A method, device, equipment and medium for target determination

Abstract

An embodiment of the present invention discloses a method, device, equipment and medium for target determination. The method includes: based on a target recognition model, performing target recognition on initial point cloud data to determine a first target point cloud set; using a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except for the first target point cloud set; determining a target according to the first target point cloud set and the second target point cloud set. Through the above technical solution, the purpose of identifying a target from point cloud data is achieved; the point cloud data is processed by a deep learning model and a clustering algorithm in sequence to obtain the targets corresponding to the dense point cloud and the sparse point cloud respectively, and then the final target is determined from the two recognition results, achieving the effects of high target recognition accuracy and good stability, and improving the target recall rate.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of autonomous driving, and in particular, to a method, device, equipment and medium for target determination. Background Art

[0002] As the driverless technology becomes more and more mature, the requirements for the perception system in the autonomous driving system are getting higher and higher; among them, the point cloud target recognition technology can identify the target three-dimensional space position information and contour information from the point cloud data obtained by the lidar.

[0003] The problems existing in the current point cloud target recognition solutions include: (1) There are relatively high requirements for the quality and density of the point cloud distribution, and there is a relatively serious dependence on the point cloud feature distribution. It has a good recognition effect for targets with short distances and dense point clouds, but it cannot stably recognize targets with long distances, sparse point clouds and irregular shapes, and problems such as misdetection and missed detection of obstacle targets are likely to occur, and the target recall rate is relatively low. (2) The target recognition process is mainly completed based on the distance information or density information of the point cloud model. In the process of point cloud target recognition, insufficient segmentation will be caused due to too short distances between point cloud targets, and over-segmentation will also be caused due to too large differences in the density information of individual point cloud targets. Summary of the Invention

[0004] The present invention provides a method, device, equipment and medium for target determination, which are used to solve the defects in the prior art, realize accurately and stably identifying targets from point cloud data, and improve the target recall rate.

[0005] In a first aspect, an embodiment of the present invention provides a method for target determination, including:

[0006] Based on a target recognition model, perform target recognition on initial point cloud data to determine a first target point cloud set;

[0007] Use a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set;

[0008] Determine a target according to the first target point cloud set and the second target point cloud set.

[0009] In a second aspect, an embodiment of the present invention provides a device for target determination, including:

[0010] A first target point cloud set determination module, configured to perform target recognition on initial point cloud data based on a target recognition model to determine a first target point cloud set;

[0011] The second target point cloud set determination module is configured to perform clustering processing on the remaining point cloud data by using a clustering algorithm to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except for the first target point cloud set;

[0012] The target determination module is configured to determine a target according to the first target point cloud set and the second target point cloud set.

[0013] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0014] One or more processors;

[0015] A storage device for storing one or more programs;

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the target determination method as described in the first aspect.

[0017] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the target determination method as described in the first aspect is implemented.

[0018] An embodiment of the present invention discloses a target determination method, apparatus, device and medium. The method includes: based on a target recognition model, performing target recognition on initial point cloud data to determine a first target point cloud set; performing clustering processing on the remaining point cloud data by using a clustering algorithm to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except for the first target point cloud set; determining a target according to the first target point cloud set and the second target point cloud set. Through the above technical solutions, the purpose of identifying a target from point cloud data is achieved; the point cloud data is processed by a deep learning model and a clustering algorithm in sequence to obtain the targets corresponding to the dense point cloud and the sparse point cloud respectively, and then the final target is determined from the two recognition results, achieving the effects of high target recognition accuracy and good stability, and improving the target recall rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In combination with the drawings and referring to the following specific embodiments, the above and other features, advantages and aspects of the various embodiments of the present disclosure will become more apparent. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic and the original elements and elements are not necessarily drawn to scale.

[0020] Figure 1 It is a flowchart of a target determination method provided in Embodiment 1 of the present invention;

[0021] Figure 2Flowchart of a method for target determination provided in the second embodiment of the present invention;

[0022] Figure 3 Structural schematic diagram of a device for target determination provided in the third embodiment of the present invention;

[0023] Figure 4 Structural schematic diagram of an electronic device provided in the fourth embodiment of the present invention. Detailed implementation manners

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. In addition, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. It should also be noted that, for the sake of description, only the parts related to the present invention are shown in the drawings instead of all the structures.

[0025] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of the steps can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the steps can be rearranged. The process can be terminated when its operation is completed, but it can also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, and so on.

[0026] It should be noted that the concepts such as "first" and "second" mentioned in the embodiments of the present invention are only used to distinguish different devices, modules, units, or other objects, and are not used to limit the order of functions performed by these devices, modules, units, or other objects or their interdependent relationships.

[0027] To better understand the embodiments of the present invention, the related technologies will be introduced below.

[0028] Embodiment 1

[0029] Figure 1The figure is a flowchart of a target determination method provided in the first embodiment of the present invention. This embodiment is applicable to the situation of point cloud target determination. Typically, it can be used to identify point cloud data to determine targets. For example, in the field of autonomous driving, obstacles in front of an autonomous vehicle are identified through point cloud data, and then an obstacle avoidance route is planned for the autonomous vehicle to control the autonomous vehicle to bypass the obstacles. Specifically, this method can be executed by a target determination device, which can be implemented in software and / or hardware and integrated in an electronic device. Further, the electronic device includes but is not limited to: desktop computers, laptop computers, smartphones, and servers, etc. Further, the server includes but is not limited to: industrial integration servers, system background servers, and cloud servers.

[0030] As Figure 1 shown, the method specifically includes the following steps:

[0031] S110, based on a target recognition model, perform target recognition on the initial point cloud data to determine a first target point cloud set.

[0032] Among them, the target recognition model can be a pre-trained deep learning network. The target recognition model includes but is not limited to being constructed based on a fully connected network structure, a convolutional neural network, and a recurrent neural network. For example, it can be constructed based on a 3D grid deep learning network, a 3D point cloud deep learning network, a 2D view deep learning network, and a graph-based deep learning network. The appropriate deep learning network construction method can be selected according to the type of point cloud data structure obtained by the lidar. It can be trained using point cloud data. The point cloud data can be obtained through a lidar sensor, a 3D laser scanner, or a photogrammetric scanner, etc. The initial point cloud data can be obtained by the lidar sensor from the surrounding environment of the autonomous vehicle. The target point cloud set can be a target set, and this set can include various obstacle targets, such as vehicles, pedestrians, non-motor vehicles, etc.

[0033] Specifically, the initial point cloud data has target point clouds with high density and target point clouds with low density. The deep learning model extracts target point clouds with high density with higher confidence. Therefore, the initial point cloud data is identified using a deep learning network, and a target set, that is, a first target point cloud set, is determined from the recognition results.

[0034] In the embodiment of the present invention, optionally, based on a target recognition model, performing target recognition on the initial point cloud data to determine a first target point cloud set includes: inputting the initial point cloud data into the target recognition model to determine the recognition target and confidence corresponding to the initial point cloud data; according to the confidence, determining the first target point cloud set from the initial point cloud data corresponding to the recognition target.

[0035] Among them, the recognition targets may be the corresponding targets in the initial point cloud data. For example, the initial point cloud data contains point cloud models of targets such as vehicles, trees, pedestrians, and roadblocks, and the targets such as vehicles, trees, pedestrians, and roadblocks therein are the recognition targets. The confidence level may be the degree of credibility of the target. For example, a certain vehicle is recognized from the target recognition model, and its credibility is seventy percent.

[0036] Specifically, after inputting the initial point cloud data into the target recognition model, the recognition targets included in the initial point cloud data and the confidence levels of each recognition target can be obtained. With the help of the confidence levels, the recognition targets are selected, and relatively reliable recognition targets are selected and put into the first target point cloud set.

[0037] In this solution, the recognition targets are reselected through the confidence levels, improving the accuracy of the recognition targets.

[0038] In an embodiment of the present invention, further, according to the confidence level, a first target point cloud set is determined from the initial point cloud data corresponding to the recognition target, including: if the confidence level of the recognition target is greater than a preset confidence level threshold, the initial point cloud data corresponding to the recognition target is determined as the first target point cloud set.

[0039] Among them, the preset confidence level threshold may be a reliability value of the confidence level set in advance, and the specific value can be adaptively determined according to the actual situation. For example, eighty percent can be set as the preset confidence level threshold.

[0040] Exemplarily, the preset confidence level threshold may be fifty percent, and this value can be fine-tuned according to actual requirements. If the confidence level of the recognition target is greater than fifty percent, it can be put into the first target point cloud set.

[0041] In this solution, by setting the confidence level threshold, the recognition targets that meet the requirements are screened out.

[0042] S120. Use a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except for the first target point cloud set.

[0043] Among them, the clustering algorithm may be a method of dividing a data set into different classes or clusters, and may be a partitioning clustering method, a density-based clustering method, a hierarchical clustering method, and various new methods, such as k-means, OPTICS, Divisive, and quantum clustering.

[0044] Exemplarily, by applying a clustering algorithm to the remaining point cloud data, the obstacle point clouds can be clustered into different clusters, and each cluster represents an obstacle target, ultimately achieving the recognition of targets such as vehicles, pedestrians, and irregular objects at a relatively long distance. The point cloud data obtained by the lidar has the characteristics that the point cloud model of the nearby target is dense and the point cloud model of the far target is sparse. For the point clouds within different distance ranges, corresponding clustering conditions can be set. For example, dividing them into three groups according to the target distance, namely 0 to 50 meters, 50 to 100 meters, and more than 100 meters, and each group corresponds to an appropriate clustering algorithm, which can improve the clustering effect of the obstacle target point clouds.

[0045] In the remaining point cloud data, the target point clouds with relatively low density or irregular target point clouds are mainly included. Such point cloud data is free from the interference of dense target point clouds, and it is easier to identify the target and has a high recognition efficiency by using the clustering algorithm.

[0046] S130. Determine the target according to the first target point cloud set and the second target point cloud set.

[0047] Specifically, the first target point cloud set is a target set obtained through deep learning, and the second target point cloud set is a target set obtained through the clustering algorithm. According to the two target sets, the final target is determined.

[0048] Exemplarily, the target can be an obstacle target, that is, an obstacle that may affect autonomous driving

[0049] The embodiments of the present invention disclose a method, device, equipment, and medium for target determination. The method includes: based on a target recognition model, performing target recognition on the initial point cloud data to determine a first target point cloud set; using a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; where the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set; determining the target according to the first target point cloud set and the second target point cloud set. Through the above technical solution, the purpose of identifying the target from the point cloud data is achieved; the point cloud data is processed by a deep learning model and a clustering algorithm successively to obtain the targets corresponding to the dense point cloud and the sparse point cloud respectively, and then the final target is determined from the two recognition results, achieving the effects of high target recognition accuracy and good stability, and improving the target recall rate.

[0050] In the embodiments of the present invention, preferably, before using the clustering algorithm to perform clustering processing on the remaining point cloud data to determine the second target point cloud set, the method further includes: determining a target ground segmentation algorithm according to the acquisition environment characteristics of the initial point cloud data; using the target ground segmentation algorithm to screen the remaining point cloud data to remove the remaining point cloud data located on the ground.

[0051] Among them, the target ground segmentation algorithm can be an algorithm used to segment the ground and non-ground. For example, the remaining point cloud data contains the point cloud data of objects such as flowers, plants, and trees on the ground and objects such as pets and vehicles on the non-ground. The target ground segmentation algorithm removes the point cloud data of the ground target.

[0052] Exemplarily, the ground segmentation algorithm specifically includes the ground segmentation algorithm based on statistics, the ground segmentation algorithm based on angles, the plane fitting algorithm, and the region growing algorithm based on facets, etc. During the implementation of this solution, it can be dynamically selected according to the road type. For example, the plane fitting algorithm is selected on a flat road, and the region growing algorithm based on facets is used on a rough road. This ensures the adaptability of the ground segmentation algorithm and improves the effect of removing ground point clouds.

[0053] In this solution, the interference of ground point clouds on the clustering process of non-ground point cloud data is avoided.

[0054] Embodiment 2

[0055] Figure 2 It is a flowchart of a target determination method provided in Embodiment 2 of the present invention. This embodiment is optimized on the basis of the above embodiment and specifically describes the target determination. It should be noted that the technical details not described in detail in this embodiment can be referred to in any of the above embodiments.

[0056] Specifically, as Figure 2 shown, the method specifically includes the following steps:

[0057] S210, based on the target recognition model, perform target recognition on the initial point cloud data to determine the first target point cloud set.

[0058] S220, use a clustering algorithm to perform clustering processing on the remaining point cloud data to determine the second target point cloud set; among them, the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set.

[0059] S230, merge the first target point cloud set and the second target point cloud set to obtain a merged target point cloud set.

[0060] Among them, the merged target point cloud set can be the merged target point cloud set, including the first target point cloud set and the second target point cloud set. For example, if the first target point cloud set has pedestrians and trees, and the second target point cloud set has vehicles and roadblocks, then the merged target point cloud set has pedestrians, trees, vehicles, roadblocks, etc.

[0061] S240, determine the target from the merged target point cloud set.

[0062] Exemplarily, the merged target point cloud set includes the integration of all obstacle target information, from which all obstacle target information can be determined, providing safety guarantee for autonomous driving.

[0063] In an embodiment of the present invention, optionally, determining a target from the merged target point cloud set includes: performing a merging process on the targets according to the distances between the targets in the merged target point cloud set.

[0064] Specifically, in the merged target point cloud set, the distances between targets may be very close. It may be that a part of the target is in the first target point cloud set and another part is in the second target point cloud set. Therefore, the final result needs to be determined according to the distances between the targets in the merged target point cloud set.

[0065] In this solution, the problem of over-segmenting targets is solved, which provides convenience for obstacle discrimination in the autonomous driving system.

[0066] In an embodiment of the present invention, further, performing a merging process on the targets according to the distances between the targets in the merged target point cloud set includes: traversing each target in the merged target point cloud set to determine the distances between the target and each other target in the merged target point cloud set; if the distance is less than a preset distance threshold, then merge the currently traversed target and the other target into the same target.

[0067] Specifically, for each target in the merged point cloud set, the following operations are performed: calculate the distance between the target and other targets, and if the distance is less than the preset distance threshold, then merge the target and the other target. For example, if the distance between target 1 and target 2 is less than the preset distance threshold, then merge target 1 and target 2 to obtain target 3. If the distance between target 3 and target 4 is less than the preset distance threshold, then merge target 3 and target 4 to obtain target 5. Or, if the distance between target 1 and target 2 is less than the preset distance threshold and the distance between target 1 and target 4 is less than the preset distance threshold, then merge target 1, target 2, and target 4. Among them, the preset distance threshold can be flexibly determined. For example, the preset distance threshold between vehicles is set to be greater than the preset distance threshold of each target in the crowd.

[0068] In this solution, the mis-segmented targets are merged to restore their original state, making the output of obstacle target information more accurate.

[0069] In an embodiment of the present invention, exemplarily, after performing merging target processing on the merged target point cloud set, a point cloud set of the final target is obtained. Then, bounding box fitting processing is performed on each point cloud set of the final target. In this solution, the autonomous driving system requires the real size of the obstacle to perform accurate calculations. Therefore, the minimum circumscribed box and the minimum bounding box are used for obstacle bounding box fitting processing. And the spatial geometric information of the obstacle target is calculated, including parameters such as the center point, centroid point, length, width, and height. After the fitting is completed, the output of the obstacle target information can be completed, providing data support for the decision-making and planning module of the autonomous driving system.

[0070] A target determination method provided in the second embodiment of the present invention is optimized based on the above embodiment. First, the mis-segmented targets are merged, and then the final target information is determined, making the finally obtained target more credible and helping the autonomous driving system to avoid obstacles.

[0071] Embodiment Three

[0072] Figure 3 It is a schematic structural diagram of a target determination device provided in the third embodiment of the present invention. The target determination device provided in this embodiment includes:

[0073] The first target point cloud set determination module 310 is configured to perform target recognition on the initial point cloud data based on the target recognition model to determine the first target point cloud set;

[0074] The second target point cloud set determination module 320 is configured to perform clustering processing on the remaining point cloud data by using a clustering algorithm to determine the second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set;

[0075] The target determination module 330 is configured to determine the target according to the first target point cloud set and the second target point cloud set.

[0076] Based on the above embodiment, the first target point cloud set determination module 310 includes:

[0077] The recognition target and confidence determination unit is configured to input the initial point cloud data into the target recognition model to determine the recognition target and confidence corresponding to the initial point cloud data;

[0078] The first target point cloud set determination unit is configured to determine the first target point cloud set from the initial point cloud data corresponding to the recognition target according to the confidence.

[0079] The first target point cloud set determination unit includes:

[0080] A first target point cloud set determination subunit, configured to determine the initial point cloud data corresponding to the recognition target as the first target point cloud set if the confidence level of the recognition target is greater than a preset confidence threshold.

[0081] The target determination module 330 includes:

[0082] A merged target point cloud set acquisition unit, configured to merge the first target point cloud set and the second target point cloud set to obtain a merged target point cloud set;

[0083] A target determination unit, configured to determine a target in the merged target point cloud set.

[0084] The target determination unit includes:

[0085] A merging processing subunit, configured to perform merging processing on the targets according to the distances between the targets in the merged target point cloud set.

[0086] The merging processing subunit further includes:

[0087] Traverse each target in the merged target point cloud set to determine the distances between the target and each other target in the merged target point cloud set;

[0088] If the distance is less than a preset distance threshold, merge the currently traversed target and the other target into the same target.

[0089] The device further includes:

[0090] A target ground segmentation algorithm determination module, configured to determine a target ground segmentation algorithm according to the acquisition environment characteristics of the initial point cloud data;

[0091] A remaining point cloud data removal module, configured to use the target ground segmentation algorithm to screen the remaining point cloud data and remove the remaining point cloud data located on the ground.

[0092] The target determination device provided in Embodiment 3 of the present invention can be used to execute the target determination method provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0093] Embodiment 4

[0094] Figure 4FIG. 0 shows a schematic structural diagram of an electronic device 10 that can be used to implement an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 10 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, user equipment, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0095] As Figure 4 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0096] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, wireless networks.

[0097] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the target determination method.

[0098] In some embodiments, the target determination method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the methods described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the target determination method by any other suitable means (e.g., by means of firmware).

[0099] The various implementations of the systems and techniques described above in this document may be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a special-purpose or general-purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0100] The computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that the computer programs, when executed by the processor, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on the remote machine or server.

[0101] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0102] To provide for interaction with a user, the systems and techniques described herein can be implemented on an electronic device 10 having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device 10. Other kinds of devices can also be used to provide for interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0103] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0104] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0105] It should be understood that various forms of processes shown above can be used, steps can be reordered, added or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0106] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A target determination method, characterized in that, it includes: Based on a target recognition model, perform target recognition on the initial point cloud data to determine a first target point cloud set; Use a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set; Determine the target according to the first target point cloud set and the second target point cloud set; wherein, the determining the target according to the first target point cloud set and the second target point cloud set includes: Merge the first target point cloud set and the second target point cloud set to obtain a merged target point cloud set; Determine the target from the merged target point cloud set; The determining the target from the merged target point cloud set includes: According to the distances between the targets in the merged target point cloud set, perform merging processing on the targets.

2. The method according to claim 1, characterized in that, Based on a target recognition model, performing target recognition on the initial point cloud data to determine a first target point cloud set includes: Input the initial point cloud data into the target recognition model to determine the recognition target and confidence level corresponding to the initial point cloud data; According to the confidence level, determine the first target point cloud set from the initial point cloud data corresponding to the recognition target.

3. The method according to claim 2, characterized in that, According to the confidence level, determining the first target point cloud set from the initial point cloud data corresponding to the recognition target includes: If the confidence level of the recognition target is greater than a preset confidence threshold, then determine the initial point cloud data corresponding to the recognition target as the first target point cloud set.

4. The method according to claim 1, characterized in that, According to the distances between the targets in the merged target point cloud set, performing merging processing on the targets includes: Traverse each target in the merged target point cloud set to determine the distances between the target and each other target in the merged target point cloud set; If the distance is less than a preset distance threshold, then merge the currently traversed target and the other target into the same target.

5. The method according to claim 1, characterized in that, Before using a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set, the method further includes: According to the acquisition environment characteristics of the initial point cloud data, determine a target ground segmentation algorithm; Use the target ground segmentation algorithm to screen the remaining point cloud data to remove the remaining point cloud data located on the ground.

6. A target determination device, characterized in that, it includes: A first target point cloud set determination module, configured to perform target recognition on the initial point cloud data based on a target recognition model to determine a first target point cloud set; A second target point cloud set determination module, configured to use a clustering algorithm to perform clustering processing on the remaining point cloud data to determine a second target point cloud set; wherein, the remaining point cloud data is the point cloud data in the initial point cloud data except the first target point cloud set; A target determination module, configured to determine the target according to the first target point cloud set and the second target point cloud set; The target determination module includes: A merged target point cloud set acquisition unit, configured to merge the first target point cloud set and the second target point cloud set to obtain a merged target point cloud set; A target determination unit, configured to determine a target from the merged target point cloud set; The target determination unit includes: A merged processing subunit, configured to perform a merging process on the targets according to the distances between the targets in the merged target point cloud set.

7. An electronic device, characterized in that it includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the target determination method according to any one of claims 1-5.

8. A computer-readable storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, it implements the target determination method according to any one of claims 1-5.

Images