A service area video data driven intelligent management method and system

Abstract

The application provides an intelligent management method and system based on service area video data driving, and belongs to the technical field of data management, and specifically comprises the following steps: when the dynamic line identification matching type of a service area does not belong to a monitoring matching service area, the dynamic adjustment of the monitoring device in the service area is performed according to the dynamic line identification matching type of the service area and the monitoring data of the traffic flow, and the service area is taken as a dynamic line identification service area; according to the dynamic line identification result of the monitoring device of the dynamic line identification service area under different monitoring angles, and in combination with the dynamic line identification result of the monitoring device of the monitoring matching service area, it is determined whether the monitoring device of the dynamic line identification service area needs to be modified, so that the reliability of the verification processing of the dynamic line identification model is improved.

Description

Technical Field

[0001] This invention belongs to the field of data management technology, and in particular relates to an intelligent management method and system driven by service area video data. Background Technology

[0002] With the increasing traffic volume at highway service areas, traditional management models are no longer sufficient to meet the demands for efficient, safe, and intelligent management. Existing service area management systems often suffer from problems such as data silos, slow response times, high reliance on manual intervention, and difficulty in real-time detection of safety hazards. Therefore, there is an urgent need for an intelligent management solution based on video data to achieve comprehensive perception, intelligent decision-making, and efficient execution of service areas.

[0003] To address the aforementioned technical issues, CN202410767079.4, "An AI-Powered Smart Operation System for Service Areas Across All Scenarios," generates virtual maps based on data collected by a data acquisition module and provides a detailed, layered, and categorized display of target behavior states. This enables online, intelligent, and digitalized management of service area safety, event monitoring, and supervision, improving service area management efficiency, service levels, and emergency response capabilities. However, the above technical solution suffers from the following technical problems:

[0004] Vehicle movement recognition often requires combining the recognition results from multiple video devices. Due to differences in traffic flow and monitoring device settings within different service areas, the reliability of movement recognition processing varies. Therefore, determining which service areas require monitoring device optimization and updates based on the reliability of existing movement recognition models in the monitoring devices is a pressing technical problem that needs to be solved. This would improve both the reliability of monitoring processing and the reliability of verification processing of movement recognition models.

[0005] Therefore, there is an urgent need for an intelligent management method and system based on service area video data. Summary of the Invention

[0006] To achieve the objectives of this invention, the following technical solution is adopted:

[0007] Specifically, this application provides an intelligent management method based on service area video data, which includes:

[0008] S1 uses the identification results of the monitoring device to determine the monitoring data of traffic flow in the service area, determines the congestion time periods in different service areas based on the monitoring data, and determines the movement identification matching type of the service area in combination with the monitoring device data in the service area. Based on the movement identification matching type of different service areas, when it is determined that movement identification analysis processing of the monitoring device is required, proceed to the next step.

[0009] S2 When the service area's movement identification matching type does not belong to the monitoring matching service area, the service area where the monitoring device in the service area is dynamically adjusted is determined based on the service area's movement identification matching type and traffic flow monitoring data, and this service area is then designated as the movement identification service area.

[0010] S3 determines whether the monitoring device of the movement recognition service area needs to be modified based on the movement recognition service area monitoring device under different monitoring angles and in combination with the movement recognition results of the monitoring matching service area monitoring device.

[0011] The beneficial effects of this invention are as follows:

[0012] Based on the different movement identification and matching types of service areas, it is determined whether movement identification analysis processing of monitoring devices is required. Thus, when the number of service areas with high movement identification matching is small, dynamic adjustment processing of monitoring devices can be carried out in some service areas, which further improves the efficiency and reliability of the verification processing of the movement identification model. At the same time, it also lays the foundation for the transformation and optimization of monitoring devices in service areas with poor movement identification reliability.

[0013] Based on the movement recognition results of the monitoring devices in the movement recognition service area from different monitoring angles, and the movement recognition results of the monitoring devices in the matching service area, it is determined whether the monitoring devices in the movement recognition service area need to be modified. This takes into account the differences in the reliability of the movement recognition results of the monitoring devices in the movement recognition service area under different monitoring angles, which leads to differences in the reliability of movement recognition processing in the movement recognition service area. Furthermore, by combining the movement recognition results of the monitoring devices in the matching service area, the reliability of the movement recognition model can be evaluated. Thus, from the perspective of the verification processing requirements of the movement recognition model and the reliability of the movement recognition processing of the movement recognition service area itself, it is possible to determine the service areas that need to modify the monitoring devices, thereby improving not only the reliability of its own monitoring processing, but also the reliability of the verification processing of the movement recognition model.

[0014] Furthermore, the traffic flow monitoring data in the service area includes the traffic flow in the service area at different time periods.

[0015] Furthermore, the congestion periods in the service area are determined based on the monitoring data from the monitoring devices in the service area. Specifically, the periods when the vehicle speed in the service area is less than a preset speed threshold are defined as congestion periods.

[0016] Furthermore, the method for determining the service area's movement identification matching type is as follows:

[0017] Based on the identification results of congestion periods in the service area, the distribution data of congestion periods in the service area on different dates is determined, and the total duration of congestion periods on different dates is determined based on the distribution data;

[0018] Using the data from the monitoring devices in the service area, determine the matching monitoring devices for different driving sub-areas of the vehicle within the service area;

[0019] The service area's traffic flow identification matching type is determined by the total duration of congestion periods on different dates and the matching monitoring device of the service area in different driving sub-areas of the vehicle.

[0020] Furthermore, determining whether the monitoring device in the aforementioned movement identification service area needs modification specifically includes:

[0021] Based on the movement recognition results of the monitoring devices in the monitoring and matching service area, determine the movement recognition matching vehicles of the monitoring devices in the monitoring and matching service area. Based on the proportion of movement recognition matching vehicles in different monitoring and matching service areas among all recognized vehicles, determine the recognition matching factor of different monitoring and matching service areas.

[0022] Based on the movement recognition results of the monitoring device in the movement recognition service area under different monitoring angles, determine the recognition matching factor of the monitoring device in the movement recognition service area under different monitoring angles.

[0023] Based on the identification matching factors of different detection matching service areas and the identification matching factors of the monitoring device of the movement identification service area under different monitoring angles, it is determined whether the monitoring device of the movement identification service area needs to be modified.

[0024] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described intelligent management method based on service area video data when running the computer program.

[0025] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0027] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0028] Figure 1 This is a flowchart of an intelligent management method driven by service area video data;

[0029] Figure 2 This is a flowchart illustrating the method for determining the matching type of traffic flow identification in service areas;

[0030] Figure 3 This is a flowchart for determining the movement identification and analysis processes required by the monitoring device;

[0031] Figure 4 This is a flowchart illustrating the method for determining service areas by identifying movement patterns within them. Detailed Implementation

[0032] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0033] Example 1

[0034] like Figure 1 As shown, this application provides an intelligent management method driven by service area video data, specifically including:

[0035] S1 uses the identification results of the monitoring device to determine the monitoring data of traffic flow in the service area, determines the congestion time periods in different service areas based on the monitoring data, and determines the movement identification matching type of the service area in combination with the monitoring device data in the service area. Based on the movement identification matching type of different service areas, when it is determined that movement identification analysis processing of the monitoring device is required, proceed to the next step.

[0036] Furthermore, the traffic flow monitoring data in the service area includes the traffic flow in the service area at different time periods.

[0037] It should be noted that the traffic flow is determined based on monitoring data from the monitoring device in the service area.

[0038] Furthermore, the congestion periods in the service area are determined based on the monitoring data from the monitoring devices in the service area. Specifically, the periods when the vehicle speed in the service area is less than a preset speed threshold are defined as congestion periods.

[0039] Scenario: In the intelligent management system of highways, it is necessary to evaluate the monitoring capabilities of vehicle movement (driving trajectories) in numerous service areas along the route. Traffic congestion frequently occurs within service areas, but the existing surveillance cameras are deployed in inconsistent ways, resulting in blind spots or poor viewing angles in some areas, making it impossible to effectively identify the causes of congestion and specific vehicle behaviors (such as illegal parking or driving against traffic).

[0040] Objective: To develop an automated assessment method that categorizes each service area into "movement identification matching types" based on historical congestion data and the coverage quality of surveillance cameras over the driving area. This classification will guide the upgrading priority of the monitoring system and the applicability of congestion analysis algorithms.

[0041] Monitoring and matching service area: The monitoring system is complete. Type 1 deviation: The monitoring has some defects and needs to be optimized or a small number of monitoring points need to be added. Type 2 deviation: The monitoring has serious defects and engineering transformation needs to be prioritized.

[0042] Driving sub-area: A single grid unit formed by dividing all areas within the service area that vehicles may pass through (lanes, parking lots, ramps, etc.) into grids with a fixed unit area (e.g., 10m×10m).

[0043] Matching monitoring device: refers to a camera whose monitoring screen can completely or partially cover a certain driving sub-area. One camera may cover multiple driving sub-areas.

[0044] Monitoring Matching Factor: This quantifies the monitoring quality of a single camera over a specific driving sub-area it covers. The calculation formula is: (Pixel area of ​​the driving sub-area in the camera's view) / (Actual physical area of ​​the driving sub-area × Calibration coefficient). A higher ratio indicates that the sub-area is clearer and occupies a larger portion of the image, resulting in better monitoring quality.

[0045] Movement path recognition matching type:

[0046] Monitoring and matching service areas: Areas with comprehensive and high-quality monitoring, or areas with severe congestion but acceptable monitoring. Detailed traffic flow analysis can be performed directly.

[0047] Type 1 deviation: The monitoring has obvious defects (blind spots or poor quality), but the overall situation is acceptable and the congestion is not serious.

[0048] Type II deviation: If there are fundamental defects in the monitoring (there are areas that are completely uncovered), the reliability of the movement analysis may be low, and there is a need for engineering modification.

[0049] Specifically, such as Figure 2 As shown, the method for determining the traffic flow identification and matching type in the service area is as follows:

[0050] The service area is divided into N = 120 driving sub-areas.

[0051] This service area possesses average vehicle speed data per minute over the past 30 days, used to identify congestion periods (speed < 5 km / h). It also has calibration parameters for all cameras (position, viewing angle, focal length), allowing for calculation of their geometric relationship with each driving sub-area.

[0052] S11 Based on the identification results of congestion periods in the service area, determine the distribution data of congestion periods in the service area on different dates, and determine the total duration of congestion periods on different dates based on the distribution data;

[0053] To determine the total congestion duration, analyze 30 days of data, identify the congested periods each day, and calculate the total daily congestion duration. Assume the calculation result: the average total congestion duration over 30 days is 1.8 hours / day.

[0054] Logic: Quantify the severity of congestion in service areas to provide a basis for subsequent classification.

[0055] S12 uses the monitoring data in the service area to determine the matching monitoring devices in different driving sub-areas of the vehicle within the service area.

[0056] To determine the matching relationship, the following steps are performed: Based on the camera calibration parameters and the location of each driving sub-region, calculate the "monitoring matching factor" between each camera and each driving sub-region. This forms a matching factor matrix of 15 (cameras) × 120 (sub-regions). Establishing a digital mapping relationship between the monitoring system and the physical space is fundamental to evaluating coverage quality.

[0057] S13 uses the total duration of congestion periods on different dates and the matching monitoring device of the service area in different driving sub-areas of the vehicle to determine the traffic flow identification matching type of the service area.

[0058] Specifically, the driving sub-region is obtained by dividing the area into units of area based on the possible driving areas of vehicles in the service area.

[0059] It should be noted that the matching monitoring device is a monitoring device capable of monitoring the driving sub-region.

[0060] It is understood that the service area's traffic flow identification matching type is determined by the total duration of congestion periods on different dates and the matching monitoring device for different driving sub-areas of the service area. Specifically, this includes:

[0061] S131 Based on the matching monitoring device of the service area in different driving sub-areas of the vehicle, determine whether all driving sub-areas can be monitored simultaneously. If yes, proceed to the next step. If no, determine that the movement identification matching type of the service area is a type II deviation.

[0062] Determine if all driving sub-regions are covered by at least one camera (monitoring matching factor > 0). Execute: Check each column of the matching factor matrix (representing a sub-region). It is found that all 120 sub-regions have a non-zero matching factor in at least one camera view (i.e., they have all been captured). Determine: Yes, all regions can be monitored. Proceed to S132.

[0063] Logical (if "No"): If any sub-region has all matching factors equal to 0 (blind zone), it is immediately classified as "Type II deviation". Because there is a completely invisible area, the reliability of the movement analysis is poor, and there is a need to modify the monitoring device.

[0064] S132 determines the monitoring matching factor of the driving sub-region based on the proportion of the image area of ​​different driving sub-regions in the matching monitoring device, and determines whether the monitoring matching factor of all driving sub-regions is greater than the preset matching factor threshold. If so, the movement recognition matching type of the service area is determined to be the monitoring matching service area. If not, proceed to the next step.

[0065] To determine if the monitoring quality of all sub-areas is "reliable" (matching factor > 0.6), perform the following: Check the matching factor matrix to see if each sub-area has at least one camera with a monitoring matching factor > 0.6. Suppose that 110 sub-areas meet the condition, but 10 sub-areas (such as some corners or remote parking spaces) have a maximum matching factor ≤ 0.6 across all cameras (too small or too off-center in the image).

[0066] Judgment: No, there are areas with poor monitoring quality. Proceed to S133.

[0067] Logic (if "yes"): If the monitoring quality of all areas is high, it is directly determined as "monitoring matching service area" without any further steps.

[0068] S133 defines the driving sub-regions with monitoring matching factors greater than the preset matching factor threshold as reliable monitoring sub-regions, and determines whether the proportion of the number of reliable monitoring sub-regions in the driving sub-regions is greater than the preset threshold. If yes, proceed to the next step; otherwise, determine that the movement identification matching type of the service area is a type of deviation.

[0069] In the above steps, determine whether the proportion of "reliable monitoring sub-regions" meets the standard (>85%). Calculate: Number of reliable monitoring sub-regions = 110. Total number of sub-regions = 120. Proportion = 110 / 120 ≈ 91.7%.

[0070] Judgment: 91.7% > 85%, condition is met. Proceed to S134.

[0071] Logic (if "No"): If the percentage of reliable monitoring is too low (≤85%), it indicates that there are too many areas with poor monitoring quality and the overall system performance is poor, and it is directly judged as "Type I deviation".

[0072] S134 determines the traffic flow identification matching type of the service area based on the total duration of congestion periods on different dates.

[0073] The final decision is based on the duration of congestion. It is known that the average daily congestion duration is 1.8 hours.

[0074] Judgment: 1.8 hours ≤ preset duration threshold (2 hours). Condition is met (average value is "not greater than" the preset threshold).

[0075] Final decision: Since the daily congestion is not severe (≤2 hours), but there are a few poor quality areas (10) in the monitoring, the system determines that the movement identification matching type of "a service area in a certain highway" is "a type of deviation".

[0076] (Comparison of decision-making scenarios):

[0077] If the average daily congestion duration is 2.5 hours (>2 hours): Even if the monitoring has the same defects, the congestion problem is more prominent. At this time, even if there is a deviation, the vehicles in the congested state move at a slower speed in the traffic flow, and there is enough time and data for traffic flow identification processing, and it is determined to be a "monitoring matching service area".

[0078] It should be noted that if the average total duration of congestion periods on different dates is greater than a preset duration threshold, the service area's movement identification matching type is determined to be a monitoring matching service area; if the average total duration of congestion periods on different dates is not greater than the preset duration threshold, the service area's movement identification matching type is determined to be a deviation type.

[0079] It should be noted that the first type of deviation is smaller than the second type of deviation.

[0080] Scenario: A provincial highway administration has completed the assessment and classification of "traffic path identification and matching type" for 120 key service areas within its jurisdiction (based on the method of the previous embodiment). The classification results are as follows:

[0081] There are 30 types of second-class deviations, 70 types of first-class deviations, and 20 monitoring and matching service areas.

[0082] The core challenge is that management plans to introduce an advanced "video movement recognition and analysis model" (such as deep learning-based vehicle tracking, trajectory clustering, and behavior recognition algorithms) to deeply analyze service area operational patterns and diagnose the root causes of congestion. However, the training, deployment, and optimization of this model require a large amount of high-quality, labeled surveillance video data as a foundation.

[0083] Decision Objective: To determine whether the existing service area monitoring matching level meets the requirements when launching a special "movement line recognition analysis and processing" task targeting video data from monitoring devices across the entire road network. In case of poor monitoring matching, the angles of monitoring devices in certain service areas will be dynamically adjusted. This will determine the reliability of the movement line recognition processing of the monitoring devices, providing data support for future upgrades to the monitoring devices, while also ensuring the reliability of the movement line recognition model's recognition and processing.

[0084] Line recognition and analysis processing: This refers to a specific data processing task. Its core action is to retrieve historical or real-time monitoring video streams of eligible service areas (monitoring matching service areas), and use basic algorithms or manual assistance to extract trajectories and label behaviors of vehicles in the videos to form a structured "vehicle movement line dataset." This is a prerequisite for training advanced AI models.

[0085] Monitoring and matching service area: This refers to a service area that has been identified as a "monitoring and matching service area" for movement recognition matching through prior evaluation. The monitoring data from this type of service area is of high quality and is a "high-quality data source" for producing reliable training data.

[0086] Preset threshold for the number of matching service areas: set to 25. This is the "data source base threshold" for starting a global data collection and analysis task.

[0087] Specifically, such as Figure 3 As shown, it is determined that the movement identification and analysis of the monitoring device needs to be performed, specifically including:

[0088] Based on the different service area movement identification and matching types, the monitoring and matching service areas are determined;

[0089] Based on the monitoring and matching service area data, determine whether it is necessary to perform movement identification and analysis processing of the monitoring device.

[0090] It should be noted that if the number of monitored matching service areas is less than the preset threshold for the number of matched service areas, it is difficult to effectively determine the reliability of the movement recognition model by relying solely on the monitoring matching service areas. Therefore, it is necessary to perform movement recognition analysis processing on the monitoring device.

[0091] In one possible specific embodiment:

[0092] Step 1: Data aggregation and classification. Execution process: The system aggregates the movement identification and matching type evaluation results of all 120 service areas. This is the basic input for decision-making, clarifying which are the ideal "data farms" (monitoring and matching service areas).

[0093] Step 2: Determine the set of "Monitoring Matching Service Areas". Execution process: Filter records of type "Monitoring Matching Service Areas" from the summary table. Result: Obtain the set S_matched = {Service Area A, Service Area B, ..., Service Area O}, with a total of 20 service areas.

[0094] Step 3: Decision-making – whether to start the movement identification and analysis process. Execution process: compare the number of “monitored matching service areas” (20) with the preset matching service area number threshold (25).

[0095] Judgment: 20 < 25, condition is met (quantity is less than the threshold), decision output: determine that the movement line identification and analysis of the monitoring device is required.

[0096] S2 When the movement identification matching type of the service area does not belong to the monitored matching service area, the movement identification service area in the service area is determined based on the movement identification matching type of the service area and the monitoring data of traffic flow.

[0097] "Traffic path recognition service area" specifically refers to service areas where traffic path recognition and analysis are performed by dynamically adjusting the camera's monitoring angle during the monitoring process. This means that the monitoring network of this type of service area has the characteristics of being "proactive and adjustable," and can focus on different driving sub-areas through means such as pan-tilt-zoom (PTZ) control to achieve higher quality traffic path recognition.

[0098] Scenario: A smart highway project plans to deploy an "active monitoring and traffic flow recognition system" in some service areas. This system not only includes a traffic flow recognition AI model, but more importantly, it requires the service area's monitoring network to have remotely controllable and angle-adjustable hardware capabilities (such as PTZ cameras, rotatable high-speed dome cameras, etc.) to achieve dynamic focused scanning of key areas.

[0099] Decision Objective: To select suitable "movement identification service areas" for deploying this proactive monitoring system from the entire network. The selection process must consider the following:

[0100] Business monitoring reliability (whether there is severe congestion requires dynamic monitoring and analysis).

[0101] Basic adaptability of the monitoring network (whether the existing monitoring system has adjustable capabilities or potential for modification).

[0102] Feasibility of project implementation (the pilot scope should not be too large to ensure that resources are focused).

[0103] Service areas with "Type II Deviation" have the lowest potential for improvement and are least likely to be "actively controllable" due to serious hardware defects in their monitoring (such as many blind spots in fixed viewpoints). They are the priority targets for basic monitoring upgrades and are therefore given priority as service areas for movement identification.

[0104] "Type I deviation" and "monitoring matching service area" are potential candidates. Among them, "Type I deviation" may have poor recognition results due to fixed monitoring angle or incomplete coverage. Since the impact is small, it needs to be optimized by "angle adjustment". Therefore, it is not the main candidate for "movement recognition service area".

[0105] Given: There are a total of 120 service areas in the entire network. Distribution of movement identification and matching types: 30 types of Class II deviation, 70 types of Class I deviation, and 20 service areas monitored and matched.

[0106] Specifically, such as Figure 4 As shown, the method for determining the service area based on the movement pattern is as follows:

[0107] S31 determines the number of service areas with different movement identification matching types based on the movement identification matching type of the service area;

[0108] S32 determines the traffic flow of the service area at different time periods based on the monitoring data of the traffic flow of the service area;

[0109] S33 determines the traffic flow identification matching type of the service area, and combines the number of service areas with different traffic flow identification matching types and the traffic flow of the service area in different time periods to determine whether the service area is a traffic flow identification service area.

[0110] It should be noted that if the service area's movement identification matching type is a type II deviation, then the service area is determined to be a movement identification service area.

[0111] (For Type II deviation): Judgment: If the service area is "Type II deviation".

[0112] Decision: The service area is determined to be a "traffic flow identification service area".

[0113] Reason: The monitoring hardware infrastructure in this type of service area is too poor, so optimization and control are needed. This would allow for targeted dynamic adjustment of the angle of the monitoring devices in the service area. However, the basic conditions for deploying an active and adjustable system are not met, and this should be classified as a basic renovation project.

[0114] Furthermore, if the service area's movement identification matching type does not belong to the second type of deviation, the following steps are included:

[0115] S331 obtains the number of service areas, and determines whether the number of service areas is greater than a preset service area number threshold. If yes, it is determined that the service area does not belong to the traffic flow recognition service area; otherwise, it proceeds to the next step.

[0116] To determine if the size of the candidate pool is too large, if the number of candidate service areas (90) is greater than the preset service area number threshold (assumed to be 40), then yes.

[0117] Decision: Determine that the service area currently being assessed does not belong to the traffic flow identification service area.

[0118] Reason: When the candidate pool is too large, the reliability of the movement recognition model is high, so it can be directly determined that all service areas do not belong to the movement recognition service area.

[0119] (Assuming the scenario changes: a simulation will be conducted)

[0120] If the number of candidate service areas is ≤ 40, proceed to S332.

[0121] S332 determines the movement recognition matching weight coefficient of different service areas based on the movement recognition matching type of different service areas, and judges whether the sum of the movement recognition matching weight coefficients of different service areas is greater than the preset weight coefficient threshold. If so, it is determined that the service area does not belong to the movement recognition service area; otherwise, it proceeds to the next step.

[0122] S333 determines whether the sum of the movement recognition matching weight coefficients of different service areas is less than a preset coefficient threshold (less than a preset weight coefficient threshold). If yes, the service area is determined to belong to the movement recognition service area; otherwise, proceed to the next step.

[0123] S332 & S333: Evaluate the overall monitoring basis of the candidate pool (used to decide whether to "select all" or "select none")

[0124] Weight settings: Weight of deviation type = 1 (deviation in monitoring reliability), weight of monitoring matching service area = 2 (already good, high monitoring reliability).

[0125] Calculate the total weight sum. Scenario A: Total weight sum > High threshold: This indicates that the existing monitoring devices have a high degree of monitoring matching, and the need for adjustment is not urgent. The entire candidate pool can be set as not belonging to the "movement recognition service area".

[0126] Scenario B: Total weights < low threshold: This indicates that the monitoring reliability in the candidate pool is not high and there is a strong need for proactive adjustment. Therefore, it is necessary to proceed to the next step to determine whether it belongs to the movement recognition service area.

[0127] Scenario C: The total weight is in the middle range: an overall decision cannot be made, proceed to the next step of detailed screening.

[0128] S334 identifies service areas whose movement identification matching type does not belong to the second type of deviation and do not belong to the monitoring matching service area as other service areas, and determines whether the proportion of the number of the other service areas in the service area is less than the preset proportion threshold. If so, it is determined that the other service areas do not belong to the movement identification service area. If not, proceed to the next step.

[0129] Determine whether the candidate pool belongs to the "minority". If the proportion of candidate service area (70) to total number (120) is greater than the preset quantity proportion threshold (assuming it is 20%), proceed to S335 (because the candidate pool is the mainstream and further refined screening is required).

[0130] Reason: If the number of non-Type II deviation service areas is small, then adjusting them will have a small impact on the reliability of the overall movement recognition model. Therefore, it can be determined that all other service areas do not belong to the movement recognition service areas.

[0131] S335 determines the congestion periods on different dates based on the traffic flow of the other service areas at different times, and determines whether the other service areas are traffic flow identification service areas based on the distribution data of the congestion periods on different dates.

[0132] S335: Final screening based on congestion conditions (refined screening). Execution: For each candidate service area, analyze its historical traffic flow data to determine whether it experiences congestion periods on different dates.

[0133] Decision-making rules:

[0134] If "yes" (continuous congestion): designate this service area as a "traffic identification service area". This is because it has the potential for monitoring adjustments (non-Type II bias), and due to the numerous periods of congestion, its monitoring reliability is also relatively high.

[0135] If "No" (not frequently congested): This service area is not a "traffic flow identification service area".

[0136] Furthermore, if the other service areas experience congestion periods on different dates, then the other service areas are identified as traffic flow identification service areas.

[0137] S3 determines whether the monitoring device of the movement recognition service area needs to be modified based on the movement recognition service area monitoring device under different monitoring angles and in combination with the movement recognition results of the monitoring matching service area monitoring device.

[0138] A provincial expressway group has completed the following tasks:

[0139] Service area classification: Monitoring and matching service areas (M category): 20 (as core verification benchmarks), movement recognition service areas (R category): 50 (targets that need to be evaluated for renovation).

[0140] Full-domain verification phase: Deployment and trial operation of the traffic flow recognition model were completed in all 70 service areas to obtain recognition matching factor data for each service area.

[0141] The fundamental logic of this method is that whether or not to monitor and modify the "movement recognition service area (R type)" depends entirely on the efficiency requirements of the current verification work.

[0142] When the verification reliability is high:

[0143] Meaning: The model validation data obtained in the M-class service area has good consistency and high confidence, and can quickly and accurately evaluate the model performance.

[0144] Current situation: The existing verification system is efficient and can support the iterative optimization of the model. Therefore, it is only necessary to determine how to modify it after the model optimization is completed.

[0145] Decision: No monitoring upgrades will be implemented for the time being. The existing verification environment is already good enough, and the marginal benefit of upgrading is low.

[0146] When the verification reliability is poor:

[0147] Meaning: The validation data obtained in the M-class service area fluctuates greatly and has low confidence, making model evaluation time-consuming, labor-intensive, and the conclusions uncertain.

[0148] Current situation: Validation work has reached a bottleneck, and the model optimization process is hindered.

[0149] Decision: Monitoring system upgrades are necessary. The core objective of these upgrades is to improve verification processing efficiency—by improving monitoring quality, obtaining clearer and more stable video data, thereby accelerating annotation speed, improving annotation consistency, and shortening the model evaluation cycle.

[0150] Key Insight: The transformation is not aimed at directly improving the recognition rate in business scenarios, but rather at breaking through the bottleneck in model development. When verification efficiency becomes the main obstacle to the evolution of AI systems, investment in monitoring hardware becomes necessary and urgent.

[0151] Identification matching factor (service area level) is defined as the percentage of vehicles correctly identified by the traffic flow recognition AI model within a service area, relative to the total number of vehicles actually passing through that service area.

[0152] Calculation formula: Identification matching factor = Number of correctly identified vehicles / Total number of vehicles actually passing through.

[0153] Meaning: This factor directly quantifies the overall recognition accuracy of the model in the overall environment of the service area. The higher the value, the more accurate the recognition.

[0154] Identification matching factor (camera-angle level), defined as: the percentage of vehicles correctly identified by a single surveillance camera at a specific preset monitoring angle out of the actual number of vehicles passing by at that angle.

[0155] Significance: Used to evaluate the instantaneous performance and reliability of specific hardware devices under specific working conditions, it is a key indicator for locating micro-level problems.

[0156] Monitoring devices that are difficult to control: refers to monitoring devices in which the identification matching factor of a single camera can exceed a high threshold (such as 0.70) under multiple (≥2) preset monitoring angles.

[0157] Significance: These cameras have excellent performance and are well-positioned for installation, enabling them to identify movement patterns from multiple angles. However, this requires frequent adjustments to the camera angle, resulting in poor reliability of the monitoring device.

[0158] Specifically, determining whether the monitoring device in the aforementioned traffic flow identification service area needs to be upgraded includes:

[0159] S41 uses the movement recognition results of the monitoring device in the monitoring and matching service area to determine the movement recognition matching vehicle in the monitoring and matching service area, and uses the proportion of movement recognition matching vehicles in different monitoring and matching service areas to determine the identification matching factor of different monitoring and matching service areas among all identified vehicles.

[0160] Calculate the matching factor for the control group to determine the reliability of the validation process for the current model.

[0161] Glossary: ​​The "Monitoring Matching Service Area" serves as a control group in this step, and its performance represents the best model performance that ideal monitoring hardware can support under current technological conditions.

[0162] Execution and Results: Calculate the identification matching factors for 20 monitored matching service areas. Assume the results are: 0.92, 0.89, 0.94, 0.86, 0.91, 0.93, 0.87, 0.90, 0.88, 0.92, 0.85, 0.83, 0.90, 0.87, 0.89, 0.91, 0.84, 0.92, 0.86, 0.88.

[0163] S42 determines the identification matching factor of the monitoring device of the movement recognition service area under different monitoring angles based on the movement recognition results of the monitoring device of the movement recognition service area under different monitoring angles;

[0164] Creating a "problem map"—calculating fine-grained identification matching factors for the target region

[0165] Definition: The "Traffic Flow Recognition Service Area" is the evaluation target of this step. Camera-angle level analysis is performed to create a microscopic performance heatmap.

[0166] Execution and Results: For 50 target service areas, the recognition matching factor of each camera at each preset angle was calculated. For example, it was found that the factors of camera CAM-003 in service area R5 were 0.35, 0.28, and 0.39 at the three angles.

[0167] Significance of the steps:

[0168] Precise problem localization: Break down the macro-level service area performance problem and locate it to specific hardware devices (which camera) and specific operating conditions (which angle).

[0169] Laying the foundation for precise transformation: providing direct data to determine "where needs to be transformed and what needs to be transformed".

[0170] S43 determines whether the monitoring device of the movement recognition service area needs to be modified based on the identification matching factors of different detection matching service areas and the identification matching factors of the monitoring device of the movement recognition service area under different monitoring angles.

[0171] Furthermore, the identification matching factor of the monitoring device under different monitoring angles is determined based on the proportion of the vehicle identified by the monitoring device under the monitoring angle among all identified vehicles under the monitoring angle.

[0172] Furthermore, based on the identification matching factors of different detection matching service areas and the identification matching factors of the monitoring devices in the movement recognition service areas under different monitoring angles, it is determined whether the monitoring devices in the movement recognition service areas need to be modified, specifically including:

[0173] S431 obtains the movement recognition time in different detection matching service areas, and determines whether the movement recognition time is greater than the preset recognition time threshold. If yes, proceed to the next step. If no, the recognition reliability of the movement recognition model cannot be determined at this time, so there is no need to modify the monitoring device for the time being.

[0174] Data sufficiency check – Identify if the duration exceeds 90 days. Judgment logic: The trial run duration is 95 days, exceeding the 90-day threshold. Decision and flow: Data is sufficient; proceed to S432.

[0175] Significance of this step: This is the first line of defense for scientific decision-making. It ensures that all subsequent analyses are based on long-term, stable, and statistically significant data, avoiding erroneous judgments due to short-term fluctuations or unforeseen events.

[0176] S432 determines whether there is a detection matching service area where the identification matching factor is less than the preset matching factor threshold. If so, it is determined that the movement recognition service area needs to be modified by the monitoring device to improve the verification processing efficiency of the recognition reliability of the movement recognition model. If not, proceed to the next step.

[0177] Verify reliability screening – Does the monitoring matching service area factor <0.85 exist? Judgment logic: Check the S41 results and find that factors 0.83 and 0.84 exist, both less than 0.85.

[0178] Decision: The conditions are met, and it is determined that the monitoring device needs to be upgraded.

[0179] Root cause diagnosis: This step revealed that even in the most ideal "gold standard" environment, the model performance did not consistently reach high standards (>0.85). This suggests that the problem may be systemic: or that the current model performance is biased and difficult to validate effectively.

[0180] The decision to "require upgrades" at this point aims to improve the overall reliability of verification processing across all motion recognition service areas.

[0181] Efficiency Goal: Through systematic transformation, improve the quality of network-wide monitoring, thereby obtaining more stable and consistent verification data, and ultimately improve the efficiency and reliability of the verification process for the entire model's recognition reliability.

[0182] S433 determines the identification matching coefficient based on the identification matching factors of different detection matching service areas, and judges whether the identification matching coefficient is greater than the preset matching coefficient threshold. If it is, no modification of the monitoring device is required; otherwise, proceed to the next step.

[0183] If all monitored service area factors are ≥0.85, proceed to S433, overall benchmark assessment—calculate and determine the matching coefficient.

[0184] Definition: The identification matching coefficient here refers to the arithmetic mean of the identification matching factors of 20 monitoring and matching service areas, representing the overall performance level of the control group.

[0185] Judgment logic: The calculated average is approximately 0.885. Determine if 0.885 is greater than C_th = 0.88? If the condition is not met (or not significantly met), the decision and flow is: proceed to S434.

[0186] Assess the robustness of the validation baseline: if the coefficients are close to but do not significantly exceed the high threshold (0.88), the validation baseline is "good". If the baseline is strong enough (coefficients > 0.88), the model is excellent and the validation reliability is high. Therefore, no modification is needed for the time being.

[0187] S434 uses the identification matching factor of the monitoring device in the movement recognition service area under different monitoring angles to determine whether there are monitoring devices with identification matching factors less than the preset value of the matching factor under different monitoring angles. If so, it is determined that the movement recognition service area needs to be modified by the monitoring device to improve the verification processing efficiency of the recognition reliability of the movement recognition model. If not, proceed to the next step.

[0188] Critical Defect Emergency Location – Existence of a Failed Camera at All Angles: Judgment Logic: Check fine-grained data from S42. It was found that, for example, the CAM-003 camera in service area R5 had a factor <0.40 at all angles.

[0189] Decision: The conditions are met, and it is determined that the monitoring device needs to be upgraded.

[0190] Significance of the steps: Identifying the "absolute weak link": These cameras represent a clear hardware failure or a serious misselection, acting as a "dead node" in the surveillance network. Emergency priority upgrade: Regardless of the overall strategy, these devices must be replaced immediately, as their presence renders any algorithm optimizations in their service area meaningless.

[0191] (Scenario simulation): If no such camera is found, proceed to the final evaluation stage S435.

[0192] S435 determines the difficult-to-control monitoring devices in the monitoring device based on the identification matching factors under different monitoring angles, and determines whether the monitoring devices in the service area need to be modified based on the data of the difficult-to-control monitoring devices in the service area.

[0193] It is understood that the control-difficult monitoring device is a monitoring device with multiple identification matching factors greater than a preset identification factor threshold for monitoring angles.

[0194] Specifically, when the proportion of difficult-to-control monitoring devices in the service area exceeds a preset threshold for the proportion of difficult-to-control monitoring devices, it is determined that the monitoring devices in the traffic flow identification service area need to be modified.

[0195] Network Health and Resilience Assessment – ​​Based on the Proportion of “Control Difficult Monitoring Devices”

[0196] Judgment Logic: Recognition: Among the 10 cameras in the traffic flow recognition service area, 5 "control difficult monitoring devices" (i.e., factor > 0.70 at ≥2 angles) are identified. Calculation ratio: 5 / 10 = 50%. Judgment: 50% > R_th = 30%, the condition is met.

[0197] Decision: It was determined that the monitoring device needed to be upgraded.

[0198] Significance of the steps (deeper logic):

[0199] Network structure assessment: "Difficult-to-control monitoring device" indicates that it may not be possible to reliably monitor all vehicles from a single angle, thus requiring dynamic adjustment of the angle, which precisely illustrates the poor coverage area of ​​the monitoring device.

[0200] The contradiction revealed by the high proportion: once the number of difficult-to-control monitoring devices is large, it is inevitable that the angle needs to be adjusted frequently in order to verify the model, which will affect the reliability and efficiency of the model verification to a certain extent. Therefore, it is necessary to carry out modification to make the installation density of monitoring devices higher.

[0201] Example 2

[0202] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-described intelligent management method based on service area video data when running the computer program.

[0203] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0204] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0205] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. An intelligent management method based on service area video data, characterized in that, Specifically, it includes: Based on the identification results of the monitoring device, the monitoring data of traffic flow in the service area is determined. Based on the monitoring data, the congestion time periods in different service areas are determined. Combined with the data from the monitoring device in the service area, the movement identification matching type of the service area is determined. Based on the movement identification matching type of different service areas, when it is determined that movement identification analysis processing of the monitoring device is required, proceed to the next step. When the service area's traffic flow identification matching type does not belong to the monitoring matching service area, the service area where the monitoring device in the service area is dynamically adjusted is determined based on the service area's traffic flow identification matching type and traffic flow monitoring data, and this service area is then used as the traffic flow identification service area. Based on the movement recognition results of the monitoring device in the movement recognition service area under different monitoring angles, and in combination with the movement recognition results of the monitoring device in the monitoring matching service area, it is determined whether the monitoring device in the movement recognition service area needs to be modified. The method for determining the traffic flow identification and matching type of the service area is as follows: Based on the identification results of congestion periods in the service area, the distribution data of congestion periods in the service area on different dates is determined, and the total duration of congestion periods on different dates is determined based on the distribution data; Using the data from the monitoring devices in the service area, determine the matching monitoring devices for different driving sub-areas of the vehicle within the service area; The service area's traffic flow identification matching type is determined by the total duration of congestion periods on different dates and the matching monitoring device for different driving sub-areas of the service area. The movement path identification and matching types include monitoring and matching service areas, type I deviation, and type II deviation. Determining whether the monitoring device in the aforementioned traffic flow identification service area needs to be upgraded includes: Based on the movement recognition results of the monitoring devices in the monitoring and matching service area, determine the movement recognition matching vehicles of the monitoring devices in the monitoring and matching service area. Based on the proportion of movement recognition matching vehicles in different monitoring and matching service areas among all recognized vehicles, determine the recognition matching factor of different monitoring and matching service areas. Based on the movement recognition results of the monitoring device in the movement recognition service area under different monitoring angles, determine the recognition matching factor of the monitoring device in the movement recognition service area under different monitoring angles. Based on the identification matching factors of different detection matching service areas and the identification matching factors of the monitoring device of the movement identification service area under different monitoring angles, it is determined whether the monitoring device of the movement identification service area needs to be modified.

2. The intelligent management method based on service area video data as described in claim 1, characterized in that, The monitoring data on traffic flow in the service area includes the traffic flow in the service area at different time periods.

3. The intelligent management method based on service area video data as described in claim 2, characterized in that, The traffic flow is determined based on monitoring data from the monitoring device in the service area.

4. The intelligent management method based on service area video data as described in claim 1, characterized in that, The congestion periods in the service area are determined based on monitoring data from the monitoring devices in the service area.

5. The intelligent management method based on service area video data as described in claim 1, characterized in that, The driving sub-regions are obtained by dividing the service area according to the area where vehicles may travel.

6. The intelligent management method based on service area video data as described in claim 1, characterized in that, The required motion identification and analysis processing for the monitoring device has been determined, specifically including: Based on the different service area movement identification and matching types, the monitoring and matching service areas are determined; Based on the monitoring and matching service area data, determine whether it is necessary to perform movement identification and analysis processing of the monitoring device.

7. The intelligent management method based on service area video data as described in claim 6, characterized in that, If the number of monitored matching service areas is less than the preset threshold for the number of matching service areas, it is determined that the movement identification and analysis of the monitoring device is required.

8. A computer system, comprising: A memory and processor connected by communication, and a computer program stored in the memory and capable of running on the processor, characterized in that, when the processor runs the computer program, it executes an intelligent management method based on service area video data as described in any one of claims 1-7.

Images