio conversion between the length of the intersection interval and the length of the original behavior time period to generate the behavior time overlap ratio value. The action tag filtering submodule extracts the action tags recorded in each behavior region based on the behavior time overlap ratio value, judges the behavior combinations with the same tags, and calculates the behavior consistency fusion coefficient by combining the time overlap ratio value and the tag consistency. The behavior linkage generation submodule determines and identifies behavior combinations with consistent behavior labels in multiple regions based on the behavior consistency fusion coefficient, establishes cross-regional time period intersection combinations, combines behavior synchronization characteristics with behavior path labels, and generates behavior linkage analysis results.

5. The office scheduling system based on behavior recognition according to claim 4, characterized in that, The scheduling and filtering module includes: The resource identifier filtering submodule obtains the behavior linkage analysis results and establishes resource number combinations by identifying and extracting the spatial number and functional number of the resource codes of associated tasks in the scheduling task. The event association comparison submodule, based on the resource number combination, calls the behavior event identifier in the scheduling task and the event number recorded in the task instruction, compares the matching of the behavior event identifier and the event number recorded in the task instruction, filters the resource scheduling set triggered by the same behavior source, and obtains the event source analysis result. The task sorting and adjustment submodule determines the execution priority of tasks and adjusts their execution status based on the analysis results of the event source and the hierarchical order of employee role numbers, thereby generating resource conflict coordination results.

6. The office scheduling system based on behavior recognition according to claim 5, characterized in that, The resource allocation module includes: The tag matching and judgment submodule obtains the action tags and target resource function tags in each task based on the resource conflict coordination results, judges the degree of association matching, and generates a tag content association degree value. The duty path analysis submodule identifies the role category corresponding to the employee identity label based on the correlation value of the label content, compares the task category information, obtains the role task participation characteristics, and extracts the spatial trajectory sequence of the employee in the task to statistically analyze the path coherence and regional distribution to obtain the behavioral coherence performance index. The intent intensity assessment submodule calculates the task execution intent intensity value based on the behavioral coherence performance index, combined with the role task participation characteristics and the degree of correlation of tag content, judges and locks the associated resources, and generates resource pre-locking information.

7. The office scheduling system based on behavior recognition according to claim 1, characterized in that, The system also includes: The preference scheduling module uses the resource pre-locking information to count the frequency of resource calls by employees in multiple time periods, determine the resource sorting index position, construct a priority resource list, adjust the corresponding resource scheduling matching order, and generate resource preference adjustment records. The resource preference adjustment record includes resource frequency statistics, time period priority index, and scheduling order adjustment sequence.

8. The office scheduling system based on behavior recognition according to claim 7, characterized in that, The preference scheduling module includes: The resource frequency statistics submodule obtains the resource pre-locking information, analyzes the resource call sequence of employees in multiple time periods, counts the number of times each resource code is called in multiple time periods, and establishes a resource frequency ranking quantity. The priority resource generation submodule, based on the resource frequency ranking, combines priority resource numbers to form a set according to the resource call characteristics of each time period, constructs the priority resource list for employees in the corresponding time period, and establishes the resource priority set data; The scheduling order adjustment submodule analyzes the correspondence between the time period of the employee's current scheduled task and the resource number of the priority set based on the resource priority set data, adjusts the resource scheduling matching order, and establishes a resource preference adjustment record.

Images