Target detection methods, devices, edge computing servers, and storage media

Abstract

This application relates to a target detection method, apparatus, edge computing server, and storage medium. The method includes: receiving real-time image data sent by multiple detection devices; determining the area captured by the real-time image data sent by any one detection device as the detection area corresponding to that detection device; identifying all targets within the detection area corresponding to any detection device; obtaining the position of any target within the detection area; determining a detection strategy for the target based on the position; and detecting the target based on the detection strategy. Therefore, it is possible to detect targets in multiple detection areas corresponding to multiple detection devices using an edge computing server, improving both security and detection efficiency.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more particularly to a target detection method, apparatus, edge computing server, and storage medium. Background Technology

[0002] With the increasing emphasis on security checks, it is now necessary to inspect people and vehicles entering and exiting various scenarios to determine their legitimacy. The scope of these inspections is expanding, and the required response time is also accelerating. Furthermore, factors such as the geographical environment of the area being inspected must often be considered during the inspection process.

[0003] In existing technologies, manual inspection is the primary method used to detect unqualified personnel or vehicles. However, due to the large number of inspection scenarios, especially in open areas with high security requirements, manual inspection is inefficient. Furthermore, it is difficult to detect individuals or vehicles that are disguised or concealed, and manual inspection can pose risks to the inspectors. Summary of the Invention

[0004] In view of this, in order to solve the above-mentioned problems of low efficiency of manual detection due to the large number of scenarios to be detected, especially in open spaces with high security requirements, and the difficulty in detecting people or vehicles when they are disguised or hidden, and the potential danger to the detection personnel, this application provides a target detection method, device, edge computing server and storage medium.

[0005] In a first aspect, embodiments of this application provide a target detection method applied to an edge computing server, the method comprising:

[0006] Receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any one of the detection devices as the detection area corresponding to the detection device;

[0007] For any detection device, determine all targets within the detection area.

[0008] For any target in the detection area, the position of the target is obtained, and the detection strategy for the target is determined based on the position;

[0009] The target is detected based on the target detection strategy.

[0010] In an optional implementation, determining the target detection strategy based on the location includes:

[0011] If the target's location is within a preset first detection line in the detection area, then a first detection strategy is determined;

[0012] The detection strategy based on the target includes detecting the target, comprising:

[0013] Based on the first detection strategy, a first early warning message is triggered, and the target is tracked.

[0014] In an optional implementation, tracking the target includes:

[0015] Obtain the morphological information of the target;

[0016] Determine whether the position and / or shape information of the target has changed within a preset time period;

[0017] If the position and / or shape information of the target does not change within a preset time, the tracking of the target will end.

[0018] If the position and / or shape information of the target changes within a preset time period, the tracking of the target will continue.

[0019] In an optional implementation, the step of determining the target detection strategy based on the location further includes:

[0020] If the distance between the target's position and a preset second detection line within the detection area is less than a preset distance, then a second detection strategy is determined, wherein a preset first detection line within the detection area is located before a preset second detection line within the detection area;

[0021] The detection strategy based on the target further includes:

[0022] Based on the second detection strategy, it is determined whether the target is a legitimate target in the preset target library;

[0023] If the target is a legitimate target in a preset target library, then the tracking of the target ends;

[0024] If the target is not a legitimate target in the preset target library, a second warning message is triggered, and image data or video data containing the target is sent to the monitoring platform.

[0025] In an optional implementation, triggering the second warning information and sending image or video data containing the target to the monitoring platform includes:

[0026] If the target's location is within a preset second detection line in the detection area, a second warning message is triggered, and image data or video data containing the target is sent to the monitoring platform.

[0027] In an optional implementation, determining all targets in the detection area includes:

[0028] Acquire real-time image data of the detection area and input the real-time image data into the target detection model;

[0029] Obtain all targets in the detection region as output by the target detection model;

[0030] The target detection model is obtained in the following way:

[0031] Acquire sample image data under a preset scenario, wherein the sample image data includes sample targets, and the sample targets have been labeled;

[0032] The sample image data is processed to obtain target sample image data; wherein the processing of the sample image data includes at least one or more of the following: appearance enhancement processing, geometric enhancement processing, and virtual sample enhancement processing.

[0033] The initial detection model is trained based on the target sample image data to obtain the target detection model.

[0034] In an optional implementation, training the initial detection model based on the target sample image data to obtain the target detection model includes:

[0035] Pruning and / or channel reduction are performed on the initial detection model;

[0036] Determine the hyperparameters in the processed initial detection model;

[0037] The processed initial detection model is trained based on the target sample image data to adjust the hyperparameters and obtain the target detection model.

[0038] Secondly, embodiments of this application provide a target detection device applied to an edge computing server, the device comprising:

[0039] Detection area determination module: used to receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device;

[0040] Target acquisition module: used to determine all targets in the detection area corresponding to any detection device;

[0041] Detection strategy determination module: used to obtain the position of any target in the detection area and determine the detection strategy of the target based on the position;

[0042] Target detection module: used to detect the target based on the target detection strategy.

[0043] Thirdly, embodiments of this application provide an edge computing server, including: a processor and a memory, wherein the processor is used to execute a program stored in the memory to implement any of the target detection methods in the first aspect.

[0044] Fourthly, embodiments of this application provide a storage medium storing one or more programs, which can be executed by one or more processors to implement any of the target detection methods in the first aspect.

[0045] The technical solution provided in this application embodiment receives real-time image data sent by multiple detection devices, determines the area captured by the real-time image data sent by any detection device as the detection area corresponding to that detection device; for the detection area corresponding to any detection device, determines all targets in the detection area; for any target in the detection area, obtains the position of the target, determines the detection strategy for the target based on the position, and detects the target based on the detection strategy. By using an edge computing server to simultaneously detect targets in multiple detection areas corresponding to multiple detection devices, and adopting different detection strategies according to the different positions of targets in each detection area, security and detection efficiency are improved. Attached Figure Description

[0046] Figure 1 A schematic diagram illustrating the implementation process of a target detection method provided in this application embodiment;

[0047] Figure 2 A schematic diagram illustrating the implementation process of another target detection method provided in this application embodiment;

[0048] Figure 3 A schematic diagram of a detection area provided in an embodiment of this application;

[0049] Figure 4 A schematic diagram illustrating the implementation process of a target tracking method provided in this application embodiment;

[0050] Figure 5 A schematic diagram illustrating the implementation process of another target detection method provided in this application embodiment;

[0051] Figure 6 A schematic diagram illustrating the implementation process of another target detection method provided in this application embodiment;

[0052] Figure 7 A schematic diagram illustrating the implementation process of a target determination method provided in this application embodiment;

[0053] Figure 8 A schematic diagram illustrating the implementation process of a target detection model training method provided in this application embodiment;

[0054] Figure 9 A schematic diagram illustrating the implementation process of another object detection model training method provided in this application embodiment;

[0055] Figure 10 This is a schematic diagram of the structure of a target detection device provided in an embodiment of this application;

[0056] Figure 11 This is a schematic diagram of the structure of an edge computing server provided in an embodiment of this application. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] In certain special target detection scenarios, such as open and remote border areas, manual detection of people and vehicles is inefficient and difficult to detect when they are camouflaged or concealed. This application provides a target detection method that uses a camera to capture image data of the area to be detected and transmits the real-time image data to an edge computing server. The edge computing server receives the real-time image data from the camera and performs target detection on it. The edge computing server can be integrated with the camera or deployed independently. In this application, image processing and analysis are performed close to the data acquisition end, offering advantages such as rapid processing and less susceptibility to network speed limitations. Furthermore, the target detection method provided in this application can be applied to various other scenarios, such as schools, tourist attractions, and wildlife parks, all of which require target detection to prevent people or vehicles from entering dangerous areas.

[0059] Figure 1 This is a schematic diagram illustrating the implementation process of a target detection method provided in an embodiment of this application. The method may include the following steps:

[0060] S101: Receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device.

[0061] In this embodiment of the application, when it is necessary to detect a scene, the size of the scene needs to be determined first, and a reasonable number of detection devices are set in the scene. Each detection device corresponds to a detection area. The edge computing server obtains the real-time image data sent by each detection device, and regards the shooting area in the real-time image data sent by each detection device as the detection area corresponding to each detection device, thereby performing target detection.

[0062] For example, when the detection device is a camera, eight cameras need to be installed in the scene to be detected. The edge computing server then obtains the real-time image data sent by the eight cameras, and regards the shooting area in the real-time image data sent by the eight cameras as the detection area corresponding to the eight cameras, thereby detecting the target in each detection area.

[0063] S102: For any detection device, determine all targets within the detection area.

[0064] In this embodiment of the application, the edge computing server performs target detection simultaneously on the detection areas corresponding to each detection device, and for a detection area corresponding to a detection device, determines all targets in that detection area.

[0065] For example, if there are 8 detection devices, there are 8 detection areas. The edge computing server performs target detection on the 8 detection areas simultaneously to determine all targets in each detection area.

[0066] S103: For any target in the detection area, obtain the target's position and determine the target's detection strategy based on the position.

[0067] In this embodiment of the application, for any target in any detection area determined in S102, the following processing is performed:

[0068] The location of the target is obtained, and the location can be represented by horizontal and vertical coordinates, which is not limited in this application. In the embodiments of this application, corresponding target detection strategies are adopted according to different target locations.

[0069] S104: Target-based detection strategy for target detection.

[0070] In the embodiments of this application, the target detection strategy is different depending on the location of the target, and different detection methods are adopted for the target according to the target detection strategy.

[0071] Based on the above description of the technical solutions provided in the embodiments of this application, the target detection method of this application determines multiple detection areas, then acquires all targets in each detection area, and determines different target detection strategies based on the positions of all targets, thereby achieving the detection of each target.

[0072] Figure 2 This is a schematic diagram illustrating the implementation process of another target detection method provided in this application embodiment. The method may include the following steps:

[0073] S201: Receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device.

[0074] S202: For any detection device, determine all targets within the detection area.

[0075] In the embodiments of this application, S201 and S202 have been described in detail in steps S101 and S102, and will not be repeated here.

[0076] S203: For any target in the detection area, obtain the target's position. If the target's position is within the preset first detection line in the detection area, then determine the first detection strategy.

[0077] In this embodiment, the preset first detection line is a detection line drawn according to the actual situation after the detection area is obtained, as shown in the example below. Figure 3 As shown, Figure 3 This is a schematic diagram of a detection area provided in an embodiment of the present application. Specifically, the detection area also includes a preset second detection line.

[0078] In this embodiment of the application, for ease of description, only one target in one detection area is described in S203 to S208, and this target is referred to as the first target. It can be understood that in the edge computing server, all targets in all detection areas are detected simultaneously.

[0079] In this embodiment of the application, the position of the first target is obtained. If the position of the first target is inside the preset first detection line in the detection area, that is, the position of the first target is between the preset first detection line and the preset second detection line, then the first detection strategy corresponding to the position is determined.

[0080] S204: Based on the first detection strategy, trigger the first early warning information and track the target.

[0081] In this embodiment of the application, the first detection strategy includes triggering a first warning message, that is, the edge computing server sends a message to the monitoring platform that the location of the first target is within a preset first detection line in the detection area, that is, the location of the first target is between the preset first detection line and the preset second detection line. At the same time, the edge computing server also sends the location and time information of the first target to the monitoring platform. Meanwhile, the edge computing server tracks the first target and sends the real-time location of the first target to the monitoring platform.

[0082] It should be noted that this application does not limit the specific information sent by the edge computing server to the monitoring platform. This information is sufficient to ensure that the monitoring platform can obtain the real-time location of the first target within the preset first detection line and between the preset first detection line and the preset second detection line within the detection area.

[0083] The method for tracking the target in S204 is as follows: Figure 4 As shown, Figure 4 This application provides a schematic diagram of the implementation process of a target tracking method, which may include the following steps:

[0084] S401: Obtain the shape information of the target.

[0085] In this application embodiment, the first target includes, but is not limited to, people and vehicles. When the first target is a person, the person can be standing, squatting, bending over, etc. When the first target is a vehicle, the vehicle door can be open or closed while keeping the position unchanged. It can also rotate in place. The edge computing server obtains the shape information of the first target through the detection device.

[0086] S402: Determine whether the target's position and / or shape information has changed within a preset time. If the target's position and / or shape information has not changed within the preset time, then execute S403. If the target's position and / or shape information has changed within the preset time, then execute S404.

[0087] S403: End tracking of the target.

[0088] S404: Maintain tracking of the target.

[0089] The following provides a unified explanation of S402 to S404.

[0090] In this embodiment of the application, the first target is tracked in real time, and it is determined whether the position and / or shape information of the first target changes within a preset time. If the position and / or shape information of the first target does not change within the preset time, the tracking of the first target ends; if the position and / or shape information of the first target changes within the preset time, that is, the position and / or shape information of the first target keeps changing, or the time without change is less than the preset time, the tracking of the first target continues.

[0091] In this embodiment of the application, if the position and / or shape information of the first target does not change within a preset time, the tracking of the first target will end, and the first target will be detected to determine whether the position and / or shape information of the first target has changed. When the position and / or shape information of the first target is detected to have changed again, the tracking of the first target will be resumed.

[0092] Figure 5 This is a schematic diagram illustrating the implementation process of another target detection method provided in this application. The method may include the following steps:

[0093] S501: Receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device.

[0094] S502: For any detection device, determine all targets within the detection area.

[0095] In the embodiments of this application, S501 and S502 have been described in detail in S101 and S102, and will not be repeated here.

[0096] S503: For any target in the detection area, obtain the target's position. If the distance between the target's position and the preset second detection line in the detection area is less than a preset distance, determine the second detection strategy, wherein the preset first detection line in the detection area is located before the preset second detection line in the detection area.

[0097] In this embodiment of the application, a preset first detection line within the detection area precedes a preset second detection line, such as... Figure 3 As shown, Figure 3 This is a schematic diagram of a detection area provided in an embodiment of the present application. After obtaining the detection area, a preset first detection line and a preset second detection line are delineated according to the actual situation.

[0098] In the embodiments of this application, for ease of description, only one target in one detection area is described in S503 to S506, and this target is referred to as the first target. It can be understood that in the edge computing server, all targets in all detection areas are detected simultaneously.

[0099] In this embodiment of the application, the position of the first target is obtained. If the position of the first target is between a preset first detection line and a preset second detection line, and the distance between the first target and the preset second detection line is determined to be less than a preset distance based on the position of the first target, then the second detection strategy corresponding to the position is determined.

[0100] For example, if the preset distance is 200 meters, when the distance between the first target and the preset second detection line is less than or equal to 200 meters, the second detection strategy is invoked to detect the first target. It should be noted that the preset distance is calculated by the edge computing server based on the real-time positions of the first target and the second detection line detected by the detection device; this application does not limit the specific calculation method.

[0101] S504: Based on the second detection strategy, determine whether the target is a legal target in the preset target library. If the target is a legal target in the preset target library, execute S505. If the target is not a legal target in the preset target library, execute S506.

[0102] S505: End tracking of the target.

[0103] S506: Triggers a second warning message and sends image or video data containing the target to the monitoring platform.

[0104] In this embodiment of the application, the second detection strategy includes comparing the first target with the legal targets in the preset target library. If the first target is included in the legal targets in the preset target library, the tracking of the first target ends. If the first target is not included in the legal targets in the preset target library, a second warning message is triggered. That is, the edge computing server sends the information that the distance between the first target and the preset second detection line is less than the preset distance to the monitoring platform, and sends the image or video containing the first target to the monitoring platform.

[0105] For example, if the first target is a person, the person's image or video data obtained by the detection device is used to perform facial recognition to determine whether the person's information has been recorded in the preset target database. If the person's information has been recorded in the preset target database, the person is determined to be a legitimate target, and the tracking of the person ends. If the person's information has not been recorded in the preset target database, the person is determined to be an illegal target, triggering a second warning message, and the image or video of the person captured by the detection device is sent to the monitoring platform.

[0106] For example, when the first target is a vehicle, the vehicle is identified by the image or video data obtained by the detection equipment. This could involve identifying the license plate number to determine whether the vehicle's information has been entered into the preset target database. If the vehicle's information has been entered into the preset target database, the vehicle is determined to be a legitimate target, and the tracking of the vehicle ends. If the vehicle's information has not been entered into the preset target database, the vehicle is determined to be an illegal target, triggering a second warning message, and the image or video data of the vehicle captured by the detection equipment is sent to the monitoring platform.

[0107] Figure 6 This is a schematic diagram illustrating the implementation process of another target detection method provided in this application embodiment. The method may include the following steps:

[0108] S601: Receives real-time image data sent by multiple detection devices and determines the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device.

[0109] S602: For any detection device, determine all targets within the detection area.

[0110] S603: For any target in the detection area, obtain the target's position. If the distance between the target's position and the preset second detection line in the detection area is less than a preset distance, determine the second detection strategy, wherein the preset first detection line in the detection area is located before the preset second detection line in the detection area.

[0111] S604: Based on the second detection strategy, determine whether the target is a legal target in the preset target library. If the target is a legal target in the preset target library, execute S605. If the target is not a legal target in the preset target library, execute S606.

[0112] S605: End tracking of the target.

[0113] In the embodiments of this application, S601 to S605 have been described in detail in S501 to S505, and will not be repeated here.

[0114] S606: Determine whether the target's position is within the preset second detection line of the detection area.

[0115] S607: If the target's location is within the preset second detection line of the detection area, a second warning message is triggered, and image data or video data containing the target is sent to the monitoring platform.

[0116] In this embodiment, the first target is still described. When the distance between the first target and the preset second detection line is less than the preset distance, and the first target is not a legal target in the preset target library, the tracking of the first target is maintained, and it is determined whether the first target is within the preset second detection line in the detection area. If the first target is within the preset second detection line in the detection area, the second warning information is triggered, and the image data or video data containing the first target is sent to the monitoring platform.

[0117] It should be noted that, taking a school as an example, when dividing the detection area, the first preset detection line is drawn 500 meters away from the school gate, and the second preset detection line is drawn at the school gate. The first preset detection line is ahead of the second preset detection line. When a target crosses the first preset detection line, a first warning message is sent, and other detections are performed according to the first detection strategy. Only after a target has crossed the first preset detection line can it cross the second preset detection line. When a user crosses the second preset detection line, a second warning message is sent, and other detections are performed according to the second detection strategy. Therefore, the second warning message is of a higher level than the first warning message. After receiving the first warning message, the monitoring platform only tracks the first target. After receiving the second warning message, the monitoring platform will acquire real-time image or video data of the first target to form the target's movement trajectory.

[0118] Through the above Figures 2 to 6 The technical solution provided in the embodiments of this application describes how this application determines different detection strategies for different behaviors of all targets in the detection area, detects each target based on different detection strategies, generates different levels of early warning information, and realizes automated detection of the detection area.

[0119] In S102, all targets in the detection area are identified, specifically as follows: Figure 7 As shown, Figure 7 This application provides a schematic diagram of the implementation process of a target determination method, which may include the following steps:

[0120] S701: Acquire real-time image data of the detection area and input the real-time image data into the target detection model.

[0121] S702: Obtain all targets in the detection area output by the target detection model.

[0122] In this embodiment of the application, for any detection area, real-time image data of the detection area is acquired, the real-time image data is input into the target detection model, and all targets in the detection area are output from the target detection model.

[0123] In S701, the target detection model in S701 can be specifically implemented through... Figure 8 The steps shown are as follows: Figure 8 This application provides a schematic diagram of the implementation process of a target detection model training method, which may include the following steps:

[0124] S801: Obtain sample image data in a preset scene, wherein the sample image data includes sample targets and the sample targets have been labeled.

[0125] In this embodiment, the preset scene includes general scenes and scenes to be detected. General scenes include, but are not limited to, highways, shopping malls, and parks. Scenes to be detected include the scene to be detected and scenes similar to it. For example, if the scene to be detected is a school, sample images of that school and other schools can be collected. It should be noted that the acquired sample image data of the preset scene can include sample image data of the preset scene under different time periods, different weather conditions, different seasons, etc.

[0126] In this embodiment, the acquired sample image data is manually annotated to mark the target objects in the sample image data. For example, when the target objects are people and vehicles, all target objects in the sample image data are marked. It can be understood that people and vehicles in a concealed or semi-concealed state should also be marked.

[0127] It should be noted that when the sample target is a person and a vehicle, the sample image data may contain only sample image data containing people, or only sample image data containing vehicles, or sample image data containing both people and vehicles. This application does not limit this.

[0128] In this embodiment of the application, the sample image data can also be obtained by first obtaining sample video data, and then performing frame extraction processing on the sample video data to obtain sample image data.

[0129] S802: Process the sample image data to obtain target sample image data; wherein, the processing of the sample image data includes at least one or more of the following: appearance enhancement processing, geometric enhancement processing, and virtual sample enhancement processing.

[0130] In this embodiment, the sample image data acquired in S801 is enhanced to obtain target sample image data, thereby expanding the sample image dataset. This includes performing appearance enhancement processing on the sample image data to obtain target sample image data, performing geometric enhancement processing on the sample image data to obtain target sample image data, and further performing virtual sample enhancement on the sample image data to obtain target sample image data.

[0131] Appearance enhancement and / or geometric enhancement increase the diversity of sample image data through data augmentation, including but not limited to: JPEG (Joint Photographic Experts Group) compression, brightness and contrast enhancement, Gamma brightness enhancement, Gaussian blur, motion blur, Gaussian noise, and rotation at random angles (blank areas are filled with the average pixels of the entire image). Virtual sample enhancement increases the diversity of sample image data through image synthesis, including but not limited to: synthesizing sample image data and constructing various types of targets to add to sample image data.

[0132] S803: Train the initial detection model based on the target sample image data to obtain the target detection model.

[0133] In this embodiment of the application, all target sample image data are input into the initial detection model for model training to obtain the target detection model.

[0134] Figure 9 This is a schematic diagram illustrating the implementation process of another object detection model training method provided in this application embodiment. The method may include the following steps:

[0135] S901: Obtain sample image data in a preset scene, wherein the sample image data includes sample targets and the sample targets have been labeled.

[0136] S902: Process the sample image data to obtain target sample image data; wherein, the processing of the sample image data includes at least one or more of the following: appearance enhancement processing, geometric enhancement processing, and virtual sample enhancement processing.

[0137] In the embodiments of this application, S901 to S902 have been described in detail in S801 to S802, and will not be repeated here.

[0138] S903: Perform pruning and / or channel reduction processing on the initial detection model.

[0139] S904: Determine the hyperparameters in the processed initial detection model.

[0140] S905: Train the processed initial detection model based on the target sample image data to adjust the hyperparameters and obtain the target detection model.

[0141] The following provides a unified explanation of S903 to S905:

[0142] In this embodiment, a YOLO V3 neural network is used to train the detection model. First, the YOLO V3 neural network is pruned and / or its channels are reduced; specifically, the number of channels in each convolutional layer is halved to obtain an initial detection model. Then, the hyperparameters in the initial detection model are determined through optimal selection. Optimal selection of hyperparameters includes obtaining an optimal solution based on existing training samples using clustering algorithms or prior experience. The target sample image data is then input into the processed initial detection model for training to adjust the hyperparameters, thereby obtaining the target detection model.

[0143] Based on the above description of the technical solutions provided in the embodiments of this application, this application expands sample image data through data augmentation to obtain diversity in multiple scenes, weather, and environments; increases sample diversity through image synthesis; balances detection accuracy and computational processing speed by pruning and / or reducing channels of the initial detection model; and optimizes the training process by selecting the optimal hyperparameters.

[0144] Figure 10 This is a schematic diagram of the structure of a target detection device provided in an embodiment of this application. The device includes: a detection area determination module 1001, a target acquisition module 1002, a detection strategy determination module 1003, and a target detection module 1004.

[0145] Detection area determination module 1001: used to receive real-time image data sent by multiple detection devices and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device;

[0146] Target acquisition module 1002: used to determine all targets in the detection area corresponding to any detection device;

[0147] Detection strategy determination module 1003: for any target in the detection area, it obtains the position of the target and determines the detection strategy of the target based on the position;

[0148] Target detection module 1004: Used to detect targets based on target detection strategies.

[0149] Figure 11 This is a schematic diagram of the structure of an edge computing server provided in an embodiment of this application. Figure 11The edge computing server 1100 shown includes at least one processor 1101, memory 1102, at least one network interface 1104, and a user interface 1103. The various components in the edge computing server 1100 are coupled together via a bus system 1105. It is understood that the bus system 1105 is used to implement communication between these components. In addition to a data bus, the bus system 1105 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 11 The general designated all buses as Bus System 1105.

[0150] The user interface 1103 may include a display, keyboard or clicking device (e.g., mouse, trackball), touchpad or touch screen, etc.

[0151] It is understood that the memory 1102 in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 1102 described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0152] In some implementations, memory 1102 stores elements, executable units or data structures, or subsets thereof, or extended sets thereof: operating system 11021 and application program 11022.

[0153] The operating system 11021 includes various system programs, such as a framework layer, a core library layer, and a driver layer, used to implement various basic business functions and handle hardware-based tasks. The application program 11022 includes various applications, such as a media player and a browser, used to implement various application functions. Programs implementing the methods of the embodiments of this application can be included in the application program 11022.

[0154] In this embodiment, by calling the program or instructions stored in memory 1102, specifically the program or instructions stored in application program 11022, processor 1101 executes the method steps provided in each method embodiment, including, for example:

[0155] The system receives real-time image data from multiple detection devices and determines the area captured by any one of the real-time image data as the detection area corresponding to that device. For any detection area, it identifies all targets within that area. For any target within the detection area, it obtains the target's location and determines a detection strategy based on that location. Finally, it detects the target based on the target's detection strategy.

[0156] The methods disclosed in the embodiments of this application can be applied to or implemented by processor 1101. Processor 1101 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 1101 or by instructions in the form of software. The processor 1101 may 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. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software units in the decoding processor. The software units may be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 1102. Processor 1101 reads the information in memory 1102 and, in conjunction with its hardware, completes the steps of the above method.

[0157] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions of this application, or combinations thereof.

[0158] For software implementation, the techniques described herein can be implemented through units that perform the functions described herein. The software code can be stored in memory and executed by a processor. The memory can be implemented within the processor or external to the processor.

[0159] The edge computing server provided in this embodiment can be as follows: Figure 11 The edge computing server shown can perform operations such as Figure 1-9 All steps of the target detection method are implemented to achieve... Figure 1-9 For details on the technical effectiveness of the target detection method, please refer to [link / reference]. Figure 1-9 The relevant descriptions are presented concisely and will not be elaborated upon here.

[0160] This application also provides a storage medium (computer-readable storage medium). This storage medium stores one or more programs. The storage medium may include volatile memory, such as random access memory; it may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive; and it may also include combinations of the above types of memory.

[0161] When one or more programs in the storage medium can be executed by one or more processors to implement the target detection method described above that is executed on the edge computing server side.

[0162] The processor is used to execute the target detection program stored in memory to implement the following steps of the target detection method executed on the edge computing server side:

[0163] The system receives real-time image data from multiple detection devices and determines the area captured by any one of the real-time image data as the detection area corresponding to that device. For any detection area, it identifies all targets within that area. For any target within the detection area, it obtains the target's location and determines a detection strategy based on that location. Finally, it detects the target based on the target's detection strategy.

[0164] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0165] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0166] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A target detection method, characterized in that, Applied to edge computing servers, the method includes: Receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any one of the detection devices as the detection area corresponding to the detection device; For any detection device, determine all targets within the detection area. For any target in the detection area, the position of the target is obtained, and the detection strategy of the target is determined based on the position, including: if the position of the target is inside a preset first detection line in the detection area, then a first detection strategy is determined; Detecting the target based on the target detection strategy includes: triggering a first warning message based on the first detection strategy, and tracking the target; The tracking of the target includes: acquiring the target's morphological information, which includes: a person's standing, crouching, or bending state, and a vehicle's door being open or closed; determining whether the target's position and / or morphological information has changed within a preset time period; if the target's position and / or morphological information has not changed within the preset time period, then ending the tracking of the target; if the target's position and / or morphological information has changed within the preset time period, then continuing the tracking of the target.

2. The method according to claim 1, characterized in that, The detection strategy for determining the target based on the location further includes: If the distance between the target's position and a preset second detection line within the detection area is less than a preset distance, then a second detection strategy is determined, wherein a preset first detection line within the detection area is located before a preset second detection line within the detection area; The detection strategy based on the target further includes: Based on the second detection strategy, it is determined whether the target is a legitimate target in the preset target library; If the target is a legitimate target in a preset target library, then the tracking of the target ends; If the target is not a legitimate target in the preset target library, a second warning message is triggered, and image data or video data containing the target is sent to the monitoring platform.

3. The method according to claim 2, characterized in that, The triggering of the second early warning information and the sending of image or video data containing the target to the monitoring platform includes: If the target's location is within a preset second detection line in the detection area, a second warning message is triggered, and image data or video data containing the target is sent to the monitoring platform.

4. The method according to claim 1, characterized in that, Determining all targets in the detection area includes: Acquire real-time image data of the detection area and input the real-time image data into the target detection model; Obtain all targets in the detection region as output by the target detection model; The target detection model is obtained in the following way: Acquire sample image data under a preset scenario, wherein the sample image data includes sample targets, and the sample targets have been labeled; The sample image data is processed to obtain target sample image data; wherein the processing of the sample image data includes at least one or more of the following: appearance enhancement processing, geometric enhancement processing, and virtual sample enhancement processing. The initial detection model is trained based on the target sample image data to obtain the target detection model.

5. The method according to claim 4, characterized in that, The step of training the initial detection model based on the target sample image data to obtain the target detection model includes: Pruning and / or channel reduction are performed on the initial detection model; Determine the hyperparameters in the processed initial detection model; The processed initial detection model is trained based on the target sample image data to adjust the hyperparameters and obtain the target detection model.

6. A target detection device, characterized in that, The device, applied to an edge computing server, includes: Detection area determination module: used to receive real-time image data sent by multiple detection devices, and determine the area captured by the real-time image data sent by any detection device as the detection area corresponding to the detection device; Target acquisition module: used to determine all targets in the detection area corresponding to any detection device; Detection strategy determination module: for any target in the detection area, to obtain the position of the target and determine the detection strategy of the target based on the position, including: if the position of the target is inside a preset first detection line in the detection area, then determine a first detection strategy; Target detection module: used to detect the target based on the target detection strategy, including: triggering a first warning message based on the first detection strategy, and tracking the target; The tracking of the target includes: acquiring the target's morphological information, which includes: a person's standing, crouching, or bending state, and a vehicle's door being open or closed; determining whether the target's position and / or morphological information has changed within a preset time period; if the target's position and / or morphological information has not changed within the preset time period, then ending the tracking of the target; if the target's position and / or morphological information has changed within the preset time period, then continuing the tracking of the target.

7. An edge computing server, characterized in that, include: A processor and a memory, the processor being configured to execute a program stored in the memory to implement any one of the target detection methods of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium stores one or more programs, which can be executed by one or more processors to implement any one of the target detection methods of claims 1 to 5.

Images