An event detection method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application provide an event detection method and device, electronic equipment and storage medium. Based on the video data stream collected for the environment to be detected, a target detection model is used to obtain the target category of each target to be detected, and for each target category, a corresponding category tracking frame set is used to track the target trajectory of each target to be detected. Each category tracking frame contains multiple tracking units, and each tracking unit can be used to track the target trajectory of one target to be detected. Based on the above method, the targets to be detected in the airport environment are tracked in stages according to the target category, avoiding the problem of tracking errors caused by information exchange between targets to be detected in different categories, thereby improving the accuracy of the tracked target trajectory and the accuracy of the corresponding detected security events.

Description

Technical Field

[0001] This invention relates to the field of target detection technology, and in particular to an event detection method, apparatus, electronic device, and storage medium. Background Technology

[0002] In smart airports, target detection and tracking are used for each target to be inspected. Combined with tripwire intrusion and preset judgment logic, various security events (such as bridge docking events and bridge removal events) that occur in the airport environment can be detected and alarmed, so that airport personnel can keep track of the key processes of flight security.

[0003] In existing technologies, for each detected target, the location coordinate information is used to correlate the consecutive video frames collected from the airport environment to track the target trajectory. This method is prone to causing information exchange problems between adjacent targets, which in turn leads to the tracking of each target trajectory being inaccurate and affects the accurate alarm of the security event. Summary of the Invention

[0004] This application provides an event detection method, apparatus, electronic device, and storage medium to improve the accuracy of event alarms.

[0005] In a first aspect, embodiments of this application provide an event detection method, including:

[0006] A target detection model is used to detect targets in the video data stream collected from the environment to be inspected, and the target category of each target in the environment to be inspected is obtained.

[0007] For each of the target categories, a corresponding category tracking box is used to track the target trajectory of each target to be inspected. Each category tracking box contains multiple tracking units, and each tracking unit is used to track the target trajectory of one target to be inspected.

[0008] Based on the positional relationship between each target trajectory and the set warning area, it is detected whether each target to be inspected has experienced a protection event, wherein the protection event is an event associated with each target to be inspected in the environment to be inspected.

[0009] In one alternative implementation, the target detection model is trained in the following manner:

[0010] Obtain a training sample set, wherein each training sample includes: input information corresponding to a training target and a true category label, wherein the training target is associated with the environment to be inspected;

[0011] Using training samples from the aforementioned training sample set, the preset detection model is trained iteratively through multiple rounds. When the convergence condition is met, the target detection model is output. During one round of iterative training, the following operations are performed:

[0012] Using the aforementioned detection model, a predicted category is obtained based on the input information in the training samples, and the parameters of the detection model are adjusted based on the adaptive loss value between the predicted category and the corresponding true category label.

[0013] In one alternative implementation, the adaptive loss value is calculated as follows:

[0014] Based on the predicted category of a training sample and the corresponding true category label of the training sample, calculate the original loss value between the predicted category and the true category label;

[0015] The original loss value is weighted using adaptive weights corresponding to the true category label from a preset weight set to obtain an adaptive loss value between the predicted category and the true category label. The adaptive weights are associated with the proportion of each training sample in the training sample set corresponding to the true category label.

[0016] In one optional implementation, the step of tracking the target trajectory of each target to be inspected using a set corresponding category tracking box for each target category includes:

[0017] For each of the target categories, perform the following operations respectively:

[0018] Based on the confidence level of at least one target to be inspected corresponding to a target category, and using a set confidence threshold, each target to be inspected with a confidence level greater than the confidence threshold is obtained from the at least one target to be inspected, and these targets are respectively the targets to be matched.

[0019] Multiple tracking units in the corresponding category tracking frame, set for a target category, are used to track the target trajectory of each target to be matched;

[0020] In response to the completion of target trajectory tracking for each of the aforementioned targets to be matched, the remaining tracking units are used to track the target trajectories of the remaining targets among the at least one target to be inspected.

[0021] In an optional implementation, after tracking the target trajectories of the remaining targets among the at least one target to be inspected, the method further includes:

[0022] Determine whether the tracking of the target trajectory of each of the remaining targets to be inspected has ended; wherein:

[0023] If so, then retain the target trajectory of each target to be inspected obtained through tracking;

[0024] Otherwise, a new tracking unit is used to continue tracking the target trajectories of the remaining targets to be inspected.

[0025] In one optional implementation, detecting whether a protection event has occurred for each of the targets to be inspected based on the positional relationship between each target trajectory and the set warning area includes:

[0026] For each of the targets to be inspected, perform any one of the following operations:

[0027] In response to the fact that the target trajectory of a target under inspection does not cross the set warning area, the current working state of the target under inspection is detected, and based on the current working state, each first image associated with the target under inspection is acquired, and based on the number of the first images, it is detected whether a protection event has occurred for the target under inspection. The first image is an image acquired after the target under inspection enters the current working state and before the target under inspection exits the current working state, and the current working state is the state associated with the protection event of the target under inspection.

[0028] In response to a target trajectory of an inspected object entering a set warning area, each second image associated with the inspected object is acquired, and based on the number of the second images, it is detected whether a protection event has occurred for the inspected object. The second images are images acquired after the target trajectory of the inspected object enters the warning area and before the target trajectory of the inspected object leaves the warning area.

[0029] In response to a target trajectory of an inspected object leaving a pre-defined warning area, third images associated with the inspected object are acquired. Based on the number of the third images, a guarantee event is detected for the inspected object. The third images are images captured after the target trajectory of the inspected object leaves the warning area and before the target trajectory of the inspected object enters the warning area. In an optional implementation, detecting the current operating status of the inspected object includes any one of the following operations:

[0030] Using the target detection model, a target detection box is obtained for the target to be detected, and the current working status of the target to be detected is detected based on the aspect ratio of the target detection box.

[0031] Using the target detection model, a target detection box is obtained for the target to be inspected, and the current working status of the target to be inspected is detected based on the positional relationship between the target detection box and the set rule box.

[0032] In an optional embodiment, after detecting whether a protection event has occurred for each of the targets to be inspected based on the positional relationship between the target trajectories and the set warning areas, the method further includes:

[0033] In response to the detection of a target to be alerted, an alarm is issued for the protection event corresponding to the target to be alerted, wherein the target to be alerted is the target to be detected in which the protection event occurs.

[0034] Secondly, embodiments of this application provide an event detection device, comprising:

[0035] The target detection module is used to perform target detection on the video data stream collected for the environment under inspection using a target detection model, and to obtain the target category of each target under inspection in the environment under inspection.

[0036] The trajectory tracking module is used to track the target trajectory of each target to be inspected using a set corresponding category tracking box for each target category. Each category tracking box contains multiple tracking units, and each tracking unit is used to track the target trajectory of one target to be inspected.

[0037] An event detection module is used to detect whether a safeguard event has occurred for each of the targets to be inspected, based on the positional relationship between each target trajectory and a set warning area. The safeguard event is an event associated with each target to be inspected in the environment to be inspected. In an optional implementation, the target detection model is trained in the following manner:

[0038] Obtain a training sample set, wherein each training sample includes: input information corresponding to a training target and a true category label, wherein the training target is associated with the environment to be inspected;

[0039] Using training samples from the aforementioned training sample set, the preset detection model is trained iteratively through multiple rounds. When the convergence condition is met, the target detection model is output. During one round of iterative training, the following operations are performed:

[0040] Using the aforementioned detection model, a predicted category is obtained based on the input information in the training samples, and the parameters of the detection model are adjusted based on the adaptive loss value between the predicted category and the corresponding true category label.

[0041] In one alternative implementation, the adaptive loss value is calculated as follows:

[0042] Based on the predicted category of a training sample and the corresponding true category label of the training sample, calculate the original loss value between the predicted category and the true category label;

[0043] The original loss value is weighted using adaptive weights corresponding to the true category label from a preset weight set to obtain an adaptive loss value between the predicted category and the true category label. The adaptive weights are associated with the proportion of each training sample in the training sample set corresponding to the true category label.

[0044] In one optional implementation, for each of the target categories, a corresponding category tracking box is used to track the target trajectory of each target to be inspected. The trajectory tracking module is used for:

[0045] For each of the target categories, perform the following operations respectively:

[0046] Based on the confidence level of at least one target to be inspected corresponding to a target category, and using a set confidence threshold, each target to be inspected with a confidence level greater than the confidence threshold is obtained from the at least one target to be inspected, and these targets are respectively the targets to be matched.

[0047] Multiple tracking units in the corresponding category tracking frame, set for a target category, are used to track the target trajectory of each target to be matched;

[0048] In response to the completion of target trajectory tracking for each of the aforementioned targets to be matched, the remaining tracking units are used to track the target trajectories of the remaining targets among the at least one target to be inspected.

[0049] In an optional implementation, after tracking the target trajectories of the remaining targets among the at least one target to be inspected, the trajectory tracking module is further configured to:

[0050] Determine whether the tracking of the target trajectory of each of the remaining targets to be inspected has ended; wherein:

[0051] If so, then retain the target trajectory of each target to be inspected obtained through tracking;

[0052] Otherwise, a new tracking unit is used to continue tracking the target trajectories of the remaining targets to be inspected.

[0053] In one optional implementation, the step of detecting whether a protection event has occurred for each of the targets to be inspected, based on the positional relationship between each target trajectory and the set warning area, is performed by the event detection module for:

[0054] For each of the targets to be inspected, perform any one of the following operations:

[0055] In response to the fact that the target trajectory of a target under inspection does not cross the set warning area, the current working state of the target under inspection is detected, and based on the current working state, each first image associated with the target under inspection is acquired, and based on the number of the first images, it is detected whether a protection event has occurred for the target under inspection. The first image is an image acquired after the target under inspection enters the current working state and before the target under inspection exits the current working state, and the current working state is the state associated with the protection event of the target under inspection.

[0056] In response to a target trajectory of an inspected object entering a set warning area, each second image associated with the inspected object is acquired, and based on the number of the second images, it is detected whether a protection event has occurred for the inspected object. The second images are images acquired after the target trajectory of the inspected object enters the warning area and before the target trajectory of the inspected object leaves the warning area.

[0057] In response to a target trajectory of an inspected object leaving a pre-defined warning area, third images associated with the inspected object are acquired, and based on the number of the third images, a protection event is detected for the inspected object. The third images are images acquired after the target trajectory of the inspected object leaves the warning area and before the target trajectory of the inspected object enters the warning area.

[0058] In one optional implementation, the event detection module is used to perform any of the following operations to detect the current working state of the target object:

[0059] Using the target detection model, a target detection box is obtained for the target to be detected, and the current working status of the target to be detected is detected based on the aspect ratio of the target detection box.

[0060] Using the target detection model, a target detection box is obtained for the target to be inspected, and the current working status of the target to be inspected is detected based on the positional relationship between the target detection box and the set rule box.

[0061] In an optional embodiment, after detecting whether a protection event has occurred for each of the targets to be inspected based on the positional relationship between the target trajectories and the set warning areas, the event detection module is further configured to:

[0062] In response to the detection of a target to be alerted, an alarm is issued for the protection event corresponding to the target to be alerted, wherein the target to be alerted is the target to be detected in which the protection event occurs.

[0063] Thirdly, an electronic device is proposed, comprising a processor and a memory, wherein the memory stores program code that, when executed by the processor, causes the processor to perform the steps of the event detection method described in the first aspect.

[0064] Fourthly, a computer-readable storage medium is proposed, comprising program code that, when executed on an electronic device, causes the electronic device to perform the steps of the event detection method described in the first aspect.

[0065] The technical effects of the embodiments of this application are as follows:

[0066] This application provides an event detection method, apparatus, electronic device, and storage medium. Based on video data streams collected from the environment to be inspected, a target detection model is used to obtain the target category of each target to be inspected. For each target category, a corresponding category tracking frame is used to track the target trajectory of each target to be inspected. Each category tracking frame contains multiple tracking units, and each tracking unit can be used to track the target trajectory of one target to be inspected. Based on the above method, each target to be inspected in the airport environment is tracked in stages according to the target category, avoiding the problem of trajectory tracking errors caused by information exchange between targets to be inspected in different categories, thereby improving the accuracy of the tracked target trajectories and the accuracy of the corresponding detected security events. Attached Figure Description

[0067] Figure 1 A flowchart of an event detection method provided in an embodiment of this application;

[0068] Figure 2 A schematic diagram of a warning area provided in an embodiment of this application;

[0069] Figure 3 A schematic diagram illustrating the detection of a safeguard event, provided as an embodiment of this application;

[0070] Figure 4a This application provides a schematic diagram of a covered bridge approaching the bridge.

[0071] Figure 4b This application provides a schematic diagram of a covered bridge that can be removed.

[0072] Figure 5 This is a schematic diagram of the structure of an event detection device provided in an embodiment of this application;

[0073] Figure 6 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0074] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this invention.

[0075] It should be noted that in the description of this application, "multiple" is understood as "at least two". "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. A connected to B can represent: A and B directly connected, or A and B connected through C. Furthermore, in the description of this application, terms such as "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order.

[0076] Furthermore, the data collection, dissemination, and use in the technical solution of this application all comply with the requirements of relevant national laws and regulations.

[0077] The design concept of this application is as follows:

[0078] In existing technologies, for each detected target, the location coordinate information is used to correlate consecutive video frames collected from the airport environment to track the target trajectory. This method is prone to causing information exchange problems between adjacent targets, which leads to insufficient accuracy in tracking the target trajectory and affects the accuracy of the detected security events.

[0079] To improve the accuracy of event alarms, embodiments of this application provide an event detection method, apparatus, electronic device, and storage medium. Based on video data streams collected from the environment under inspection, a target detection model is used to detect the target category of each target under inspection. For each target category, a corresponding category tracking box is used to track the target trajectory of each target under inspection. Each category tracking box contains multiple tracking units, and each tracking unit can be used to track the target trajectory of one target under inspection. Based on the above method, each target under inspection in the airport environment is tracked in stages according to its target category, avoiding the information exchange problem between targets under inspection of different categories, thereby improving the accuracy of the tracked target trajectories and the corresponding detection of security events. On the other hand, parallel tracking using category tracking boxes for each target category further improves the tracking efficiency of target trajectories.

[0080] The event detection method provided in this application will now be described and explained in detail with reference to the accompanying drawings.

[0081] First, the event detection method provided in this application is specifically applied in the field of smart airport technology, and it can be integrated into electronic devices, such as terminals and servers.

[0082] The terminals can include Android devices, iOS devices, mobile internet devices (MIDs), wireless terminal devices in smart cities, vehicle cameras, mobile phones, tablets, smart Bluetooth devices, laptops, or personal computers, etc.

[0083] A server can be a single server or a server cluster consisting of multiple servers. The servers include, but are not limited to, servers used to provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0084] Furthermore, the algorithms related to the event detection method provided in this application can also be deployed on network cameras (IP cameras, IPCs) or network video recorders (NVRs), or on decision analysis systems in smart airports, such as Airport Collaborative Decision Making (ACDM).

[0085] It is understood that the event detection method in this application embodiment can be executed on a terminal, on a server, or jointly by a terminal and a server. The above examples should not be construed as limiting this application.

[0086] Furthermore, based on the above description, the event detection method provided in this application can be applied to a real environment that requires target detection and / or target tracking, such as an airport environment or a road traffic environment. For ease of understanding, an airport environment will be used as an example. Furthermore, the assurance event can be an event associated with the environment under inspection, such as a critical event for flight assurance. Specifically, it can include, but is not limited to, one or more of the following events:

[0087] For events related to aircraft entering a parking space, aircraft leaving a parking space, jet bridge docking, jet bridge removal, support vehicle arrival, and support vehicle departure, the aforementioned support vehicles may include, but are not limited to, one or more of the following: fuel trucks, catering trucks, baggage trucks, guide vehicles, and tractor vehicles.

[0088] Taking the aforementioned security events as examples, in this embodiment of the application, aircraft, landing gear, jet bridges, fuel trucks, catering trucks, baggage carts, guide vehicles, and tractor vehicles in the airport environment are regarded as various targets to be inspected in the airport environment, and multiple collected images or video frames containing the aforementioned targets to be inspected are regarded as video data streams collected for the airport environment.

[0089] See Figure 1 As shown, the event detection method provided in this application specifically includes:

[0090] S101: Use a target detection model to perform target detection on the video data stream collected for the environment under inspection, and obtain the target category of each target under inspection in the environment under inspection.

[0091] Optionally, cameras and webcams deployed in the airport environment can be used to collect video data streams related to each target to be inspected, and the target category of each target to be inspected can be detected through a target detection model.

[0092] Specifically, the target category to be inspected can be one or more of the above-mentioned aircraft, landing gear, jet bridge, fuel truck, catering truck, baggage truck, guide vehicle, and tractor.

[0093] In one optional implementation, to improve the accuracy of the predicted category, the following steps are used to train the object detection model, including:

[0094] Step 1: Obtain a training sample set, wherein each training sample includes: input information corresponding to a training target and a true category label, wherein the training target is associated with the environment to be inspected.

[0095] Step 2: Using training samples from the training sample set, perform multiple rounds of iterative training on the preset detection model. When the convergence condition is met, output the target detection model. In one round of iterative training, perform the following operations: using the detection model, obtain the predicted category based on the input information in the training samples, and adjust the parameters of the detection model based on the adaptive loss value between the predicted category and the corresponding real category label.

[0096] Specifically, images or video frames containing each training target are used as training samples for the detection model. Each training target is associated with the airport environment. For example, images or video frames containing aircraft, landing gear, jet bridges, fuel trucks, catering trucks, baggage carts, guide vehicles, and tractor vehicles are used as training samples. The predicted category of each training target is obtained through the detection model.

[0097] Optionally, to avoid class prediction bias caused by the long-tailed distribution of sample data (meaning that the number of individuals with extremely low values ​​accounts for the majority of the population), this application may use an adaptive loss value to iteratively adjust the parameters of the detection model during model training.

[0098] Specifically, the adaptive loss value during one round of training is calculated using the following steps:

[0099] Step 1: Based on the predicted class of a training sample and the corresponding true class label of the training sample, calculate the original loss value between the predicted class and the true class label.

[0100] Step 2: Using the adaptive weights corresponding to the true category labels in the preset weight set, the original loss value is weighted to obtain the adaptive loss value between the predicted category and the true category label. The adaptive weights are associated with the proportion of each training sample in the training sample set corresponding to the true category label.

[0101] Based on the above method, this application adaptively weights different prediction categories, thereby driving the model to learn from a smaller number of individuals, so as to improve the detection accuracy of the target detection model for training targets such as guide vehicles and conveyor belt vehicles in airport environments where corresponding sample images are difficult to collect; corresponding to the adaptive loss value calculated above, the adaptive loss function adopted in this application is as shown in the following equation (1):

[0102]

[0103] Where n is the total number of training samples, c is the number of predicted classes, y is the true class label for a training target, p is the predicted class, and α is the adaptive weight. n i This represents the number of training samples corresponding to the true category labels in the training sample set.

[0104] Based on the above method, this application can detect each target to be detected in the airport environment through the target detection model after training, and use the predicted category output by the model as the target category of the target to be detected.

[0105] S102: For each target category, the corresponding category tracking box is set to track the target trajectory of each target to be inspected.

[0106] Specifically, based on the target categories output by the target detection model, different categories of tracking boxes are set to track the targets to be detected separately. Each category tracking box can contain multiple tracking units, and each tracking unit is used to track the target trajectory of one target to be detected.

[0107] For example, assuming in the embodiments of this application, for the targets to be inspected 1 to n, their corresponding target categories (assumed to be OD_Res1 to OD_Res) K Any one of the following (K is at least one) and the corresponding category tracking boxes can be shown in Table 1 below:

[0108] Table 1

[0109] Target to be inspected Target Category Category Tracking Box Target to be inspected 1 <![CDATA[Aircraft / OD_Res1]]> <![CDATA[TRK_Res1]]> Target to be inspected 2 <![CDATA[Aircraft / OD_Res1]]> <![CDATA[TRK_Res1]]> Target 3 to be inspected <![CDATA[Covered Bridge / OD_Res2]]> <![CDATA[TRK_Res2 <!-- 7 -->]]> Target 4 to be inspected <![CDATA[Guided vehicle / OD_Res3]]> <![CDATA[TRK_Res3]]> … … … Target n to be inspected <![CDATA[Aircraft / OD_Res1]]> <![CDATA[TRK_Res1]]>

[0110] Based on the above method, this application uses different category tracking boxes to track different categories of targets to be inspected, avoiding information exchange problems between different categories of targets to be inspected; optionally, for each target category, the following operations are performed:

[0111] Step 1: Based on the confidence level of at least one target to be inspected corresponding to a target category, and using a set confidence threshold, obtain each target to be inspected whose confidence level is greater than the confidence threshold from at least one target to be inspected, which are the targets to be matched.

[0112] Optionally, a trained target detection model is used to obtain the detection results, and the target category and confidence level are obtained based on the detection results. The confidence level threshold can be adjusted according to the actual situation.

[0113] For example, taking the above-mentioned targets 1, 2, and n as examples, assuming the confidence threshold is 0.5 and the confidence levels are as shown in Table 2 below:

[0114] Table 2

[0115] Target to be inspected Target Category Confidence Target to be inspected 1 <![CDATA[Aircraft / OD_Res1]]> 0.6 Target to be inspected 2 <![CDATA[Aircraft / OD_Res1]]> 0.3 Target n to be inspected <![CDATA[Aircraft / OD_Res1]]> 0.7

[0116] Based on the above method, target 1 and target n to be inspected are respectively taken as targets to be matched.

[0117] Step 2: Using multiple tracking units in the corresponding category tracking frame set for a target category, track the target trajectory of each target to be matched.

[0118] Optionally, the category tracking box can carry multiple tracking units. Each tracking unit can use a target tracking algorithm to track the corresponding target to be detected, such as using the Kalman filter algorithm and the Hungarian matching algorithm to track the corresponding target to be detected.

[0119] For example, multiple trackers in the category tracking box TRK_Res1 are used to track the above-mentioned target 1 and target n.

[0120] Step 3: In response to the end of the target trajectory tracking of each target to be matched, the remaining tracking units are used to track the target trajectory of each of the remaining targets to be inspected in at least one of the targets to be inspected.

[0121] Specifically, after tracking the targets with high confidence levels, the remaining tracking units in the category tracking box are used to continue tracking the targets with low confidence levels.

[0122] For example, the remaining tracking unit is used to continue tracking the above-mentioned target 2 to be inspected.

[0123] Optionally, after using the remaining tracking units for tracking, it is determined whether the target trajectory of each target to be inspected in the current target category has been tracked. If the target trajectory has been obtained, it is retained; if the complete target trajectory of each target to be inspected has not been obtained, the newly added tracking unit under the tracking box of that category is used to continue tracking.

[0124] Based on the above method, this application tracks each target of the same target category separately according to the confidence level in the detection results, prioritizes the target trajectory of each target with higher confidence level, and continues to track each target with lower confidence level (e.g., occluded targets) to ensure the robustness of the target tracking algorithm.

[0125] S103: Based on the positional relationship between each target trajectory and the set warning area, detect whether each target to be inspected has experienced a protection event.

[0126] Optionally, warning zones can be set up based on the actual working area of ​​the airport environment. There can be one or more warning zones, which are used to detect events for different targets to be inspected.

[0127] For example, see Figure 2 As shown, a warning zone for security incidents is set up using solid lines on the airport apron.<A,B,C,D,E,F,G,H,I> The warning area further includes a sub-warning area a, which is used to warn of the parking status of the aircraft landing gear.

[0128] Optionally, to avoid false alarms caused by invalid targets at adjacent positions, logical operations and time constraints are used to detect safeguard events occurring at each target to be inspected. For details, please refer to [link / reference]. Figure 3 As shown, for a target to be inspected, perform any one of the following operations:

[0129] 1. In response to the fact that the target trajectory of a target to be inspected does not cross the set warning area, the current working status of the target to be inspected is detected, and based on the current working status, each first image associated with the target to be inspected is acquired, and based on the number of each first image, it is detected whether a protection event has occurred for the target to be inspected.

[0130] Specifically, the current working state refers to the state associated with a safeguard event of a target to be inspected. The first image is an image collected after a target to be inspected enters the current working state and before a target to be inspected exits the current working state.

[0131] For example, assuming a target to be inspected is a covered walkway, the current working state of the target to be inspected may be: a bridge-on state associated with a "bridge-on event", then the first image is an image captured after the covered walkway enters the bridge-on state and before it exits the bridge-on state; or, for another example, the current working state of the target to be inspected may also be a bridge-removal state associated with a "bridge-removal event", then the first image is an image captured after the covered walkway enters the bridge-removal state and before it exits the bridge-removal state.

[0132] In one optional implementation, the current working state of the target to be inspected can be obtained through any of the following detection operations:

[0133] (1) Using a target detection model, a target detection box for a target to be detected is obtained, and the current working status of a target to be detected is detected based on the aspect ratio of the target detection box.

[0134] Specifically, the target detection box can be a detection box generated for a single target when the target detection model is used to detect a target in a video data stream, or it can be a corresponding detection box generated when the target detection model is used to detect any image or video frame containing the single target. Based on the aspect ratio of the target detection box and the set aspect ratio threshold, the current working state of the single target can be determined.

[0135] For example, see Figure 4a As shown, assuming a target to be inspected is a bridge, and the aspect ratio threshold is set to 1.5 for this target, the target detection model is used to perform target detection on the image containing the target. Based on the fact that the aspect ratio of the generated target detection box is greater than the set aspect ratio threshold, the current working state of the target is determined to be the bridge-adjacent state.

[0136] For example, see Figure 4bAs shown, assuming a target to be inspected is a bridge, and the aspect ratio threshold is set to 1.5 for this target, the target detection model is used to perform target detection on the image containing the target. Based on the fact that the aspect ratio of the generated target detection box is less than or equal to the set aspect ratio threshold, the current working state of the target is determined to be the bridge removal state.

[0137] (2) Using a target detection model, obtain the target detection box for a target to be inspected, and detect the current working status of a target to be inspected based on the positional relationship between the target detection box and the set rule box.

[0138] Specifically, a rule box is set in the actual working area of ​​the inspection environment for a target to be inspected. The rule box can be one or more.

[0139] For example, suppose a target to be inspected is an aircraft, and for this target, the rule box is set as follows: Figure 2 The sub-warning area a shown uses a target detection model to detect the target in the image. If the generated target detection box intersects with the set rule box, the current working state of the target is determined to be in the in-position state. If the generated target detection box does not intersect with the set rule box, the current working state of the target is determined to be out of position.

[0140] Based on the above method, the image collected and associated with the corresponding current working state is taken as the first image of the target to be inspected. Based on the number of the first images, a set first threshold is used to determine whether a protection event has occurred. If the number of the first images is greater than the first threshold, it is determined that a protection event has occurred for the target to be inspected; otherwise, no protection event has occurred.

[0141] For example, assuming the target to be inspected is a covered walkway, and the number of corresponding first images collected for the covered walkway in the docked state is 30, then based on the set first threshold of 50, it is determined that the covered walkway has not experienced the corresponding "docked event".

[0142] 2. In response to a target trajectory of an inspected object entering a set warning area, acquire each second image associated with the inspected object, and based on the number of each second image, detect whether a protection event has occurred for the inspected object.

[0143] Specifically, the second image is an image captured after a target to be inspected enters the warning area and before a target to be inspected leaves the warning area.

[0144] For example, assuming a target to be inspected is an aircraft, the image of the target within the warning area after it enters the warning area and before it leaves the warning area is used as the second image.

[0145] Based on the above method, a set second threshold is used to determine whether a protection event has occurred based on the number of second images. If the number of second images is greater than the second threshold, it is determined that a protection event has occurred for the target to be inspected; otherwise, no protection event has occurred.

[0146] For example, assuming the target to be inspected is an aircraft, and the number of corresponding second images collected after the aircraft enters the warning area and before it leaves the warning area is 60, then based on the set second threshold of 40, it is determined that the aircraft has experienced a corresponding "entry event".

[0147] 3. In response to a target trajectory of an inspected object leaving a set warning area, acquire each third image associated with the inspected object, and based on the number of each third image, detect whether a protection event has occurred for the inspected object.

[0148] Specifically, the third image is an image captured after a target to be inspected leaves the warning area and before a target to be inspected enters the warning area.

[0149] For example, assuming a target to be inspected is an aircraft, the image outside the warning area after the target has left the warning area and before it has entered the warning area is used as the third image.

[0150] Based on the above method, a set third threshold is used to determine whether a protection event has occurred based on the number of third images. If the number of third images is greater than the third threshold, it is determined that a protection event has occurred for the target to be inspected; otherwise, no protection event has occurred.

[0151] For example, assuming the target to be inspected is an aircraft, and the number of corresponding third images collected after the aircraft leaves the warning area and before it re-enters the warning area is 50, then based on the set third threshold of 30, it is determined that the aircraft has experienced a corresponding "departure event".

[0152] In one optional embodiment, for the target trajectory of the target to be inspected to enter and / or exit a pre-set warning zone, the occurrence of the corresponding protection event can be quickly detected by the following logical operation state equation, which is shown in the following formula:

[0153]

[0154] Where (x,y)inArea represents that the target to be inspected is within the set warning area, (x,y)outArea represents that the target to be inspected is outside the set warning area, and Flag trip Num represents whether there is an entry or exit relationship between the target trajectory of the target under inspection and the warning area. Frame It represents the relationship between the number of corresponding images and a preset threshold.

[0155] Specifically, when the target to be inspected is within the set warning area, if the target's trajectory enters the set warning area, Flag... trip The value is 1. If the target trajectory does not enter the set warning area, Flag will be set to 1. trip The value is 0; when the target to be inspected is outside the set warning area, if the target trajectory moves out of the set warning area, Flag... trip The value is 1. If the target trajectory does not leave the set warning area, Flag will be set to 1. trip The value of is 0.

[0156] When the target trajectory of the target to be inspected enters the set warning area, if the number of acquired second images is greater than a preset second threshold, Num Frame The value of Num is 1. If the number of second images obtained is not greater than the preset second threshold, then Num... Frame The value is 0; when the target trajectory of the target to be detected leaves the set warning area, if the number of acquired third images is greater than the preset third threshold, Num Frame The value of Num is 1. If the number of acquired third images is not greater than the preset third threshold, then Num... Frame The value of is 0.

[0157] Based on the above method, by acquiring the number of corresponding images, a time constraint is imposed on the target to be inspected entering or leaving the warning area, so as to avoid detection errors in the protection event caused by the accidental entry of targets from adjacent camera positions.

[0158] In an optional embodiment, after detecting whether a protection event has occurred for each target based on the positional relationship between each target trajectory and the set warning area, the method further includes:

[0159] In response to the detection of a target to be alerted, an alert is issued for the corresponding protection event of the target.

[0160] Among them, the targets to be alarmed are the targets to be inspected in the event of a security incident.

[0161] For example, assuming that the targets to be inspected in the environment to be inspected are target 1, target 2, ... target 10, respectively, each of the above targets to be inspected will be taken as a target to be alarmed, so as to obtain the event detection status in the environment to be inspected in real time.

[0162] The following specific embodiments will be provided to illustrate the event detection method provided in this application in detail.

[0163] 1) Detection of aircraft entering and leaving the parking space.

[0164] In one specific embodiment, assuming a target to be inspected is detected as an aircraft, for the target to be inspected, when its target trajectory enters the set warning area, corresponding second images are acquired, and when the number of second images is greater than a preset second threshold, it is determined that the aircraft has entered the position event and an aircraft entry alarm is triggered; similarly, for the target to be inspected, when its target trajectory leaves the set warning area, corresponding third images are acquired, and when the number of third images is greater than a preset third threshold, it is determined that the aircraft has left the position event and an aircraft departure alarm is triggered.

[0165] 2) Detection of incidents involving the bridge being pulled close to or removed from the covered bridge.

[0166] In one specific embodiment, assuming a target to be inspected is detected as a boarding bridge, for the target to be inspected, if its current working state is a docked state, corresponding first images are acquired, and when the number of first images is greater than a preset first threshold, it is determined that the boarding bridge has docked and a boarding bridge docking alarm is triggered; similarly, for the target to be inspected, if its current working state is a dismounted state, corresponding first images are acquired, and when the number of first images is greater than a preset first threshold, it is determined that the aircraft has dismounted and a boarding bridge dismounting alarm is triggered.

[0167] 3) Detection of arrival and departure events of support vehicles.

[0168] In one specific embodiment, assuming a target to be inspected is detected as a support vehicle, wherein the support vehicle can be any one of aviation fuel vehicle, aviation catering vehicle, baggage vehicle, guide vehicle, and tractor vehicle, then for the target to be inspected, when its target trajectory enters the set warning area, corresponding second images are acquired, and when the number of second images is greater than a preset second threshold, it is determined that the support vehicle has arrived and a support vehicle arrival alarm is triggered; similarly, for the target to be inspected, when its target trajectory leaves the set warning area, corresponding third images are acquired, and when the number of third images is greater than a preset third threshold, it is determined that the support vehicle has left and a support vehicle departure alarm is triggered.

[0169] Based on the same technical concept, embodiments of this application also provide an event detection device, which is used to implement the above-described method flow of embodiments of this application. See also... Figure 5 As shown, the device includes: a target detection module 501, a trajectory tracking module 502, and an event detection module 503, wherein:

[0170] The target detection module 501 is used to perform target detection on the video data stream collected for the environment to be inspected using a target detection model, and to obtain the target category of each target to be inspected in the environment to be inspected.

[0171] The trajectory tracking module 502 is used to track the target trajectory of each target to be inspected using a set corresponding category tracking box for each target category. Each category tracking box contains multiple tracking units, and each tracking unit is used to track the target trajectory of one target to be inspected.

[0172] The event detection module 503 is used to detect whether a protection event has occurred for each of the targets to be inspected based on the positional relationship between each target trajectory and the set warning area, wherein the protection event is an event associated with each target to be inspected in the environment to be inspected.

[0173] In one alternative implementation, the target detection model is trained in the following manner:

[0174] Obtain a training sample set, wherein each training sample includes: input information corresponding to a training target and a true category label, wherein the training target is associated with the environment to be inspected;

[0175] Using training samples from the aforementioned training sample set, the preset detection model is trained iteratively through multiple rounds. When the convergence condition is met, the target detection model is output. During one round of iterative training, the following operations are performed:

[0176] Using the aforementioned detection model, a predicted category is obtained based on the input information in the training samples, and the parameters of the detection model are adjusted based on the adaptive loss value between the predicted category and the corresponding true category label.

[0177] In one alternative implementation, the adaptive loss value is calculated as follows:

[0178] Based on the predicted category of a training sample and the corresponding true category label of the training sample, calculate the original loss value between the predicted category and the true category label;

[0179] The original loss value is weighted using adaptive weights corresponding to the true category label from a preset weight set to obtain an adaptive loss value between the predicted category and the true category label. The adaptive weights are associated with the proportion of each training sample in the training sample set corresponding to the true category label.

[0180] In one optional implementation, for each of the target categories, a corresponding category tracking box is used to track the target trajectory of each target to be inspected. The trajectory tracking module 502 is used for:

[0181] For each of the target categories, perform the following operations respectively:

[0182] Based on the confidence level of at least one target to be inspected corresponding to a target category, and using a set confidence threshold, each target to be inspected with a confidence level greater than the confidence threshold is obtained from the at least one target to be inspected, and these targets are respectively the targets to be matched.

[0183] Multiple tracking units in the corresponding category tracking frame, set for a target category, are used to track the target trajectory of each target to be matched;

[0184] In response to the completion of target trajectory tracking for each of the aforementioned targets to be matched, the remaining tracking units are used to track the target trajectories of the remaining targets among the at least one target to be inspected.

[0185] In an optional implementation, after tracking the target trajectories of the remaining targets among the at least one target to be inspected, the trajectory tracking module 502 is further configured to:

[0186] Determine whether the tracking of the target trajectory of each of the remaining targets to be inspected has ended; wherein:

[0187] If so, then retain the target trajectory of each target to be inspected obtained through tracking;

[0188] Otherwise, a new tracking unit is used to continue tracking the target trajectories of the remaining targets to be inspected.

[0189] In one optional implementation, the event detection module 503 is used to detect whether a protection event has occurred for each of the targets to be inspected, based on the positional relationship between each target trajectory and the set warning area.

[0190] For each of the targets to be inspected, perform any one of the following operations:

[0191] In response to the fact that the target trajectory of a target under inspection does not cross the set warning area, the current working state of the target under inspection is detected, and based on the current working state, each first image associated with the target under inspection is acquired, and based on the number of the first images, it is detected whether a protection event has occurred for the target under inspection. The first image is an image acquired after the target under inspection enters the current working state and before the target under inspection exits the current working state, and the current working state is the state associated with the protection event of the target under inspection.

[0192] In response to a target trajectory of an inspected object entering a set warning area, each second image associated with the inspected object is acquired, and based on the number of the second images, it is detected whether a protection event has occurred for the inspected object. The second images are images acquired after the target trajectory of the inspected object enters the warning area and before the target trajectory of the inspected object leaves the warning area.

[0193] In response to a target trajectory of an inspected object leaving a pre-defined warning area, third images associated with the inspected object are acquired, and based on the number of the third images, a protection event is detected for the inspected object. The third images are images acquired after the target trajectory of the inspected object leaves the warning area and before the target trajectory of the inspected object enters the warning area.

[0194] In one optional implementation, the event detection module 503 is used to perform any of the following operations when detecting the current working state of the target to be inspected:

[0195] Using the target detection model, a target detection box is obtained for the target to be detected, and the current working status of the target to be detected is detected based on the aspect ratio of the target detection box.

[0196] Using the target detection model, a target detection box is obtained for the target to be inspected, and the current working status of the target to be inspected is detected based on the positional relationship between the target detection box and the set rule box.

[0197] In an optional embodiment, after detecting whether a protection event has occurred for each of the targets to be inspected based on the positional relationship between the target trajectories and the set warning areas, the event detection module 503 is further configured to:

[0198] In response to the detection of a target to be alerted, an alarm is issued for the protection event corresponding to the target to be alerted, wherein the target to be alerted is the target to be detected in which the protection event occurs.

[0199] Based on the same inventive concept as the embodiments described above, this application also provides an electronic device that can be used for event detection. In one embodiment, the electronic device can be a server, a terminal device, or other electronic equipment. In this embodiment, the structure of the electronic device can be as follows: Figure 6 As shown, it includes a memory 601, a communication interface 603, and one or more processors 602.

[0200] The memory 601 is used to store computer programs executed by the processor 602. The memory 601 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0201] Memory 601 may be volatile memory, such as random-access memory (RAM); memory 601 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 601 may be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 601 may be a combination of the above-described memories.

[0202] Processor 602 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 602 is used to implement the above-described event detection method when it calls the computer program stored in memory 601.

[0203] Communication interface 603 is used to communicate with terminal devices and other servers.

[0204] This application embodiment does not limit the specific connection medium between the memory 601, the communication interface 603, and the processor 602. This application embodiment... Figure 6 The memory 601 and the processor 602 are connected via a bus 604, and the bus 604 is in Figure 6 The connections between other components are shown in thick lines only and are not intended to be limiting. The 604 bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0205] Based on the same inventive concept, embodiments of this application also provide a storage medium storing computer instructions that, when executed on a computer, cause the computer to perform an event detection method as described above.

[0206] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0207] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0208] This application provides an event detection method, apparatus, electronic device, and storage medium. Based on video data streams collected from the environment to be inspected, a target detection model is used to obtain the target category of each target to be inspected. For each target category, a corresponding category tracking frame is used to track the target trajectory of each target to be inspected. Each category tracking frame contains multiple tracking units, and each tracking unit can be used to track the target trajectory of one target to be inspected. Based on the above method, each target to be inspected in the airport environment is tracked in stages according to the target category, avoiding the problem of trajectory tracking errors caused by information exchange between targets to be inspected in different categories, thereby improving the accuracy of the tracked target trajectories and the accuracy of the corresponding detected security events.

[0209] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0210] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a server, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0211] Program code for performing the operations of this application can be written using any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0212] In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0213] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0214] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0215] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An event detection method, characterized in that, include: A target detection model is used to detect targets in the video data stream collected for the environment under inspection, and the target category of each target in the environment under inspection is obtained. For each of the target categories, perform the following operations respectively: Based on the confidence level of at least one target to be inspected corresponding to a target category, and using a set confidence threshold, each target to be inspected with a confidence level greater than the confidence threshold is obtained from the at least one target to be inspected, and these targets are respectively the targets to be matched. Multiple tracking units in a corresponding category tracking box, set for a target category, are used to track the target trajectory of each target to be matched. Each category tracking box contains multiple tracking units, and each tracking unit is used to track the target trajectory of a target to be detected. In response to the completion of target trajectory tracking for each of the targets to be matched, the remaining tracking units are used to track the target trajectories of the remaining targets among the at least one target to be inspected. Based on the positional relationship between each target trajectory and the set warning area, it is detected whether each target to be inspected has experienced a protection event, wherein the protection event is an event associated with each target to be inspected in the environment to be inspected.

2. The method as described in claim 1, characterized in that, The target detection model was trained in the following manner: Obtain a training sample set, wherein each training sample includes: input information corresponding to a training target and a true category label, wherein the training target is associated with the environment to be inspected; Using training samples from the aforementioned training sample set, the preset detection model is trained iteratively through multiple rounds. When the convergence condition is met, the target detection model is output. During one round of iterative training, the following operations are performed: Using the aforementioned detection model, a predicted category is obtained based on the input information in the training samples, and the parameters of the detection model are adjusted based on the adaptive loss value between the predicted category and the corresponding true category label.

3. The method as described in claim 2, characterized in that, The adaptive loss value is calculated in the following way: Based on the predicted category of a training sample and the corresponding true category label of the training sample, calculate the original loss value between the predicted category and the true category label; The original loss value is weighted using adaptive weights corresponding to the true category label from a preset weight set to obtain an adaptive loss value between the predicted category and the true category label. The adaptive weights are associated with the proportion of each training sample in the training sample set corresponding to the true category label.

4. The method according to any one of claims 1-3, characterized in that, After tracking the target trajectories of the remaining targets in the at least one target to be inspected, the method further includes: Determine whether the tracking of the target trajectory of each of the remaining targets to be inspected has ended; wherein: If so, then retain the target trajectory of each target to be inspected obtained through tracking; Otherwise, a new tracking unit is used to continue tracking the target trajectories of the remaining targets to be inspected.

5. The method according to any one of claims 1-3, characterized in that, The step of detecting whether a protection event has occurred for each of the targets to be inspected, based on the positional relationship between each target trajectory and the set warning area, includes: For each of the targets to be inspected, perform any one of the following operations: In response to the fact that the target trajectory of a target under inspection does not cross the set warning area, the current working state of the target under inspection is detected, and based on the current working state, each first image associated with the target under inspection is acquired, and based on the number of the first images, it is detected whether a protection event has occurred for the target under inspection. The first image is an image acquired after the target under inspection enters the current working state and before the target under inspection exits the current working state, and the current working state is the state associated with the protection event of the target under inspection. In response to a target trajectory of an inspected object entering a set warning area, each second image associated with the inspected object is acquired, and based on the number of the second images, it is detected whether a protection event has occurred for the inspected object. The second images are images acquired after the target trajectory of the inspected object enters the warning area and before the target trajectory of the inspected object leaves the warning area. In response to a target trajectory of an inspected object leaving a pre-defined warning area, third images associated with the inspected object are acquired, and based on the number of the third images, a protection event is detected for the inspected object. The third images are images acquired after the target trajectory of the inspected object leaves the warning area and before the target trajectory of the inspected object enters the warning area.

6. The method as described in claim 5, characterized in that, The detection of the current working status of the target to be inspected includes any one of the following operations: Using the target detection model, a target detection box is obtained for the target to be detected, and the current working status of the target to be detected is detected based on the aspect ratio of the target detection box. Using the target detection model, a target detection box is obtained for the target to be inspected, and the current working status of the target to be inspected is detected based on the positional relationship between the target detection box and the set rule box.

7. The method as described in claim 6, characterized in that, After detecting whether a protection event has occurred for each of the targets based on the positional relationship between the target trajectories and the set warning areas, the method further includes: In response to the detection of a target to be alerted, an alarm is issued for the protection event corresponding to the target to be alerted, wherein the target to be alerted is the target to be detected in which the protection event occurs.

8. An event detection device, characterized in that, include: The target detection module is used to perform target detection on the video data stream collected for the environment under inspection using a target detection model, and to obtain the target category of each target under inspection in the environment under inspection. The trajectory tracking module is configured to perform the following operations for each of the target categories: based on the confidence level of at least one target to be inspected corresponding to a target category, using a set confidence threshold, obtain each target to be inspected whose confidence level is greater than the confidence threshold from the at least one target to be inspected, which are respectively the targets to be matched; use multiple tracking units in the corresponding category tracking box set for the target category to track the target trajectory of each target to be matched, wherein each category tracking box contains multiple tracking units, each tracking unit is used to track the target trajectory of one target to be inspected; and in response to the end of the tracking of the target trajectory of each target to be matched, use the remaining tracking units to track the target trajectory of the remaining targets to be inspected in the at least one target to be inspected. The event detection module is used to detect whether a safeguard event has occurred for each of the targets under inspection based on the positional relationship between each target trajectory and the set warning area, wherein the safeguard event is an event associated with each target under inspection in the environment under inspection.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.

Images