A method for controlling pedestrian and vehicle passing at a zebra crossing

Abstract

The application provides a zebra crossing pedestrian and vehicle passing control method, which comprises the following steps: an image acquisition device collects video stream data, wherein the video stream data contains pedestrian and vehicle image information in a crosswalk area; pedestrian and vehicle detection and tracking are performed on the video stream data to obtain the number of pedestrians, age structure, moving intention and vehicle queue length; the number of pedestrians staying per unit time, moving speed and vehicle queue length are calculated based on the pedestrian and vehicle detection and tracking results to form pedestrian and vehicle flow data; a dynamic signal timing scheme is generated by processing the pedestrian and vehicle flow data based on a double-layer control model, and the dynamic signal timing scheme is issued to a signal control device through an edge computing node.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for controlling pedestrian and vehicle traffic at zebra crossings. Background Technology

[0002] In the field of urban traffic management, ensuring the efficiency and safety of traffic at intersections and pedestrian crossings is of paramount importance. This area directly relates to the smoothness of citizens' daily travel and the safety of their lives and property, especially in densely populated urban environments where the proper control of traffic signals is key to alleviating congestion and reducing accidents. However, with the acceleration of urbanization and the increasing complexity of traffic scenarios, traditional management methods are gradually revealing their shortcomings, urgently requiring technological innovation to address new challenges.

[0003] Currently, video surveillance-based traffic control methods face significant limitations in practical applications. Many existing solutions can identify pedestrians and vehicles well under ideal conditions, but the quality of video data deteriorates drastically when encountering low light conditions at night, reduced visibility due to rain or snow, or equipment vibrations caused by wind. In such situations, the system often fails to accurately capture targets, or even misjudgments or omissions occur, leading to a lack of basis for traffic signal adjustments and consequently affecting traffic efficiency and safety.

[0004] A deeper technical challenge lies in ensuring the stability and clarity of video capture under complex environments and adverse conditions. This core factor directly determines the accuracy of subsequent target recognition. If the video image is blurry due to insufficient light or out of focus due to equipment shake, even the most advanced recognition technology will struggle to accurately determine the number of pedestrians and the specific details of vehicle queues. For example, at pedestrian crossings at night, pedestrians may be overlooked by the system due to dim lighting, causing traffic lights to fail to switch in time and resulting in long waits for pedestrians. In rainy or snowy weather, the length of vehicle queues may be misestimated due to unclear images, leading to unreasonable signal timing and exacerbating traffic congestion. This series of problems demonstrates that the stability of video capture quality has become a bottleneck affecting the overall effectiveness of traffic control.

[0005] Therefore, ensuring stable and clear video footage under adverse conditions such as nighttime, rain, and snow, as well as in complex traffic scenarios, and accurately identifying the dynamic information of pedestrians and vehicles based on this footage, has become a key issue in improving the rationality of traffic signal control and traffic safety. Summary of the Invention

[0006] This invention provides a method for controlling pedestrian and vehicle traffic at zebra crossings, mainly including:

[0007] Image acquisition equipment acquires video stream data, which includes image information of pedestrians and vehicles in the crosswalk area. Pedestrian and vehicle detection and tracking are performed on the video stream data to obtain the number of pedestrians, age structure, movement intentions, and vehicle queue lengths. Based on the pedestrian and vehicle detection and tracking results, the number of pedestrians remaining, their movement speed, and the vehicle queue lengths per unit time are calculated to form pedestrian and vehicle flow data. A dynamic signal timing scheme is generated by processing the pedestrian and vehicle flow data based on a two-layer control model, and the dynamic signal timing scheme is distributed to the signal control equipment through an edge computing node. Further, the pedestrian and vehicle detection and tracking of the video stream data includes: a target detection module performing target segmentation on the crosswalk area in the video stream data to identify pedestrian and vehicle targets; and a tracking module generating trajectory information based on the pedestrian and vehicle targets to determine the number of pedestrians, age structure, movement intentions, and vehicle queue lengths. Furthermore, the calculation of the number of pedestrians remaining, their moving speed, and the length of vehicle queues per unit time based on the pedestrian and vehicle detection and tracking results includes: a pixel-level segmentation module performing segmentation processing on the video stream data to obtain pedestrian remaining areas; an optical flow calculation module determining the moving speed based on the pedestrian remaining areas, and statistically analyzing the number of pedestrians remaining and the length of vehicle queues per unit time to form the pedestrian and vehicle traffic data. Furthermore, the generation of a dynamic signal timing scheme based on the pedestrian and vehicle traffic data using a two-layer control model includes: a rule engine determining whether the pedestrian waiting time in the pedestrian and vehicle traffic data exceeds a preset threshold; if it does, a green light extension command is triggered; a timing prediction network receiving the rule engine output and the pedestrian and vehicle traffic data, predicting the signal duration, and generating the dynamic signal timing scheme; and the two-layer control model fusing the green light extension command and the signal duration to determine the final timing scheme. Furthermore, the target detection module performs target segmentation on the pedestrian crossing area in the video stream data, including: a feature extraction layer optimizing and extracting pedestrian features for small target pedestrians; a target classification layer distinguishing pedestrian and vehicle targets based on the pedestrian and vehicle features; and a tracking module combining the pedestrian and vehicle targets to eliminate occlusion interference and generate trajectory information. Furthermore, the edge computing node sends the dynamic signal timing scheme to the signal control device, including: the edge computing node receiving the video stream data and performing local pedestrian and vehicle detection; and the edge computing node transmitting control commands to the signal control device through a communication interface according to the dynamic signal timing scheme. Furthermore, the image acquisition device acquires video stream data, including: the image acquisition device integrating a fill light module to acquire nighttime video stream data; and the image acquisition device integrating an image stabilization module to acquire rain and snow scene video stream data, forming the video stream data.Furthermore, the step of calculating the number of pedestrians staying, their movement speed, and the length of vehicle queues per unit time based on the pedestrian and vehicle detection and tracking results includes: the state filtering module receiving the instantaneous flow rate output by the optical flow calculation module and correcting it to form the pedestrian and vehicle flow data; the pedestrian and vehicle flow data is fused with meteorological information and historical flow data.

[0008] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0009] This invention discloses a method for controlling pedestrian and vehicle traffic at zebra crossings. It addresses the problem that video acquisition at urban intersections or pedestrian crossings is susceptible to interference from insufficient light, equipment vibration, and target occlusion in adverse weather conditions such as nighttime, rain, and snow, as well as complex traffic scenarios. This leads to inaccurate pedestrian and vehicle detection, consequently affecting the rationality of signal timing. This invention utilizes an image acquisition device integrating a supplementary lighting module and an image stabilization module to ensure stable and clear video stream data at night and in rainy or snowy weather. Simultaneously, it employs a feature extraction layer optimized for small targets and multi-target tracking technology to accurately identify the number of pedestrians, their age structure, movement intentions, and vehicle queue lengths. Furthermore, it integrates meteorological and historical data to form reliable pedestrian and vehicle traffic flow data. Based on this, this invention employs a two-layer control model. A rule engine responds in real-time to situations where pedestrian waiting times exceed a threshold, triggering a green light extension. A time-series prediction network predicts traffic trends, and a dynamic signal timing scheme is generated. This is then processed and distributed locally through edge computing nodes, ultimately achieving dynamic optimization and adjustment of traffic signals, effectively improving the efficiency and safety of pedestrian and vehicle traffic. Attached Figure Description

[0010] Figure 1 This is a flowchart of a zebra crossing pedestrian and vehicle traffic control method according to the present invention.

[0011] Figure 2 This is a schematic diagram of a zebra crossing pedestrian and vehicle traffic control method according to the present invention.

[0012] Figure 3 This is another schematic diagram of a zebra crossing pedestrian and vehicle traffic control method according to the present invention. Detailed Implementation

[0013] 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 in 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.

[0014] Example 1

[0015] Embodiment 1 of the present invention provides a method for controlling pedestrian and vehicle traffic at zebra crossings, comprising the following steps:

[0016] The image acquisition device acquires video stream data, which includes image information of pedestrians and vehicles in the crosswalk area;

[0017] Perform pedestrian and vehicle detection and tracking on the video stream data to obtain the number of pedestrians, age structure, movement intentions, and vehicle queue length;

[0018] Based on the pedestrian and vehicle detection and tracking results, the number of pedestrians staying, their speed of movement, and the length of vehicle queues per unit time are calculated to form pedestrian and vehicle flow data.

[0019] The pedestrian and vehicle traffic data is processed based on a two-layer control model to generate a dynamic signal timing scheme, which is then distributed to the signal control equipment via an edge computing node.

[0020] Example 2

[0021] like Figures 1-3 This embodiment of a method for controlling pedestrian and vehicle traffic at a zebra crossing may specifically include:

[0022] This invention provides a method for controlling pedestrian and vehicle traffic at zebra crossings, aiming to dynamically adjust traffic signals through video stream data analysis to improve the efficiency and safety of pedestrian and vehicle traffic. The technical solution of this invention is described in detail below with reference to specific embodiments to make the objectives, technical solutions, and advantages of this invention clearer. In one embodiment, the method provided by this invention mainly targets urban road intersections or pedestrian crossing areas. It acquires real-time video stream data through an image acquisition device and performs pedestrian and vehicle detection, traffic flow analysis, and signal timing optimization based on the video stream data. The method includes multiple steps, covering the entire process from data acquisition to signal control, and is applicable to various traffic scenarios, such as traffic management during peak hours, low-traffic periods at night, and under special weather conditions. The following will describe these steps step by step. Step S1: The image acquisition device acquires video stream data, which includes image information of pedestrians and vehicles in the pedestrian crossing area. Specifically, the image acquisition device is usually installed at a fixed location near the intersection or pedestrian crossing, such as a traffic light pole or a dedicated bracket, ensuring that the camera's field of view covers the pedestrian crossing and adjacent lane areas. The video stream data acquired by the device is a continuous sequence of image frames. The frame rate can be adjusted according to actual needs, such as 25 or 30 frames per second, to ensure the real-time performance and accuracy of subsequent analysis. During the acquisition process, the device needs to adapt to different lighting conditions and weather environments to ensure that the image clarity meets the detection requirements. Step S11: The image acquisition device integrates a supplementary lighting module to acquire nighttime video stream data. It should be noted that the imaging quality of the image acquisition device may decrease at night or in low-light scenarios, affecting the accuracy of subsequent pedestrian and vehicle detection. To address this, the device has a built-in supplementary lighting module, typically employing infrared or white light supplementary lighting technology, which automatically activates when light is insufficient to provide additional light source support. For example, in road sections without streetlights at night, the infrared supplementary lighting module can capture the outline information of pedestrians and vehicles by emitting invisible light, avoiding image blurring or target loss due to insufficient light. The power and illumination range of the supplementary lighting module can be dynamically adjusted according to the installation location and coverage area to reduce energy consumption and avoid glare interference to pedestrians or drivers. Step S12: The image acquisition device integrates a stabilization module to acquire video stream data for rain and snow scenes, forming the video stream data. Specifically, rain, snow, or other harsh environments can cause image acquisition equipment to shake due to wind or vibration, affecting the stability of video stream data. To address this issue, devices incorporate image stabilization modules, typically using hardware stabilization or software algorithms to achieve image stabilization. Hardware stabilization uses a gyroscope sensor to detect device shaking in real time and adjust the lens position, while software stabilization uses image processing technology to correct shaking frames. For example, in windy or rainy weather, the stabilization module can effectively reduce image blur, ensuring that the outlines of pedestrians and vehicles are clearly distinguishable in the acquired video stream data. The resulting video stream data will serve as the basis for subsequent pedestrian and vehicle detection.In one possible implementation, the installation location and parameter settings of the image acquisition device need to be optimized according to the specific scenario. For example, at busy intersections in the city center, the device can be installed on top of traffic light poles with a downward angle of about 30 degrees, covering the entire pedestrian crossing and the two adjacent lanes. In suburban areas or sections of road with low pedestrian traffic, the device can be installed at a lower height, focusing on the core area of ​​the pedestrian crossing to reduce the collection of invalid data. Furthermore, the device supports remote parameter adjustments, such as focal length and frame rate, to adapt to changes in traffic flow at different times. Through these methods, the acquired video stream data can comprehensively reflect the dynamics of people and vehicles, providing a reliable basis for subsequent analysis. It should be noted that during the acquisition of video stream data, the device also needs to consider the real-time nature of data transmission. Typically, the device transmits video stream data to edge computing nodes or cloud servers for processing via wired or wireless networks. To ensure transmission efficiency, the data can be transmitted in a compressed format, and each frame can be timestamped to ensure accurate time-point correspondence during subsequent analysis. For example, during peak hours, the device can prioritize transmitting video stream data from key areas, reducing the amount of data from non-key areas, thereby improving transmission speed and processing efficiency. Step S2 involves performing pedestrian and vehicle detection and tracking on the video stream data to obtain the number of pedestrians, age structure, movement intentions, and vehicle queue lengths. Specifically, this step analyzes each frame of the video stream data to identify pedestrians and vehicles within the crosswalk area and further extracts relevant feature information. The detection and tracking process is typically completed in edge computing nodes or local processing units to reduce data transmission latency and ensure real-time performance. The obtained pedestrian number, age structure, movement intentions, and vehicle queue lengths will serve as important bases for subsequent traffic flow analysis. Step S21 involves the target detection module performing target segmentation on the crosswalk area in the video stream data to identify pedestrians and vehicles. It should be noted that the target detection module first preprocesses each frame of the video stream data, such as denoising and contrast enhancement, to improve image quality. Subsequently, the module performs target segmentation on the crosswalk area, dividing the image into multiple regions and identifying pedestrians and vehicles within each region. During the segmentation process, the module performs preliminary classification based on the shape, size, and texture features of the targets. For example, pedestrian targets typically appear as small vertical outlines, while vehicle targets appear as larger rectangular outlines. In this way, the module can accurately distinguish target objects from complex backgrounds. Step S211: The feature extraction layer optimizes the extraction of pedestrian features for small-target pedestrians. Specifically, in pedestrian crossing areas, pedestrian targets may appear as small targets due to distance or occlusion, making them prone to being missed by traditional detection methods. Therefore, the feature extraction layer is specifically optimized for small-target pedestrians, capturing local features such as head and limb contours through multi-scale feature analysis.For example, in video stream data, the feature extraction layer can magnify pedestrian targets that are far from the camera and extract their subtle features by combining high-resolution feature maps, thereby improving detection accuracy. In addition, the feature extraction layer also analyzes the dynamic features of pedestrian targets, such as gait and direction of movement, providing support for subsequent judgments of age structure and movement intention. In step S212, the target classification layer distinguishes between pedestrian and vehicle targets based on the pedestrian and vehicle features. It should be noted that the target classification layer receives feature information output by the feature extraction layer and distinguishes targets using a pre-trained classification model. The classification model is typically trained on a large amount of labeled data and can accurately determine the target type based on feature differences. For example, pedestrian features usually include more curvature changes and dynamic gait information, while vehicle features are mainly straight-line contours and stable movement trajectories. The classification layer integrates these features, classifying targets into two main categories: pedestrians and vehicles, and assigning a unique identifier to each target for subsequent tracking. In step S213, the tracking module combines the pedestrian and vehicle targets to eliminate occlusion interference and generate trajectory information. Specifically, the tracking module receives target information output by the target classification layer and generates the movement trajectory of each target through inter-frame correlation analysis. To address occlusion issues, the module employs multi-target tracking technology, predicting the movement path and speed of targets to infer the position of occluded targets. For example, in a pedestrian crossing area, a pedestrian may be briefly occluded by other pedestrians or vehicles. The tracking module predicts the current position based on the trajectory information of the previous few frames and corrects it using the detection results of subsequent frames. In this way, the module can generate continuous trajectory information, providing accurate data for subsequent analysis. Step S22: The tracking module generates trajectory information based on the pedestrian and vehicle targets, determining the number of pedestrians, age structure, movement intention, and vehicle queue length. Specifically, based on the generated trajectory information, the tracking module counts the number of pedestrians in the pedestrian crossing area and further infers the age structure through pedestrian characteristics. For example, children typically have smaller body sizes and faster gait frequencies, while older adults exhibit slower movement speeds and more stable gaits. Movement intent is determined through trajectory direction and speed changes. For example, a pedestrian's trajectory pointing towards a crosswalk with a stable speed indicates an intention to cross. Vehicle queue length is calculated by detecting the distribution and trajectory information of vehicles within the lane, typically expressed as the number of vehicles or lane occupancy. Through the above analysis, the tracking module can output comprehensive dynamic information on pedestrians and vehicles. In one possible implementation, the determination of pedestrian age structure can be achieved by combining multiple features for comprehensive analysis. For example, in video stream data, the module can perform preliminary classification based on the height ratio, gait frequency, and limb movement characteristics of pedestrian targets. For children, the feature extraction layer focuses on analyzing their smaller size and rapid gait changes; for elderly targets, it focuses on their slower movement speed and stable body posture. In addition, the module can also combine clothing features for auxiliary judgment, such as school uniforms potentially corresponding to children or teenagers.By analyzing multi-dimensional features, the accuracy of age structure determination can be significantly improved, providing more detailed data support for subsequent signal timing. It should be noted that determining movement intent is crucial in vehicle-pedestrian cooperative control. For example, in pedestrian crossing areas, some pedestrians may simply linger on the roadside without intending to cross. If the signal timing scheme incorrectly responds to such situations, it may lead to a decrease in vehicle traffic efficiency. Therefore, the tracking module performs intent analysis by comprehensively considering multiple indicators such as trajectory direction, speed changes, and dwell time. For example, if a pedestrian's trajectory direction consistently faces the pedestrian crossing and their speed remains stable, it is determined that they intend to cross; if their trajectory direction changes repeatedly or they linger for extended periods, it is determined that they do not intend to cross. Through this refined analysis, the system can respond more accurately to actual needs and avoid invalid signal adjustments. In one embodiment, the calculation of vehicle queue length can be optimized according to different lane types. For example, at multi-lane intersections, the tracking module counts the number of vehicles in each lane and calculates the queue length by combining this with the spacing information between vehicles. For straight-ahead lanes, the module focuses on analyzing the longitudinal distribution of vehicles; for left-turn or right-turn lanes, it determines whether vehicles belong to a queue based on their trajectory direction. Furthermore, the module can correct current calculation results using historical data. For example, if a lane frequently experiences queuing during peak hours, the module will appropriately increase the weight of queue length to ensure that the signal timing scheme better reflects actual traffic conditions. Step S3 calculates the number of pedestrians stationed, their movement speed, and vehicle queue length per unit time based on the pedestrian and vehicle detection and tracking results, forming pedestrian and vehicle flow data. Specifically, this step further processes the pedestrian and vehicle detection and tracking results to extract dynamic flow information per unit time, providing a quantitative basis for subsequent signal timing. The calculation process is typically completed in edge computing nodes to ensure real-time performance. The resulting flow data will comprehensively reflect the traffic demand and congestion status of the pedestrian crossing area. See also. Figure 2 The specific execution steps of step S3 include steps S31-S32;

[0023] Step S31: The pixel-level segmentation module performs segmentation processing on the video stream data to obtain pedestrian dwelling areas. It should be noted that the pixel-level segmentation module performs fine-grained segmentation on each frame of the video stream data to identify the specific location of pedestrians within the crosswalk area. The segmentation process first models the background of the image, distinguishing between static backgrounds and dynamic targets. Then, it performs pixel-level analysis on the dynamic target areas to determine the clustered areas of pedestrian dwellings. For example, in the waiting areas on both sides of the crosswalk, the module marks the distribution range of high-density pedestrian targets and excludes moving pedestrian targets. In this way, the module can accurately obtain the location and range of dwelling areas, providing a basis for subsequent quantity statistics. Step S32: The optical flow calculation module determines the movement speed based on the pedestrian dwelling areas, and calculates the number of pedestrians dwelling per unit time and the length of vehicle queues to form the pedestrian and vehicle flow data. Specifically, the optical flow calculation module calculates the movement speed of pedestrian targets by analyzing the pixel changes between consecutive frames in the video stream data. For the designated area, the module counts the number of pedestrians within a unit of time, typically summarizing per minute or every 5 minutes. Vehicle queue length is further corrected based on the aforementioned tracking results to ensure data accuracy. For example, during peak hours, the module can use optical flow analysis to determine if pedestrian movement speed is below normal. If the speed decreases significantly, it indicates a potentially high number of pedestrians, requiring priority response. The resulting pedestrian and vehicle flow data will serve as a key input for signal timing adjustment. In step S321, the state filtering module receives the instantaneous flow rate output by the optical flow calculation module and corrects it to form the pedestrian and vehicle flow data. It should be noted that the instantaneous flow rate output by the optical flow calculation module may fluctuate due to noise or transient anomalies (such as sudden acceleration of pedestrians). Therefore, the state filtering module smooths the instantaneous flow rate and corrects outliers. For example, in video stream data, if a frame's image shows a pedestrian target detection error due to lighting changes, the state filtering module will interpolate and correct it using flow data from previous and subsequent frames to ensure the continuity and stability of the output. In this way, the corrected pedestrian and vehicle flow data can more accurately reflect traffic conditions. Step S322: The pedestrian and vehicle traffic data is fused with meteorological information and historical traffic data. Specifically, to further improve the reliability of the traffic data, the system fuses and analyzes the current calculation results with meteorological information and historical traffic data. For example, in rainy or snowy weather, pedestrian movement speed may generally decrease. The system will adjust the current traffic data with weights based on meteorological information to avoid misjudgments caused by weather factors. Simultaneously, the system will also refer to historical traffic data, such as the average traffic value for the same period over the past week, to determine whether the current traffic is abnormal. By fusing multi-source information, the system can generate more comprehensive and accurate pedestrian and vehicle traffic data, providing reliable support for subsequent signal timing. In one possible implementation, the fusion of meteorological information can be based on different adjustment strategies according to different weather types.For example, in light rain, the system can reduce the weight of pedestrian movement speed calculations and prioritize changes in the number of people lingering. In heavy rain or snow, the system will further increase the weight of lingering pedestrians and, combined with historical data, determine whether signal duration needs to be adjusted in advance. Furthermore, the system can dynamically update its strategies by accessing real-time weather warnings. For instance, if a short-term heavy rainfall warning is received, the system will immediately increase pedestrian priority to ensure that the signal timing scheme can respond promptly to the traffic demand caused by weather changes. It should be noted that the fusion of historical traffic data is particularly important when dealing with periodic traffic patterns. For example, during weekday morning rush hours, a large number of pedestrians often linger at crosswalks at intersections. The system can predict the peak traffic flow for the current period based on historical data and correct the real-time calculation results. If the current traffic flow is significantly lower than the historical average, the system will further analyze whether it is due to a special event and verify this by combining video stream data. In this way, the system can avoid signal timing errors caused by the bias of a single data source. In one embodiment, the formation of pedestrian and vehicle traffic data can also be combined with a time dimension for hierarchical analysis. For example, the system can divide a day into multiple time periods, such as morning peak, off-peak, evening peak, and late-night low-traffic periods, and statistically analyze the pedestrian and vehicle traffic characteristics for each time period. During the morning peak, the system focuses on the number of pedestrians waiting and the length of vehicle queues, prioritizing pedestrian safety; during late-night periods, it focuses more on vehicle traffic efficiency, reducing unnecessary signal switching. Through time-segmented analysis, the system can generate pedestrian and vehicle traffic data that better reflects actual needs, providing accurate support for subsequent dynamic signal timing. This hierarchical analysis approach effectively improves the flexibility and targeting of traffic management. In one possible implementation, the statistics on the number of pedestrians waiting can also be optimized by incorporating spatial dimensions. For example, in a pedestrian crossing area, the system can divide the waiting area into multiple sub-areas and statistically analyze the number of pedestrians waiting in each sub-area. If the number of pedestrians waiting in a certain sub-area is consistently higher than in other areas, the system will determine whether there is local congestion and include this information in the pedestrian and vehicle traffic data. In addition, the system can also analyze the spatial distribution of the number of pedestrians waiting to determine whether pedestrians are concentrated on one side of the pedestrian crossing; if so, it may indicate that the signal duration is insufficient to meet traffic demand. Through refined spatial analysis, the system can gain a more comprehensive understanding of traffic dynamics, providing multi-faceted support for signal adjustments. It's important to note that the steps outlined above, from video stream data acquisition to the formation of pedestrian and vehicle traffic data, form the foundation for dynamic pedestrian-vehicle collaborative control. Through multi-level analysis of video stream data, the system can accurately capture the passage demand and congestion status in pedestrian crossing areas, laying a solid foundation for the subsequent generation of signal timing schemes. Especially in complex traffic scenarios, such as peak hours or in adverse weather conditions, the system, through multi-dimensional data fusion and refined analysis, can effectively improve the response speed and accuracy of traffic management.In one embodiment, the system can also perform optimized analysis for the passage needs of special groups. For example, at intersections near schools, the system can identify characteristics of student groups through video stream data, such as short stature and group movement, and combine this with time information to determine whether it is a peak time for school arrival and departure. If it is determined to be a peak time for student passage, the system will automatically increase the weight of pedestrian dwell time to ensure that the signal duration can meet the needs of students for safe passage. In addition, the system can also analyze the movement speed and trajectory direction of student groups to determine whether there are unsafe behaviors such as jaywalking, and incorporate relevant information into the traffic flow data so that safety is prioritized in subsequent signal adjustments. In one possible implementation, the calculation of vehicle queue length can also be differentiated based on lane function. For example, at intersections with dedicated bus lanes, the system will prioritize the calculation of bus lane queue length and assign higher passage weight to buses. If the bus lane queue length is long, the system will highlight this information in the pedestrian and vehicle traffic flow data to ensure that subsequent signal timing schemes can prioritize the passage efficiency of buses. At the same time, the system can also analyze the arrival time of buses to predict their impact on signal duration, thereby further optimizing the accuracy of traffic flow data. It should be noted that the various analysis methods and optimization strategies involved in the above steps are all aimed at improving the comprehensiveness and reliability of pedestrian and vehicle traffic data. Through in-depth mining of video stream data and fusion of multi-source information, the system can generate high-quality traffic data, providing solid support for subsequent dynamic signal timing. Especially in urban traffic management, this video-based analysis method can effectively cope with complex traffic scenarios and improve the efficiency and safety of intersections. In one embodiment, the system can also dynamically update pedestrian and vehicle traffic data through a real-time feedback mechanism. For example, during video stream data analysis, if the system detects a sudden event, such as a traffic accident or a large-scale crowd gathering, the system will immediately adjust the traffic data calculation strategy, prioritizing the statistics of the number of people staying and the speed of movement in the relevant areas, and updating the results to the subsequent processing stages in real time. Through this dynamic feedback mechanism, the system can quickly respond to abnormal traffic conditions, ensuring the timeliness and accuracy of traffic data and providing timely basis for signal timing adjustments. In one possible implementation, the system can also combine user feedback information to correct pedestrian and vehicle traffic data. For example, at some intersections, pedestrians or drivers may submit traffic condition feedback through mobile applications, such as excessively long signal waiting times or severe lane congestion. The system compares this feedback information with the video stream data analysis results. If a significant difference is found, a data correction process is triggered to recalculate the traffic data for the relevant area. By incorporating user feedback, the system can further improve the reliability of traffic data, ensuring that subsequent signal timing schemes better meet actual needs. It should be noted that the above steps, from data acquisition to traffic analysis, fully demonstrate the advantages of video-based human-vehicle collaborative control methods.By performing multi-level processing and multi-dimensional analysis of video stream data, the system can comprehensively grasp the traffic dynamics of pedestrian crossing areas, laying the foundation for subsequent signal timing optimization. Especially in complex traffic environments, this method can effectively improve the precision of traffic management and ensure the safe passage of pedestrians and vehicles. In one embodiment, the system can also optimize traffic flow data based on seasonal traffic characteristics. For example, in winter, rain and snow may lead to a general decrease in pedestrian movement speed; the system will seasonally adjust the traffic flow calculation results based on historical data and meteorological information to ensure that the data accurately reflects actual traffic demand. Simultaneously, the system can also predict traffic peaks on special dates by analyzing traffic flow changes during holidays and adjust the weighting of traffic flow data in advance. Through this seasonal and temporal analysis, the system can generate more realistic traffic flow data, providing accurate support for signal timing. In one possible implementation, the system can also improve the coverage of traffic flow data through multi-device collaboration. For example, at large intersections, a single image acquisition device may not be able to cover all areas; the system can jointly analyze video stream data collected by multiple devices to generate comprehensive traffic flow data. Data alignment between devices is achieved through timestamps and spatial coordinates to ensure the continuity and consistency of the analysis results. Through multi-device collaboration, the system can effectively address traffic analysis needs in complex scenarios, providing more reliable data support for subsequent signal control. Step S4 involves processing the pedestrian and vehicle traffic data based on a two-layer control model to generate a dynamic signal timing scheme, which is then distributed to the signal control equipment via edge computing nodes. Specifically, this step involves comprehensively analyzing the aforementioned pedestrian and vehicle traffic data and using a two-layer control model to generate a signal timing scheme adapted to the current traffic conditions. The two-layer control model combines rule-based judgment and predictive analysis, ensuring real-time response while also considering long-term optimization. The generated timing scheme will be transmitted to the signal control equipment via edge computing nodes to achieve dynamic adjustment of traffic lights, ensuring the efficiency and safety of pedestrian and vehicle traffic. See also... Figure 3 The specific execution steps of step S4 include steps S41-S43;

[0024] Step S41: The rule engine determines whether the pedestrian waiting time in the pedestrian and vehicle traffic data exceeds a preset threshold. If it does, it triggers a green light extension command. It should be noted that the rule engine is the first layer of the two-layer control model, primarily responsible for quickly judging pedestrian and vehicle traffic data based on preset rules. Pedestrian waiting time is usually calculated by statistically analyzing the number of pedestrians and the duration of their stay in the designated area. If the pedestrian waiting time exceeds a preset threshold within a certain period, such as exceeding 60 seconds, the rule engine will immediately trigger a green light extension command to ensure safe pedestrian passage. For example, during peak hours, if a large number of pedestrians gather at a crosswalk and the waiting time is long, the rule engine will prioritize responding to pedestrian needs and generate a command to extend the green light duration, preventing unsafe behavior due to prolonged waiting. Step S42: The time-series prediction network receives the output of the rule engine and the pedestrian and vehicle traffic data, predicts signal duration, and generates the dynamic signal timing scheme. Specifically, the time-series prediction network is the second layer of the two-layer control model, responsible for predicting signal duration requirements for a future period based on historical and current pedestrian and vehicle traffic data. The temporal prediction network analyzes the changing trends of traffic data and combines the initial instructions output by the rule engine to generate a more refined signal timing scheme. For example, if it predicts that the queue length will continue to increase within the next 5 minutes, the network will appropriately extend the green light duration for vehicle traffic while balancing pedestrian traffic demand. Through this prediction method, the system can anticipate traffic changes and improve the foresight of signal adjustments. In step S43, the dual-layer control model integrates the green light extension instruction with the signal duration to determine the final timing scheme. It should be noted that the dual-layer control model generates the final dynamic signal timing scheme by combining the real-time instructions from the rule engine and the prediction results from the temporal prediction network. During the integration process, the system will weight and adjust the two results according to the current traffic conditions. For example, if the rule engine detects that the pedestrian waiting time has exceeded the threshold, the system will prioritize executing the green light extension instruction; if the prediction network indicates that vehicle traffic is about to surge, the system will appropriately increase the vehicle passage time while ensuring pedestrian safety. Through this integration mechanism, the system can find a balance between real-time response and long-term optimization, ensuring the rationality and effectiveness of the timing scheme. In one possible implementation, the threshold settings of the rule engine can be dynamically adjusted according to different scenarios. For example, at intersections near schools, the system can set the pedestrian waiting time threshold to a lower value, such as 30 seconds, to prioritize the safety of students crossing the road; in commercial or office areas, the threshold can be set to a higher value, such as 90 seconds, to balance vehicle traffic efficiency. Furthermore, the threshold can be tiered according to time periods, such as lowering the threshold during morning and evening peak hours and appropriately raising it during off-peak hours. Through this flexible threshold adjustment, the rule engine can better adapt to the needs of different traffic scenarios and improve the targeting of signal timing. In one embodiment, the prediction process of the time-series prediction network can be optimized by incorporating multi-dimensional data.For example, the system can combine current pedestrian and vehicle traffic data with historical traffic data, meteorological information, and special event information to generate more accurate prediction results. In rainy or snowy weather, the system predicts that pedestrian movement speed may decrease, thus appropriately extending pedestrian crossing time; during holidays or large events, the system predicts that traffic peaks will arrive earlier, adjusting signal duration accordingly. Through comprehensive analysis of multi-dimensional data, the prediction network can effectively cope with complex traffic conditions, ensuring the foresight and adaptability of the timing scheme. It should be noted that the design of the two-layer control model fully considers the balance between real-time performance and predictability. The rule engine is responsible for quickly responding to current traffic demands, ensuring the system can react to emergencies in a short time; while the time-series prediction network optimizes the long-term effect of signal timing by analyzing traffic trends. The collaborative work of the two models enables the system to minimize vehicle queuing time and improve the overall traffic efficiency of the intersection while ensuring pedestrian safety. Step S421: The edge computing node receives the video stream data and performs local pedestrian and vehicle detection. Specifically, the edge computing node is typically deployed in devices near the intersection, responsible for receiving video stream data transmitted by the image acquisition device and performing preliminary pedestrian and vehicle detection tasks locally. Local detection mainly includes target recognition and trajectory tracking, similar to the aforementioned steps, but with a greater emphasis on real-time performance. Edge computing nodes reduce the pressure of data transmission to the cloud through local processing, ensuring the system can complete detection tasks under low latency conditions. For example, in areas with poor network conditions, edge computing nodes can independently extract the number of people and vehicles and their trajectory information, providing timely data support for subsequent signal timing. In step S422, the edge computing node transmits control commands to the signal control device through the communication interface according to the dynamic signal timing scheme. It should be noted that after generating or receiving the dynamic signal timing scheme, the edge computing node will send control commands to the signal control device through the communication interface. The communication interface typically uses wired or wireless methods to ensure the stability and reliability of command transmission. For example, at urban intersections, the edge computing node can transmit control commands to the traffic light controller through a fiber optic network. The commands include green light duration, red light duration, and phase switching sequence. In this way, the signal control device can adjust the traffic light status in real time according to the commands, ensuring smooth traffic flow. In one possible implementation, the local detection function of the edge computing node can be dynamically allocated according to computing resources. For example, with sufficient computing resources, nodes can perform complete detection and tracking tasks on video stream data, generating detailed pedestrian and vehicle traffic data. When resources are limited, nodes prioritize processing video data from critical areas, simplifying the detection process to ensure real-time performance. Furthermore, nodes can collaborate with cloud servers to upload some complex computing tasks to the cloud for processing, thus maintaining system efficiency even with limited resources. Through this dynamic allocation mechanism, edge computing nodes can adapt to application needs under different hardware conditions.In one embodiment, the communication process between the edge computing node and the signal control equipment can be equipped with multiple safeguards. For example, the system can use dual-channel communication to ensure reliable transmission of control commands. If the primary channel is interrupted due to network failure, the system will automatically switch to the backup channel to avoid command loss. Furthermore, the system can encrypt control commands to prevent external interference or malicious attacks, ensuring the accuracy of commands received by the signal control equipment. Through these safeguards, the system can operate stably in complex environments, ensuring the continuity and security of traffic signal adjustments. It should be noted that the edge computing node plays a crucial role in the distribution of dynamic signal timing schemes. Through local processing and real-time communication, the node can effectively reduce system latency, ensuring that the signal control equipment can respond promptly to changes in traffic conditions. Especially during peak hours or when unexpected events occur, the rapid response capability of the edge computing node can significantly improve the efficiency of traffic management and reduce intersection congestion. In one possible implementation, the edge computing node can also adaptively adjust and optimize the command transmission frequency. For example, during periods of stable traffic flow, nodes can reduce the frequency of instruction transmission to minimize communication resource consumption; during periods of drastic traffic changes, such as morning and evening rush hours, nodes will increase the transmission frequency to ensure that signal control equipment can receive the latest timing scheme in real time. Through this adaptive adjustment, the system can save communication resources as much as possible while ensuring response speed, thereby improving overall operational efficiency. In one embodiment, the system can also achieve collaborative control of multiple intersections through edge computing nodes. For example, on urban arterial roads, edge computing nodes at multiple adjacent intersections can be interconnected through a network to share pedestrian and vehicle traffic data and timing schemes. If a certain intersection experiences severe congestion, the system will notify the nodes at adjacent intersections to adjust the signal duration, guide vehicle diversion, and alleviate local pressure. Through this collaborative control method, the system can optimize traffic flow distribution at the regional level and improve the overall road network capacity. It should be noted that the generation and distribution process of the dynamic signal timing scheme fully demonstrates the flexibility and efficiency of the present invention in traffic management. Through the collaborative work of the two-layer control model and edge computing nodes, the system can respond to current needs in real time while taking into account future traffic flow trends, ensuring the scientific and rational nature of signal adjustments. Especially in complex traffic scenarios, this method can effectively improve intersection management and provide strong support for the smooth operation of urban traffic. In one possible implementation, the system can also monitor the operational status of signal control equipment in real time through edge computing nodes. For example, nodes can periodically receive operational data from the signal control equipment, such as whether traffic light switching is normal and whether the green light duration matches the instructions. If an anomaly is detected, the node will immediately generate an alarm message and notify maintenance personnel, while simultaneously attempting to adjust the signal status using backup instructions. Through this monitoring mechanism, the system can promptly detect and resolve equipment failures, ensuring the smooth execution of the signal timing scheme.In one embodiment, the system can also set an emergency timing mode for special traffic events. For example, in the event of a traffic accident or road construction, edge computing nodes can detect abnormal traffic distribution based on video stream data and automatically switch to emergency timing mode. In emergency mode, the system prioritizes the traffic efficiency of roads surrounding the accident area, guiding vehicles to detour by adjusting signal duration, while reserving fast lanes for rescue vehicles. Through this emergency response mechanism, the system can quickly adjust signal status when an emergency occurs, reducing the impact of the event on traffic. It should be noted that the above steps, through the collaborative work of a two-layer control model and edge computing nodes, realize the generation and distribution of dynamic signal timing schemes. The system can flexibly adjust signal duration and phase switching sequence based on real-time pedestrian and vehicle traffic data and prediction results, ensuring efficient operation of intersections under different traffic conditions. Especially in urban traffic management, this dynamic adjustment method can significantly reduce pedestrian waiting time and vehicle queue length, improving overall traffic efficiency. In one possible implementation, the system can also periodically optimize the signal timing scheme through a two-layer control model. For example, the system can analyze historical signal timing schemes and corresponding traffic data weekly or monthly to evaluate the effectiveness of each scheme in different scenarios. If the system finds that a particular signal timing scheme is ineffective during a certain period, it will automatically adjust the threshold of the rule engine or the weight settings of the time-series prediction network to improve the adaptability of subsequent schemes. Through this periodic optimization, the system can continuously improve the signal timing effect and adapt to long-term changes in urban traffic. In one embodiment, the system can also set differentiated signal timing strategies for different types of intersections. For example, at a T-junction, the system will focus on optimizing the traffic efficiency of vehicles on the main road while ensuring the basic needs of vehicles and pedestrians on side roads; at a roundabout, the system will dynamically adjust the signal duration at each entrance to balance traffic flow in all directions and avoid congestion within the roundabout. Through this differentiated strategy, the system can generate signal timing schemes that better meet actual needs based on the specific shape and traffic characteristics of the intersection. It should be noted that the generation process of dynamic signal timing schemes fully considers the diverse needs of pedestrian and vehicle traffic. Through the collaborative analysis of the rule engine and the time-series prediction network, the system can maximize vehicle traffic efficiency while ensuring pedestrian safety. Especially during peak hours or in adverse weather conditions, this dynamic adjustment method can effectively alleviate traffic congestion and provide urban residents with a more convenient travel experience. In one possible implementation, the system can also personalize signal timing schemes through a two-layer control model. For example, at intersections near hospitals or nursing homes, the system can identify special groups, such as the elderly or those with mobility impairments, based on video stream data and automatically extend pedestrian crossing time to ensure their safe passage. Furthermore, the system can analyze the travel patterns of special groups using historical data and adjust signal durations in advance during specific time periods. Through this personalized customization, the system can better serve special groups and improve the inclusivity of traffic management.In one embodiment, the system can also interact with other modules of the intelligent transportation system through edge computing nodes. For example, nodes can share dynamic signal timing schemes with the vehicle-road cooperative system, providing autonomous vehicles with real-time signal status information to assist them in planning their travel routes. If the signal timing scheme shows that the green light duration in a certain direction is short, the autonomous vehicle can adjust its speed in advance or choose an alternate route to avoid unnecessary waiting. Through this linkage mechanism, the system can further improve the intelligence level of traffic management and lay the foundation for the future development of intelligent transportation. It should be noted that the above steps achieve intelligent control of traffic signals through in-depth analysis of pedestrian and vehicle flow data and flexible adjustment of dynamic signal timing. The system can quickly respond to traffic demands in different scenarios, ensuring the efficiency and safety of pedestrian and vehicle traffic. Especially in complex urban traffic environments, this video-based collaborative control method can significantly improve the management effect of intersections and provide an effective solution for alleviating traffic congestion. In one possible implementation, the system can also verify the signal timing scheme in real time through a two-layer control model. For example, after the timing scheme is issued, the system continuously monitors traffic flow changes at intersections through video stream data. If a significant deviation is found between the actual traffic flow and the predicted results, the system immediately triggers a scheme correction process to regenerate the timing scheme. Through this real-time verification mechanism, the system can promptly identify and resolve potential problems in the timing scheme, ensuring that signal adjustments always align with actual traffic conditions. In one embodiment, the system can also set specific timing strategies for different weather conditions. For example, in foggy or rainy weather, the system automatically extends the green light duration in all directions based on video stream data and meteorological information to ensure the visibility safety of drivers and pedestrians; in hot weather, the system shortens pedestrian waiting time to avoid the health risks of prolonged exposure to high temperatures. Through such specific strategies, the system can provide more humane signal adjustment schemes under special weather conditions, improving the adaptability of traffic management. It should be noted that the generation and issuance process of the dynamic signal timing scheme demonstrates the innovation and practicality of the present invention in traffic management. By conducting multi-level analysis of pedestrian and vehicle traffic data and collaboratively optimizing a two-layer control model, the system can ensure real-time response while also considering long-term traffic trends, thus guaranteeing the scientific and rational nature of signal adjustments. Particularly in urban traffic management, this method effectively addresses traffic demands in complex scenarios, providing crucial support for improving intersection efficiency and safety. In one possible implementation, the system can also achieve local backup of signal timing schemes through edge computing nodes. For example, in the event of a network outage or cloud server failure, edge computing nodes can invoke locally stored backup timing schemes to ensure the continuous operation of signal control equipment. Backup schemes are typically based on historical traffic data and common scenario presets, covering most routine traffic conditions.Through this local backup mechanism, the system can maintain stability under abnormal conditions, avoiding signal control interruptions caused by communication failures. In one embodiment, the system can also perform multi-objective optimization of the signal timing scheme through a two-layer control model. For example, the system can simultaneously consider multiple objectives such as pedestrian safety, vehicle traffic efficiency, and energy consumption, and generate a comprehensive optimal timing scheme through weighted adjustments. In terms of energy consumption, the system will minimize the frequent switching of traffic lights to reduce energy consumption; in terms of traffic efficiency, the system will dynamically adjust the duration of each direction based on traffic flow data to ensure overall traffic balance. Through this multi-objective optimization, the system can meet traffic demand while minimizing resource waste and improving the sustainability of traffic management. It should be noted that the above steps, through the generation and distribution of dynamic signal timing schemes, realize intelligent adjustment of traffic signals. The system can flexibly adjust signal duration and phase switching sequence based on real-time and predicted pedestrian and vehicle traffic data, ensuring efficient operation of intersections in different scenarios. Especially in urban traffic management, this video-based dynamic control method can significantly reduce congestion and provide residents with a smoother travel environment. In one possible implementation, the system can also perform feedback analysis on the execution effect of signal control equipment through edge computing nodes. For example, nodes can monitor traffic changes after signal adjustment through video stream data, such as whether the number of pedestrians has decreased or the length of vehicle queues has shortened. If the adjustment effect is found to be unsatisfactory, the node will upload the feedback information to the system core module, triggering further optimization of the timing scheme. Through this feedback analysis mechanism, the system can continuously improve the signal timing effect, ensuring continuous improvement in traffic management. In one embodiment, the system can also set personalized signal timing schemes for the travel habits of different groups. For example, at intersections near residential areas, the system can analyze residents' peak travel times based on video stream data and historical data, such as the time for walking dogs in the morning or taking walks in the evening, and appropriately extend the pedestrian crossing time during these times. In addition, the system can also identify special groups such as the elderly or children by analyzing pedestrian movement speed and provide them with longer green light times. Through this personalized scheme, the system can better meet the daily travel needs of residents and improve the service level of traffic management. It should be noted that the generation and execution process of the dynamic signal timing scheme fully demonstrates the flexibility and efficiency of the method in traffic management. Through the collaborative work of the two-layer control model and edge computing nodes, the system can respond to current needs in real time while taking into account future traffic flow trends, ensuring the scientific and rational nature of signal adjustments. Especially in complex urban traffic scenarios, this method can effectively improve the management level of intersections and provide important support for the construction of intelligent transportation. In one possible implementation, the system can also adapt the signal timing scheme to specific scenarios through the two-layer control model.For example, during low-traffic periods at night, the system can automatically switch to energy-saving mode, reducing the frequency of traffic light switching and extending the traffic light cycle to reduce energy consumption. During peak daytime periods, the system switches to high-response mode, shortening the signal cycle and responding quickly to changes in traffic flow. Through this scenario-based adaptation, the system can provide the most suitable timing scheme at different times, improving the flexibility and efficiency of traffic management. In one embodiment, the system can also dynamically expand the signal timing scheme through edge computing nodes. For example, during urban road expansion or temporary traffic control, nodes can automatically adjust the coverage and adjustment logic of the timing scheme according to the new road layout and traffic distribution. If new lanes or pedestrian crossing areas are added, the nodes will re-divide the detection area through video stream data and generate corresponding timing schemes. Through this dynamic expansion mechanism, the system can adapt to the continuous changes in the urban traffic environment, ensuring the continuous effectiveness of signal control.

[0025] If the technical solution of this application involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, the personal information processing rules are clearly informed through signs / information, and authorization is obtained through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

[0026] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A method for controlling pedestrian and vehicle traffic at a zebra crossing, characterized in that, include: The image acquisition device acquires video stream data, which includes image information of pedestrians and vehicles in the crosswalk area; Perform pedestrian and vehicle detection and tracking on the video stream data to obtain the number of pedestrians, age structure, movement intentions, and vehicle queue length; Based on the pedestrian and vehicle detection and tracking results, the number of pedestrians staying, their speed of movement, and the length of vehicle queues per unit time are calculated to form pedestrian and vehicle flow data. The pedestrian and vehicle traffic data is processed based on a two-layer control model to generate a dynamic signal timing scheme, which is then distributed to the signal control equipment via an edge computing node.

2. The method as described in claim 1, characterized in that, The step of performing pedestrian and vehicle detection and tracking on the video stream data includes: a target detection module performing target segmentation on the pedestrian crossing area in the video stream data to identify pedestrian and vehicle targets; and a tracking module generating trajectory information based on the pedestrian and vehicle targets to determine the number of pedestrians, their age structure, their movement intentions, and the length of the vehicle queue.

3. The method as described in claim 1, characterized in that, The calculation of the number of pedestrians staying, their moving speed, and the length of vehicle queues per unit time based on the pedestrian and vehicle detection and tracking results includes: a pixel-level segmentation module performing segmentation processing on the video stream data to obtain the pedestrian staying area; and an optical flow calculation module determining the moving speed based on the pedestrian staying area, and statistically analyzing the number of pedestrians staying and the length of vehicle queues per unit time to form the pedestrian and vehicle flow data.

4. The method as described in claim 1, characterized in that, The method of generating a dynamic signal timing scheme based on the pedestrian and vehicle traffic data using a two-layer control model includes: a rule engine determining whether the pedestrian waiting time in the pedestrian and vehicle traffic data exceeds a preset threshold; if it does, triggering a green light extension command; a timing prediction network receiving the output of the rule engine and the pedestrian and vehicle traffic data, predicting the signal duration, and generating the dynamic signal timing scheme; and the two-layer control model integrating the green light extension command and the signal duration to determine the final timing scheme.

5. The method as described in claim 2, characterized in that, The target detection module performs target segmentation on the pedestrian crossing area in the video stream data, including: a feature extraction layer for optimizing the extraction of pedestrian features for small target pedestrians; a target classification layer for distinguishing pedestrian and vehicle targets based on the pedestrian and vehicle features; and a tracking module for generating trajectory information by combining the pedestrian and vehicle targets to eliminate occlusion interference.

6. The method as described in claim 4, characterized in that, The edge computing node sends the dynamic signal timing scheme to the signal control device, including: the edge computing node receiving the video stream data and performing local human and vehicle detection; the edge computing node transmitting control commands to the signal control device through a communication interface according to the dynamic signal timing scheme.

7. The method as described in claim 1, characterized in that, The image acquisition device acquires video stream data, including: the image acquisition device integrates a fill light module to acquire nighttime video stream data; the image acquisition device integrates an image stabilization module to acquire rain and snow scene video stream data, forming the video stream data.

8. The method as described in claim 3, characterized in that, The calculation of the number of pedestrians staying, their speed of movement, and the length of vehicle queues per unit time based on the pedestrian and vehicle detection and tracking results includes: the state filtering module receiving the instantaneous flow rate output by the optical flow calculation module and correcting it to form the pedestrian and vehicle flow data; the pedestrian and vehicle flow data is fused with meteorological information and historical flow data.

Images