Target tracking method and device, storage medium and cloud computing platform

Abstract

Embodiments of the present application provide a target tracking method and device, a storage medium and a cloud computing platform. The method comprises: acquiring point information of a plurality of single points; generating a fusion recognition result corresponding to each point information according to the plurality of point information; and generating a tracking result according to a preset point division result and the plurality of fusion recognition results. Thus, the cloud computing platform can fuse the information of a single point, fuse the fused information of a plurality of single points again, and continuously track a target object across a plurality of points when the target object moves across the plurality of points.

Description

[0001] Embodiments of the present application relate to the technical field of target tracking, and in particular to a target tracking method and device, a storage medium and a cloud computing platform.

[0002] In the related art, the road side radar data and video data at the same single point are fused by a computing unit to achieve a unique perception ID number for the same target object. If the trajectory of the target object crosses multiple single points, the same target object has a unique perception ID number at different single points, but the multiple perception ID numbers are not the same, which results in the inability to obtain the running trajectory of the target object crossing multiple single points.

[0003] Therefore, embodiments of the present application provide a target tracking method, device, storage medium and cloud computing platform to solve the problem of the inability to obtain the running trajectory of the target object crossing multiple single points in the prior art.

[0004] In a first aspect, embodiments of the present application provide a target tracking method, comprising:

[0005] obtaining point information of multiple single points;

[0006] generating a fusion recognition result corresponding to each point information according to multiple point information;

[0007] generating a tracking result according to a preset point division result and multiple fusion recognition results.

[0008] In a possible implementation, the generating a fusion recognition result corresponding to each point information according to multiple point information comprises:

[0009] when the point information includes a camera information, identifying the camera information by an identification algorithm to generate an identification result, and taking the identification result as the fusion recognition result; or

[0010] when the point information includes camera information and radar information, identifying the camera information by an identification algorithm to generate an identification result, generating a single point fusion result according to the identification result and the radar information by a specific fusion algorithm and a distance threshold, and generating the fusion recognition result according to the single point fusion result and the identification result; or

[0011] ​​​When the point information includes multiple camera information, the multiple camera information is identified by an identification algorithm to generate an identification result corresponding to each camera information; the camera fusion identification result is generated from the multiple identification results by a specific fusion algorithm; and the fusion identification result is generated from the camera fusion identification result and the multiple identification results.

[0012] In a possible implementation, the single point includes a continuous area point or a non-continuous area point, the point division result includes at least one continuous area point and / or at least one non-continuous area point corresponding to each single point, and the tracking result includes a continuous cross-point result corresponding to the multiple continuous area points and a non-continuous cross-point result corresponding to the multiple non-continuous area points; and the generation of the tracking result from the preset point division result and the multiple fusion identification results includes:

[0013] According to the point division result, a target continuous area point corresponding to the continuous area point is obtained.

[0014] According to the fusion identification result corresponding to the continuous area point and the fusion identification result corresponding to the target continuous area point, the continuous cross-point result is generated.

[0015] According to the point division result and the obtained predicted trajectory, a target non-continuous area point corresponding to each non-continuous area point is obtained.

[0016] According to the fusion identification result corresponding to the non-continuous area point and the fusion identification result corresponding to the target non-continuous area point, the non-continuous cross-point result is generated.

[0017] The tracking result is generated from the at least one continuous cross-point result and the at least one non-continuous cross-point result.

[0018] In a possible implementation, the fusion identification result includes a camera fusion result, and the camera fusion result includes target coordinates and appearance features of at least one target object; and the generation of the continuous cross-point result from the fusion identification result corresponding to the continuous area point and the fusion identification result corresponding to the target continuous area point includes:

[0019] From the first camera fusion result corresponding to the continuous area point, a first target coordinate of a first target object in an overlapping area is found out by a pre-divided overlapping area.

[0020] From the second camera fusion result corresponding to the target continuous area point, a second target coordinate of a second target object in the overlapping area is found out by the overlapping area.

[0021] determine the second target object corresponding to the first target object according to the first target coordinate of the at least one first target object and the second target coordinate of the at least one second target object by a bidirectional matching target handover method;

[0022] generate a first matching result according to a first appearance feature of the first target object and a second appearance feature of the second target object corresponding to the first target object;

[0023] generate the continuous cross-point result according to the first matching result, the first camera fusion result and the second camera fusion result.

[0024] In a possible implementation, the first appearance feature includes a first appearance image, and the second appearance feature includes a second appearance image; and the generating the first matching result according to the first appearance feature of the first target object and the second appearance feature of the second target object corresponding to the first target object includes:

[0025] generating a first color model in HSV format according to the first appearance image;

[0026] generating a second color model in HSV format according to the second appearance image;

[0027] generating a target similarity according to the first color model and the second color model;

[0028] generating the first matching result according to the target similarity.

[0029] In a possible implementation, the fusion recognition result includes a camera fusion result, the camera fusion result includes an appearance feature of at least one target object, and the generating the non-continuous cross-point result according to the fusion recognition result corresponding to the non-continuous area point and the fusion recognition result corresponding to the target non-continuous area point includes:

[0030] generating a second matching result corresponding to the third target object and the fourth target object according to a third appearance feature of the third target object corresponding to the non-continuous area point and a fourth appearance feature of the fourth target object corresponding to the target non-continuous area point;

[0031] generating the non-continuous cross-point result according to the second matching result, a third camera fusion result corresponding to the third target object and a fourth camera fusion result corresponding to the fourth target object.

[0032] In a possible implementation, the tracking result includes a display number of at least one target object, and after the generating the tracking result, the method further includes:

[0033] acquire a driving feature of the target object;

[0034] generate a time threshold according to the driving feature;

[0035] determine whether a second target object or a fourth target object matching the target object exists within the time threshold;

[0036] if the second target object or the fourth target object matching the target object does not exist, delete the display number corresponding to the target object.

[0037] In a second aspect, an embodiment of the present application provides a target tracking device, comprising:

[0038] a first acquisition module configured to acquire point information of a plurality of single points;

[0039] a first generation module configured to generate a fusion recognition result corresponding to each point information according to the plurality of point information;

[0040] a second generation module configured to generate a tracking result according to a preset point division result and the plurality of fusion recognition results.

[0041] In a third aspect, an embodiment of the present application provides a storage medium, which comprises a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the target tracking method in the first aspect or any possible implementation manner of the first aspect.

[0042] In a fourth aspect, an embodiment of the present application provides a cloud computing platform, comprising a memory and a processor, wherein the memory is configured to store information comprising program instructions, and the processor is configured to control the execution of the program instructions, and when the program instructions are loaded and executed by the processor, the steps of the target tracking method in the first aspect or any possible implementation manner of the first aspect are implemented.

[0043] In the technical scheme of the target tracking method, device, storage medium and cloud computing platform provided by the embodiment of the present application, the point information of a plurality of single points is acquired; the fusion recognition result corresponding to each point information is generated according to the plurality of point information; and the tracking result is generated according to the preset point division result and the plurality of fusion recognition results, so that the cloud computing platform can realize the fusion of the information of the single points, the fusion of the fused information of the plurality of single points, and the multi-point cross-domain continuous tracking of the target object when the target object moves across a plurality of points. BRIEF DESCRIPTION OF DRAWINGS

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 A flowchart illustrating a target tracking method provided in an embodiment of the present invention;

[0046] Figure 2 A point location diagram provided for an embodiment of the present invention;

[0047] Figure 3 A tracking schematic diagram provided for an embodiment of the present invention;

[0048] Figure 4 A schematic diagram illustrating an updated set of appearance features provided in an embodiment of the present invention;

[0049] Figure 5 A schematic diagram of the structure of a target tracking device provided in an embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram of a cloud computing platform provided in an embodiment of the present invention.

Detailed Implementation Methods

[0051] To better understand the technical solution of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0052] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0053] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0054] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0055] It should be understood that, although the terms first, second, third, etc. can be employed in describing various elements in the embodiments of the present application, these elements should not be limited to these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element without departing from the scope of the embodiments of the present application.

[0056] The word "if" can be interpreted as meaning "when" or "upon" or "in response to a determination" or "in response to a detection," as used herein, depending on the context. Similarly, the phrase "if it is determined" or "if a detected (stated condition or event)" can be interpreted as meaning "upon a determination" or "in response to a determination" or "when a detected (stated condition or event)" or "in response to a detected (stated condition or event)," depending on the context.

[0057] Figure 1 A flowchart of a target tracking method provided by an embodiment of the present application is shown in FIG. 1, which includes the following steps. Figure 1

[0058] In step 101, a cloud computing platform acquires point information of a plurality of single points.

[0059] In an embodiment of the present application, sensors can be arranged on roads to record behaviors of traffic participants, for example, the sensors are devices with recording functions such as cameras, radars, etc. The radars include laser radars or millimeter wave radars. The cloud computing platform can pre-divide a geographical area where the sensors are located into one or more single points according to the distribution of the sensors. Different single points can be provided with different sensors, and each single point includes at least one sensor.

[0060] However, each single point includes a camera, for example, the single point includes at least one camera, and the point information includes at least one camera information. The single point corresponding to a relatively complex intersection can also be provided with a laser radar and / or a millimeter wave radar, for example, the single point includes at least one camera and at least one laser radar, and the point information includes at least one camera information and at least one laser radar information; or the single point includes at least one camera and at least one millimeter wave radar, and the point information includes at least one camera information and at least one millimeter wave radar information; or the single point includes at least one camera, at least one laser radar and at least one millimeter wave radar, and the point information includes at least one camera information, at least one laser radar information and at least one millimeter wave radar information.

[0061] Figure 2 A point diagram provided by an embodiment of the present application is shown in FIG. 2, in which each frame represents one single point, Figure 2 Figure 2 ​​The diagram shows eight single-point locations. Borders 1, 3, 6, and 8 each contain only cameras, recording road conditions purely visually. Borders 2, 4, and 5 each include both cameras and millimeter-wave radar, using a combination of radar and visual methods to record road conditions. Border 7 includes a camera, lidar, and millimeter-wave radar, employing three sensors to record the conditions of pedestrian crossings and lanes. Cameras collect video data, including information such as the appearance, lane, and location of traffic participants. Millimeter-wave radar is unaffected by environmental conditions such as rain, fog, dust, and snow, and collects millimeter-wave radar data, including parameters such as the position, height, speed, posture, and shape of traffic participants. Lidar collects data by emitting laser beams, including parameters such as the position, height, speed, posture, and shape of traffic participants, treating them as target objects.

[0062] Figure 3 A tracking diagram provided for an embodiment of the present invention, such as... Figure 3 As shown, each individual point may include at least one sensor, but each individual point must include a camera. The LiDAR and millimeter-wave radar within the dashed boxes are optional sensors. Point information from multiple points is uploaded to a cloud computing platform. This platform includes a fusion perception algorithm calculation module, which comprises a feature extraction module, a feature matching module, a feature union processing module, a feature storage module, and a vehicle unique ID continuous tracking module. This enables the recording of vehicle trajectory information, visualization of the trajectory, and cooperative autonomous driving.

[0063] However, at least two types of sensors are typically deployed only in busy or complex road sections, such as areas with more than four lanes in both directions and high traffic volume. In these cases, a combination of radar-visual fusion and pure vision sensors is usually deployed alternately. Figure 2 As shown, borders 1 to 4 employ an alternating deployment of radar-visual fusion and pure visual methods. Borders 1 and 3 use a pure visual method, while borders 2 and 4 use a radar-visual fusion method. When traffic flow is very low, a continuous pure visual method can be used.

[0064] Step 102: The cloud computing platform generates the fusion recognition result corresponding to each location information based on multiple location information.

[0065] In the embodiment of the present application, the fusion recognition result includes a fusion result and a recognition result. When a single point position includes one camera, the cloud computing platform takes the recognition result of the camera information as the fusion recognition result. When the single point position includes multiple sensors, the cloud computing platform fuses the information collected by the multiple sensors to generate a fusion result, and takes the fusion result as the fusion recognition result. Thus, the identification and tracking of multiple target objects of at least one sensor of the single point position can be realized.

[0066] If the single point position includes multiple sensors, the multiple sensors include multiple cameras, or the multiple sensors include at least one camera and at least one radar device. As shown in FIG. 1, the frame 1 includes only one camera, the cloud computing platform identifies the camera information to generate a recognition result; the frame 2 includes a camera and a millimeter wave radar, and the cloud computing platform fuses the camera information and the millimeter wave radar information to generate a fusion result. Figure 2

[0067] In step 103, the cloud computing platform generates a tracking result according to the preset point position division result and the multiple fusion recognition results.

[0068] In the embodiment of the present application, the single point position includes a continuous area point position or a non-continuous area point position. Since each single point position includes a camera, if the cameras of adjacent single point positions have an overlapping area, the single point positions of the adjacent cameras with the overlapping area are taken as the continuous area point positions; if the cameras of adjacent single point positions do not have an overlapping area, the single point positions of the adjacent cameras without the overlapping area are taken as the non-continuous area point positions. The point position division result includes at least one continuous area point position and / or at least one non-continuous area point position corresponding to each single point position, and the tracking result includes a continuous cross-point position result corresponding to the multiple continuous area point positions and a non-continuous cross-point position result corresponding to the multiple non-continuous area point positions. Thus, the identification and tracking of the same target object across multiple point positions can be realized.

[0069] The single point position can be both a continuous area point position and a non-continuous area point position. As shown in FIG. 1, if the camera shown in the frame 3 and the camera shown in the frame 7 have a common overlapping shooting area, the single point position corresponding to the frame 3 is a continuous area point position, and the single point position corresponding to the frame 7 is a target continuous area point position. As shown in FIG. 1, if the camera shown in the frame 3 and the camera shown in the frame 4 do not have a common overlapping shooting area, the single point position corresponding to the frame 3 is a non-continuous area point position, and the single point position corresponding to the frame 7 is a target non-continuous area point position. Figure 2 Figure 2

[0070] ​​​The embodiment of the present application provides a tracking method of a target, a cloud computing platform acquires point information of a plurality of single points; generates a fusion recognition result corresponding to each point information according to the plurality of point information; generates a tracking result according to a preset point division result and the plurality of fusion recognition results, so that the cloud computing platform can realize fusion of information of the single point, fusion of the fused information of the plurality of single points again, and multi-point cross-domain continuous tracking of a target object when the target object moves across a plurality of points.

[0071] In a possible implementation manner, the step 101 comprises: the cloud computing platform collects the point information of the single point based on a unified synchronization time.

[0072] In the embodiment of the present application, each single point can include at least one sensor, and data and coding output by different sensors are different, so that the cloud computing platform needs to synchronize the time of the sensors in time to ensure the data fusion result of the single point, so that the time of collecting data by the sensors is consistent. When the cloud computing platform collects the point information of the single point, the cloud computing platform needs to add a time stamp to the data collected by the sensors, for example, the video data collected by the camera includes a plurality of frames of data, and each frame of data corresponds to a collection time. The cloud computing platform comprises a network time protocol (NTP) server, and the NTP server time-ensures the sensors corresponding to each single point before the cloud computing platform collects the point data of the plurality of single points, so as to ensure the time synchronization of the plurality of single points.

[0073] In a possible implementation manner, when the point information comprises camera information, the step 102 comprises: the cloud computing platform identifies the camera information by using an identification algorithm to generate an identification result; and the identification result is taken as the fusion recognition result.

[0074] In the embodiment of the present application, the camera information includes at least one target object, and the recognition result includes at least one of the detection frame, the category recognition result, the first confidence, the first identification number, the lane, the pixel coordinates based on the pixel coordinate system, the image coordinates based on the image coordinate system, the camera relative coordinates based on the camera coordinate system, and the world coordinates based on the world coordinate system of the at least one target object. For example, the cloud computing platform identifies the category of the at least one target object according to the obtained data set and the camera information by using the recognition algorithm, and generates the category recognition result and the first confidence of the target object. The world coordinate system refers to a coordinate system based on the real latitude and longitude, and the camera coordinate system refers to a coordinate system with the camera as the origin. Since the camera information includes at least one frame of image, each image includes a plurality of pixels, when the cloud computing platform identifies the target object, the pixel coordinates of the target object based on the pixel coordinate system are first obtained; then the pixel coordinates are converted into the image coordinates based on the image coordinate system; then the image coordinates are converted into the camera relative coordinates based on the camera coordinate system; and finally, the camera relative coordinates are converted into the world coordinates based on the actual latitude and longitude.

[0075] The recognition algorithm includes a YOLO v5 algorithm, and the cloud computing platform can realize the identification, tracking and positioning of the target object by using the YOLO v5 algorithm. When identifying the target object, the cloud computing platform generates a plurality of detection frames, the first confidence corresponding to each detection frame and the category recognition result of the same target object by using the YOLO v5 algorithm; arranges the first confidences from high to low according to the scores of the plurality of first confidences; takes the first confidence with the highest score as the first confidence of the target object, and takes the category recognition result corresponding to the first confidence with the highest score as the category recognition result of the target object. The YOLO v5 algorithm can also be used to track the target object, and the cloud computing platform assigns a unique first identification number to each identified target object. If the target object appears in a plurality of images, the first identification number of the same target object in each image is the same.

[0076] In a possible implementation, when the point information includes the camera information and the radar information, the step 102 includes: step 1022, the cloud computing platform identifies the camera information by using the recognition algorithm to generate a recognition result; step 1023, the cloud computing platform generates a single-point fusion result according to the recognition result and the radar information by using the obtained distance threshold and a specific fusion algorithm; and step 1024, the cloud computing platform generates a fusion recognition result according to the single-point fusion result and the recognition result.

[0077] In the embodiment of the present application, the step 1022 can refer to the step 1021 described above.

[0078] In step 1023, the radar information includes lidar information and / or millimeter wave radar information. The radar information includes at least one target object in the radar coordinate system, world coordinates in the world coordinate system, object speed, heading, three-dimensional size, second confidence, and the like. The cloud computing platform pre-establishes a radar coordinate system with the radar device as the origin, converts the radar coordinates in the radar coordinate system into world coordinates in the world coordinate system, thereby realizing the alignment of the target objects of different sensors in the same coordinate system. The specific fusion algorithm can be a fusion algorithm based on the data of multiple sensors in the same single point. The specific fusion algorithm includes a radar-vision fusion algorithm, a light-vision fusion algorithm, a radar-vision-light fusion algorithm, or a pure vision fusion algorithm, and the cloud computing platform can select different specific fusion algorithms according to the types of sensors included in the single point. The single point fusion result includes the fusion result of the target objects in the overlapping detection area of multiple sensors.

[0079] The sensor can correspond to a distance perception threshold, and the distance perception threshold is the maximum value of the range of data that the sensor can collect. Among the distance perception thresholds corresponding to the multiple sensors, the distance threshold is equal to the distance perception threshold with the smallest value. For example, the distance perception threshold of the camera is 200 meters, the distance perception threshold of the 32-line lidar is 150 meters, and the distance perception threshold of the millimeter wave radar is 400 meters. When the single point includes a camera, a 32-line lidar, and a millimeter wave radar, the distance perception threshold with the smallest value is 150 meters, and the distance threshold is 150 meters. The cloud computing platform filters at least one first single point target object within the distance threshold from the recognition result according to the distance threshold; filters at least one second single point target object within the distance threshold from the radar information according to the distance threshold; and determines the second single point target object corresponding to the first single point target object based on the world coordinate system, according to the first world coordinates of the at least one first single point target object and the second world coordinates of the at least one second single point target object. Based on the world coordinate system, the target objects in the recognition result and the target objects in the radar information can be matched, the fusion of the recognition information and the radar information of the matched target objects is realized, and the single point fusion result is generated.

[0080] The single point fusion result includes a fusion confidence, the recognition information includes a first confidence, and the radar information includes a second confidence. Step 1023 includes: the cloud computing platform generates a fusion confidence according to the first confidence, the second confidence, and the fusion coefficient of the different sensors obtained.

[0081] In the embodiment of the present application, the cloud computing platform assigns fusion coefficients to different sensors according to the accuracy of sensors in identification, and the sum of the fusion coefficients of multiple sensors of a single point is 1. The second confidence level includes a millimeter wave radar confidence level and / or a laser radar confidence level. The cloud computing platform calculates a first fusion confidence parameter according to the first confidence level and the fusion coefficient corresponding to the first confidence level, and calculates a second fusion confidence parameter according to the second confidence level and the fusion coefficient corresponding to the second confidence level; and takes the sum of the first fusion confidence parameter and the second fusion confidence as the fusion confidence.

[0082] For example, for the same target object, the first confidence level is 75%, the first confidence level corresponds to a camera, the fusion coefficient of the camera is 0.4, and the fusion coefficient corresponding to the first confidence level is 0.4; the millimeter wave radar confidence level is 60%, the millimeter wave radar confidence level corresponds to a millimeter wave radar, the fusion coefficient of the millimeter wave radar is 0.6, and the fusion coefficient corresponding to the millimeter wave radar confidence level is 0.6, and the fusion confidence is 75%*0.4+85%*0.6=0.81.

[0083] In step 1024, the cloud computing platform updates the camera information according to the fusion result of the single point to generate camera fusion information. The camera fusion information includes at least one of the detection frame of the target object, the category identification result, the fusion confidence, the object number, the lane, the pixel coordinates based on the pixel coordinate system, the image coordinates based on the image coordinate system, the camera relative coordinates based on the camera coordinate system, the world coordinates based on the world coordinate system, the object speed, the heading, and the three-dimensional size. The cloud computing platform can record multiple target movement coordinates according to the movement of the same target object to generate a target movement coordinate set, and the target movement coordinates include at least one of the pixel coordinates based on the pixel coordinate system, the image coordinates based on the image coordinate system, the camera relative coordinates based on the camera coordinate system, and the world coordinates based on the world coordinate system of the target object.

[0084] In a possible implementation, when the point information includes multiple camera information, step 102 includes: step 1025, the cloud computing platform identifies each camera information through an identification algorithm to generate an identification result corresponding to each camera information; step 1026, the cloud computing platform generates camera fusion identification results according to multiple identification results through a specific fusion algorithm; and step 1027, the cloud computing platform generates a fusion identification result according to the camera fusion identification result and the multiple identification results.

[0085] In the embodiment of the present application, step 1025 can refer to step 1021 described above. In step 1026, the specific fusion algorithm includes a pure vision fusion algorithm. The cloud computing platform performs data fusion on the different camera data of the single point position according to the pure vision fusion algorithm to achieve the purpose of data deduplication. In step 1027, the cloud computing platform updates the recognition result according to the camera fusion recognition result to generate a camera fusion result.

[0086] In a possible implementation, the fusion recognition result includes an appearance feature set of at least one target object. After step 1021, step 1024, or step 1027, the cloud computing platform further generates an appearance feature set of at least one target object according to the recognition result; and saves the appearance feature set of at least one target object. The fusion recognition result includes an object number of at least one target object, and the second identification number of the target object can be included in the radar information; and after step 1021, step 1024, or step 1027, the cloud computing platform further generates an object number according to the matched first identification number and second identification number of the target object, so that the same target object in different sensors at a single point position has a unique object number; and saves the object number of the target object.

[0087] In the embodiment of the present application, the appearance feature set includes at least one appearance feature of the target object and an appearance feature identification number corresponding to each appearance feature.

[0088] In a possible implementation, the single point position includes a continuous area point position or a non-continuous area point position, the point position division result includes at least one continuous area point position and / or at least one non-continuous area point position corresponding to each single point position, and the tracking result includes a continuous cross-point position result corresponding to a plurality of continuous area point positions and a non-continuous cross-point position result corresponding to a plurality of non-continuous area point positions.

[0089] In the embodiment of the present application, the single point position can be both a continuous area point position and a non-continuous area point position. As shown in Figure 2 If the camera shown by the frame 3 and the camera shown by the frame 7 have a common overlapping shooting area, the single point position corresponding to the frame 3 is a continuous area point position, and the single point position corresponding to the frame 7 is a target continuous area point position. As shown in Figure 2 If the camera shown by the frame 3 and the camera shown by the frame 4 do not have a common overlapping shooting area, the single point position corresponding to the frame 3 is a non-continuous area point position, and the single point position corresponding to the frame 7 is a target non-continuous area point position.

[0090] In step 103, step 1031 can specifically include: acquiring, by the cloud computing platform, a target continuous area point position corresponding to the continuous area point position according to the point position division result.

[0091] In the embodiment of the present application, the target continuous area point is a point corresponding to the continuous area point in the plurality of points.

[0092] In step 1032, the cloud computing platform generates a continuous cross-point result according to the fusion recognition result corresponding to the continuous area point and the fusion recognition result corresponding to the target continuous area point.

[0093] In the embodiment of the present application, the fusion recognition result includes a camera fusion result. When a single point includes only one camera, the recognition result is taken as the camera fusion result. The camera fusion result includes target coordinates, appearance features, and object numbers of at least one target object. Step 1032 can specifically include:

[0094] In step 1032A, the cloud computing platform finds out first target coordinates of a first target object in the overlapping area from the first camera fusion result corresponding to the continuous area point through the pre-divided overlapping area.

[0095] In the embodiment of the present application, the cloud computing platform can pre-divide an overlapping area in the shooting area of the camera. The overlapping area refers to an area that can be simultaneously detected by the sensor of the continuous area point and the sensor of the target continuous area point. The first camera fusion result is the camera fusion result corresponding to the continuous area point. The first target object is a target object in the overlapping area of the first camera fusion result. The first target coordinates are pixel coordinates, image coordinates, or camera relative coordinates of the first target object.

[0096] In step 1032B, the cloud computing platform finds out second target coordinates of a second target object in the overlapping area from the second camera fusion result corresponding to the target continuous area point through the overlapping area.

[0097] In the embodiment of the present application, the second target object is a target object in the overlapping area. The second target coordinates are pixel coordinates, image coordinates, or camera relative coordinates of the second target object.

[0098] In step 1032C, the cloud computing platform determines the second target object corresponding to the first target object according to the first target coordinates of the at least one first target object and the second target coordinates of the at least one second target object through the bidirectional matching target handover method.

[0099] In the embodiment of the present application, the two-way matching target handover method refers to a coordinate transformation relationship model between sensors of different point locations. For example, the first target coordinate is a first camera relative coordinate of a camera coordinate system based on a continuous area point location, the second target object is a second camera relative coordinate of a camera coordinate system based on a target continuous area point location, and the cloud computing platform can convert the first target coordinate into a third camera relative coordinate of a camera coordinate system based on the target continuous area point location through the two-way matching target handover method. Whether the third camera relative coordinate matches the second camera relative coordinate is determined. If it is determined that the third camera relative coordinate matches the second camera relative coordinate, it is determined that the first target object corresponding to the third camera relative coordinate corresponds to the second target object corresponding to the second camera relative coordinate. Wherein, whether the third camera relative coordinate matches the second camera relative coordinate refers to whether the distance between the third camera relative coordinate and the second camera relative coordinate exceeds an error range.

[0100] In step 1032D, the cloud computing platform generates a first matching result according to the first appearance feature of the first target object and the second appearance feature of the second target object corresponding to the first target object.

[0101] In the embodiment of the present application, the first matching result includes matching or not matching. The first appearance feature includes a first appearance image, and the second appearance feature includes a second appearance image.

[0102] The cloud computing platform performs step 1032D to determine whether the target objects at different point locations are the same target object. If the target object is a vehicle, before step 1032D1, the cloud computing platform further includes: determining whether the first appearance image includes a first license plate identifier, and determining whether the second appearance image includes a second license plate identifier; if it is determined that the first appearance image includes the first license plate identifier and the second appearance image includes the second license plate identifier, it is determined whether the first license plate identifier and the second license plate identifier are the same; if it is determined that the first license plate identifier and the second license plate identifier are the same, it is determined that the first matching result is matching; if it is determined that the first license plate identifier and the second license plate identifier are not the same, it is determined that the first matching result is not matching; if it is determined that the first appearance image does not include the first license plate identifier and / or the second appearance image does not include the second license plate identifier, step 1032D1 is performed. If the target object is not a vehicle, the cloud computing platform performs step 1032D1.

[0103] The first appearance image includes the color, texture, contour, etc. of the first target object. The second appearance image includes the color, texture, contour, etc. of the second target object. The cloud computing platform can measure the similarity of the appearance of different target objects according to the feature histogram, to ensure the accuracy of target handover.

[0104] Step 1032D includes: step 1032D1, the cloud computing platform generates a first color model in HSV format according to the first appearance image.

[0105] In the embodiment of the application, the camera collects an image in RGB format based on a red, green, blue (RGB) color model. The cloud computing platform can convert the RGB color model into an HSV (Hue, Saturation, Value) color model, and convert the image in RGB format into a first color model in HSV format.

[0106] Step 1032D2, the cloud computing platform generates a second color model in HSV format according to the second appearance image.

[0107] In the embodiment of the application, step 1032D2 can refer to step 1032D1 described above.

[0108] Step 1032D3, the cloud computing platform generates a target similarity according to the first color model and the second color model.

[0109] In the embodiment of the application, the cloud computing platform compares the similarity of the first target object and the second target object based on three visual features of hue, saturation and brightness, and obtains the target similarity.

[0110] Step 1032D4, the cloud computing platform generates a first matching result according to the target similarity.

[0111] In the embodiment of the application, the cloud computing platform can determine whether the first target object and the second target object match according to the target similarity.

[0112] Step 1032E, the cloud computing platform generates a continuous cross-point result according to the first matching result, a first camera fusion result and a second camera fusion result.

[0113] In the embodiment of the application, the camera fusion result includes an object number and a target moving coordinate set. The first camera fusion result includes a first camera fusion result and a first target moving coordinate set, and the second camera fusion result includes a second object number and a second target moving coordinate set. If the first target object and the second target object match, it means that the first target object and the second target object are the same target object, the cloud computing platform assigns the same number to the same target object, determines the display number of the target object as the first object number or the second object number, and the cloud computing platform can determine the first object number or the second object number according to the generation time. The cloud computing platform saves the display number of the target object to the first fusion recognition result and the second fusion recognition result. The cloud computing platform updates the target moving coordinate set of the target object according to the first target moving coordinate set and the second target moving coordinate set.

[0114] In step 1033, the cloud computing platform obtains a target non-continuous region point corresponding to each non-continuous region point according to the point division result and the obtained predicted trajectory.

[0115] In the embodiment of the application, before step 1033, the cloud computing platform further obtains the driving feature of the target object; and generates the predicted trajectory according to the driving feature corresponding to the third target object. The driving feature includes at least one of the plurality of target movement coordinates of the target object and the object speed. The object speed of the target object can be the object speed in the radar information, or can be calculated by the cloud computing platform according to the plurality of target movement coordinates and the time point corresponding to the target movement coordinates.

[0116] The cloud computing platform fits the transfer time distribution of the target object according to the plurality of target movement coordinates and the object speed of the target object by means of space-time transfer modeling, trajectory fitting function, etc., predicts the trajectory of the target object, and takes the next single point that the target object can reach as the target non-continuous region point according to the predicted trajectory. The cloud computing platform obtains at least one single point that meets the transfer time distribution based on the time stamp and the transfer time distribution, and takes the single point that meets the transfer time distribution as the target non-continuous region point.

[0117] In step 1034, the cloud computing platform generates a non-continuous cross-point result according to the fusion recognition result corresponding to the non-continuous region point and the fusion recognition result corresponding to the target non-continuous region point.

[0118] In the embodiment of the application, in step 1034A, the cloud computing platform generates a second matching result corresponding to the third target object and the fourth target object according to the third appearance feature of the third target object corresponding to the non-continuous region point and the fourth appearance feature of the fourth target object corresponding to the target non-continuous region point; and in step 1034B, the cloud computing platform generates the non-continuous cross-point result according to the second matching result, the third camera fusion result corresponding to the third target object, and the fourth camera fusion result corresponding to the fourth target object. Step 1034A can refer to step 1032D described above, and step 1034B can refer to step 1032E described above.

[0119] In a possible implementation, after step 1032D, if it is determined that the first target object and the second target object match, the appearance feature set corresponding to the first target object is updated according to the first appearance feature of the first target object and the second appearance feature of the second target object. In a possible implementation, after step 1034A, if it is determined that the third target object and the fourth target object match, the appearance feature set corresponding to the third target object is updated according to the third appearance feature of the third target object and the fourth appearance feature of the fourth target object.

[0120] In the embodiment of the present application, the cloud computing platform realizes the target feature perfecting mechanism based on the perception experience sharing of multi-access edge computing (MEC), that is, in the multi-point recognition process, the new appearance features obtained by the target object are saved.

[0121] Figure 4 An updating appearance feature set diagram provided by the embodiment of the present application is shown in FIG. 6. As shown in FIG. 6, it is assumed that 4 features are extracted by the point 1 camera, 5 features are extracted by the point 2 camera, and 4 features are extracted by the point 3 camera. The results obtained by the point 1 and 2 cameras are matched based on the point 1 camera as the reference, and a new number is given to the feature extracted by the point 2 camera but not extracted by the point 1 camera. The results obtained by the point 2 and 3 cameras are matched based on the point 2 camera as the reference, and a new number is given to the feature extracted by the point 3 camera but not extracted by the point 2 camera. Finally, the union of the continuous point camera feature sequences is taken as the appearance feature set of the target object, so that the features of each traffic participant can be gradually perfected in the continuous recognition and storage feature process, and the purpose of cross-point tracking of multiple points is achieved. Figure 4

[0122] Step 1035, the cloud computing platform generates a tracking result according to at least one continuous cross-point result and at least one non-continuous cross-point result.

[0123] In the embodiment of the present application, the cloud computing platform can create multiple threads, each thread corresponding to an execution step of a continuous cross-point result or an execution step of a non-continuous cross-point result, so that the tracking of the trajectories of multiple target objects can be realized.

[0124] In the embodiment of the present application, after step 1035, the cloud computing platform further displays the tracking result. The cloud computing platform displays the picture processed by each camera information. The picture includes at least one of the detection box of the target object, the confidence, the category recognition result, the display number, the trajectory line and other information. The trajectory line can be a digital twin of the movement trajectory of the target object according to the tracking result, and a three-dimensional perspective visual trajectory line of the movement state of the target object is constructed.

[0125] In a possible implementation manner, the tracking result includes the display number of at least one target object, and after step 103, the cloud computing platform further acquires the driving feature of the target object, generates a time threshold value according to the driving feature, judges whether there is a second target object or a fourth target object matching the target object within the time threshold value, and deletes the display number corresponding to the target object if there is no second target object or fourth target object matching the target object. In a possible implementation manner, the tracking result includes the display number of at least one target object, and after step 103, the cloud computing platform further acquires the driving feature of the target object, generates a time threshold value according to the driving feature, judges whether there is a second target object or a fourth target object matching the target object within the time threshold value, and deletes the display number corresponding to the target object if there is no second target object or fourth target object matching the target object.

[0126] In the embodiment of the present application, the driving feature includes at least one of the speed, the heading, and the trajectory information of the object. The cloud computing platform generates a first time according to the average value of the speed of the object and the distance between the two single points; generates a second time according to the rated speed of the lane where the target object is located and the distance between the two single points; and generates a time threshold according to the first time and the second time. For example, the time threshold is the first time, the second time, or the sum of the first time and the second time. The time threshold can also be twice the larger one of the first time and the second time. For example, on a highway, both points are non-continuous area points, and there is an area of about 50 meters between the two points that cannot be photographed. If the time threshold is 2 minutes, and the cloud computing platform does not find a fourth target object corresponding to the target object within 2 minutes, the display number of the target object is deleted.

[0127] Figure 5 A structure diagram of a target tracking device provided by the embodiment of the present application is shown in FIG. 1. The device includes a first acquisition module 11, a first generation module 12, and a second generation module 13. The first acquisition module 11 is connected to the first generation module 12, and the first generation module 12 is connected to the second generation module 13. Figure 5

[0128] The first acquisition module 11 is configured to acquire point information of a plurality of single points. The first generation module 12 is configured to generate a fusion recognition result corresponding to each point information according to the plurality of point information. The second generation module 13 is configured to generate a tracking result according to a preset point division result and the plurality of fusion recognition results.

[0129] In a possible implementation, the first generation module 12 is specifically configured to, when the point information includes camera information, identify the camera information through an identification algorithm to generate an identification result, and take the identification result as the fusion recognition result; or, when the point information includes the camera information and radar information, identify the camera information through the identification algorithm to generate an identification result, generate a single point fusion result according to the identification result and the radar information through a distance threshold and a specific fusion algorithm, and generate the fusion recognition result according to the single point fusion result and the identification result; or, when the point information includes a plurality of camera information, identify the plurality of camera information through the identification algorithm to generate an identification result corresponding to each camera information, generate a camera fusion recognition result according to the plurality of identification results through a specific fusion algorithm, and generate the fusion recognition result according to the camera fusion recognition result and the plurality of identification results.

[0130] ​In a possible implementation, the single point site includes a continuous area point site or a non-continuous area point site, the point site division result includes at least one continuous area point site and / or at least one non-continuous area point site corresponding to each single point site, and the tracking result includes a continuous cross-point site result corresponding to a plurality of continuous area point sites and a non-continuous cross-point site result corresponding to a plurality of non-continuous area point sites; the second generation module 13 is specifically configured to: acquire a target continuous area point site corresponding to the continuous area point site according to the point site division result; generate the continuous cross-point site result according to the fusion recognition result corresponding to the continuous area point site and the fusion recognition result corresponding to the target continuous area point site; acquire a target non-continuous area point site corresponding to each non-continuous area point site according to the point site division result and the acquired predicted trajectory; generate the non-continuous cross-point site result according to the fusion recognition result corresponding to the non-continuous area point site and the fusion recognition result corresponding to the target non-continuous area point site; and generate the tracking result according to the at least one continuous cross-point site result and the at least one non-continuous cross-point site result.

[0131] In a possible implementation, the fusion recognition result includes a camera fusion result, and the camera fusion result includes target coordinates and appearance features of at least one target object; the second generation module 13 is specifically configured to: find, from the first camera fusion result corresponding to the continuous area point site, first target coordinates of a first target object in an overlapping area through the overlapping area pre-divided; find, from the second camera fusion result corresponding to the target continuous area point site, second target coordinates of a second target object in the overlapping area through the overlapping area; determine a second target object corresponding to the first target object according to the first target coordinates of the at least one first target object and the second target coordinates of the at least one second target object through a bidirectional matching target handover method; generate a first matching result according to a first appearance feature of the first target object and a second appearance feature of the second target object corresponding to the first target object; and generate the continuous cross-point site result according to the first matching result, the first camera fusion result, and the second camera fusion result.

[0132] In a possible implementation, the first appearance feature includes a first appearance image, and the second appearance feature includes a second appearance image; the second generation module 13 is specifically configured to: generate a first color model in HSV format according to the first appearance image; generate a second color model in HSV format according to the second appearance image; generate a target similarity according to the first color model and the second color model; and generate the first matching result according to the target similarity.

[0133] In a possible implementation, the fusion recognition result includes a camera fusion result, the camera fusion result includes appearance features of at least one target object, and the second generation module 13 is specifically configured to generate a second matching result corresponding to a third target object and a fourth target object according to a third appearance feature of the third target object corresponding to the non-continuous region point and a fourth appearance feature of the fourth target object corresponding to the target non-continuous region point; and generate a non-continuous cross-point result according to the second matching result, a third camera fusion result corresponding to the third target object, and a fourth camera fusion result corresponding to the fourth target object.

[0134] In a possible implementation, the device further includes a second acquisition module 14, a fourth generation module 15, a judgment module 16, and a deletion module 17. The second acquisition module 14 is connected with the second generation module 13 and the fourth generation module 15, the fourth generation module 15 is connected with the judgment module 16, the judgment module 16 is connected with the deletion module 17.

[0135] The second acquisition module 14 is configured to acquire a driving feature of a target object; the fourth generation module 15 is configured to generate a time threshold according to the driving feature; the judgment module 16 is configured to judge whether there is a second target object or a fourth target object matched with the target object within the time threshold; and the deletion module 17 is configured to delete a display number corresponding to the target object if there is no second target object or fourth target object matched with the target object.

[0136] Embodiments of the present application provide a target tracking device, point information of a plurality of single points is acquired; a fusion recognition result corresponding to each point information is generated according to the plurality of point information; and a tracking result is generated according to a preset point division result and the plurality of fusion recognition results, so that the cloud computing platform can realize fusion of information of a single point, and the information of the plurality of single points after fusion is fused again, and when a target object moves across a plurality of points, multi-point cross-domain continuous tracking of the target object is realized.

[0137] Embodiments of the present application provide a storage medium, the storage medium includes a stored program, wherein when the program runs, the device where the storage medium is located is controlled to perform each step of the embodiments of the target tracking method, and specific descriptions can be referred to the embodiments of the target tracking method.

[0138] Embodiments of the present application provide a cloud computing platform, including a memory and a processor, the memory is used to store information including program instructions, the processor is used to control the execution of the program instructions, the program instructions are loaded and executed by the processor to realize each step of the embodiments of the target tracking method, and specific descriptions can be referred to the embodiments of the target tracking method.

[0139] Figure 6A schematic diagram of a cloud computing platform is provided for an embodiment of the present application. As shown in Figure 6 The cloud computing platform 30 of this embodiment includes a processor 31, a memory 32, and a computer program 33 stored in the memory 32 and executable on the processor 31, which, when executed by the processor 31, implements the tracking method applied to a target in the embodiment. To avoid repetition, details are not described herein. Alternatively, the computer program, when executed by the processor 31, implements the functions of each model / unit in the tracking device applied to a target in the embodiment. To avoid repetition, details are not described herein.

[0140] The cloud computing platform 30 includes, but is not limited to, the processor 31 and the memory 32. Those skilled in the art can understand that Figure 6 The cloud computing platform 30 is only an example and does not constitute a limitation on the cloud computing platform 30, and can include more or fewer components than shown in the diagram, or combine certain components, or include different components, for example, the cloud computing platform 30 can also include an input / output device, a network access device, a bus, etc.

[0141] The processor 31 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0142] The memory 32 can be an internal storage unit of the cloud computing platform 30, such as a hard disk or a memory of the cloud computing platform 30. The memory 32 can also be an external storage device of the cloud computing platform 30, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 32 can include both an internal storage unit and an external storage device of the cloud computing platform 30. The memory 32 is used to store computer programs and other programs and data required by the cloud computing platform 30. The memory 32 can also be used to temporarily store data that has been output or will be output.

[0143] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0144] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be implemented by other manners. For example, the above-described device embodiments are only schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0145] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0146] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware, or in the form of hardware plus software functional unit.

[0147] The integrated unit realized in the form of software functional unit can be stored in a computer readable storage medium. The software functional unit stored in a storage medium includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute part of the steps of the method of each embodiment of the present application. The foregoing storage medium includes a variety of program code storage media such as a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0148] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of tracking a target, characterized by, The method comprises the following steps: acquiring point information of a plurality of single points; generating a fusion recognition result corresponding to each point information according to a plurality of the point information; generating a tracking result according to a preset point division result and a plurality of the fusion recognition result; the single point includes a continuous area point or a non-continuous area point, the point division result includes at least one continuous area point and / or at least one non-continuous area point corresponding to each single point, and the tracking result includes a continuous cross-point result corresponding to a plurality of the continuous area points and a non-continuous cross-point result corresponding to a plurality of the non-continuous area points; the step of generating a tracking result according to a preset point division result and a plurality of the fusion recognition result comprises the following steps: acquiring a target continuous area point corresponding to a continuous area point according to the point division result; generating the continuous cross-point result according to a fusion recognition result corresponding to the continuous area point and a fusion recognition result corresponding to the target continuous area point; acquiring a target non-continuous area point corresponding to each non-continuous area point according to the point division result and a predicted trajectory acquired; generating the non-continuous cross-point result according to a fusion recognition result corresponding to the non-continuous area point and a fusion recognition result corresponding to the target non-continuous area point; generating the tracking result according to at least one continuous cross-point result and at least one non-continuous cross-point result.

2. The method of claim 1, wherein, the step of generating a fusion recognition result corresponding to each point information according to a plurality of the point information comprises the following steps: when the point information includes camera information, generating an identification result by identifying the camera information through an identification algorithm; taking the identification result as the fusion recognition result; or when the point information includes camera information and radar information, generating an identification result by identifying the camera information through an identification algorithm; generating a single point fusion result according to the identification result and the radar information through a specific fusion algorithm and a distance threshold acquired; generating the fusion recognition result according to the single point fusion result and the identification result; or when the point information includes a plurality of camera information, generating an identification result corresponding to each camera information by identifying a plurality of the camera information through an identification algorithm; generating a camera fusion identification result according to a plurality of the identification result through a specific fusion algorithm; and generating the fusion recognition result according to the camera fusion identification result and a plurality of the identification result.

3. The method of claim 1, wherein, the fusion recognition result includes a camera fusion result, and the camera fusion result includes target coordinates and appearance features of at least one target object; the step of generating the continuous cross-point result according to a fusion recognition result corresponding to the continuous area point and a fusion recognition result corresponding to the target continuous area point comprises the following steps: finding out first target coordinates of a first target object in the overlapping area from the first camera fusion result corresponding to the continuous area point through a pre-divided overlapping area; find out the second target coordinates of the second target object in the overlapping area from the second camera fusion result corresponding to the target continuous area point; determine the second target object corresponding to the first target object according to the first target coordinates of the at least one first target object and the second target coordinates of the at least one second target object by a bidirectional matching target handover method; the bidirectional matching target handover method is a coordinate transformation relationship model between sensors at different points; generate a first matching result according to the first appearance feature of the first target object and the second appearance feature of the second target object corresponding to the first target object; generate the continuous cross-point result according to the first matching result, the first camera fusion result and the second camera fusion result.

4. The method of claim 3, wherein, The first appearance feature includes a first appearance image, and the second appearance feature includes a second appearance image; the first matching result is generated according to the first appearance feature of the first target object and the second appearance feature of the second target object corresponding to the first target object, including: generating a first color model in HSV format according to the first appearance image; generating a second color model in HSV format according to the second appearance image; generating a target similarity according to the first color model and the second color model; generating the first matching result according to the target similarity.

5. The method of claim 3, wherein, The fusion recognition result includes a camera fusion result, the camera fusion result includes an appearance feature of at least one target object, and the non-continuous cross-point result is generated according to the fusion recognition result corresponding to the non-continuous area point and the fusion recognition result corresponding to the target non-continuous area point, including: generating a second matching result of the third target object and the fourth target object corresponding to the third appearance feature of the third target object corresponding to the non-continuous area point and the fourth appearance feature of the fourth target object corresponding to the target non-continuous area point; generating the non-continuous cross-point result according to the second matching result, the third camera fusion result corresponding to the third target object and the fourth camera fusion result corresponding to the fourth target object.

6. The method of claim 5, wherein, The tracking result includes a display number of at least one target object, and after the tracking result is generated, the following steps are further included: obtaining a driving feature of a target object; generating a time threshold according to the driving feature; determining whether there is the second target object or the fourth target object matched with the target object within the time threshold; if there is no second target object or fourth target object matched with the target object, deleting the display number corresponding to the target object.

7. A tracking device of a target, characterized in that including: a first acquisition module, configured to acquire point information of a plurality of single points; a first generation module, configured to generate a fusion recognition result corresponding to each point information according to a plurality of the point information; a second generation module, configured to generate a tracking result according to a preset point division result and a plurality of the fusion recognition result; The single point position includes a continuous area point position or a discontinuous area point position, the point position division result includes at least one continuous area point position and / or at least one discontinuous area point position corresponding to each single point position, and the tracking result includes a continuous cross-point position result corresponding to a plurality of continuous area point positions and a discontinuous cross-point position result corresponding to a plurality of discontinuous area point positions; The second generation module is specifically configured to: according to the point position division result, obtain a target continuous area point position corresponding to a continuous area point position; according to a fusion recognition result corresponding to the continuous area point position and a fusion recognition result corresponding to the target continuous area point position, generate the continuous cross-point position result; according to the point position division result and the obtained predicted trajectory, obtain a target discontinuous area point position corresponding to each discontinuous area point position; According to a fusion recognition result corresponding to the discontinuous area point position and a fusion recognition result corresponding to the target discontinuous area point position, the discontinuous cross-point position result is generated. The tracking result is generated according to at least one continuous cross-point position result and at least one discontinuous cross-point position result.

8. A storage medium, characterized by The storage medium includes a stored program, wherein the program controls the device where the storage medium is located to execute the tracking method of any one of claims 1 to 6 when the program is running.

9. A cloud computing platform comprising a memory for storing information including program instructions and a processor for controlling execution of the program instructions, characterized in that, The program instructions are loaded and executed by the processor to realize the tracking method steps of any one of claims 1 to 6.

Images