Target detection-based message prompting method and device, and storage medium

Abstract

The application discloses a target detection-based message prompting method and device and a storage medium. The target detection-based message prompting method comprises the following steps: performing target detection processing on a current detection scene to obtain a target object in the current detection scene and a target event triggered by the target object; if the target object triggers the target event for the first time, performing message prompting processing on the target event, and creating an event track chain of the target object; if the target object triggers the target event for the second time, analyzing event change information of the target object in the event track chain; and determining whether to perform message prompting processing on the target event according to the event change information. The above scheme can optimize the message prompting effect of target detection.

Description

Technical Field

[0001] This application relates to the field of target detection technology, and in particular to a message prompting method, device and storage medium based on target detection. Background Technology

[0002] Object detection is a core task in the field of computer vision. It often involves analyzing visual data collected by image acquisition devices using object detection models to ultimately output object detection results.

[0003] Currently, object detection technology is commonly used to detect objects and / or events within a region, generating corresponding prompt messages when a target event is detected within that region. However, existing message prompting methods generally suffer from problems such as duplicate prompts and missing information.

[0004] While a typical approach to addressing the issue of repeated notifications for the same event is "message aggregation," which combines multiple identical events occurring within a certain timeframe into a single notification, this method still has several drawbacks and struggles to balance the demands for timeliness, redundancy, and completeness in target detection. Summary of the Invention

[0005] This application provides at least one message prompting method, apparatus, device, and computer-readable storage medium based on target detection.

[0006] The first aspect of this application provides a message prompting method based on object detection, comprising: performing object detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object; if the target object triggers the target event for the first time, performing message prompting processing on the target event and creating an event trajectory chain of the target object; if the target object triggers the target event for the first time, analyzing the event change information of the target object in the event trajectory chain; and determining whether to perform message prompting processing on the target event based on the event change information.

[0007] In one embodiment, the step of performing message prompting processing on the target event and creating the event trajectory chain of the target object if the target object triggers the target event for the first time includes: obtaining the target detection result corresponding to the target object if the target object triggers the target event for the first time; performing message prompting processing based on the target detection result; and creating the event trajectory chain based on the target detection result.

[0008] In one embodiment, after the step of performing message notification processing on the target event and creating an event trajectory chain for the target object when the target object triggers the target event for the first time, the method further includes: performing continuous detection processing on the target object to obtain a continuous detection result corresponding to the target object; and updating the event trajectory chain based on the continuous detection result.

[0009] In one embodiment, analyzing the event change information of the target object in the event trajectory chain includes: obtaining the event attributes of historical events triggered by the target object at different time nodes in the event trajectory chain; and determining the event change information based on the event attributes of the historical events.

[0010] In one embodiment, determining the event change information based on the event attributes of the historical events includes: acquiring the event attributes of each historical event, the event attributes including the event triggering sequence and the event risk level; determining the security level change information of each historical event in sequence according to the event triggering sequence and the event risk level; and determining the event change information of the target object based on the security level change information.

[0011] In one embodiment, after determining the event change information of the target object based on the security level change information, the method includes: if the event change information indicates that the event risk level of the target event is greater than the event risk level of the historical event, then message prompting is performed on the target event; if the event change information indicates that the event risk level of the target event is less than or equal to the event risk level of the historical event, then message silence is performed on the target event.

[0012] In one embodiment, determining the event change information based on the event attributes of the historical events includes: acquiring the event attributes of each historical event, the event attributes including the event triggering sequence and the event triggering location; determining the spatial location change information of each historical event in sequence based on the event triggering sequence and the event triggering location; and determining the event change information of the target object based on the spatial location change information.

[0013] In one embodiment, determining the spatial position change information of each historical event in sequence according to the event triggering time sequence and the event triggering location includes: obtaining a topological relationship map between each target region in the current detection scene; determining a target topological path according to the target region corresponding to each event triggering location and the event triggering time sequence in the topological relationship map; and comparing the target topological path with the obtained reference topological path to obtain the spatial position change information.

[0014] In one embodiment, analyzing the event change information of the target object in the event trajectory chain includes: obtaining the event risk level and event triggering location of the target event currently triggered by the target object, and the event risk level of historical events triggered by the target object in the event trajectory chain; the event triggering sequence of the historical events is earlier than the event triggering sequence of the currently triggered target event; if the target area corresponding to the event triggering location in the current detection scene is connected to a high-risk area in the current detection scene, the event risk level of the currently triggered target event is upgraded to obtain an upgraded event risk level; the event change information is determined based on the upgraded event risk level and the event risk level of the historical events.

[0015] In one embodiment, after performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object, the method further includes: if the target object is not triggering the target event for the first time, then obtaining the distance between the event triggering position of the target event and the boundary of the high-risk area in the current detection scene; if the distance is less than the obtained warning distance threshold, then performing message prompt processing on the target event.

[0016] In one embodiment, before processing the target event with a message notification if the distance is less than the obtained warning distance threshold, the method further includes: obtaining the movement direction of the target object; if the movement direction points to the high-risk area, increasing the warning distance threshold.

[0017] In one embodiment, after performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object, the method further includes: if there are multiple target events, determining the trigger distance between each target event based on the event triggering position of each target event; clustering each target event according to the trigger distance to obtain an event clustering result; if the event clustering result indicates that the number of target events clustered within a preset range in the current detection scene is greater than a preset number threshold, then performing message prompt processing on the target events clustered within the preset range.

[0018] In one embodiment, after determining whether to perform message prompting processing on the target event based on the event change information, the method further includes: if the detection of the target object ends, obtaining the detection results corresponding to each historical event in the event trajectory chain, the detection results including spatiotemporal information and event content; and generating a complete prompt message corresponding to the event trajectory chain based on the spatiotemporal information and the event content for message prompting processing.

[0019] A second aspect of this application provides a message prompting device based on target detection, comprising: a target detection module, configured to perform target detection processing on a current detection scene to obtain a target object in the current detection scene and a target event triggered by the target object; a first-time triggering module, configured to perform message prompting processing on the target event and create an event trajectory chain of the target object if the target object triggers the target event for the first time; a non-first-time triggering module, configured to analyze the event change information of the target object in the event trajectory chain if the target object triggers the target event for the second time; and an event judgment module, configured to determine whether to perform message prompting processing on the target event based on the event change information.

[0020] A third aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is configured to execute program instructions stored in the memory to implement the above-described message prompting method based on target detection.

[0021] The fourth aspect of this application provides a computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, implement the above-described message prompting method based on target detection.

[0022] The above scheme, by performing target detection processing on the current detection scene, can obtain the target objects in the current detection scene and the target events triggered by the target objects. The target object may be triggering the target event for the first time or not. If the target object is triggering the target event for the first time, it is considered a new event, and message notification processing can be selected, and an event trajectory chain for the target object can be created. If the target object is triggering the target event not for the first time, it is considered an old event, and it is necessary to analyze the event change information of the target object in the event trajectory chain; based on the event change information, it is determined whether message notification processing is worthwhile. Therefore, message notification can be initiated when the target object triggers the target event for the first time, ensuring the timeliness of target detection; when the target object triggers the target event not for the first time, the event change information in the event trajectory chain is analyzed before deciding whether to issue a message notification, reducing the redundancy of target detection; and message notification can be issued based on the entire event trajectory chain, ensuring the completeness of target detection, thus optimizing the message notification effect of target detection.

[0023] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0025] Figure 1 This is a flowchart illustrating an exemplary embodiment of the message prompting method based on object detection according to this application; Figure 2 This is a schematic diagram of an exemplary application scenario of the message prompting method based on object detection in this application; Figure 3 This is a block diagram illustrating a target detection-based message prompting device in an exemplary embodiment of this application; Figure 4 This is a schematic diagram of the structure of an embodiment of the electronic device of this application; Figure 5 This is a schematic diagram of the structure of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0026] The solution of this embodiment will now be described in detail with reference to the accompanying drawings.

[0027] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0028] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. Moreover, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of objects. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0029] For ease of understanding, one of the applicable scenarios of this application is illustrated by example.

[0030] Object detection is a core task in the field of computer vision. It often involves analyzing visual data collected by image acquisition devices using object detection models to ultimately output object detection results.

[0031] Currently, object detection technology is commonly used to detect objects and / or events within a region, generating corresponding prompt messages when a target event is detected within that region. However, existing message prompting methods generally suffer from problems such as duplicate prompts and missing information.

[0032] To address the issue of repeated notifications for the same event, a typical solution is "message aggregation," which combines multiple identical events occurring within a certain time period into a single message notification.

[0033] For example, when the same target object triggers alarm messages consecutively in different sub-regions within the same screen or multiple related screens, the target detection system first identifies this as a multi-region alarm event. Then, the target detection system compares the start time of the target object's appearance in each sub-region to determine its last appearance region. Only when the event in the aforementioned "last appearance region" ends will a single alarm message be reported.

[0034] Traditional methods aim to merge multiple alarms into a single message to reduce redundancy and conserve communication and storage resources. However, these message notification methods still have many shortcomings and struggle to balance the demands for timeliness, redundancy, and completeness in target detection.

[0035] In one possible implementation of this application, a hierarchical, modular design can be adopted for the target detection system. Its core data object is an "event trajectory chain," a structure dynamically created core container for each detected independent target event, which may include, but is not limited to, the following key fields: 1. event_id: A globally unique event identifier.

[0036] 2. target_info: Object information of the target object, which may include anonymous ID, target feature vector (which may be a single feature or fused feature of the target object, etc.), associated identity identifier (such as face ID, etc.) and dynamically updated risk attribute labels (such as "unarmed", "armed", "key focus object", etc.).

[0037] 3. first_alert: The notification message for the first event trigger, which may include, but is not limited to, the trigger time, trigger location, target snapshot, and alarm level.

[0038] 4. `trajectory_chain`: An ordered list used to record the continuous motion state of a target object in the world coordinate system. Each node may include, but is not limited to: timestamp, camera / region ID, 2D / 3D world coordinates, target feature snapshot, and motion state (such as entering, staying, leaving, etc.). This chain is dynamically and irreversibly updated during the duration of the event, with nodes strictly ordered by timestamps to ensure that the trajectory is continuous and traceable in physical space.

[0039] 5. alert_upgrade_records: A list that can be used to record all upgrade notification messages triggered during the event's lifecycle, including but not limited to: time, location, upgrade reason (rule / attribute), and details.

[0040] 6. status: Event status, such as: active, ended, pending report, archived, etc.

[0041] Please see Figure 1 , Figure 1 This is a flowchart illustrating an exemplary embodiment of the message prompting method based on object detection according to this application. Specifically, it may include the following steps: Step S110: Perform target detection processing on the current detection scene to obtain the target objects in the current detection scene and the target events triggered by the target objects.

[0042] The target detection process may include target object detection and / or target event detection. Specific detection methods can be found in the relevant methods in this technical field, and will not be elaborated here.

[0043] Target detection processing can be performed periodically, in real time, or in response to preset conditions, etc., and there are no restrictions here.

[0044] It should be noted that this application does not limit the detectable target objects and target events. The specific settings can be configured as needed according to the actual application scenario, which will not be elaborated here.

[0045] The current detection scene can be a scene determined by all or part of the field of view of a single or multiple image acquisition devices.

[0046] In some possible implementations, the target object can be a living or non-living object, and can be a physical object (such as an animal, a vehicle, etc.) or a virtual object (such as the ambient brightness of the current detection scene, weather conditions, etc.), which will not be elaborated here.

[0047] For ease of understanding, this application primarily uses the example of a living entity as the target object. It is understood that in other application scenarios, the target object may be of other types as shown in the example above; therefore, the judgment conditions in the subsequent examples can be adjusted as needed based on the actual application scenario, and will not be elaborated upon here.

[0048] There can be one or more target events.

[0049] It should also be noted that when a target object is detected, the target object may or may not trigger the target event, or the detection of the target object and / or the non-detection of the target object can also be set as the target event.

[0050] Furthermore, the target detection method of this application can simultaneously detect multiple target objects and / or multiple target events existing in the current detection scene. For details, please refer to the relevant methods in this technical field, which are distinguished by setting a unique identifier, which will not be elaborated here.

[0051] For ease of understanding and explanation, the embodiments of this application mainly use the example of detecting that the target object has triggered a target event in the current detection scene.

[0052] Step S120: If the target object triggers the target event for the first time, then the target event is processed with a message prompt, and the event trajectory chain of the target object is created.

[0053] For example, the target event could be "entering the target area of ​​the current detection scene." That is, the target event is triggered for the first time when the target object enters the target area. Therefore, message notification processing can be performed on this target event, and an event trajectory chain of the target object can be created.

[0054] The message prompting process may include, but is not limited to, directly using the event detection result of the target event as the prompt message, or generating the corresponding prompt message based on the preset message generation method and the event detection result before proceeding with the message prompting process, etc., which will not be elaborated here.

[0055] Specifically, when a target object first enters any target area (or enters any target area and triggers a target event), the target detection system can immediately (supporting millisecond-level) create an event_id, initialize the event trajectory chain of the target object, and record event information.

[0056] It is understood that the target event can occur within or outside the target area; no specific limitation is made here.

[0057] Simultaneously, it can push notification messages to devices with communication connections, such as upper-layer applications or user terminals. These notification messages can include: event_id, event type, first trigger time, trigger location, target snapshot, etc., thus resolving the potential delay issue in message notifications.

[0058] Furthermore, after sending a message to the target event that is triggered for the first time, you can choose to perform continuous monitoring on the target object.

[0059] For example, you can refer to Figure 2 As shown, Figure 2 This is an exemplary application scenario diagram of the message prompting method based on object detection in this application. Regardless of whether the target object moves between different sub-regions within the same camera or between different cameras, the object detection system can initiate a tracking and detection mode based on the object ID of the target object.

[0060] Step S130: If the target object is not triggering the target event for the first time, analyze the event change information of the target object in the event trajectory chain.

[0061] In conjunction with the steps described above, if the target object is not triggering the target event for the first time, it is necessary to analyze the event change information of the target object in the event trajectory chain to determine whether there are any changes in the currently triggered target event compared to the historically triggered target events in the event trajectory chain. This will facilitate the determination in subsequent steps whether the currently triggered target event is worth sending a message notification.

[0062] It should be noted that the target event in this application may include one or more events. Therefore, the target event that is not triggered for the first time by the target object in step S130 and the target event that is triggered for the first time by the target object in step S120 may be the same type of event or different types of event.

[0063] In other words, the event trajectory chain may record multiple target events triggered by the target object at different time points. These multiple target events can be of the same type or different types, and the multiple target events can be triggered at the same time or at different times. This is not limited here.

[0064] For example, event change information may include, but is not limited to, changes in event risk level, changes in event trigger location, etc. The specific settings can be configured as needed according to the actual situation, and are not limited here.

[0065] Step S140: Determine whether to process the target event with a message based on the event change information.

[0066] For example, in response to changes in event risk level, corresponding regional risk levels can be pre-set for multiple target detection areas in the current detection scenario. If the target object moves within multiple target areas with the same regional risk level (or moves from a target area with a higher regional risk level to a target area with a lower regional risk level), the event change information can be used to determine that the currently triggered target event is a relatively routine process movement.

[0067] Understandably, given the changes in such event information, it can be determined that the currently triggered target event is not worth prompting again. Therefore, alarm triggering can be strictly suppressed, and no message prompts should be made for the currently triggered target event.

[0068] Optionally, even if the target detection system does not process message prompts for the currently triggered target events, it can silently update the "event trajectory chain" of the target object without generating any additional message pushes, thus eliminating most of the repeated alarm interference in the target detection process and reducing the redundancy of message prompts.

[0069] In another example, if a target object moves from a target area with a lower risk level to a target area with a higher risk level, it can be determined from the event change information that the currently triggered target event is an unusual movement.

[0070] Therefore, based on the information regarding changes in such events, it can be determined that the currently triggered target event warrants a further message notification. Similarly, referring to the method used in the previous example of providing a message notification for the first triggered event, we can select the currently triggered target event for message notification processing. This also avoids the problems of missing important messages and timeliness issues.

[0071] As can be seen, this application, by performing target detection processing on the current detection scene, can obtain the target objects in the current detection scene and the target events triggered by the target objects. The target object may be triggering the target event for the first time or not. If the target object triggers the target event for the first time, it is considered a new event, and message notification processing can be selected, and an event trajectory chain for the target object can be created. If the target object triggers the target event not for the first time, it is considered an old event, requiring analysis of the event change information of the target object in the event trajectory chain; based on the event change information, it is determined whether message notification processing is worthwhile. Therefore, message notification can be initiated when the target object triggers the target event for the first time, ensuring the timeliness of target detection; when the target object triggers the target event not for the first time, the event change information in the event trajectory chain is analyzed before deciding whether to issue a message notification, reducing the redundancy of target detection; and message notification can be issued based on the entire event trajectory chain, ensuring the integrity of target detection, thus optimizing the message notification effect of target detection.

[0072] Based on the above embodiments, this embodiment should also explain that an auxiliary data structure—a detection area topology graph—can be introduced into the technical solution of this application. This graph uses the detection area or target area defined within the target detection system as nodes (which may include the field of view area of ​​the physical camera and virtual sub-regions set in the software, etc.), and uses information such as physical connectivity, adjacency relationship, and line-of-sight reachability between each area as edges to construct a directed or undirected graph, such as G=(V,E).

[0073] Each edge can be assigned attributes, including but not limited to: the expected time for a target object to traverse the edge, the risk transmission coefficient, and whether it belongs to a normal passage path. This topology map can be pre-generated through system configuration (e.g., manual editing based on the actual layout of the detection scenario) or automatic learning (e.g., analyzing historical trajectory data of a large number of target detection results to identify frequently traversed paths). This allows for dynamic updates based on actual triggered events during the operation of the target detection system. For example, if a path has recently experienced multiple intrusion events, its risk transmission coefficient can be increased.

[0074] The regional topology graph can work in conjunction with the event trajectory chain. For example, each "region node (or the region corresponding to the event node)" in the event trajectory chain can be mapped to a vertex in the topology graph, enabling the target detection system to analyze whether the behavior of the target object between regions conforms to the expected pattern, thereby providing richer decision-making basis for the target detection strategy and message prompt strategy. This will not be elaborated here.

[0075] Furthermore, this application can also utilize regional topological relationship graphs and spatial location geometric calculations to achieve more intelligent message prompting decisions. This can include at least two core solutions: regional topological collaborative analysis and spatial location relationship-driven event escalation prompts and message prompt suppression.

[0076] Firstly, regional topological collaborative analysis.

[0077] Based on the region topology diagram mentioned in the above embodiments, the target detection system can perform real-time analysis of the region access sequence in the global event trajectory chain of the target event triggered by the target object: 1. Normal path pattern matching: The target detection system can pre-store or automatically learn typical behavior patterns, such as "normal passage path sequence" (e.g., entrance → corridor → office), "abnormal detour path sequence" (e.g., entrance → restricted area perimeter → office), and "loitering pattern" (moving back and forth between multiple adjacent areas).

[0078] When performing silent detection on a target object, the target detection system can perform sequence matching between the region access sequence in the current event trajectory chain and the event pattern library (or historical behavior pattern library) (the algorithm is not limited, such as the longest common subsequence (LCS) or dynamic time warping (DTW) algorithm).

[0079] If a "normal travel path" is matched, silent detection continues, and message prompts can be suppressed for target events triggered in all intermediate areas of that path. If an "abnormal detour" or "wandering" pattern is matched, an "abnormal behavior escalation prompt" can be triggered immediately and highlighted in the final report.

[0080] 2. Path prediction and deviation detection: Based on the current location of the target object and the topology map, predict the next area that the target object may move to (based on historical probability or shortest path methods, which will not be elaborated here).

[0081] If the actual movement path of the target object deviates from the predicted path (for example, suddenly turning from a normal passage area to a restricted area), the target detection system can immediately trigger a "path deviation warning" without waiting to enter a high-risk area before issuing a message.

[0082] 3. Cross-regional joint notification strategy: (1) Intermediate area message suppression: If the target object moves along the normal topology path, the system can choose to suppress the message prompts when the target object triggers the target event in all intermediate areas on the path, and only retain the message prompts and final reports for the first trigger of the target event, and describe the movement process of the target object in a concise way in the final report.

[0083] (2) Regional Risk Linkage Upgrade: When a target object enters a low-risk area, but this area has a strong connection with a high-risk area in the topology (for example, the low-risk area is the only preceding path to the high-risk area), the system can pre-upgrade the event risk level of the target event triggered by the target object. Thus, even if the target object has not yet entered the high-risk area, a "pre-linkage event risk upgrade" message prompt can be triggered.

[0084] Secondly, the suppression of event escalation prompts and message prompts driven by spatial location relationships.

[0085] This module can introduce upgrade triggering conditions based on precise geometric calculations, which is fundamentally different from the traditional techniques in this field that rely solely on time or static levels.

[0086] 1. Preset key / normal areas and parameter configurations: For example, administrators can manually draw key areas (such as warehouses and server racks), normal areas (such as corridors and lobbies), and restricted areas in the monitoring screen or world coordinate system. Each area can be independently configured with corresponding basic risk level, warning distance threshold (e.g., 5 meters), directional sensitivity, and other parameters.

[0087] 2. Pre-upgrade based on distance and direction: The system calculates the shortest distance d between the current coordinates of the target object and the boundaries of all key areas in real time.

[0088] If d is less than the preset "warning distance threshold", a "spatial proximity pre-escalation alarm" message can be triggered even if the target object has not yet entered the key area. Example message: "Target object is approaching the core restricted area, approximately 3 meters away. Please be aware."

[0089] Meanwhile, the target detection system can also support direction perception: if the target object's movement direction is pointing towards the key area (for example, the angle between the target object's velocity vector and the area's normal is less than 45°), the warning distance threshold can be adaptively and dynamically increased (for example, from 5 meters to 10 meters) to improve the sensitivity of event detection; conversely, if the direction deviates, the pre-escalation strategy can be suppressed.

[0090] 3. Multi-Event Spatial Clustering Detection and Upgrade: The target detection system can also maintain an "active event space index," which can calculate the Euclidean distance between the locations of any two target events within a preset time period in real time or periodically. Density clustering algorithms (such as DBSCAN) are used to cluster the locations of target events within the preset time period.

[0091] When the number of target events gathered within a preset range (e.g., radius R = 10 meters) exceeds a preset threshold N (e.g., 3 events), and the event semantics of these target events are similar (which can be obtained by analyzing event attribute information and calculating similarity, which will not be elaborated here), and if the target event types involved are similar (e.g., both are "personnel intrusion"), then "gathering alarm escalation" can be triggered.

[0092] This upgrade strategy can be implemented by sending a comprehensive alert message to users, rather than targeting a single event, such as: "[Clustering Risk] Three intrusion events have occurred within 10 meters of a certain area. Please take immediate action." Simultaneously, each event's event trajectory chain within that area will be marked as "in a clustering risk zone."

[0093] Based on the above embodiments, this embodiment describes the method in step S120.

[0094] Specifically, in this embodiment, if the target object triggers the target event for the first time, the method for processing the target event with a message and creating the event trajectory chain of the target object includes at least the following steps S210 to S220: Step S210: If the target object triggers the target event for the first time, then obtain the target detection result corresponding to the target object.

[0095] The target detection results may include, but are not limited to, object detection results and / or event detection results. Object detection results include information related to the target object (e.g., object type, appearance features, location information, motion information, etc.), while event detection results include information related to the target event (e.g., event type, event risk level, event trigger location, event trigger time, etc.). Specific settings for the detectable information can be found in various target detection methods within this technical field, and will not be elaborated upon here.

[0096] Step S220: Process message prompts based on the target detection results and create an event trajectory chain based on the target detection results.

[0097] Based on the steps described above, after obtaining the target detection result, the target detection result can be pushed out or a prompt message can be generated based on the target detection result (such as through natural language processing) and then pushed out, thus realizing message prompt processing.

[0098] In addition, for scenarios where the target object triggers the target event for the first time, it is also necessary to create an event trajectory chain corresponding to the target object based on the target detection results, so that the target object's behavior can be continuously recorded and analyzed over a period of time through the event trajectory chain.

[0099] Based on the above embodiments, this embodiment should also note that, optionally, for the target event and event trajectory chain of this application, a finite state machine centered on the target event can be constructed. Each target event triggered by a target object typically experiences the following states sequentially during its lifecycle, such as: 1. Initialization: When the target object triggers the target event rule for the first time, a global trajectory chain is created and the object enters the "first alarm" state.

[0100] 2. First Alarm (Output): Pushes an immediate alarm when the target event is triggered for the first time. The event status can be transferred to "Silent Continuous Detection".

[0101] 3. Silent continuous detection (no output): The target detection system can continuously update the event trajectory chain, but does not generate message pushes.

[0102] In this state, the target detection system can detect in real time two types of event triggering conditions, including but not limited to the following examples: ① The regional risk level of the area the target is currently entering is higher than the regional risk level of the area where the target event was first triggered; ② The object attributes (such as item possession attributes, identity attributes, movement attributes, etc.) undergo a significant risk escalation.

[0103] If any condition is met, the state can be transitioned to "Event Escalation Alarm"; otherwise, the state can be transitioned to "End Judgment" when the current event ends (when the detection of the target object ends).

[0104] 4. Escalation Alarm (Output): When the risk level of a target event escalates, an escalation alarm can be pushed and the escalation information recorded. The event status can then return to "Silent Continuous Monitoring" to continue updating the event trajectory chain.

[0105] 5. Termination Judgment: When the target event ends or the detection of the target object ends (e.g., the target object leaves all target areas and after a stable delay), the event state can be transferred to the generation of the "Final Report".

[0106] 6. Final Report (Output): Generate and push a complete notification message, and archive the events in this event trajectory chain.

[0107] Therefore, this application, by introducing an event-level state machine, achieves fine-grained control over the entire lifecycle of a target event. While maintaining complete silence (avoiding the harassment caused by frequent message pushes), it also ensures timely response to dynamically occurring event risk escalations. This realizes a composite message push paradigm of "silent detection as the primary method and event escalation as a secondary method."

[0108] Based on the above embodiments, this embodiment describes the method after step S130.

[0109] Specifically, in this embodiment, the method after processing the target event with a message and creating the event trajectory chain of the target object when the target object triggers the target event for the first time includes at least the following steps S310 to S320: Step S310: Perform continuous detection processing on the target object to obtain the continuous detection result corresponding to the target object.

[0110] Step S320: Update the event trajectory chain based on the continuous detection results.

[0111] As illustrated in the foregoing embodiments, after creating the event trajectory chain of the target object, continuous target detection can be performed silently on the target object.

[0112] The continuous detection results are based on the same principle as the target detection results in the previous example; they are simply named differently here to distinguish the temporal order of the two detection results. Optionally, the two detection results may include the same or different data content, which is not limited here.

[0113] Optionally, the continuous detection result can be the detection result when the target object triggers a target event, the detection result when the target object triggers a non-target event, or the detection result when the target object does not trigger a target event (or does not trigger a non-target event), which is not limited here.

[0114] Furthermore, after obtaining the continuous detection results corresponding to the target object, the event trajectory chain corresponding to the target object can be updated based on the continuous detection results.

[0115] Therefore, the events recorded in the event trajectory chain can include only the target event, or include both the target event and non-target events, or it can also be selected to record the trajectory of the target object when the target event is not triggered again, etc., without any limitation here.

[0116] For example, in order to ensure the accuracy, continuity and spatiotemporal consistency of the event trajectory chain, re-identification algorithms, including but not limited to, based on deep feature fusion (such as weighted fusion of shallow and deep features in CNN), can be used, combined with motion model prediction (such as Kalman filtering), to perform cross-regional and / or cross-camera target object association, ensuring that the IDs (object identifiers) of the detected target objects in the event trajectory chain are consistent, and achieving highly robust target association (object and event association).

[0117] Additionally, it is understandable that when a target object triggers a target event, the location information of the target object and the target event may be the same or different; this is not limited here.

[0118] Among them, camera calibration parameters can be used to map the target pixel coordinates (which may be the pixel coordinates corresponding to the target event and / or the pixel coordinates corresponding to the target object) of the corresponding acquired image in each event node to a unified world coordinate system (such as the UTM coordinate system), so that the position information of each event and / or each object in the event trajectory chain is continuous in physical space.

[0119] For example, for a target within a single camera, an efficient inter-frame target detection and association algorithm can be used to ensure that the target object's identifier (target_info) in the trajectory_chain remains consistent as the target object moves between virtual sub-regions.

[0120] Furthermore, for issues such as brief tracking loss or occlusion that may occur during target detection, this application can also employ techniques such as motion model prediction and target re-identification to perform optimal matching and stitching of event trajectory segments under the constraint of world coordinates, thereby ensuring the spatiotemporal continuity of the trajectory_chain.

[0121] Based on the above embodiments, this embodiment describes the method in step S130.

[0122] Specifically, the method for analyzing event change information of a target object in an event trajectory chain in this embodiment includes at least the following steps S410 to S420: Step S410: Obtain the event attributes of historical events triggered by the target object at different time points in the event trajectory chain.

[0123] It should be noted that if the target object is detected as triggering the target event for the first time in step S130, then the detected target event is the current event. Once the current event is recorded in the event trajectory chain, it can be recorded as a historical event. This can be done either immediately after the current event is detected, or after analyzing whether to push a notification message for the current event (or after pushing a notification message for the current event has been completed), without limitation here.

[0124] Historical events refer to target events or non-target events in an event trajectory chain that occur earlier than the current event. An event trajectory chain can contain one or more historical events; this is not limited here.

[0125] Event attributes can be set as needed, such as including but not limited to: event trigger time, event trigger location, event risk level, etc., which are not limited here.

[0126] Historical events and current events (as well as various historical events) may have the same or different event attributes, which is not limited here.

[0127] Step S420: Determine event change information based on the event attributes of historical events.

[0128] Among them, event change information can be obtained by analyzing the differences between different events on one or more of the same event attributes.

[0129] In a brief example, common object detection applications pre-set corresponding event risk levels for the target events to be detected. Therefore, by analyzing the differences in event risk levels between two adjacent (or non-adjacent) events, information about event changes can be obtained (such as the event risk level increasing (upgrading), remaining unchanged (stable), or decreasing (downgrading), etc., which are not limited here).

[0130] For example, in some object detection applications, the movement of target objects is detected. Therefore, by analyzing the differences in the triggering time and location of two adjacent (or non-adjacent) events, the speed of the target object between the two events can be determined, which can also serve as information on event changes (such as whether the target object is speeding, moving at a constant speed, or decelerating in the current event).

[0131] In this embodiment, event change information can be determined by comparing the event attributes of multiple historical events; alternatively, it can be determined by comparing the event attributes of historical events with the event attributes of the currently triggered target event. Specific examples in this application can be referenced similarly, and will not be elaborated upon here.

[0132] Based on the above embodiments, this embodiment should also explain that the event trajectory chain can record the target events triggered by the target object based on information such as time nodes and / or event nodes and / or location nodes.

[0133] Therefore, in some implementations, event change information can be determined by the event attributes of historical events triggered by the target object at different time points in the event trajectory chain. Since multiple target events may be triggered simultaneously at the same time point, event change information can also be determined by the event attributes of the target object at different event nodes (or different position nodes) in the event trajectory chain, which will not be elaborated upon here.

[0134] Based on the above embodiments, this embodiment describes the method in step S420.

[0135] Specifically, the method for determining event change information based on the event attributes of historical events in this embodiment includes at least the following steps S510 to S530: Step S510: Obtain the event attributes of each historical event. The event attributes include the event trigger sequence and the event risk level.

[0136] Step S520: Determine the security level change information of each historical event in sequence according to the event triggering sequence and event risk level.

[0137] Step S530: Determine the event change information of the target object based on the security level change information.

[0138] In conjunction with the foregoing embodiments, in one possible implementation, the security level change information of each historical event on the event trajectory chain can be determined sequentially by the event triggering sequence and the event risk level.

[0139] For example, if there are three historical events in the event trajectory chain (first historical event, second historical event, and third historical event), it can be seen from the event triggering sequence of the historical events that the first historical event is earlier than the second historical event, and the second historical event is earlier than the third historical event.

[0140] Therefore, based on the event risk level of the first historical event and the event risk level of the second historical event, the change information of the security level of the target event from the timeline of the first historical event to the timeline of the second historical event can be determined.

[0141] The event risk level can be set as needed in numerical form and / or natural language form. For example, a higher risk level value indicates a greater risk. Therefore, if the risk level value of the second historical event minus the risk level value of the first historical event (i.e., the resulting security level change information) is a positive number, then the security level change information indicates that the risk level has been increased.

[0142] In scenarios where event risk levels are described in natural language, the risk levels can be compared between the second historical event and the first historical event. For example, if the risk level of the second historical event is high and the risk level of the first historical event is low, then the information indicating a change in security level represents an increase in risk.

[0143] It is understandable that there are many other methods for analyzing event change information, and the specific methods can be set according to the actual application scenario, which will not be elaborated here.

[0144] Based on the above embodiments, this embodiment describes the method after step S530.

[0145] Specifically, the method in this embodiment after determining the event change information of the target object based on the security level change information includes at least the following steps S610 to S620: Step S610: If the event risk level of the target event, as indicated by the event change information, is greater than the event risk level of the historical events, then a message notification is sent to the target event.

[0146] Step S620: If the event risk level of the target event represented by the event change information is less than or equal to the event risk level of the historical event, the target event is silenced.

[0147] Referring to the foregoing embodiments, in application scenarios that focus on the risk level of an event, if the event risk level of the target event (or the current event) is greater than that of historical events, then message notification processing can be selected for the target event.

[0148] If the event risk level of the target event, as indicated by the event change information, is less than or equal to the event risk level of the historical event, then the target event can be silenced.

[0149] Message silence processing refers to not providing message notifications for the target event. However, it is optional to record the event detection result corresponding to the target event in the event trajectory chain, or to ignore the target event (not record it in the event trajectory chain). No restrictions are imposed here.

[0150] Based on the above embodiments, this embodiment should also explain that, in addition to determining whether to provide a message notification for a target event that is not triggered for the first time based on event change information, it is also possible to determine whether the event attributes of the target event that is not triggered for the first time meet preset event attribute requirements (compared with preset conditions rather than with historical events) and then decide whether to provide a message notification (for example, if the requirements are met, a message notification is provided; if the requirements are not met, no message notification is provided; the reverse is also true); or it is also possible to first determine whether the event attributes of the target event that is not triggered for the first time meet preset event attribute requirements, and then analyze the event change information to determine whether to provide a message notification for the target event that is not triggered for the first time, which will not be elaborated here.

[0151] Based on the above embodiments, this embodiment describes the method in step S420.

[0152] Specifically, the method for determining event change information based on the event attributes of historical events in this embodiment includes at least the following steps S710 to S730: Step S710: Obtain the event attributes of each historical event. The event attributes include the event triggering sequence and the event triggering location.

[0153] Step S720: Determine the spatial location change information of each historical event in sequence according to the event triggering time sequence and event triggering location.

[0154] Step S730: Determine the event change information of the target object based on the spatial location change information.

[0155] The foregoing embodiments illustrate how to determine event change information based on the event risk level of different target events in the event trajectory chain, and then decide whether to provide a message notification for target events that are not triggered for the first time.

[0156] In this embodiment, another possible implementation is to determine the event change information of the target object based on the spatial location change information, and then decide whether to provide a message prompt for the target event that is not triggered for the first time.

[0157] The event trigger location refers to the location information when the event is triggered. It may be the same as or different from the location information of the target object when the event is triggered. No limitation is made here.

[0158] For example, in this scenario, the event attributes need to include the event triggering sequence and the event triggering location. Then, based on the event triggering sequence and the event triggering location, the spatial location change information of each historical event in the event trajectory chain can be determined sequentially.

[0159] In some implementation methods, when the spatial location change information indicates that the triggering locations of the two events are different, a message prompt is given to the currently triggered target event; when the spatial location change information indicates that the triggering locations of the two events are the same, a message silence is given to the currently triggered target event. Similarly, the reverse condition can also be true, which will not be elaborated here.

[0160] In other possible implementations, a reference position can be set in the current detection scene. If the spatial position change information indicates that the target object is moving closer to the reference position, then a message prompt can be selected for the currently triggered target event; if the spatial position change information indicates that the target object is moving away from the reference position, then a message silence can be selected for the currently triggered target event. Similarly, the opposite of the judgment condition can also be true, which will not be elaborated here.

[0161] In other possible implementations, the event interval time between triggering each historical event can be calculated by using the spatial location change information of each historical event and the event triggering sequence (or event triggering time).

[0162] Then, you can choose to provide a message notification for the currently triggered target event when the event interval is less than a preset time threshold; or to silence the currently triggered target event when the event interval is greater than or equal to the preset time threshold. Similarly, the opposite of the condition can also be true, which will not be elaborated here.

[0163] Another example is that the event triggering location can also be bound to the event risk level in the aforementioned embodiments. For example, different locations correspond to different event risk levels, thereby enabling the determination of the security level change information of each target event based on the change in the event triggering location of each target event, which will not be elaborated here.

[0164] It should also be noted that, in this application, when processing message prompts for the currently triggered target event, one may choose to process message prompts only for the target event (the current event), or one may choose to process message prompts in conjunction with other historical events in the event trajectory chain of the same target object based on the current event, or one may choose to process message prompts in conjunction with the current events and / or historical events in the event trajectory chains of different target objects based on the current event. No limitation is made here.

[0165] Based on the above embodiments, this embodiment describes the method in step S720.

[0166] Specifically, the method for determining the spatial location change information of each historical event in sequence according to the event triggering time and event triggering location in this embodiment includes at least the following steps S810 to S830: Step S810: Obtain the topological relationship diagram between each target region in the current detection scene.

[0167] The principle of the topology diagram can be found in the explanation of the foregoing embodiments, and will not be repeated here.

[0168] Step S820: Determine the target topology path based on the target area corresponding to each event trigger location and the event trigger sequence in the topology relationship diagram.

[0169] The target topology path can be a historical path formed by the target area that the target object has moved through when each event is triggered in sequence, and / or it can be a predicted path obtained by predicting the target area that the target object has moved through when each event is triggered in sequence. There is no limitation here.

[0170] Alternatively, the target region and time sequence can be determined in the topology graph based on the movement information (location and time) detected by the target object, thereby determining the target topology path (or predicting the target topology path), which will not be elaborated here.

[0171] Step S830: Compare the target topology path with the obtained reference topology path to obtain spatial location change information.

[0172] The reference topology path can be preset or obtained in real time; no limitation is made here.

[0173] For example, when using the historical path through which the target object has moved as the target topological path, the reference topological path can be preset. That is, a corresponding normal path (as the reference topological path) can be preset based on some normal behaviors in the actual application scenario, and the target topological path is compared with the reference topological path to obtain spatial position change information.

[0174] If the spatial location change information indicates that the target topological path is the same as the reference topological path, no message prompting is required. If the spatial location change information indicates that the target topological path is different from the reference topological path, then a message prompting can be performed.

[0175] In another example, when using the predicted path of the target object as the target topological path, the reference topological path can be acquired in real time. That is, the target object can be continuously detected, and the continuously detected actual path can be used as the reference topological path. The target topological path can be compared with the reference topological path to obtain spatial location change information.

[0176] If the spatial location change information indicates that the target topological path is the same as the reference topological path, then no message prompt is needed. Alternatively, factors such as whether the predicted path involves high-risk areas can be pre-judged before deciding whether to prompt a message; this will not be elaborated upon here. If the spatial location change information indicates that the target topological path differs from the reference topological path (the actual path deviates from the predicted path), then a message prompt can be prompted.

[0177] Based on the above embodiments, this embodiment describes the method in step S130.

[0178] Specifically, the method for analyzing event change information of a target object in an event trajectory chain in this embodiment includes at least the following steps S910 to S930: Step S910: Obtain the event risk level and event triggering location of the target event currently triggered by the target object, as well as the event risk level of the historical events triggered by the target object in the event trajectory chain; the event triggering sequence of the historical events is earlier than the event triggering sequence of the currently triggered target event.

[0179] Among them, the target event currently triggered by the target object refers to the event when the target object is detected to be triggering a target event for the first time (also referred to as the current triggering event).

[0180] For the explanation of event risk level and event trigger location, please refer to the event attribute explanation in the previous embodiments, which will not be repeated here.

[0181] Step S920: If the target area corresponding to the event trigger location in the current detection scene is connected to the high-risk area in the current detection scene, then the event risk level of the currently triggered target event is upgraded to obtain the upgraded event risk level.

[0182] For example, the connectivity of each target region can be determined by obtaining the topology diagram in the foregoing embodiments. Each target region may have a corresponding regional risk level, thereby enabling the classification of target regions into high-risk regions, low-risk regions, etc., without limitation here.

[0183] It is understood that various events in this application can be pre-set with corresponding event risk levels. Therefore, in some feasible implementations, event change information can be determined by comparing the event risk level of the currently triggered event with the event risk levels of its adjacent historical events.

[0184] Therefore, in the process of comparing event risk levels, the event risk level that is preset for the current triggering event can be used, or the event risk level that is obtained after making certain adjustments to the preset event risk level that is preset for the current triggering event can be used.

[0185] For example, based on the topology diagram and the target area corresponding to the trigger location of the current event, it is determined whether there is a connectivity or strong connectivity with a high-risk area in the current detection scenario. If so, the preset event risk level corresponding to the current event needs to be upgraded; if not, no upgrade is performed.

[0186] The conditions for determining connectivity may include, but are not limited to: the existence of connecting edges (which may include direct and / or indirect connections), or the number of connecting edges between two regions being less than or equal to a preset edge number threshold, or the proportion of connecting edges between two regions to the total number of connecting edges in the target region corresponding to the event trigger location being greater than a preset connecting edge proportion threshold, or the target region corresponding to the event trigger location being the only preceding path to a high-risk region, etc., which are not limited here.

[0187] Step S930: Determine event change information based on the upgraded event risk level and the event risk level of historical events.

[0188] Referring to the preceding steps, if the risk level of the currently triggered event is upgraded, the event change information is determined based on the upgraded risk level and the risk levels of historical events. The specific details can be found in the description of the aforementioned embodiments, and will not be repeated here.

[0189] Based on the above embodiments, this embodiment describes the method after step S110.

[0190] Specifically, in this embodiment, the method for performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object includes at least the following steps S1010 to S1020: Step S1010: If the target object is not triggering the target event for the first time, then obtain the distance between the event trigger location of the target event and the boundary of the high-risk area in the current detection scene.

[0191] Step S1020: If the distance is less than the obtained warning distance threshold, then a message prompt is made for the target event.

[0192] Referring to the foregoing embodiments, when it is detected that the target object has triggered the target event for the first time, the event change information between the target event and historical events can be obtained through the foregoing embodiments to determine whether to perform message prompt processing on the target event; alternatively, the event attributes of the target event can be compared with some obtained attribute thresholds (which may be preset and / or obtained in real time, etc.), and then the comparison results can be used to determine whether to perform message prompt processing on the target event. No limitation is made here.

[0193] For example, if the triggering location of the target event is within the target area or in a high-risk area (other risk levels can also be set as needed; this is just an example and not a limitation), then you can choose to provide a message notification for the target event.

[0194] If the event trigger location of the target event is not within the target area or the event trigger location of the target event is not within the high-risk area, then the distance between the event trigger location and the boundary of the high-risk area (or target area) in the current detection scene (or the distance between other locations in the area, such as the geometric center of the area) can be obtained.

[0195] If the distance is less than the obtained warning distance threshold, it indicates that the target object is very close to a high-risk area (or the target area), and therefore, a message notification can be sent to the target event. Conversely, if the distance is greater than the threshold, the message can be left silent, or the decision to send a message can be made based on the event change information between the target event and historical events; these details will not be elaborated here.

[0196] Alternatively, one could first determine whether to process a message notification based on the event change information between the target event and historical events. If not, one could then compare the event attributes of the target event with some obtained attribute thresholds, and then determine whether to process a message notification for the target event based on the comparison results.

[0197] Based on the above embodiments, this embodiment describes the method prior to step S1020.

[0198] Specifically, in this embodiment, the method prior to message notification processing of the target event if the distance is less than the obtained warning distance threshold includes at least the following steps S1110 to S1120: Step S1110: Obtain the motion direction of the target object.

[0199] Step S1120: If the direction of movement points to a high-risk area, increase the warning distance threshold.

[0200] The warning distance threshold in the foregoing embodiments can be preset or flexibly calculated and obtained in real time according to the actual application scenario.

[0201] For example, when performing target detection processing on the current detection scene, the direction of motion of the target object can be obtained when the target object is detected.

[0202] It is understandable that the direction of movement of a target object to some extent represents its possible future direction of movement. Therefore, if the direction of movement points to a high-risk area (or target area), the warning distance threshold can be adaptively increased (the specific increase method can be set as needed, such as increasing it by an increment based on a preset distance threshold, etc., which will not be elaborated here), so that the distance judgment in the aforementioned embodiments is more sensitive.

[0203] Among them, the criteria for determining whether the direction of movement points to a high-risk area may include, but are not limited to, the angle between the direction of movement (such as the velocity vector) and the normal of the area being less than a preset angle threshold (such as 45°).

[0204] Furthermore, the angle between the direction of movement and the region's normal can also be negatively correlated with the increase in the warning distance threshold. That is, the smaller the angle between the direction of movement and the region's normal, the more likely the target object is to move towards a high-risk area. Therefore, the warning distance threshold can be increased more significantly to expedite the execution of message notifications for the target event. Conversely, the same configuration can be chosen to support reducing the warning distance threshold, which will not be elaborated here.

[0205] Based on the above embodiments, this embodiment describes the method after step S110.

[0206] Specifically, in this embodiment, the method for performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object includes at least the following steps S1210 to S1230: Step S1210: If there are multiple target events, determine the trigger distance between each target event based on the event triggering position of each target event.

[0207] Step S1220: Cluster each target event according to the trigger distance to obtain the event clustering result.

[0208] In conjunction with the foregoing embodiments, during the target detection process of the current detection scene, this application may detect one or more target events. Multiple target events may be triggered by the same target object or different target objects, and multiple target events may be triggered at the same location or different locations, without limitation here.

[0209] Therefore, when a single target event is detected, the method described in the foregoing embodiments can be applied similarly, and will not be repeated here. When multiple target events are detected, the trigger distance between each target event can be determined based on its trigger location, and the target events can be clustered based on the trigger distance to obtain the event clustering result.

[0210] Step S1230: If the event clustering result indicates that the number of target events clustered within a preset range in the current detection scene is greater than a preset number threshold, then message prompts are processed for the target events clustered within the preset range.

[0211] Furthermore, if the event clustering results indicate that the number of target events clustered within a preset range (e.g., a radius of 10 meters) in the current detection scene is greater than a preset threshold (e.g., 3), then message prompting can be performed on the target events clustered within the preset range. Otherwise, message silence can be performed on the target events, or the method provided in the aforementioned embodiments can be used to determine whether to perform message prompting based on the event change information between the target events and historical events; this will not be elaborated upon here.

[0212] Alternatively, one can first determine whether to perform message notification processing based on the event change information between the target event and historical events. If not, then perform clustering processing, and then determine whether to perform message notification processing for the target event based on the event clustering results.

[0213] One alternative implementation method after obtaining the event clustering results may include, but is not limited to, obtaining the event similarity between target events in a preset range (which can be calculated using one or more event attributes such as location, time, event semantics, and event risk level). If the number of target events in the cluster with an event similarity greater than a preset similarity threshold is greater than a preset similar event number threshold (or the proportion of target events with an event similarity greater than the preset similarity threshold relative to the total number of target events in the same cluster is greater than a preset similarity number proportion threshold), then message notification processing can be selected for the target events. Conversely, the same principle applies to the description of the foregoing embodiments, which will not be repeated here.

[0214] Based on the above embodiments, this embodiment describes the method after step S140.

[0215] Specifically, the method in this embodiment after determining whether to perform message notification processing on the target event based on event change information includes at least the following steps S1310 to S1320: Step S1310: If the detection of the target object is completed, obtain the detection results corresponding to each historical event in the event trajectory chain. The detection results include spatiotemporal information and event content.

[0216] Step S1320: Based on the spatiotemporal information and event content, generate a complete prompt message corresponding to the event trajectory chain and perform message prompt processing.

[0217] Referring to the foregoing embodiments, if the detection of the target object is completed, the detection results of each historical event in the event trajectory chain corresponding to the target object can be obtained. These detection results may include, but are not limited to, spatiotemporal information and event content at the time of event triggering.

[0218] Therefore, based on the spatiotemporal information and content of each historical event, a complete prompt message corresponding to the event trajectory chain can be generated for message prompting, ensuring the integrity of information during the target detection process.

[0219] For example, the detection of the target object can be terminated when the target object leaves the current detection scene or leaves the target area in the current detection scene.

[0220] If no target object is detected immediately or after a pre-configurable stable delay (e.g., 5-30 seconds), the target detection system determines to end the detection process.

[0221] Subsequently, the target detection system can automatically generate and report a structured "complete event report" (i.e., a complete alert message) based on the complete trajectory_chain and alert_upgrade_records information in the event trajectory chain of the target object.

[0222] For example, the report may include an overall intelligence summary of the event trajectory chain, the core content of which may include, but is not limited to: an event overview, historical events arranged in chronological order, textual descriptions and visual trajectory diagrams of movement sequences based on the world coordinate system, key statistical information, event risk escalation markers, and related prompts generated after comparison with a historical behavior pattern database, thereby providing a complete review view without missing information.

[0223] For example, the generated complete prompt message may also include, but is not limited to: 1. Target object identity information fusion: If the target detection system integrates modules such as object recognition and facial recognition, the recognized identity ID can be bound to target_info. The target object's identity information can then be included in the notification message for the first triggering of a target event or in the notification message following an escalation of the event's risk level.

[0224] 2. Target Object Behavior Pattern Analysis: The target detection system can maintain a "historical trajectory and event pattern library". For example, when generating a complete prompt message, the target object's trajectory_chain can be compared with the preset event pattern library (similarity measurement can be performed using algorithms such as DTW), and the similar events determined by the comparison can be appended to the complete prompt message. For example: "The similarity between this trajectory and the 'abnormal loitering' pattern is 78%".

[0225] 3. Visualized Intelligence Output: Complete alert messages can include more than just textual modal information. The target detection system can also automatically generate dynamic trajectory heatmaps or animation playback links overlaid on monitor floor plans or maps. This allows users to intuitively review the entire event process with a single click. Furthermore, event nodes with upgraded risk levels can be marked in a more prominent manner (e.g., flashing red dots).

[0226] Furthermore, it should be noted that the complete notification message in this application does not simply list and push the recorded trajectory nodes, but rather transforms the mechanical "trajectory data" into highly readable "behavioral intelligence" through a series of intelligent processes, such as: 1. Structured Description of Spatiotemporal Information: The target detection system can automatically generate semantic descriptions that conform to human reading habits based on the world coordinates in the event nodes and the preset target area semantics (such as "Waiting Area 1", "Core Restricted Area", etc.). For example, "The target object moved from Waiting Area 1 to Counter Area 2 at 09:32:15, taking about 3 minutes, staying for 2 minutes, and then leaving." This description avoids the mechanical nature of outputting raw coordinates and greatly improves the readability of the detection report.

[0227] 2. Intelligent Behavioral Pattern Comparison: The target detection system maintains a historical trajectory and event pattern database, which stores trajectory features of pre-labeled typical behavioral patterns (such as "loitering," "tailgating," and "rapid crossing"). When generating the final complete alert message, the global event trajectory chain of the current event can be dynamically normalized and compared with the pattern database to calculate the similarity. Intelligent prompts are added to the report, such as "The similarity between this trajectory and the 'abnormal loitering' pattern reaches 78%" or "The similarity between this trajectory and the trajectory of a historical high-risk event reaches 65%." This provides valuable decision-making references for some typical target detection scenarios (such as security scenarios).

[0228] 3. Automatic Generation of Visual Intelligence: Based on the world coordinate sequence in the trajectory chain, the system automatically generates a dynamic trajectory heatmap or animation playback link that can be overlaid on the monitoring floor plan or GIS map, and prominently marks escalation alarm points on the map (e.g., flashing red dots). This allows users to intuitively review the entire event process without needing to retrieve multiple video playbacks, significantly reducing manual investigation costs.

[0229] Based on the foregoing embodiments, and for ease of understanding, this embodiment will illustrate some possible implementations of the method of this application in conjunction with some specific application scenarios.

[0230] Application Scenario 1: Single camera detection of multiple virtual surveillance areas.

[0231] 1. Scenario: A panoramic camera is deployed in a bank branch. The target detection software divides the current detection scene into three virtual warning zones: No. 1 Customer Waiting Area (Level: Medium), No. 2 Cash Transaction Counter Area (Level: Medium), and No. 3 Internal Office Door Restricted Area (Level: High).

[0232] 2. Implementation process: (1) 09:30:00 - Target event detected for the first time: A target person enters the No. 1 customer waiting area, triggering an area intrusion alarm. The target detection system executes: Create a target event (identifier EVT20250810001), initialize the first node of the event trajectory chain [Area 1, 09:30:00, Enter]. Millisecond-level push of the first triggered target event instant alarm: "[Instant Alarm] Target person detected in [Business Hall - No. 1 waiting area], time 09:30:00, event ID: EVT20250810001".

[0233] (2) 09:32:15 - Cross-regional movement, silent continuous detection: The target person moves to the No. 2 cash business counter area. Target detection system executes: Continuously detects the target object through a single camera and updates its event trajectory chain to [Area 1, 09:32:15, Leave] -> [Area 2, 09:32:15, Enter]. No new alarm messages are pushed during this process.

[0234] (3) 09:35:40 - Entering a high-risk area triggers an event escalation alarm: This person attempted to enter the restricted area of ​​internal office door No. 3 (the risk level of the area is higher than the target area when the target event was first triggered). System execution: Update the event trajectory chain. Due to entering a higher-risk area, an "Event Escalation Alarm" message is triggered and pushed, such as: "[Escalation Alarm] Event ID: EVT20250810001, the target is attempting to enter the core restricted area [Restricted Area of ​​Door No. 3], time 09:35:40, please pay close attention!"

[0235] (4) 09:36:20 - Event End and Final Report: The target person leaves the camera's field of view, and the event is determined to be over after a 5-second delay. System Execution: Generate and push a complete prompt message, describing in detail the target person's complete cross-regional trajectory from Zone 1 to Zone 3, the duration of stay in Zone 2, highlighting the escalation alarm point at 09:35:40, and providing trajectory playback.

[0236] Application Scenario 2: Hazardous materials detection.

[0237] It should be noted that this application scenario may be in the same environment or a different environment than application scenario 1 in the previous example, which will not be elaborated here. For example, it could be an addition of the detection of hazardous materials to application scenario 1.

[0238] 1. Scenario: The relevant algorithms for "hazardous materials detection" can be added and implemented in the environment of application scenario 1.

[0239] 2. Implementation process: (1) 10:00:00 - First alarm: The man enters Zone 1, triggering the target event for the first time and immediately alarming, creating EVT20250810002.

[0240] (2) 10:02:30 - Normal movement, silent detection: The man enters Zone 2, and the system silently updates the corresponding event trajectory chain.

[0241] (3) 10:05:15 - Dangerous item detected, triggering risk escalation alarm: The dangerous item detection algorithm detected a suspected knife at the man's waist, triggering the target event. System execution: Update target_info to "Holding a weapon", immediately trigger the "escalation alarm" for the target event, such as: "[Emergency escalation alarm] Event ID: EVT20250810002, the target was detected holding a sharp weapon at 10:05:15 in [Counter Area 2]! Risk level: High risk!"

[0242] (4) 10:05:40 - Entering the restricted area, triggering a risk escalation alarm again: The armed man rushed towards Restricted Area No. 3. System execution: Update event trajectory chain. Because the target's current attribute is "armed" and it has entered a high-risk area, an escalation alarm can be triggered again.

[0243] (5) 10:06:00 - Event ends: The target escapes and the event ends. In the final report, the trajectory description can clearly mark the two escalation warning points, and the message prompt can be: "This event has experienced a serious risk escalation. The target is armed and attempted to break into the core area."

[0244] Application Scenario 3: The target object quickly leaves the current detection scene (boundary case).

[0245] 1. Scenario: The target object quickly leaves the current detection scene or target area after triggering the target event for the first time.

[0246] 2. Implementation process: (1) 11:00:00: The target briefly appears under the perimeter camera, triggering the first perimeter intrusion target event alarm. The system creates event EVT20250810003 and pushes the first real-time alarm.

[0247] (2) 11:00:05: The target has left the camera's field of view. After a short wait (e.g., 10 seconds), the system determines that the detection of the target object has ended.

[0248] (3) System execution: Generate a complete alert message, which may read: "The target appeared at 11:00:00 on the [east side of the perimeter] and then quickly disappeared." Thus, the solution of this application provides both immediate early warning and complete record for short-lived events.

[0249] Application Scenario 4: Cross-camera network tracking scenario (comprehensive application).

[0250] 1. Scenario: A multi-camera network covering gates, roads, and buildings in a smart park.

[0251] 2. Implementation process: (1) 14:00:00: The target object appears at the gate, triggering an immediate alarm for the "entering the target area" target event for the first time.

[0252] (2) 14:02:00-14:15:00: The target object continuously crosses multiple target areas. The system silently constructs a continuous global event trajectory chain by cross-camera re-identification and world coordinate mapping.

[0253] (3) 14:16:00: The target object disappears in the warehouse area, and the event ends. The system can report a complete prompt message, which can include: a text description and animation of the spatiotemporal trajectory such as "gate -> main road -> R&D building -> warehouse", and can prompt abnormal behavior patterns.

[0254] Application Scenario 5: Progressive alarm based on multi-level spatial early warning areas.

[0255] 1. Scenario: A core equipment area (such as a data center or precision instrument room) within a large industrial park is designated as the target area, with the highest risk level. The system sets up a ring-shaped "early warning zone" (10 meters wide, i.e., a ring-shaped area 5 to 15 meters from the boundary of the core equipment area) around the core equipment area. Different alarm strategies are configured for the early warning zone and the core equipment area.

[0256] 2. Implementation process: 1. First trigger of target event alarm: When a target person enters the park perimeter (ordinary area), an immediate alarm for the "perimeter intrusion" target event is triggered for the first time, creating event EVT20250810001, and entering a silent continuous monitoring state for the target object.

[0257] 2. Entering the warning zone triggers event pre-escalation: If the person continues to move towards the core equipment area, and their coordinates enter the warning zone (less than 15 meters and more than 5 meters from the boundary of the core equipment area), the system calculates the shortest distance to be 8 meters, and the direction of movement is towards the core equipment area, then a "pre-escalation alarm" can be triggered immediately. Content: "[Spatial Warning] The target is approaching the core equipment area, currently approximately 8 meters away. Please be vigilant."

[0258] Meanwhile, the system can record this pre-upgrade event in the event trajectory chain, such as the probability_alerts field, and can choose to maintain a silent detection state (without reporting the intermediate process).

[0259] 3. Leaving the warning zone and returning to complete silence: If the personnel leave the warning zone but do not enter the core equipment area, no alarm will be generated when the system detects a distance greater than 15 meters.

[0260] 4. Re-entering the warning zone and intruding into the core equipment area formally triggers a risk escalation of the target event: Several minutes later, the person re-enters the warning zone, and the system triggers a pre-escalation alarm again (second pre-escalation). Subsequently, the person crosses the boundary of the core equipment area (distance < 5 meters), and the system immediately triggers a "formal escalation alarm," with content such as: "[Urgent] Target has intruded into the core equipment area! Immediate action!"

[0261] The alert message can optionally reference the corresponding pre-upgrade record to form a complete chain of evidence for "approaching → breaking in".

[0262] 5. Event Termination and Final Report: Once the target object leaves the target area after being stopped by security personnel, the event is considered terminated, and the system can generate a complete report. The report may include information such as a "spatial approximation analysis" graph, marking the pre-escalation trigger time (8 meters away) and the formal escalation trigger time (0 meters away), and can also calculate the "reaction time" for handling the target object (or the target event) (such as the time difference from the first pre-escalation to the formal escalation), used to evaluate the event response efficiency (or security response efficiency, etc.).

[0263] This embodiment employs a two-tiered spatial design of "early warning zone" and "core equipment zone" to achieve a progressive alarm system of "early warning upon approach, emergency response upon intrusion." This maintains daily silence (avoiding the nuisance of frequent message pushes) while proactively and in stages providing alerts when risks approach, thus buying valuable reaction time for security. The risk escalation process of this solution can be dynamically linked to spatial location, resulting in higher timely warnings and scenario adaptability.

[0264] It should be further noted that the execution entity of the object detection-based message prompting method can be an object detection-based message prompting device. For example, the object detection-based message prompting method can be executed by a terminal device, a server, or other processing devices. The terminal device can be a user equipment (UE), computer, mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, in-vehicle device, wearable device, etc. In some possible implementations, the object detection-based message prompting method can be implemented by the processor calling computer-readable instructions stored in memory.

[0265] Figure 3 This is a block diagram illustrating a target detection-based message prompting device, as shown in an exemplary embodiment of this application. Figure 3 As shown, this exemplary target detection-based message notification device 300 includes: a target detection module 310, a first-time trigger module 320, a non-first-time trigger module 330, and an event judgment module 340. Specifically: The target detection module 310 is used to perform target detection processing on the current detection scene to obtain the target objects in the current detection scene and the target events triggered by the target objects.

[0266] The first-trigger module 320 is used to process the target event and create the event trajectory chain of the target object if the target object triggers the target event for the first time.

[0267] The non-first-time trigger module 330 is used to analyze the event change information of the target object in the event trajectory chain if the target object triggers the target event for the first time.

[0268] The event judgment module 340 is used to determine whether to process the target event with a message based on the event change information.

[0269] In this exemplary object detection-based message notification device, by performing object detection processing on the current detection scene, the target objects in the current detection scene and the target events triggered by the target objects can be obtained. The target object may be triggering the target event for the first time or not. If the target object triggers the target event for the first time, it is considered a new event, and message notification processing can be performed on the target event, creating an event trajectory chain for the target object. If the target object triggers the target event not for the first time, it is considered an old event, requiring analysis of the event change information of the target object in the event trajectory chain; based on the event change information, it is determined whether message notification processing is worthwhile. Therefore, message notification can be initiated when the target object triggers the target event for the first time, ensuring the timeliness of object detection; when the target object triggers the target event not for the first time, the event change information in the event trajectory chain is analyzed before deciding whether to perform message notification, reducing the redundancy of object detection; and message notification can be performed based on the entire event trajectory chain, ensuring the integrity of object detection, thus optimizing the message notification effect of object detection.

[0270] It should be noted that the apparatus and method provided in the above embodiments belong to the same concept, and the specific ways in which each module and unit performs operations have been described in detail in the method embodiments, and will not be repeated here. In practical applications, the apparatus provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the apparatus can be divided into different functional modules to complete all or part of the functions described above, and this is not a limitation.

[0271] The functions of each module can be found in the implementation example of the message prompting method based on target detection, and will not be repeated here.

[0272] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an embodiment of the electronic device of this application. The electronic device 100 includes a memory 101 and a processor 102. The processor 102 is used to execute program instructions stored in the memory 101 to implement the steps in any of the above-described embodiments of the message prompting method based on target detection. In a specific implementation scenario, the electronic device 100 may include, but is not limited to, a microcomputer or a server. In addition, the electronic device 100 may also include mobile devices such as laptops and tablets, which are not limited here.

[0273] Specifically, processor 102 controls itself and memory 101 to implement the steps in any of the above-described embodiments of the target detection-based message prompting method. Processor 102 may also be referred to as a CPU (Central Processing Unit). Processor 102 may be an integrated circuit chip with signal processing capabilities. Processor 102 may also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor. Furthermore, processor 102 may be implemented using integrated circuit chips.

[0274] In this exemplary electronic device, by performing target detection processing on the current detection scene, the target objects in the current detection scene and the target events triggered by the target objects can be obtained. The target object may be triggering the target event for the first time or not. If the target object is triggering the target event for the first time, then the target event is a new event, and message notification processing can be selected for the target event, and an event trajectory chain of the target object can be created. If the target object is triggering the target event not for the first time, then the target event is an old event, and it is necessary to analyze the event change information of the target object in the event trajectory chain; based on the event change information, it is determined whether it is worthwhile to process message notification for the target event. Therefore, message notification can be initiated when the target object triggers the target event for the first time, ensuring the timeliness of target detection; when the target object is triggering the target event not for the first time, the event change information in the event trajectory chain is analyzed before deciding whether to process message notification, reducing the redundancy of target detection; and message notification can be based on the entire event trajectory chain, ensuring the integrity of target detection, thus optimizing the message notification effect of target detection.

[0275] Please see Figure 5 , Figure 5 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present application. The computer-readable storage medium 110 stores program instructions 111 that can be executed by a processor. The program instructions 111 are used to implement the steps in any of the above-described embodiments of the message prompting method based on target detection.

[0276] In this exemplary storage medium, by running the program instructions within the storage medium, target detection processing is performed on the current detection scene, which yields the target objects in the current detection scene and the target events triggered by the target objects. The target object may be triggering the target event for the first time or not. If the target object is triggering the target event for the first time, it is considered a new event, and message notification processing can be selected, creating an event trajectory chain for the target object. If the target object is triggering the target event not for the first time, it is considered an old event, requiring analysis of the event change information of the target object in the event trajectory chain; based on the event change information, it is determined whether message notification processing is worthwhile. This allows for message notification when the target object triggers the target event for the first time, ensuring the timeliness of target detection; when the target object is triggering the target event not for the first time, analyzing the event change information in the event trajectory chain before deciding whether to issue a message notification reduces the redundancy of target detection; and it also allows for message notification based on the entire event trajectory chain, ensuring the integrity of target detection, thus optimizing the message notification effect for target detection.

[0277] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0278] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0279] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0280] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A message prompting method based on object detection, characterized in that, The method includes: Perform target detection processing on the current detection scene to obtain the target objects in the current detection scene and the target events triggered by the target objects; If the target object triggers the target event for the first time, then the target event is processed with a message prompt, and the event trajectory chain of the target object is created; If the target object is not triggering the target event for the first time, then analyze the event change information of the target object in the event trajectory chain; Determine whether to provide a message notification for the target event based on the event change information.

2. The method according to claim 1, characterized in that, If the target object triggers the target event for the first time, then message notification processing is performed on the target event, and an event trajectory chain of the target object is created, including: If the target object triggers the target event for the first time, then the target detection result corresponding to the target object is obtained; The message prompt is processed based on the target detection result, and the event trajectory chain is created based on the target detection result.

3. The method according to claim 1, characterized in that, After the step of processing the target event with a message and creating the event trajectory chain of the target object when the target object triggers the target event for the first time, the method further includes: The target object is continuously detected to obtain the continuous detection result corresponding to the target object; The event trajectory chain is updated based on the continuous detection results.

4. The method according to claim 1, characterized in that, The analysis of the event change information of the target object in the event trajectory chain includes: Obtain the event attributes of historical events triggered by the target object at different time points in the event trajectory chain; The event change information is determined based on the event attributes of the historical events.

5. The method according to claim 4, characterized in that, Determining the event change information based on the event attributes of the historical events includes: Obtain the event attributes of each historical event, including the event trigger sequence and the event risk level; Based on the event triggering sequence and the event risk level, determine the security level change information of each historical event in sequence; The event change information of the target object is determined based on the security level change information.

6. The method according to claim 5, characterized in that, After determining the event change information of the target object based on the security level change information, the method includes: If the event change information indicates that the event risk level of the target event is greater than the event risk level of the historical event, a message notification is sent to the target event. If the event risk level of the target event, as indicated by the event change information, is less than or equal to the event risk level of the historical event, then the target event is silenced.

7. The method according to claim 4, characterized in that, Determining the event change information based on the event attributes of the historical events includes: Obtain the event attributes of each historical event, including the event triggering sequence and the event triggering location; Based on the event triggering sequence and the event triggering location, determine the spatial location change information of each historical event in sequence; The event change information of the target object is determined based on the spatial location change information.

8. The method according to claim 7, characterized in that, The step of determining the spatial location change information of each historical event in sequence according to the event triggering time sequence and the event triggering location includes: Obtain the topological relationship diagram between each target region in the current detection scene; Based on the target area corresponding to each event triggering position and the event triggering sequence in the topology diagram, determine the target topology path; The target topological path is compared with the obtained reference topological path to obtain the spatial location change information.

9. The method according to claim 1, characterized in that, The analysis of the event change information of the target object in the event trajectory chain includes: Obtain the event risk level and event triggering location of the target event currently triggered by the target object, as well as the event risk level of the historical events triggered by the target object in the event trajectory chain; the event triggering sequence of the historical events is earlier than the event triggering sequence of the currently triggered target event; If the target area corresponding to the event triggering location in the current detection scenario is connected to a high-risk area in the current detection scenario, then the event risk level of the currently triggered target event is upgraded to obtain an upgraded event risk level. The event change information is determined based on the upgraded event risk level and the event risk level of the historical events.

10. The method according to claim 1, characterized in that, After performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object, the method further includes: If the target object is not triggering the target event for the first time, then obtain the distance between the event triggering location of the target event and the boundary of the high-risk area in the current detection scene; If the distance is less than the obtained warning distance threshold, then a message prompt will be sent to the target event.

11. The method according to claim 10, characterized in that, Before the step of sending a message to the target event if the distance is less than the obtained warning distance threshold, the method further includes: Obtain the direction of motion of the target object; If the direction of movement points towards the high-risk area, then the warning distance threshold is increased.

12. The method according to claim 1, characterized in that, After performing target detection processing on the current detection scene to obtain the target object in the current detection scene and the target event triggered by the target object, the method further includes: If there are multiple target events, the trigger distance between each target event is determined based on the event triggering location of each target event; Cluster the target events based on the trigger distance to obtain the event clustering results; If the event clustering result indicates that the number of target events clustered within a preset range in the current detection scenario is greater than a preset threshold, then message prompts will be processed for the target events clustered within the preset range.

13. The method according to claim 1, characterized in that, After determining whether to provide a message notification for the target event based on the event change information, the method further includes: If the detection of the target object is completed, the detection results corresponding to each historical event in the event trajectory chain are obtained, and the detection results include spatiotemporal information and event content. Based on the spatiotemporal information and the event content, a complete prompt message corresponding to the event trajectory chain is generated for message prompting processing.

14. An electronic device, characterized in that, The method includes a memory and a processor, the processor being configured to execute program instructions stored in the memory to implement the method according to any one of claims 1 to 13.

15. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by the processor, they implement the method described in any one of claims 1 to 13.

Images