Dynamic target tracking method and device, equipment and storage medium

Abstract

The invention provides a dynamic target tracking method and device, equipment and a storage medium, and relates to the field of intelligent transportation. The method comprises the following steps: acquiring vehicle information of a plurality of first vehicles and a target detection result obtained by identifying an external image through an optimized YOLOv8 target detection model; generating a vehicle-target mapping table according to the vehicle information of each first vehicle and the target detection result; determining a task vehicle corresponding to the target according to the vehicle-target mapping table, and obtaining target feature information and target movement information sent by the task vehicle; if the task vehicle loses the target, determining a plurality of second vehicles according to the newly detected target feature information and target movement information of the task vehicle, and updating the vehicle-target mapping table; and based on the updated vehicle-target mapping table, re-determining the task vehicle, and obtaining target feature information and target movement information sent by the task vehicle. According to the method, the large-range moving coverage capability of dynamic target tracking can be improved, and the situation that the tracking target is lost is reduced.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation, and in particular to a dynamic target tracking method, apparatus, device, and storage medium. Background Technology

[0002] As cities expand and transportation networks become more complex, the need for real-time monitoring and trajectory tracking of dynamic targets on the road surface is becoming increasingly urgent. Target tracking technologies based on image acquisition and intelligent recognition are gradually developing, among which target detection models relying on deep learning in the field of artificial intelligence provide core support for the accurate recognition of target images.

[0003] Existing road target tracking technologies mainly rely on two types of equipment: fixed-location monitoring devices, which track targets by acquiring images of specific areas; and monitoring devices on mobile vehicles, which acquire images as the vehicle moves and combine them with positioning information to complete initial tracking. Existing technologies have significant limitations: fixed equipment has blind spots and incomplete monitoring coverage; the monitoring range of mobile vehicles is limited by the travel route, and targets are easily lost when moving across these areas, affecting the real-time performance and accuracy of tracking.

[0004] Therefore, how to improve target detection models, enhance the ability to cover a wide range of moving targets in dynamic target tracking, and reduce the occurrence of target loss has become an urgent problem to be solved by existing technologies. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a dynamic target tracking method, apparatus, device, and storage medium to improve the wide-range mobile coverage capability of dynamic target tracking and reduce the occurrence of target loss.

[0006] In a first aspect, embodiments of the present invention provide a dynamic target tracking method, comprising: Acquire vehicle information of multiple first vehicles, and target detection results obtained by recognizing the external images of the multiple first vehicles; Based on the vehicle information of each first vehicle and the target detection results, the first vehicle that detected the target is determined, and a vehicle-target mapping table is generated; Based on the vehicle-target mapping table, determine the task vehicle corresponding to the target, and obtain the target feature information and target movement information sent by the task vehicle; If the mission vehicle loses the target, then based on the latest detected target feature information and target movement information of the mission vehicle, multiple second vehicles are determined, and the vehicle-target mapping table is updated based on the vehicle information of the multiple second vehicles and the target detection results. Based on the updated vehicle-target mapping table, the task vehicle corresponding to the target is re-determined, and the target feature information and target movement information sent by the task vehicle are obtained.

[0007] In one possible implementation, the target detection result includes detection matching degree and detection time; the vehicle information includes vehicle ID, vehicle location, and computing power load. The step of determining the first vehicle that detected the target based on the vehicle information of each first vehicle and the target detection results, and generating a vehicle-target mapping table, includes: Based on the detection matching degree of each first vehicle, determine the first vehicle that detected the target; The vehicle ID, vehicle location, computing power load, detection matching degree, and detection time of the first vehicle that detected the target are associated with the target to obtain the vehicle-target mapping table.

[0008] In one possible implementation, determining the task vehicle corresponding to the target based on the vehicle-target mapping table includes: When there is only one first vehicle in the vehicle-target mapping table, the first vehicle is determined to be the mission vehicle; When there are two or more first vehicles in the vehicle-target mapping table, the first vehicle with the highest detection matching degree is determined as the task vehicle.

[0009] In one possible implementation, the method further includes: If the computing power load of the task vehicle exceeds the preset load saturation threshold, and the task vehicle is simultaneously the task vehicle for two or more targets, then according to the preset target priority, the vehicle is determined to be the task vehicle for the highest priority target among the multiple targets. For each of the multiple targets, in the vehicle-target mapping table of that other target, select the other vehicle that detected that other target as the task vehicle for that other target.

[0010] In one possible implementation, if the mission vehicle loses the target, then based on the latest detected target feature information and target movement information of the mission vehicle, multiple second vehicles are determined, including: Based on the latest target feature information and target movement information detected by the mission vehicle, combined with road data, the predicted area of ​​the target is determined; Vehicles located within the predicted area are identified as the second vehicle.

[0011] In one possible implementation, the method for determining whether the mission vehicle has lost its target includes: If the mission vehicle does not detect a target in any of the consecutive preset frames of exterior images, and the distance between the current position of the mission vehicle and the last detected target position is greater than a preset distance threshold, then it is determined that the mission vehicle has lost the target.

[0012] In one possible implementation, the target movement information includes the current movement speed, the current movement direction, and a continuous movement trajectory.

[0013] Secondly, embodiments of the present invention provide a dynamic target tracking device, comprising: The data acquisition module is used to acquire vehicle information of multiple first vehicles, as well as target detection results obtained by recognizing the external images of the multiple first vehicles; The mapping table generation module is used to determine the first vehicle that detected the target based on the vehicle information of each first vehicle and the target detection results, and generate a vehicle-target mapping table. The task allocation module is used to determine the task vehicle corresponding to the target according to the vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle; The task coordination module is used to determine multiple second vehicles based on the latest detected target feature information and target movement information of the task vehicle if the task vehicle loses the target, and update the vehicle-target mapping table based on the vehicle information of the multiple second vehicles and the target detection results. The task reassignment module is used to redetermine the task vehicle corresponding to the target based on the updated vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle.

[0014] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor executes the steps of the method as described in the first aspect or any implementation thereof.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in the first aspect or any implementation thereof.

[0016] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: In this embodiment of the invention, the first vehicle that detected the target is determined by using the vehicle information and target detection results obtained from image recognition of multiple first vehicles, and a vehicle-target mapping table is generated. This integrates multi-vehicle detection data and quickly locks onto vehicles that can track targets. Based on this mapping table, the corresponding task vehicle is determined, which can accurately filter vehicles and assign tracking tasks, effectively avoiding duplicate tracking of multiple vehicles or gaps in tracking tasks. At the same time, by acquiring the target feature information and movement information sent by the task vehicle, accurate monitoring of the target is ensured. When the task vehicle loses the target, multiple second vehicles are determined based on its latest detected target information, and the vehicle-target mapping table is updated. This can expand the range of tracking vehicles, make up for the limitations of tracking a single vehicle, maintain the effectiveness of the vehicle-target association, and prevent the tracking from being completely interrupted after the target is temporarily lost. Based on the updated mapping table, the task vehicle is re-determined, which can realize the rapid connection of tracking tasks, ensuring that there is a dedicated task vehicle continuously tracking the target. At the same time, the target feature information and target movement information sent by the re-determined task vehicle are acquired, ensuring the continuity and consistency of tracking data. This invention achieves continuous tracking of dynamic targets by associating and dynamically coordinating multi-vehicle data with targets, effectively reducing target loss and significantly improving tracking accuracy and efficiency. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the implementation process of a dynamic target tracking method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a dynamic target tracking device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The present application will be described more clearly below with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the function of the present application, but do not limit the present application in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0019] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0020] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0021] In the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0023] Furthermore, the term "multiple" mentioned in the embodiments of this application should be interpreted as two or more.

[0024] With the acceleration of urbanization, the demand for real-time monitoring and trajectory tracking of dynamic targets such as motor vehicles, non-motor vehicles, and pedestrians is gradually increasing, currently relying mainly on fixed monitoring equipment. While fixed monitoring equipment, such as high-definition cameras at intersections and millimeter-wave radar, can continuously monitor specific areas for extended periods, they are limited by installation location, exhibiting significant blind spots due to obstructions from buildings and bridges. Furthermore, their fixed monitoring angles make it difficult to cover target trajectories in complex terrain. Simultaneously, the detection data from different devices are stored on independent platforms, preventing direct interoperability. Cross-device access often requires additional calibration, easily leading to data delays and loss of target features. Existing technologies lack a unified collaborative scheduling mechanism, resulting in susceptibility to data lag and trajectory breaks in fixed equipment tracking. When multiple devices simultaneously monitor the same target, the lack of data analysis and target determination mechanisms easily leads to resource waste and data conflicts. These problems make existing technologies insufficient to meet the demands, frequently interrupting the tracking of key targets across regions and impacting processing efficiency.

[0025] Therefore, developing a dynamic target tracking method with wide coverage and mobile tracking capability has significant practical implications and application value.

[0026] See Figure 1 The present invention provides a dynamic target tracking method, which is described in detail below: Step S101: Obtain vehicle information of multiple first vehicles, and target detection results obtained by recognizing the external images of multiple first vehicles.

[0027] The first vehicle refers to a vehicle that is pre-connected and equipped with an on-board terminal with image acquisition and processing functions, such as a police patrol car, a compliant ride-hailing vehicle, a bus or urban operating vehicle with data network communication capabilities, etc.; the external image refers to the real-time acquisition of images of the external environment in front, to the side and behind the vehicle during the vehicle's operation through the camera integrated on the first vehicle's on-board terminal; the target refers to a dynamic object that needs to be tracked in real time, such as a hit-and-run vehicle, a missing elderly person who needs to be closely monitored, or a mobile carrier transporting specific goods, etc.

[0028] In some embodiments, when the first vehicle identifies targets outside the vehicle, it is susceptible to image noise caused by light color differences and vehicle vibration. Furthermore, directly processing the entire image can easily lead to computational overload, thereby affecting the real-time performance and accuracy of target detection. Therefore, the first vehicle can identify targets from external images through an onboard terminal to obtain target detection results.

[0029] The vehicle-mounted terminal is installed on the first vehicle and can integrate a camera, an edge computing module, a positioning module, and a communication module. After the vehicle-mounted terminal starts normally, the edge computing module loads a pre-trained optimized YOLOv8 model after system initialization and obtains the target detection result based on the external images collected by the camera; the positioning module collects the positioning data of the first vehicle.

[0030] Before identifying the target, the edge computing module can perform image preprocessing on the images captured by the camera, setting a 5×5 matrix Gaussian filter to denoise the images and eliminate noise caused by factors such as light color difference and vehicle vibration; it can retain key frames of the image and set to extract one frame after a certain number of frames to balance computing power consumption and real-time performance, and prevent computing power overload and data redundancy; it can also perform image grayscale processing to reduce the amount of computation and improve the model processing speed.

[0031] The optimized YOLOv8 model was pre-trained using a dataset consisting of 810,000 images of vehicle exteriors and 93 target types. The distances between targets and camera devices in the training dataset images were labeled, enabling the pre-trained model to effectively identify targets such as human features, vehicles, and objects, obtain target detection results, and determine the position of the target relative to the vehicle.

[0032] In this embodiment of the invention, preprocessing such as Gaussian filtering for noise reduction and keyframe extraction optimizes computational power and image quality. By utilizing a pre-trained YOLOv8 model, the target can be accurately identified and its relative position determined, balancing detection efficiency and accuracy.

[0033] In one possible implementation, the target detection results include the detection matching degree and the detection time; the vehicle information includes the vehicle ID, vehicle location, and computing load.

[0034] Step S102: Based on the vehicle information of each first vehicle and the target detection results, determine the first vehicle that detected the target and generate a vehicle-target mapping table.

[0035] In this embodiment of the invention, the first vehicle in which the target was detected is determined by the vehicle information and image recognition results of multiple first vehicles, and a vehicle target mapping table is generated. This can integrate multiple vehicle detection data and quickly lock onto the vehicle with a trackable target.

[0036] In one possible implementation, based on the vehicle information of each first vehicle and the target detection results, the first vehicle that detected the target is determined, and a vehicle-target mapping table is generated, including: Based on the detection matching degree of each first vehicle, determine the first vehicle that detected the target; The vehicle ID, vehicle location, computing load, detection matching degree, and detection time of the first vehicle that detected the target are associated with the target to obtain a vehicle-target mapping table.

[0037] Among them, the detection matching degree is the quantitative value of the feature similarity between the object and the target to be tracked after the first vehicle identifies the object in the external image. It reflects the credibility of the recognition result. According to the preset detection matching degree threshold, when the detection matching degree of the first vehicle is higher than the threshold, it indicates that the recognition result of the target is reliable and it can be determined that the first vehicle has detected the target. Therefore, the first vehicle that has detected the target can be selected based on the detection matching degree.

[0038] In this embodiment of the invention, the first vehicle that has been detected as a target is accurately screened by detecting the matching degree, and the vehicle-target mapping table is associated with key information such as the vehicle's ID, location, and computing power load, which not only ensures the reliability of the first vehicle screening, but also provides comprehensive data support for system scheduling.

[0039] Step S103: Based on the vehicle-target mapping table, determine the task vehicle corresponding to the target, and obtain the target feature information and target movement information sent by the task vehicle.

[0040] In this embodiment of the invention, the task vehicle corresponding to the target is determined based on the mapping table, which can accurately filter vehicles and assign tracking tasks, effectively avoiding the occurrence of multiple vehicles tracking repeatedly or tracking task gaps. At the same time, by obtaining the target feature information and movement information sent by the task vehicle, accurate monitoring of the target is ensured.

[0041] In one possible implementation, the task vehicle corresponding to the target is determined based on a vehicle-target mapping table, including: When there is only one first vehicle in the vehicle-target mapping table, that first vehicle is determined to be the mission vehicle; When there are two or more first vehicles in the vehicle-target mapping table, the first vehicle with the highest detection matching degree is identified as the task vehicle.

[0042] In this embodiment of the invention, the task vehicle is flexibly determined based on the status of the first vehicle in the vehicle-target mapping table, which not only ensures efficient and simple task allocation, but also guarantees the accuracy of target tracking based on high matching degree.

[0043] For example, target movement information includes current movement speed, current movement direction, and continuous movement trajectory.

[0044] For example, target feature information includes distinctive appearance or attribute information of the target. For instance, if the target is a vehicle, it may include vehicle color, model, and license plate number; if the target is a pedestrian, it may include clothing color, hairstyle, and body shape.

[0045] This invention achieves continuous tracking of dynamic targets by associating and dynamically coordinating multi-vehicle data with targets, effectively reducing target loss and significantly improving tracking accuracy and efficiency.

[0046] Step S104: If the mission vehicle loses the target, then based on the latest detected target feature information and target movement information of the mission vehicle, determine multiple second vehicles, and update the vehicle-target mapping table based on the vehicle information of the multiple second vehicles and the target detection results.

[0047] In this embodiment of the invention, when a task vehicle loses its target, multiple second vehicles are determined based on its latest detected target information, and the vehicle-target mapping table is updated. This can expand the range of tracked vehicles, make up for the limitations of tracking a single vehicle, maintain the effectiveness of the association between the vehicle and the target, and prevent the tracking from being completely interrupted after the target is briefly lost.

[0048] For example, if the target disappears at a certain moment, and the target feature information and target movement information show the results of the last report from the task vehicle, then the vehicle located in the extended area of ​​the target's last movement direction and within a certain range around the predicted trajectory can be identified as the second vehicle, ensuring that the second vehicle can cover the area where the target may appear for continued tracking.

[0049] Step S105: Based on the updated vehicle-target mapping table, redetermine the task vehicle corresponding to the target, and obtain the target feature information and target movement information sent by the task vehicle.

[0050] In this embodiment of the invention, the task vehicle is re-determined based on the updated mapping table, which enables rapid connection of tracking tasks, ensures that a dedicated task vehicle continuously tracks the target, and obtains the target feature information and target movement information sent by the re-determined task vehicle, thus ensuring the continuity and consistency of tracking data.

[0051] In some embodiments, during the tracking process, the task vehicle may lose the target due to factors such as building obstruction, rapid change of target direction, or the task vehicle itself moving out of the monitoring range. Blindly expanding the candidate vehicle selection range would generate significant data redundancy and could also lead to the complete loss of the target due to vehicles being too far apart to be tracked in a timely manner. Therefore, if the task vehicle loses the target, multiple second vehicles are determined based on the latest detected target feature information and target movement information, which may include: Based on the latest target feature information and target movement information detected by the mission vehicle, combined with road data, the predicted area of ​​the target is determined; Vehicles within the predicted area are identified as the second vehicle.

[0052] Among these features, target characteristic information can clearly identify the target type (such as vehicles or pedestrians). Different types of targets have different activity ranges; vehicles mostly travel along roads, while pedestrians can enter areas such as buildings and parks. By combining target movement information with road data, the potential activity range of the target can be accurately determined, making the prediction area more comprehensively cover the range where the target may appear.

[0053] In this embodiment of the invention, by combining target features, movement information, and road data to determine the prediction area, the potential appearance range of the target can be accurately located, avoiding resource waste caused by ineffective screening. Selecting vehicles within the prediction area as secondary vehicles can significantly improve the efficiency of re-tracking the target, quickly connect tracking tasks, and ensure tracking continuity.

[0054] In some embodiments, if the method for determining whether a mission vehicle has lost the target relies on a single condition, such as the absence of a target in consecutive frames or only referring to distance, it is prone to misjudgment. For example, a brief occlusion might be mistakenly interpreted as the target disappearing and becoming untrackable, or the risk of tracking failure might not be promptly reported when the target has moved away, affecting the timeliness and accuracy of tracking transitions. Therefore, the method for determining whether a mission vehicle has lost the target may include: If the mission vehicle does not detect the target in any of the preset frames of external images, and the distance between the current position of the mission vehicle and the last detected target position is greater than a preset distance threshold, then the mission vehicle is determined to have lost the target.

[0055] In this embodiment of the invention, by combining the dual conditions of no target detection in consecutive preset frames and distance exceeding a threshold, misjudgment caused by a single condition can be effectively avoided, and the occurrence of high loss risk situations can be accurately identified, providing a reliable basis for quickly starting the second vehicle screening and connecting tracking tasks.

[0056] In some embodiments, when a task vehicle simultaneously undertakes the tracking tasks of multiple targets, the computing load may reach a saturation threshold, leading to data processing delays, decreased target monitoring accuracy, and even affecting the tracking stability of high-priority targets. Therefore, the dynamic target tracking method provided by the present invention may further include: If the computing power load of the task vehicle exceeds the preset load saturation threshold, and the task vehicle is the task vehicle corresponding to two or more targets at the same time, then the vehicle is determined as the task vehicle of the highest priority target among the multiple targets according to the preset target priority. For each other target among multiple targets, select the other vehicle that detected the other target from the vehicle-target mapping table of that other target, and use it as the task vehicle for that other target.

[0057] In this process, after the current vehicle retains the tracking task for the highest priority target, the tasks for other targets need to be transferred to distribute the computing power pressure. From the vehicle-target mapping table of each other target, the first vehicle that detects the target (not the current vehicle) is selected to take over the task. This can effectively distribute the load while ensuring that the tracking of these targets is not interrupted.

[0058] In this embodiment of the invention, the tracking authority of the task vehicle for the highest priority target is retained according to the preset priority, while the task vehicle is reassigned to other targets. This can effectively alleviate the computing power pressure on the task vehicle, ensure the tracking quality of high priority targets, ensure that the tracking tasks of all targets are not interrupted, and improve the stability and availability of the overall tracking system.

[0059] In this embodiment of the invention, a vehicle-target mapping table is generated by integrating data from multiple vehicles, and the task vehicle is dynamically determined and adjusted. This can effectively address situations such as target loss and excessive vehicle load, enabling continuous and accurate tracking of dynamic targets and improving the stability and efficiency of tracking.

[0060] This invention provides a dynamic target tracking method, detailed below: (1) Obtain information on multiple vehicles and target detection results.

[0061] The first vehicle is equipped with an in-vehicle terminal, which integrates a camera, an edge computing module, a positioning module, and a communication module. After the in-vehicle terminal is started, the edge computing module first completes the system initialization, and then loads the pre-trained optimized YOLOv8 model. The camera collects images of the outside of the vehicle in real time and transmits them to the edge computing module. The positioning module simultaneously collects the real-time location data of the first vehicle as the vehicle information of the first vehicle.

[0062] The edge computing module performs preprocessing operations on the acquired vehicle exterior images: it uses a 5×5 matrix Gaussian filter to eliminate image noise caused by light color difference and vehicle vibration; it sets a rule to extract one frame every multiple frames to retain key frames of the image, balance computing power consumption and tracking real-time performance, and avoid computing power overload and data redundancy; it performs grayscale processing on the images to reduce the amount of model calculation and improve the target recognition speed.

[0063] The optimized YOLOv8 model is pre-trained on a specific dataset: the dataset contains 93 target types (including motor vehicles, pedestrians, etc.) and 810,000 images of the exterior of the vehicle, and the distance between the target and the camera in each image is labeled; the model performs target recognition based on the pre-processed exterior images and outputs target detection results, including detection matching degree, detection time, and the position of the target relative to the vehicle; the vehicle information of the first vehicle also includes the vehicle ID and the current computing load.

[0064] Each vehicle uploads its own vehicle information and target detection results to the system server through the communication module of its on-board terminal.

[0065] (2) Generate a vehicle-target mapping table.

[0066] The system server uses each vehicle-mounted terminal as an MQTT client and assigns a unique ID "B-license plate number" to each client to establish a multi-vehicle real-time communication network. After balancing real-time performance with resource consumption, each vehicle-mounted terminal first temporarily stores the identification results in its internal storage module, and then uploads them to the system server via a 5G module. The uploaded data includes target feature information and target movement information.

[0067] After receiving the data uploaded by each first vehicle, the system server uses the detection matching degree as the core filtering condition to determine the first vehicle that has detected the target: vehicles with a detection matching degree higher than a preset threshold are determined to be the first vehicles that have detected the target.

[0068] The vehicle information of the first vehicle that detected the target is associated with the target detection result to generate a vehicle-target mapping table. The associated content includes the vehicle ID, vehicle location, computing power load, detection matching degree, detection time of the first vehicle that detected the target, as well as the target information detected by the vehicle, such as target type and target relative position. The vehicle-target mapping table is stored in real time on the system server for task vehicle scheduling planning and target tracking.

[0069] (3) Based on the vehicle-target mapping table, determine the task vehicle corresponding to the target, and obtain the target feature information and target movement information sent by the task vehicle.

[0070] The system server filters the target vehicle according to the vehicle-target mapping table: if there is only one first vehicle that has detected the target in the vehicle-target mapping table, the vehicle is directly identified as the task vehicle; if there are two or more first vehicles that have detected the target in the mapping table, the vehicle with the highest detection matching degree is selected as the task vehicle, so as to ensure tracking accuracy.

[0071] The mission vehicle collects real-time dynamic data of the target through its onboard terminal, including target feature information such as the target vehicle's color and model, or the target pedestrian's clothing characteristics; and target movement information such as the target's current speed, current direction of movement, and continuous movement trajectory. This dynamic data is then uploaded to the server via a communication module.

[0072] If a task vehicle experiences a computing load exceeding a preset saturation threshold, and that vehicle is simultaneously a task vehicle for two or more targets, the server initiates priority scheduling: based on the preset target priority, the vehicle retains its tracking permission for the highest priority target; for the remaining targets, the first vehicle that detected the target (not currently a task vehicle) is selected from the vehicle-target mapping table corresponding to each target.

[0073] The preset target priority can be determined by the system server based on the target's ID and matched against the preset target task data. For example, the priority of an emergency tracking target is higher than that of a normal target.

[0074] (4) Target loss determination and second vehicle identification.

[0075] The system server monitors the target detection data of the mission vehicle in real time and uses two conditions to determine whether the mission vehicle has lost the target: Condition 1, no target is detected in the vehicle's external images in consecutive preset frames; Condition 2, the distance between the mission vehicle's current position and the last detected target position is greater than a preset distance threshold; when both conditions are met, the mission vehicle is determined to have lost the target.

[0076] If the target is determined to be lost, the system server determines the target's predicted area based on the target feature information and target movement information last uploaded by the task vehicle, combined with road data (such as road direction, building distribution, park or residential area range, etc.): Based on the target feature information, the target type (such as vehicle, pedestrian) is identified, and combined with the target movement information (speed, direction, trajectory) and road data, the range that the target may enter is predicted. For example, the vehicle prediction area is centered on the road, and the pedestrian prediction area may include the area around the road, park and building.

[0077] The system server queries all vehicles currently within the prediction area, identifies these vehicles as second vehicles, and obtains real-time vehicle information and target detection results for each second vehicle.

[0078] In one possible implementation, the system server can perform edge computing analysis based on the target location, target movement direction, target movement trajectory, and the characteristics of each intersection the task vehicle has passed through in the last 50 seconds, which are uploaded by the task vehicle, to determine the target's most probable direction of deviation.

[0079] In one possible implementation, the system server can select a vehicle traveling along a specific route as the second vehicle based on the target's direction of movement. For example, if the target's direction of movement matches the planned route of a bus line, a bus of that line traveling within a certain range of the target can be called as the second vehicle, thereby improving target tracking efficiency.

[0080] (5) Update the mapping table and re-determine the mission vehicle.

[0081] Based on the vehicle information and target detection results uploaded by vehicles in the second vehicle whose target detection matching degree is higher than the threshold, update the vehicle location, computing load, detection matching degree and other information in the vehicle-target mapping table.

[0082] Based on the updated vehicle-target mapping table, the task vehicles are re-selected to determine the task vehicles corresponding to the targets; priority can be given to selecting the second vehicle with high detection matching degree and sufficient computing power.

[0083] The new mission vehicle initiates target tracking, collects and uploads target feature information and target movement information in real time, and the server continuously monitors the tracking status. If the target is lost again, a second vehicle is identified and the mission vehicle is selected to achieve continuous target tracking.

[0084] In one possible implementation, if the target is not in the vehicle-target mapping table of all second vehicles, the filtering range of the second vehicles is expanded. If the radius of the expanded range exceeds 1 kilometer and a match still cannot be found, the system server reports that the target has temporarily disappeared.

[0085] When the target is temporarily missing, the priority of the target tracking task is reduced, while a low-priority tracking and identification task is issued to all vehicles, and the tracking and identification operation continues to be started until the target is successfully identified and located.

[0086] In one possible implementation, the system server can interface with the command system: The system server organizes information by target based on the mapping relationship between vehicles and targets. For the same target, the identification information of different vehicles at the same time is selected based on the data with the higher similarity. Table 1 shows a partial report record of the identification information for target ID 128 on a certain day.

[0087] Table 1 Target Feature Information Recognition Record Table

[0088] According to the data processing rules, for the data in Table 1, data for B-country A12345, B-country B67890, and B-country A12345 were obtained at AM 09:11:23, AM 09:11:33, and AM 09:11:43 respectively, and used as the motion trajectory dependency data for target ID 128. First, the time change (AM 09:11:23 to AM 09:11:43, a total of 30 seconds) was calculated, and then the position change and coordinates were calculated. to Based on the distance, the moving speed of target ID128 can be calculated to be approximately 1.112 meters per second, or 4.003 kilometers per hour.

[0089] The system server packages the target detection results, target feature information, and target movement information into structured data, which is then pushed to the command system's large screen or terminal APP via a secure API interface. The command system then visualizes key information such as the target's movement trajectory, the distribution of surrounding vehicles, and predicted locations on a GIS map. The command system can also issue commands to the system server, such as retrieving real-time video from vehicles. The system server forwards the commands to the vehicle-mounted terminal, which then pushes real-time images or video streams from outside the vehicle to the system server, thereby displaying the real-time dynamics of target tracking on the command system.

[0090] In this embodiment of the invention, the accuracy and efficiency of target detection are improved by preprocessing images on the vehicle terminal and optimizing the YOLOv8 model. Combined with the mechanisms of multi-vehicle collaborative scheduling, dual target loss judgment and prediction area screening, the problems of computing power overload and target tracking interruption are effectively solved. At the same time, the tracking dynamic visualization is realized by linking with the command system, ensuring the continuous, accurate and efficient control of target tracking.

[0091] See Figure 2 This invention provides a dynamic target tracking device 2, comprising: The data acquisition module 21 is used to acquire vehicle information of multiple first vehicles and target detection results obtained by recognizing the external images of multiple first vehicles. The mapping table generation module 22 is used to determine the first vehicle that detected the target based on the vehicle information of each first vehicle and the target detection results, and generate a vehicle-target mapping table; The task allocation module 23 is used to determine the task vehicle corresponding to the target according to the vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle. The task coordination module 24 is used to determine multiple second vehicles based on the latest detected target feature information and target movement information of the task vehicle if the task vehicle loses the target, and update the vehicle-target mapping table based on the vehicle information of the multiple second vehicles and the target detection results. The task reassignment module 25 is used to redetermine the task vehicle corresponding to the target based on the updated vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle.

[0092] In one possible implementation, the mapping table generation module 22 is used to determine the first vehicle that detected the target based on the detection matching degree of each first vehicle; and associate the vehicle ID, vehicle location, computing power load, detection matching degree and detection time of the first vehicle that detected the target with the target to obtain a vehicle-target mapping table.

[0093] In one possible implementation, the task allocation module 23 is used to determine the first vehicle as the task vehicle when there is only one first vehicle in the vehicle-target mapping table; when there are two or more first vehicles in the vehicle-target mapping table, the first vehicle with the highest detection matching degree is determined as the task vehicle.

[0094] In one possible implementation, the dynamic target tracking device 2 is used to determine, according to the preset target priority, the vehicle as the task vehicle of the highest priority target among the multiple targets if the computing power load of the task vehicle exceeds the preset load saturation threshold and the task vehicle is the task vehicle corresponding to two or more targets at the same time; for each other target among the multiple targets, other vehicles that have detected the other target are selected in the vehicle-target mapping table of the other target as the task vehicle of the other target.

[0095] In one possible implementation, the task coordination module 24 is used to determine the target's predicted area based on the latest detected target feature information and target movement information of the task vehicle, combined with road data; and to identify vehicles within the predicted area as second vehicles.

[0096] In one possible implementation, the dynamic target tracking device 2 is used to determine that the task vehicle has lost the target if the task vehicle does not detect the target in any of the consecutive preset frames of vehicle exterior images, and the distance between the current position of the task vehicle and the last detected target position is greater than a preset distance threshold.

[0097] In this embodiment of the invention, the modules of the dynamic target tracking device 2 work together to integrate multi-vehicle data to generate a vehicle-target mapping table, dynamically allocate and flexibly adjust task vehicles, and can efficiently cope with scenarios such as target loss. Relying on a precise vehicle screening and task allocation mechanism, it achieves continuous tracking of dynamic targets, significantly improving the overall tracking system's operational quality and adaptability.

[0098] See Figure 3 The diagram shows a schematic of the electronic device 3 provided in an embodiment of the present invention, which is described in detail below: like Figure 3 As shown, the electronic device 3 in this embodiment includes a processor 30 and a memory 31. The memory 31 stores a computer program 32. When the processor 30 executes the computer program 32, it implements the steps in the various method embodiments described above. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module in the various device embodiments described above.

[0099] For example, computer program 32 may be divided into one or more modules / units, which are stored in memory 31 and executed by processor 30 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 32 in electronic device 3.

[0100] Electronic device 3 may include, but is not limited to, processor 30 and memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 3 may also include input / output devices, network access devices, buses, etc.

[0101] The processor 30 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0102] The memory 31 can be an internal storage unit of the electronic device 3, such as a hard disk or memory of the electronic device 3. The memory 31 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 3. Furthermore, the memory 31 can include both internal and external storage units of the electronic device 3. The memory 31 is used to store the computer program 32 and other programs and data required by the electronic device 3. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0103] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0104] This invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0105] This invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the methods described in the above-described method embodiments.

[0106] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0107] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0108] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A dynamic target tracking method, characterized in that, include: Acquire vehicle information of multiple first vehicles, and target detection results obtained by recognizing the external images of the multiple first vehicles; Based on the vehicle information of each first vehicle and the target detection results, the first vehicle that detected the target is determined, and a vehicle-target mapping table is generated; Based on the vehicle-target mapping table, determine the task vehicle corresponding to the target, and obtain the target feature information and target movement information sent by the task vehicle; If the mission vehicle loses the target, then based on the latest detected target feature information and target movement information of the mission vehicle, multiple second vehicles are determined, and the vehicle-target mapping table is updated based on the vehicle information of the multiple second vehicles and the target detection results. Based on the updated vehicle-target mapping table, the task vehicle corresponding to the target is re-determined, and the target feature information and target movement information sent by the task vehicle are obtained.

2. The dynamic target tracking method according to claim 1, characterized in that, The target detection results include detection matching degree and detection time; the vehicle information includes vehicle ID, vehicle location, and computing power load. The step of determining the first vehicle that detected the target based on the vehicle information of each first vehicle and the target detection results, and generating a vehicle-target mapping table, includes: Based on the detection matching degree of each first vehicle, determine the first vehicle that detected the target; The vehicle ID, vehicle location, computing power load, detection matching degree, and detection time of the first vehicle that detected the target are associated with the target to obtain the vehicle-target mapping table.

3. The dynamic target tracking method according to claim 2, characterized in that, The step of determining the mission vehicle corresponding to the target based on the vehicle-target mapping table includes: When there is only one first vehicle in the vehicle-target mapping table, the first vehicle is determined to be the mission vehicle; When there are two or more first vehicles in the vehicle-target mapping table, the first vehicle with the highest detection matching degree is determined as the task vehicle.

4. The dynamic target tracking method according to claim 3, characterized in that, The method further includes: If the computing power load of the task vehicle exceeds the preset load saturation threshold, and the task vehicle is simultaneously the task vehicle for two or more targets, then according to the preset target priority, the vehicle is determined to be the task vehicle for the highest priority target among the multiple targets. For each of the multiple targets, in the vehicle-target mapping table of that other target, select the other vehicle that detected that other target as the task vehicle for that other target.

5. The dynamic target tracking method according to any one of claims 1 to 4, characterized in that, If the mission vehicle loses the target, then based on the latest detected target feature information and target movement information of the mission vehicle, multiple second vehicles are determined, including: Based on the latest target feature information and target movement information detected by the mission vehicle, combined with road data, the predicted area of ​​the target is determined; Vehicles located within the predicted area are identified as the second vehicle.

6. The dynamic target tracking method according to any one of claims 1 to 4, characterized in that, The method for determining whether the mission vehicle has lost the target includes: If the mission vehicle does not detect a target in any of the consecutive preset frames of exterior images, and the distance between the current position of the mission vehicle and the last detected target position is greater than a preset distance threshold, then it is determined that the mission vehicle has lost the target.

7. The dynamic target tracking method according to any one of claims 1 to 4, characterized in that, The target movement information includes the current movement speed, current movement direction, and continuous movement trajectory.

8. A dynamic target tracking device, characterized in that, include: The data acquisition module is used to acquire vehicle information of multiple first vehicles, as well as target detection results obtained by recognizing the external images of the multiple first vehicles; The mapping table generation module is used to determine the first vehicle that detected the target based on the vehicle information of each first vehicle and the target detection results, and generate a vehicle-target mapping table. The task allocation module is used to determine the task vehicle corresponding to the target according to the vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle; The task coordination module is used to determine multiple second vehicles based on the latest detected target feature information and target movement information of the task vehicle if the task vehicle loses the target, and update the vehicle-target mapping table based on the vehicle information of the multiple second vehicles and the target detection results. The task reassignment module is used to redetermine the task vehicle corresponding to the target based on the updated vehicle-target mapping table, and to obtain the target feature information and target movement information sent by the task vehicle.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images