Target detection method, target detection device, and computer-readable storage medium

Abstract

The present application provides a target detection method, a target detection device, and a computer-readable storage medium. The target detection method includes: obtaining the current monitoring frames of several cameras in different viewing angles of the monitoring area, and converting the target coordinates of all current monitoring frames to coordinates of the same coordinate system; identifying the main target and the auxiliary target from the current monitoring frame of each camera; obtaining the main target of the current monitoring frame of each camera, and the auxiliary targets of the current monitoring frames of other cameras that match the main target, to construct a target cluster; taking the main target in the target cluster as a reference, fusing the information of all auxiliary targets to obtain a global target. Through the above method, the target detection device improves the positioning accuracy of target detection by dividing the corresponding targets of the actual target in multiple cameras into one main and multiple auxiliary targets, and using the main target to construct a target cluster to generate a global target.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a target detection method, a target detection device, and a computer-readable storage medium. Background Art

[0002] With the advancement of science and technology, autonomous driving has gradually become possible. From assisted intelligent driving to fully autonomous driving solutions, major manufacturers around the world have invested huge human and financial resources. Vehicle-road collaboration solutions are a popular direction. Vehicle-road collaboration requires not only vehicle-side intelligence, but also road-side intelligence. Even in the absence of vehicle-side perception, road-side equipment can help the vehicle complete autonomous driving. Road-side intelligence involves the collaborative cooperation of multiple devices. The supporting technology is how to coordinate multiple devices to achieve full-field target detection and trajectory perception. Multi-sensor target fusion and trajectory fusion are the core, and how to ensure that only one actual target is included in the final result is a difficult problem in fusion technology and remains an urgent issue for researchers in this field. Summary of the Invention

[0003] The present application provides a target detection method, a target detection device, and a computer-readable storage medium.

[0004] The present application provides a target detection method, which includes:

[0005] Obtain the current monitoring frames of several cameras in different viewing directions of the monitoring area, and convert the target coordinates of all current monitoring frames to the coordinates of the same coordinate system;

[0006] Identify the primary target and the secondary target from the current monitoring frame of each camera;

[0007] Obtain the main target of the current monitoring frame of each camera and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and build a target cluster;

[0008] Taking the main target in the target cluster as the benchmark, the information of all auxiliary targets is integrated to obtain the global target.

[0009] The identifying of the primary target and the secondary target from the current monitoring frame of each camera includes:

[0010] The target passing through the preset stop line along the direction from near to far of the camera in the current monitoring frame is set as the main target;

[0011] All targets other than the primary target in the current monitoring frame are set as secondary targets.

[0012] The identifying of the primary target and the secondary target from the current monitoring frame of each camera includes:

[0013] Traverse all targets in the current monitoring frame and query whether there is a corresponding historical main target;

[0014] If so, setting the target with the historical main target as the main target;

[0015] All targets other than the primary target in the current monitoring frame are set as secondary targets.

[0016] After identifying the primary target and the secondary target from the respective current monitoring frames based on the target division area of ​​each camera, the target detection method further includes:

[0017] Based on the main target, query whether there is a corresponding historical main target in the historical monitoring frame;

[0018] If it exists, use the historical main target to set the label of the main target;

[0019] Based on the auxiliary target, query whether there is a corresponding historical auxiliary target in the historical monitoring frame;

[0020] If present, the tag of the secondary object is set using the historical secondary object.

[0021] Wherein, after obtaining the global target by taking the main target in the target cluster as a reference and fusing the information of all auxiliary targets, the target detection method further includes:

[0022] Obtaining a first global target set of a previous monitoring frame and a second global target set of a current monitoring frame;

[0023] Acquire a first global goal that exists in the first global goal set and does not exist in the second global goal set;

[0024] Acquire a primary target label of a first primary target based on the first global target;

[0025] Query the secondary target information of the matched secondary target in the current monitoring frame based on the primary target tag;

[0026] The auxiliary target information is used for fusion to generate a second global target corresponding to the first global target in the current monitoring frame, and put it into the second global target set.

[0027] The querying of the secondary target information of the matched secondary target in the current monitoring frame based on the primary target tag includes:

[0028] The number of successful matches of the secondary targets matched based on the primary target tag query;

[0029] The secondary target that exists in the current monitoring frame and has the highest number of successful matches is used as the first secondary target, and the remaining secondary targets that exist in the current monitoring frame and are associated with the primary target tag are used as the second secondary targets;

[0030] The utilizing the auxiliary target information to perform fusion to generate a second global target corresponding to the first global target in the current monitoring frame includes:

[0031] Taking the first auxiliary target as a benchmark, information of all second auxiliary targets is integrated to obtain the second global target.

[0032] The number of successful matches of the secondary target based on the primary target tag query includes:

[0033] Determine whether there are one or more successfully matched auxiliary targets whose number of successful matches is greater than a valid match threshold;

[0034] If not, it is determined that the first global target is lost in the current monitoring frame.

[0035] The present application also provides a target detection device, which includes: an acquisition module, a recognition module, a matching module, and a fusion module; wherein,

[0036] The acquisition module is used to acquire current monitoring frames of several cameras in different viewing directions of the monitoring area, and convert the target coordinates of all current monitoring frames into coordinates of the same coordinate system;

[0037] The recognition module is used to identify the main target and the auxiliary target from the current monitoring frame of each camera;

[0038] The matching module is used to obtain the main target of the current monitoring frame of each camera and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and construct a target cluster;

[0039] The fusion module is used to fuse the information of all auxiliary targets based on the main target in the target cluster to obtain the global target.

[0040] The present application also provides another target detection device, which includes a processor and a memory, wherein program data is stored in the memory, and the processor is used to execute the program data to implement the target detection method as described above.

[0041] The present application also provides a computer-readable storage medium, which is used to store program data. When the program data is executed by a processor, it is used to implement the above-mentioned target detection method.

[0042] The beneficial effects of the present application are as follows: the target detection device obtains the current monitoring frames of several cameras in different viewing directions of the monitoring area, and converts the target coordinates of all current monitoring frames to coordinates of the same coordinate system; identifies the main target and the auxiliary target from the current monitoring frame of each camera; obtains the main target of the current monitoring frame of each camera, and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and constructs a target cluster; takes the main target in the target cluster as the reference, fuses the information of all auxiliary targets, and obtains the global target. In the above manner, the target detection device improves the positioning accuracy of target detection by dividing the corresponding targets of the actual target in multiple cameras into one main and multiple auxiliary targets, and using the main target to construct a target cluster to generate a global target. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. 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 inventive efforts. Among them:

[0044] Figure 1 This is a schematic diagram of an embodiment of video frames in four directions of an intersection provided by the present application;

[0045] Figure 2 This is a schematic diagram of an embodiment of transforming the perspective of a bird's-eye view of an image provided by this application;

[0046] Figure 3 This is a schematic diagram of the positional relationship of targets in monitoring frames in various directions provided by this application in the same coordinate system;

[0047] Figure 4 This is a flow chart of an embodiment of a target detection method provided by the present application;

[0048] Figure 5 This is a schematic diagram of the overall process of the multi-eye fusion method provided by this application;

[0049] Figure 6 This is a schematic diagram of the stop line that vehicles must pass when entering an intersection, provided by this application;

[0050] Figure 7 is a flow chart of another embodiment of the target detection method provided by the present application;

[0051] Figure 8 This is a schematic diagram of the overall process of the global target trajectory relay method provided by this application;

[0052] Figure 9 This is a schematic structural diagram of an embodiment of a target detection device provided by the present application;

[0053] Figure 10 is a structural diagram of another embodiment of the target detection device provided by the present application;

[0054] Figure 11 It is a structural diagram of an embodiment of a computer-readable storage medium provided by this application. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0056] The problem to be solved in this application is to match and fuse targets between cameras at traffic intersections and road sections, obtain the complete trajectory of the target from appearance to disappearance within the camera monitoring range, and ensure that there is only one actual target in the fusion result.

[0057] Traffic intersections are equipped with surveillance cameras in every direction. In vehicle-road collaboration and holographic intersection projects, the target positioning information in the image is mapped to a bird's-eye view, and the targets of different cameras in a unified coordinate system can be obtained. The purpose of target fusion is to match and associate the same targets in different cameras to obtain the complete trajectory of the actual target in the real world from entering the intersection to disappearing from the intersection.

[0058] Specifically, in a specific embodiment, the image frames of the four directions of the intersection are as follows: Figure 1 As shown, Figure 1 This is a schematic diagram of an embodiment of video frames in four directions of an intersection provided by the present application.

[0059] exist Figure 1 In the four images, the targets marked with boxes are the same targets at the same time in different cameras. By performing bird's-eye view transformation on the image frames in different camera images, the following is obtained: Figure 2 The image shown, Figure 2 This is a schematic diagram of an embodiment of transforming the perspective of a bird's-eye view of an image provided by this application.

[0060] The target detection device will Figure 2 The target positioning mapping values ​​in each perspective are in the same coordinate system, and the position information of the target in each camera's image frame in the same scene can be obtained. Figure 3 As shown, Figure 3 This is a schematic diagram of the positional relationship of targets in monitoring frames in various directions provided by this application in the same coordinate system.

[0061] like Figure 3As shown, the target in the black circle represents the position of the same target in different cameras mapped to the unified coordinate system.

[0062] Specifically, in Figure 3 Each form of expression, such as color, label, etc., represents the mapping result of different monitoring frame targets in a unified coordinate system. It can be seen that due to camera calibration, the accuracy problem of target positioning will cause the position of the same target in a unified coordinate system under different perspectives to deviate. What this application needs to do is to fuse the same targets under different perspectives, and finally realize that there is only one positioning of the same target in the real world in this scene.

[0063] This application uses the prior knowledge that targets must appear from the scene boundary and disappear from the boundary. In the video frame, the target starting from the camera side is divided into the main target and other targets are auxiliary targets. The main target is used as the reference target, and the auxiliary targets of other cameras are matched and fused as the fusion target output. Based on this, the repeated positioning information of the same target in different cameras and the false detection targets can be filtered out to ensure that the same target is not output repeatedly. In addition, the main target is associated with the auxiliary targets of other cameras. When the main target is not detected, the auxiliary target is used to relay. Based on this, the trajectory of the target at the intersection can be guaranteed to be continuous and uninterrupted.

[0064] Please refer to the following for details: Figure 4 and Figure 5 , Figure 4 This is a flow chart of an embodiment of the target detection method provided by this application. Figure 5 This is a schematic diagram of the overall process of the multi-eye fusion method provided in this application.

[0065] The target detection method of the present application is applied to a target detection device, wherein the target detection device of the present application can be a server or a system comprising a server and a terminal device. Accordingly, the various components of the target detection device, such as the various units, subunits, modules, and submodules, can be all provided in the server or separately provided in the server and the terminal device.

[0066] Furthermore, the above-mentioned server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server is software, it can be implemented as multiple software or software modules, such as software or software modules for providing distributed servers, or it can be implemented as a single software or software module, which is not specifically limited here. In some possible implementations, the target detection method of the embodiment of the present application can be implemented by a processor calling computer-readable instructions stored in a memory.

[0067] Specifically, if Figure 4As shown, the target detection method of the embodiment of the present application specifically includes the following steps:

[0068] Step S11: obtaining current monitoring frames of several cameras in different viewing directions of the monitoring area, and converting target coordinates of all current monitoring frames into coordinates of the same coordinate system.

[0069] In an embodiment of the present application, the target detection device uses a positioning algorithm and a tracking algorithm to obtain the positioning information and tracking information of the target on the ground plane, and can obtain a unique identifier for each target in each camera: the sensor (camera) number (sensor_id) plus the tracking number (track_id).

[0070] Furthermore, the target detection device converts the ground-based positioning coordinates of each target in the current monitoring frame for each viewing direction from its own image coordinate system to a bird's-eye view coordinate system. This allows targets in different images to be mapped to a unified bird's-eye view coordinate system, facilitating the fusion of identical targets, the fusion of targets with historical trajectories, and the prediction of target motion trajectories.

[0071] Step S12: Identify the main target and the auxiliary target from the current monitoring frame of each camera.

[0072] In an embodiment of the present application, when the target detection device divides each current monitoring frame into a primary target and a secondary target, it may first query whether the target in the current monitoring frame is a recorded target.

[0073] Specifically, the target detection device extracts a matching table A of global targets and primary targets in each monitoring frame. It then uses the sensor ID sensor_id and the tracking ID track_id to query Table A to see if the target in the current monitoring frame has a corresponding historical primary target. If so, it indicates that the target has already been output in the historical monitoring frame. If not, it indicates that the target is likely new.

[0074] Among them, record table A records the matching relationship between the global target world_id and the target track_id in each image frame. Each global target is matched with only one main target in the camera.

[0075] The target detection device sets the old target retrieved from the record table A as the main target of the current monitoring frame. In addition, it can further determine whether the new target is the main target based on the prior knowledge mentioned above.

[0076] Specifically, when a vehicle enters an intersection, it must pass the corresponding stop line under each camera’s perspective, and this is the moment when the target is closest to the camera, such as Figure 6 Therefore, the target detection device can classify the vehicle target that newly enters from the stop line as a new main target.

[0077] In summary, after the target detection device sets the main target of the current monitoring frame according to the new main target and the old main target, it sets the remaining targets as auxiliary targets.

[0078] Step S13: Obtain the main target of the current monitoring frame of each camera and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and construct a target cluster.

[0079] In the embodiment of the present application, the target detection device constructs a target cluster from the main target, which corresponds to the actual target and is recorded in the record table A. Among multiple cameras, only the target in one camera will construct a corresponding target cluster. When the main target is far away from the camera it belongs to at an intersection and may be blocked, the target may not be detected or the positioning accuracy may decrease. In this case, other cameras can detect the target and the target will be classified as an auxiliary target in other monitoring frames.

[0080] Furthermore, the target detection device regards the main target and the new target that have appeared in the monitoring frame as the main targets of the monitoring frame, and the other targets as auxiliary targets. The main target in the monitoring frame is matched and associated with the auxiliary targets in other monitoring frames, and the matched auxiliary targets are added to the target cluster constructed by the main target, and recorded in record table B.

[0081] Among them, record table B records the matching relationship between the main target track_id in the camera and the auxiliary target track_id in other monitoring frames, as well as the number of successful matches.

[0082] Step S14: Taking the main target in the target cluster as the benchmark, the information of all auxiliary targets is integrated to obtain the global target.

[0083] In an embodiment of the present application, the target detection device takes the main target in the target cluster as the reference target, integrates the effective auxiliary target positioning, category and other information, and finally integrates it into a new global target. If it is a new target, it is assigned a new world_id. If it is an old target, it is assigned the corresponding world_id and historical trajectory information.

[0084] In an embodiment of the present application, a target detection device obtains the current monitoring frames of several cameras with different viewing angles in the monitoring area, and converts the target coordinates of all current monitoring frames into coordinates of the same coordinate system; identifies the main target and the auxiliary target from the current monitoring frame of each camera; obtains the main target of the current monitoring frame of each camera, and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and constructs a target cluster; takes the main target in the target cluster as a reference, and fuses the information of all auxiliary targets to obtain a global target. In the above manner, the target detection device improves the positioning accuracy of target detection by dividing the corresponding targets of the actual target in multiple cameras into one main and multiple auxiliary targets, and using the main target to construct a target cluster to generate a global target.

[0085] Further, in Figure 4 Based on the target detection method shown, after the target detection device obtains the global fusion target of the current multiple monitoring frames, it must also consider that some targets disappear at the camera's visible boundary or are obscured, but are still visible from other camera perspectives. At this time, other cameras need to be used for relay monitoring.

[0086] Please refer to the following for details: Figure 7 and Figure 8 , Figure 7 is a flow chart of another embodiment of the target detection method provided by this application, Figure 8 This is a schematic diagram of the overall process of the global target trajectory relay method provided by this application.

[0087] Specifically, if Figure 7 As shown, the target detection method of the embodiment of the present application specifically includes the following steps:

[0088] Step S21: Acquire the first global target set of the previous monitoring frame and the second global target set of the current monitoring frame.

[0089] In an embodiment of the present application, the target detection device queries the first global target set of a monitoring frame and the second global target set of the current monitoring frame from the record table A.

[0090] Step S22: Acquire a first global goal that exists in the first global goal set and does not exist in the second global goal set.

[0091] In an embodiment of the present application, the target detection device obtains a global target that exists in the previous frame fusion but does not exist in the current frame fusion.

[0092] The reason why the global target suddenly disappears in the current frame is that the main target in the target cluster corresponding to the actual target is not detected. There are two main reasons: 1. The main target is blocked and cannot be detected. 2. The main target has moved out of the visual range of the corresponding camera.

[0093] Step S23: Obtain a main target label of the first main target based on the first global target.

[0094] In the embodiment of the present application, the target detection device can use the record table A and the world_id of the first global target to obtain the sensor_id and track_id of the main target that is not detected in the current monitoring frame.

[0095] Step S24: query the secondary target information of the matched secondary target in the current monitoring frame based on the primary target tag.

[0096] In the embodiment of the present application, the target detection device can use the record table B to search for the sensor_id and track_id of the auxiliary targets in other cameras in the target cluster constructed by the disappeared main target, as well as the number of matches match_count between the main and auxiliary targets.

[0097] By setting the effective matching threshold to T, the target detection device takes the auxiliary target corresponding to the maximum match_count that satisfies match_count>T as the fusion reference target. When there is no auxiliary target that satisfies match_count>T, it can be considered that the global target is lost in the current monitoring frame.

[0098] Step S25: Utilize the auxiliary target information for fusion to generate a second global target corresponding to the first global target in the current monitoring frame, and put it into the second global target set.

[0099] In an embodiment of the present application, the target detection device uses the auxiliary target with the largest number of matches as the reference target of the target cluster, integrates the positioning, category and other information of the valid auxiliary targets that match it in other cameras, and finally integrates them into a new global target, and assigns the corresponding world_id and historical trajectory information.

[0100] In this embodiment, a method is proposed to divide the actual target corresponding to multiple cameras into one primary and multiple secondary targets, and to construct a target cluster using the primary target. A method is proposed to determine whether a newly appeared target is a real target based on the intersection boundary. The real target must enter and exit the boundary. New targets in the middle of the intersection are considered false detections or tracking string IDs. A method is proposed to ensure trajectory continuity by matching the primary and secondary targets in the target cluster when the real target switches between different cameras. Utilizing the matching relationship between the primary and secondary targets, the actual target information can be output as long as the primary target or valid secondary target exists.

[0101] Those skilled in the art will understand that in the above-mentioned method of the specific implementation method, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0102] In order to implement the target detection method of the above embodiment, this application also proposes a target detection device, please refer to Figure 9 , Figure 9 It is a structural diagram of an embodiment of the target detection device provided by this application.

[0103] The target detection device 300 of the embodiment of the present application includes an acquisition module 31 , a recognition module 32 , a matching module 33 and a fusion module 34 .

[0104] The acquisition module 31 is used to acquire current monitoring frames of several cameras in different viewing angles of the monitoring area, and convert the target coordinates of all current monitoring frames into coordinates of the same coordinate system.

[0105] The identification module 32 is configured to identify a primary target and an auxiliary target from a current monitoring frame of each camera.

[0106] The matching module 33 is configured to obtain a main target in a current monitoring frame of each camera and auxiliary targets in current monitoring frames of other cameras that match the main target, and construct a target cluster.

[0107] The fusion module 34 is used to fuse the information of all auxiliary targets based on the main target in the target cluster to obtain a global target.

[0108] In order to implement the target detection method of the above embodiment, this application also proposes another target detection device, please refer to Figure 10 , Figure 10 It is a structural diagram of another embodiment of the target detection device provided by this application.

[0109] The target detection device 400 according to the embodiment of the present application includes a memory 41 and a processor 42 , wherein the memory 41 and the processor 42 are coupled.

[0110] The memory 41 is used to store program data, and the processor 42 is used to execute the program data to implement the target detection method described in the above embodiment.

[0111] In this embodiment, the processor 42 may also be referred to as a CPU (Central Processing Unit). The processor 42 may be an integrated circuit chip having signal processing capabilities. The processor 42 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor, or the processor 42 may be any conventional processor.

[0112] To implement the target detection method of the above embodiment, the present application also provides a computer-readable storage medium, such as Figure 11 As shown, the computer-readable storage medium 500 is used to store program data 51. When the program data 51 is executed by the processor, it is used to implement the target detection method described in the above embodiment.

[0113] The present application also provides a computer program product, wherein the computer program product includes a computer program, and the computer program is operable to enable a computer to execute the target detection method as described in the embodiments of the present application. The computer program product can be a software installation package.

[0114] The target detection method described in the above embodiment of the present application, when implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a device, such as a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, 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.

[0115] The above description is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A target detection method, characterized in that: The target detection method comprises: Obtain the current monitoring frames of several cameras in different viewing directions of the monitoring area, and convert the target coordinates of all current monitoring frames to the coordinates of the same coordinate system; Identify the primary target and the secondary target from the current monitoring frame of each camera; Obtain the main target of the current monitoring frame of each camera and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and build a target cluster; wherein the target detection device builds a target cluster from the main target, corresponding to the actual target, the main target and the new target that have appeared in the current monitoring frame are regarded as the main target of the current monitoring frame, and the other targets are regarded as auxiliary targets. The main target in the monitoring frame is matched and associated with the auxiliary targets in the other monitoring frames, and the matched auxiliary targets are added to the target cluster built by the main target; Taking the main target in the target cluster as the benchmark, the information of all auxiliary targets is integrated to obtain the global target; The identifying of the primary target and the secondary target from the current monitoring frame of each camera includes: The target passing through the preset stop line along the direction from near to far of the camera in the current monitoring frame is set as the main target; All targets other than the primary target in the current monitoring frame are set as secondary targets.

2. The target detection method according to claim 1, wherein: The identifying of the primary target and the secondary target from the current monitoring frame of each camera includes: Traverse all targets in the current monitoring frame and query whether there is a corresponding historical main target; If so, setting the target with the historical main target as the main target; All targets other than the primary target in the current monitoring frame are set as secondary targets.

3. The target detection method according to claim 1, wherein: After identifying the primary target and the secondary target from the respective current monitoring frames based on the target division area of ​​each camera, the target detection method further includes: Based on the main target, query whether there is a corresponding historical main target in the historical monitoring frame; If it exists, use the historical main target to set the label of the main target; Based on the auxiliary target, query whether there is a corresponding historical auxiliary target in the historical monitoring frame; If present, the tag of the secondary object is set using the historical secondary object.

4. The target detection method according to claim 1 or 3, characterized in that: After obtaining the global target by taking the main target in the target cluster as a reference and fusing the information of all auxiliary targets, the target detection method further includes: Obtaining a first global target set of a previous monitoring frame and a second global target set of a current monitoring frame; Acquire a first global goal that exists in the first global goal set and does not exist in the second global goal set; Acquire a primary target label of a first primary target based on the first global target; Query the secondary target information of the matched secondary target in the current monitoring frame based on the primary target tag; The auxiliary target information is used for fusion to generate a second global target corresponding to the first global target in the current monitoring frame, and put it into the second global target set.

5. The target detection method according to claim 4, characterized in that: The querying of the secondary target information of the matched secondary target in the current monitoring frame based on the primary target tag includes: The number of successful matches of the secondary targets matched based on the primary target tag query; The secondary target that exists in the current monitoring frame and has the highest number of successful matches is used as the first secondary target, and the remaining secondary targets that exist in the current monitoring frame and are associated with the primary target tag are used as the second secondary targets; The utilizing the auxiliary target information to perform fusion to generate a second global target corresponding to the first global target in the current monitoring frame includes: Taking the first auxiliary target as a benchmark, information of all second auxiliary targets is integrated to obtain the second global target.

6. The target detection method according to claim 5, characterized in that: The number of successful matches of the secondary target based on the primary target tag query includes: Determine whether there are one or more successfully matched auxiliary targets whose number of successful matches is greater than a valid match threshold; If not, it is determined that the first global target is lost in the current monitoring frame.

7. A target detection device, characterized in that: The target detection device includes: an acquisition module, a recognition module, a matching module and a fusion module; wherein, The acquisition module is used to acquire current monitoring frames of several cameras in different viewing directions of the monitoring area, and convert the target coordinates of all current monitoring frames into coordinates of the same coordinate system; The recognition module is used to identify the main target and the auxiliary target from the current monitoring frame of each camera; The matching module is used to obtain the main target of the current monitoring frame of each camera and the auxiliary targets of the current monitoring frames of other cameras that match the main target, and construct a target cluster; wherein the target detection device constructs a target cluster from the main target, corresponds to the actual target, and regards the main target and the new target that have appeared in the current monitoring frame as the main target of the current monitoring frame, and the other targets as auxiliary targets. The main target in the monitoring frame is matched and associated with the auxiliary targets in the other monitoring frames, and the matched auxiliary targets are added to the target cluster constructed by the main target; The fusion module is used to fuse the information of all auxiliary targets based on the main target in the target cluster to obtain the global target; The recognition module is further configured to set the target in the current monitoring frame that passes through a preset stop line along the direction from near to far of the camera as the main target; and set all targets other than the main target in the current monitoring frame as auxiliary targets.

8. A target detection device, characterized in that: The target detection device includes a processor and a memory, wherein program data is stored in the memory, and the processor is configured to execute the program data to implement the target detection method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program data, and when the program data is executed by the processor, it is used to implement the target detection method according to any one of claims 1 to 6.

Images