Video monitoring method and system based on edge-cloud collaboration

Abstract

This invention provides a video surveillance method and system based on edge-cloud collaboration, relating to the technical field of video surveillance. The method includes: receiving trigger events reported by image acquisition devices at the edge; performing quantitative analysis on the frame images corresponding to the trigger events based on trigger conditions; generating alarm levels according to the quantitative analysis results; and generating graded alarm signals according to preset alarm rules. The edge-cloud collaborative video surveillance method and system provided by this invention allows the image acquisition devices at the edge to run only simple trigger condition detection logic during monitoring, while the high-computing-power quantitative analysis task is implemented on the cloud server, effectively reducing the hardware cost of the image acquisition devices. Simultaneously, the cloud server can perform quantitative analysis and alarm signal grading, effectively improving the accuracy and efficiency of video surveillance.

Description

Technical Field

[0001] This invention relates to the technical field of video surveillance, and in particular to a video surveillance method and system based on edge-cloud collaboration. Background Technology

[0002] Currently, the management of positions such as supermarket cashiers, guard booth operators, warehouse staff, company receptionists, and swimming lifeguards mainly relies on manual patrols or simple video surveillance. Moreover, existing technical solutions are mainly divided into two categories: monitoring methods using ordinary cameras without AI functions and monitoring methods using smart cameras with AI functions. Among them, the monitoring method using ordinary cameras suffers from low efficiency and difficulty in providing 24 / 7 coverage, while the monitoring method using smart cameras often results in higher costs, making it difficult to support large-scale deployment, which also affects monitoring efficiency. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a video surveillance method and system based on edge-cloud collaboration, which can improve video surveillance efficiency, achieve cost control, and facilitate large-scale deployment, thereby alleviating the above-mentioned technical problems.

[0004] In a first aspect, embodiments of the present invention provide a video surveillance method based on edge-cloud collaboration, applied to a cloud server; the cloud server communicates with at least one image acquisition device at the edge; the method includes: receiving a trigger event reported by the image acquisition device at the edge; wherein the trigger event is reported by the image acquisition device when it determines that an image in the acquired video stream meets a preset trigger condition; performing quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition; generating an alarm level according to the quantitative analysis result, and generating a graded alarm signal according to a preset alarm rule.

[0005] In conjunction with the first aspect, this embodiment of the invention provides a first possible implementation of the first aspect, wherein the reporting content carried in the triggering event includes frame images that satisfy the triggering conditions; the triggering conditions include at least one of the following: the pixel change rate between adjacent frame images exceeds a preset threshold; the image acquisition device detects a human-shaped target, and the confidence level of the human-shaped target exceeds a preset confidence level threshold; the image acquisition device detects the disappearance of the original human-shaped target, and the duration of the disappearance reaches a preset duration threshold.

[0006] In conjunction with the first possible implementation of the first aspect, this embodiment of the invention provides a second possible implementation of the first aspect, wherein the quantitative analysis includes personnel identity association analysis; the step of quantitatively analyzing the frame image corresponding to the triggering event based on the triggering condition includes: extracting the human-shaped region contained in the frame image, extracting features from the human-shaped target in the human-shaped region, and generating a feature vector corresponding to the human-shaped target; calculating the similarity between the feature vector and each known feature vector in a preset personnel database; if there is a known feature vector whose similarity is greater than a preset first similarity threshold, then determining the personnel identity of the human-shaped target by the personnel identity corresponding to the known feature vector.

[0007] In conjunction with the second possible implementation of the first aspect, this embodiment of the invention provides a third possible implementation of the first aspect, wherein the above method further includes: if there is no known feature vector with a similarity greater than the first similarity threshold, then the humanoid target is identified as an unknown person, and the unknown person is recorded.

[0008] In conjunction with the first aspect, this embodiment of the invention provides a fourth possible implementation of the first aspect, wherein the quantitative analysis further includes cross-camera tracking analysis; the step of quantitatively analyzing the frame image corresponding to the triggering event based on the triggering condition further includes: if there are multiple image acquisition devices, extracting the appearance feature vectors detected in each of the image acquisition devices; calculating the similarity of the appearance feature vectors extracted by different image acquisition devices within the same time window; when the similarity is greater than a preset second similarity threshold, determining the appearance feature vectors extracted by different image acquisition devices as the appearance feature vectors of the same person; associating the occurrence records of the appearance feature vectors of the same person between different image acquisition devices, and tracking the trajectory of the person based on the association result.

[0009] In conjunction with the second possible implementation of the first aspect, this embodiment of the invention provides a fifth possible implementation of the first aspect, wherein the steps of generating alarm levels based on quantitative analysis results and generating graded alarm signals according to preset alarm rules include: if the identity of the humanoid target is determined and the humanoid target is detected to appear in a preset target area, a first alarm signal is generated at a preset time interval; the first alarm signal is used to indicate that the humanoid target is in the preset target area; if the identity of the humanoid target is determined and the humanoid target is detected to disappear from the target area, the duration of disappearance is recorded; if the duration of disappearance reaches a preset duration threshold, a second alarm message is generated; the second alarm message is used to indicate that the humanoid target has disappeared from the target area; after generating the second alarm message, the duration of disappearance is continuously recorded; if the duration of disappearance reaches a preset second duration threshold, a third alarm message is generated, and a reminder message is sent to the user terminal based on the third alarm message.

[0010] In conjunction with the first aspect, this embodiment of the invention provides a sixth possible implementation of the first aspect, wherein the above method further includes: aggregating the graded alarm signals according to a preset statistical period, and calculating multi-dimensional quantitative indicators based on the graded alarm signals within the preset statistical period; generating a statistical report containing the multi-dimensional quantitative indicators.

[0011] Secondly, embodiments of the present invention also provide a video surveillance method based on edge-cloud collaboration, applied to communication of at least one image acquisition device at the edge, wherein the image acquisition device is deployed in the area to be monitored, and the image acquisition device communicates with a cloud server. The method includes: continuously acquiring video streams of the area to be monitored; determining whether the images contained in the video stream meet preset trigger conditions; if so, sending a trigger event generated when the trigger conditions are met to the cloud server, so that the cloud server performs quantitative analysis on the frame image corresponding to the trigger event based on the trigger conditions, generates an alarm level according to the quantitative analysis results, and generates a graded alarm signal according to preset alarm rules.

[0012] In conjunction with the second aspect, the present invention provides a first possible implementation of the second aspect, wherein the step of determining whether the image contained in the video stream satisfies the preset triggering conditions includes: if it is determined that the image contained in the video stream satisfies one of the following triggering conditions, then it is determined that the image contained in the video stream satisfies the preset triggering conditions: the pixel change rate between adjacent frame images exceeds a preset threshold; the image acquisition device detects a human-shaped target, and the confidence level of the human-shaped target exceeds a preset confidence level threshold; the image acquisition device detects the disappearance of the original human-shaped target, and the duration of the disappearance reaches a preset duration threshold.

[0013] Secondly, embodiments of the present invention also provide a video surveillance system based on edge-cloud collaboration. The system includes: an edge terminal, comprising at least one image acquisition device deployed in a monitored area for continuously acquiring video streams of the monitored area; and determining whether the images contained in the video stream meet preset trigger conditions. If so, the edge terminal sends a trigger event generated when the trigger conditions are met to a cloud server. The cloud server establishes communication with the image acquisition device at the edge terminal, receives the trigger event, performs quantitative analysis on the frame images corresponding to the trigger event based on the trigger conditions, generates an alarm level based on the quantitative analysis results, and generates graded alarm signals according to preset alarm rules.

[0014] The embodiments of the present invention bring the following beneficial effects: This invention provides a video surveillance method and system based on edge-cloud collaboration, capable of receiving trigger events reported by image acquisition devices at the edge. These trigger events are reported by the image acquisition devices when they determine that an image in the acquired video stream meets preset trigger conditions. The system performs quantitative analysis on the frame image corresponding to the trigger event based on the trigger conditions, generates alarm levels according to the quantitative analysis results, and generates graded alarm signals according to preset alarm rules. During monitoring, the image acquisition devices at the edge only need to run simple trigger condition detection logic, while the high-computing quantitative analysis task is implemented on the cloud server, effectively reducing the hardware cost of the image acquisition devices. Simultaneously, the cloud server can perform quantitative analysis and alarm signal grading, effectively improving the accuracy and efficiency of video surveillance.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a video surveillance method based on edge-cloud collaboration provided in an embodiment of the present invention; Figure 2 A flowchart illustrating another edge-cloud collaborative video surveillance method provided in this embodiment of the invention; Figure 3 A structural block diagram of a video surveillance system based on edge-cloud collaboration provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a video surveillance device based on edge-cloud collaboration provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of another edge-cloud collaborative video surveillance device provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Currently, the management of positions such as supermarket cashiers, guard booth operators, warehouse staff, company receptionists, and swimming lifeguards mainly relies on manual patrols or simple video surveillance. Furthermore, existing technical solutions primarily fall into two categories: (1) Monitoring methods using ordinary cameras without AI functionality; This type of monitoring equipment only has motion detection capabilities, meaning it determines whether the image has changed by comparing the pixel differences between adjacent video frames. Its logic typically involves setting a pixel change threshold (e.g., 5%-10%); when the difference between frames exceeds this threshold, an alarm is triggered. The drawback of this method is that it cannot distinguish whether the change is caused by human intervention or environmental interference, such as swaying curtains or changes in lighting, leading to a high false alarm rate. Users often disable the alarm function due to frequent invalid alarms, rendering the monitoring system ineffective.

[0021] (2) Monitoring methods using smart cameras with AI capabilities; This type of surveillance equipment can deploy lightweight AI models on cameras for target detection. However, it has the following drawbacks: low-power AI chips (such as below 0.5 TOPS) can only run 1-2 simple algorithms and cannot perform complex behavior analysis; while high-power AI chips (such as above 2 TOPS) can run complex models, the cost per device increases significantly, making large-scale deployment impractical. Furthermore, after the algorithm model is burned into the firmware, the update and upgrade cycle is long, making it difficult to flexibly adapt to new inspection needs.

[0022] In addition, existing monitoring solutions typically only provide instantaneous alarm message pushes and cannot conduct long-term, multi-dimensional statistical analysis of on-duty status, such as statistical compliance rate and cumulative off-duty time, making it difficult to provide data support for management decisions.

[0023] Therefore, there is a lack of a job inspection solution in the existing technology that can control the cost of front-end hardware, achieve high-precision analysis, and have in-depth statistical capabilities.

[0024] Based on this, the video surveillance method and system based on edge-cloud collaboration provided by the embodiments of the present invention can improve video surveillance efficiency and achieve high-precision analysis while keeping costs under control and facilitating large-scale deployment.

[0025] To facilitate understanding of this embodiment, a video surveillance method based on edge-cloud collaboration disclosed in this embodiment of the invention will first be described in detail.

[0026] In one possible implementation, embodiments of the present invention provide a video surveillance method based on edge-cloud collaboration, applied to a cloud server; the cloud server communicates with at least one image acquisition device at the edge, such as... Figure 1 The flowchart shown illustrates a video surveillance method based on edge-cloud collaboration, which includes the following steps: Step S102: Receive the trigger event reported by the image acquisition device at the edge; In this embodiment of the invention, the triggering event is reported by the image acquisition device when it determines that the image in the acquired video stream meets the preset triggering conditions. Step S104: Perform quantitative analysis on the frame image corresponding to the triggering event based on the triggering conditions; Step S106: Generate alarm levels based on quantitative analysis results, and generate graded alarm signals according to preset alarm rules.

[0027] In practical use, the edge end in this embodiment of the invention includes at least one image acquisition device. The image acquisition device is deployed in the area to be monitored and can continuously acquire video streams. The image acquisition device in this embodiment of the invention has at least the following functions: (a) motion detection capability, that is, detecting changes in the screen through the frame difference method; (b) basic human detection capability, that is, human targets can be identified through a lightweight target detection model deployed on one side of the edge end.

[0028] Furthermore, the image acquisition device establishes communication with the cloud server, enabling the image acquisition device to report a trigger event to the cloud server when it determines that the image in the acquired video stream meets the preset trigger conditions, thereby realizing quantitative analysis and alarm signal classification on the cloud server.

[0029] This invention provides a video surveillance method based on edge-cloud collaboration, capable of receiving trigger events reported by image acquisition devices at the edge. These trigger events are reported by the image acquisition devices when they determine that an image in the acquired video stream meets preset trigger conditions. The method performs quantitative analysis on the frame image corresponding to the trigger event based on the trigger conditions, generates alarm levels according to the quantitative analysis results, and generates graded alarm signals according to preset alarm rules. During monitoring, the image acquisition devices at the edge only need to run simple trigger condition detection logic, while the high-computing quantitative analysis task is implemented on the cloud server, effectively reducing the hardware cost of the image acquisition devices. Simultaneously, the cloud server can perform quantitative analysis and alarm signal grading, effectively improving the accuracy and efficiency of video surveillance.

[0030] In practical use, in this embodiment of the invention, the reporting content carried in the trigger event reported by the image acquisition device includes frame images that meet the trigger conditions; wherein, the trigger conditions in this embodiment of the invention include at least one of the following: the pixel change rate between adjacent frame images exceeds a preset threshold; the image acquisition device detects a human-shaped target, and the confidence level of the human-shaped target exceeds a preset confidence level threshold; the image acquisition device detects the disappearance of the original human-shaped target, and the duration of the disappearance reaches a preset duration threshold.

[0031] Based on the aforementioned triggering conditions, the image acquisition device can continuously acquire video streams of the monitored area while also executing simple judgment logic. For ease of understanding, Table 1 shows a judgment logic table for an image acquisition device. Table 1 uses examples of image acquisition devices including ordinary cameras and smart cameras for illustration, as shown in Table 1 below: Table 1:

[0032] Among them, ordinary cameras refer to cameras with motion detection capabilities, while smart cameras are equipped with AI chips and can perform basic human detection capabilities on the basis of the motion detection capabilities of ordinary cameras. That is, the deployed lightweight target detection model can identify human targets.

[0033] Based on the above triggering conditions, the image acquisition device does not need to continuously upload video streams to the cloud server. Instead, it only reports a trigger event when an image in the video stream meets any of the above triggering conditions. For example, when a human-shaped target is detected and the confidence level of the human-shaped target exceeds a preset confidence threshold (e.g., ≥0.6), a trigger event is reported. In this case, the reporting content carried in the trigger event includes the frame image at the triggering time and the coordinates of the detection box of the human-shaped target. For example, when the image acquisition device detects the disappearance of an existing human-shaped target and the duration of the disappearance reaches a preset duration threshold (e.g., ≥30 seconds), a trigger event is reported. In this case, the reporting content carried in the trigger event includes the disappearance of the human-shaped target and the frame image of the last detected human-shaped target.

[0034] The aforementioned image acquisition device reports trigger events according to trigger conditions, which actually implements an event-driven reporting process. It does not require continuous uploading of video streams, which can significantly reduce system bandwidth usage and cloud concurrency pressure.

[0035] Furthermore, the cloud server can listen for trigger events reported by the image acquisition devices at the edge via a message queue. Upon receiving a trigger event, quantitative analysis is performed. Specifically, the quantitative analysis in this embodiment includes personnel identity association analysis. When the cloud server performs quantitative analysis on the frame image corresponding to the trigger event based on the trigger conditions, it can call the face recognition model or appearance feature extraction model deployed on the cloud server to extract features from the human-shaped region in the frame image, obtaining feature vectors, such as generating 128-dimensional or 256-dimensional feature vectors. The specific quantitative analysis process is as follows: The process involves extracting human-shaped regions from the frame image, extracting features from the human targets within these regions, and generating feature vectors corresponding to the human targets. The similarity between these feature vectors and known feature vectors in a pre-defined personnel database is then calculated. If a known feature vector exists with a similarity greater than a pre-defined first similarity threshold, the person corresponding to that known feature vector is identified as the human target. If no known feature vector exists with a similarity greater than the first similarity threshold, the human target is identified as an unknown person and recorded.

[0036] In practical implementation, the aforementioned similarity can be represented by cosine similarity. That is, the cosine similarity between the feature vector and each known feature vector in a pre-defined personnel database is calculated. When the cosine similarity is greater than a first similarity threshold (e.g., greater than or equal to 0.75), the current humanoid target can be associated with the person's identity corresponding to the known feature vector. If the first similarity threshold is not reached (e.g., less than 0.75), the humanoid target is marked as "unknown person" and recorded. In actual use, the personnel database pre-records the association between known feature vectors and person identities. When a similarity greater than the first similarity threshold exists, it indicates that the quantitative analysis process is associated with a specific person's identity, confirming that the person has been detected in the monitored area.

[0037] Furthermore, considering that multiple image acquisition devices may be deployed in the area to be monitored, the quantitative analysis in this embodiment of the invention also includes cross-camera tracking analysis. Specifically, when performing quantitative analysis, if there are multiple image acquisition devices, the appearance feature vectors detected by each image acquisition device are extracted; the similarity of appearance feature vectors extracted by different image acquisition devices within the same time window is calculated; when the similarity is greater than a preset second similarity threshold, the appearance feature vectors extracted by different image acquisition devices are determined as the appearance feature vectors of the same person; the occurrence records of the appearance feature vectors of the same person in different image acquisition devices are associated, and the trajectory of the person is tracked based on the association results.

[0038] In the cross-camera tracking analysis process, the similarity of appearance feature vectors can also be represented by cosine similarity. For example, the cosine similarity of appearance feature vectors extracted by different image acquisition devices within the same time window (±5 seconds) can be calculated pairwise. When the cosine similarity is greater than the second similarity threshold, such as greater than or equal to 0.7, it is determined to be the same person, and its appearance records in different image acquisition devices are associated. Based on the association results, the continuous trajectory tracking of the person across cameras can be realized, avoiding the misjudgment of leaving the post due to the person moving to an adjacent area.

[0039] Furthermore, the process of generating alarm levels in step S106 above implements a hierarchical alarm process. In specific implementation, different alarm rules can be set for different triggering conditions, and different alarm levels can be determined to achieve in-depth analysis of alarm hierarchy.

[0040] For ease of understanding, Table 2 below illustrates the judgment logic for alarm classification, including alarm level, alarm rules, and response time requirements. (See Table 2 below.) Table 2:

[0041] Based on Table 2 above, when generating a graded alarm signal, if the identity of the humanoid target is determined and the humanoid target is detected to appear in a preset target area, a first alarm signal is generated at a preset time interval. This first alarm signal is used to indicate that the humanoid target is in the preset target area. The first alarm signal corresponds to the green alarm in Table 2 above, that is, the identity of the humanoid target is determined and the humanoid target is detected to appear in the preset target area. For example, during a post inspection, after determining the identity of the humanoid target, it can be further determined that the humanoid target is in the preset post area. At this time, the first alarm information can be pushed to the user terminal. For example, the first alarm information can be pushed to the user terminal every first time interval (e.g., 20 seconds). The user terminal refers to the user terminal that establishes communication with the cloud server, including the user terminal of the humanoid target whose identity has been determined, and may also include the user terminal of the management personnel, to realize the post inspection and monitoring.

[0042] Furthermore, if the identity of the humanoid target is determined, and the humanoid target is detected to disappear from the target area, the duration of the disappearance is recorded. If the duration of the disappearance reaches a preset duration threshold, a second alarm message is generated. This second alarm message is used to indicate that the humanoid target has disappeared from the target area. The second alarm message corresponds to the alarm level of the yellow alarm in Table 2 above. For example, if the humanoid target disappears from the preset work area and the duration of the disappearance reaches the preset duration threshold, such as 60 seconds, a second alarm message can be generated. This second alarm message can also be pushed to the user terminal. For example, when the humanoid target is detected to disappear from the target area and the duration of the disappearance reaches the preset duration threshold, such as 60 seconds, it can be pushed to the user terminal every second time interval (such as 10 seconds) and a telephone reminder can be triggered, such as sending a voice prompt to the user terminal corresponding to the humanoid target.

[0043] Furthermore, after generating the second alarm information, the duration of disappearance is continuously recorded. If the duration of disappearance reaches the preset second duration threshold, a third alarm information is generated, and a reminder message is sent to the user terminal based on the third alarm information. The process of generating the third alarm information corresponds to the alarm level of the red alarm in Table 2 above. For example, if a human target is not detected to return within the preset load period (e.g., within 5 minutes) after the yellow alarm is triggered, a red alarm is generated, that is, the third alarm information is generated. At the same time, a telephone reminder can be triggered immediately, such as sending a voice prompt to the user terminal corresponding to the human target. In addition, the sound and light equipment in the monitored area can be linked to provide on-site alarm prompts.

[0044] In addition, if no human targets are detected in the preset target area for a period of time, no alarm information will be generated, and only the data will be written to the database for statistical query.

[0045] Furthermore, in this embodiment of the invention, the graded alarm signals can be aggregated according to a preset statistical period, and multi-dimensional quantitative indicators can be calculated based on the graded alarm signals within the preset statistical period; a statistical report containing multi-dimensional quantitative indicators can be generated. For example, the cloud server can be configured with a report generation module, which can aggregate graded alarm signals according to a preset statistical period, such as daily, weekly, or monthly, and calculate the following quantitative indicators: Job compliance rate = (Total on-duty hours / Total required on-duty hours) × 100%; Number of times leaving the post = Number of times a yellow alarm is triggered within a preset statistical period; Cumulative absence duration = the sum of the duration of disappearance of all records within the preset statistical period; Furthermore, it can arrange hierarchical alarm information in chronological order, generate anomaly event timelines, and create anomaly event lists; it can also perform shift handover period analysis, such as generating on-duty status curves during shift handover periods, which can visualize the on-duty status of personnel. The above statistical results can be displayed on the user terminal in the form of visual charts.

[0046] Furthermore, corresponding to the aforementioned execution process of the cloud server, this embodiment of the invention also provides a video surveillance method based on edge-cloud collaboration, applied to communication between at least one image acquisition device at the edge, the image acquisition device being deployed in the area to be monitored, and the image acquisition device communicating with the cloud server, such as... Figure 2 The flowchart shown is for another video surveillance method based on edge-cloud collaboration, which includes: Step S202: Continuously collect video streams of the area to be monitored; Step S204: Determine whether the images contained in the video stream meet the preset triggering conditions; Step S206: If yes, send the trigger event generated when the trigger condition is met to the cloud server, so that the cloud server can perform quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition, generate an alarm level according to the quantitative analysis result, and generate a graded alarm signal according to the preset alarm rules.

[0047] In step S204 above, when determining whether the image contained in the video stream meets the preset triggering conditions, if it is determined that the image contained in the video stream meets one of the following triggering conditions, then it is determined that the image contained in the video stream meets the preset triggering conditions: The pixel change rate between adjacent frame images exceeds a preset threshold; The image acquisition device detected a human-shaped target, and the confidence level of the human-shaped target exceeded the preset confidence level threshold; The image acquisition device detected that the original human-shaped target had disappeared, and the duration of the disappearance reached a preset duration threshold.

[0048] In other words, based on the above triggering conditions, the image acquisition device can report the triggering event only when the image in the video stream meets any of the above triggering conditions, instead of continuously uploading the video stream, thereby realizing the event-driven reporting process, which can significantly reduce system bandwidth usage and cloud concurrency pressure.

[0049] Furthermore, after receiving the trigger event reported by the image acquisition device, the cloud server can execute the above... Figure 1 The edge-cloud collaborative video surveillance method shown can be found in the following reference. Figure 1 The process shown will not be repeated here.

[0050] Furthermore, embodiments of the present invention also provide a video surveillance system based on edge-cloud collaboration, such as... Figure 3 The diagram shows a structural block diagram of a video surveillance system based on edge-cloud collaboration. The system includes a cloud server 300 and an edge terminal 302 connected to the cloud server 300. The edge terminal includes at least one image acquisition device 303. (For ease of understanding...) Figure 3 Multiple image acquisition devices 303 are shown, and their specific number can be set according to actual use. This embodiment of the invention does not limit this.

[0051] Furthermore, the image acquisition device is deployed in the area to be monitored to continuously acquire video streams of the area to be monitored; and to determine whether the images contained in the video stream meet preset trigger conditions. If so, a trigger event generated when the trigger conditions are met is sent to the cloud server. The cloud server establishes communication with the image acquisition device at the edge, and is used to receive the trigger event, perform quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition, generate an alarm level according to the quantitative analysis result, and generate a graded alarm signal according to the preset alarm rules.

[0052] Furthermore, the cloud server provided in this embodiment of the invention is also configured with the following functional modules: Event Listening Module: Used to listen for triggered events reported by the edge terminal through a business subscription mechanism; Deep analytics module: Deployed with at least one deep AI analytics model, including one or more of the following: pose estimation model, face recognition model, and object re-identification (ReID) model; Alarm grading module: Generates alarm signals of different levels based on quantitative analysis results, that is, generates alarm levels and generates graded alarm signals according to preset alarm rules; Report generation module: Aggregates historical alarm data within a preset statistical period, such as tiered alarm signals within the preset statistical period, and generates multi-dimensional statistical reports.

[0053] Furthermore, in this embodiment of the invention, the cloud server also establishes communication with the user terminal, which is used to receive tiered alarm signals and view statistical reports.

[0054] The specific communication process between the cloud server and the image acquisition device can be found above. Figure 1 and Figure 2 The corresponding process will not be elaborated here.

[0055] In summary, the video surveillance method and system based on edge-cloud collaboration provided by the embodiments of the present invention have the following beneficial effects: (1) Cost-benefit optimization; By centralizing high-computing quantitative analysis tasks on cloud servers, edge image acquisition devices only need to run extremely lightweight detection logic (such as motion detection or basic human detection, etc.), enabling the entire system to achieve high-quality inspections using low-cost ordinary cameras or low-computing-power smart cameras. Compared with all high-computing-power image acquisition device solutions, the cost is significantly reduced.

[0056] (2) The false alarm rate has been significantly reduced; A dual verification mechanism, employing edge triggering and cloud server verification, is used. High-sensitivity triggering at the edge ensures no false alarms are missed, while the cloud server can deploy deep learning models for secondary screening of triggered events. Real-world testing shows that this mechanism significantly reduces false alarm rates caused by environmental disturbances such as swaying curtains and changes in lighting.

[0057] (3) Supports large-scale deployment; This allows for an event-driven mechanism to trigger event reporting, ensuring that the cloud server's computing power is only used when necessary. For example, assuming N workstations simultaneously deploy the image acquisition device from this embodiment, since the state changes at each workstation are discrete, the cloud server only needs to maintain a concurrent processing capacity far less than N, resulting in significantly better system scalability than the frame-driven scheme.

[0058] (4) Strong decision support capabilities; Breaking away from the limitations of traditional solutions that only provide "what happened," this invention can output management indicators such as compliance rate, cumulative off-duty time, and shift handover analysis through the aggregation and analysis of long-term data, providing data-driven decision support for personnel assessment and resource allocation.

[0059] (5) The algorithm iterates flexibly; Since the core quantitative analysis is deployed on cloud servers, there is no need to modify the device firmware deployed at the edge during updates and iterations. New features (such as uniform recognition, sleep detection, and smoking detection) can be deployed on cloud servers and become fully effective within a few days.

[0060] Furthermore, this embodiment of the invention also provides a video surveillance device based on edge-cloud collaboration, applied to a cloud server; the cloud server communicates with at least one image acquisition device at the edge; such as Figure 4 The diagram shows a structural schematic of a video surveillance device based on edge-cloud collaboration. The device includes: The receiving module 40 is used to receive trigger events reported by the image acquisition device at the edge end; wherein, the trigger event is reported by the image acquisition device when it determines that the image in the acquired video stream meets the preset trigger conditions; Analysis module 42 is used to perform quantitative analysis on the frame image corresponding to the triggering event based on the triggering conditions; The alarm module 44 is used to generate alarm levels based on quantitative analysis results and generate graded alarm signals according to preset alarm rules.

[0061] Furthermore, this embodiment of the invention also provides another edge-cloud collaborative video surveillance device, applied to communication between at least one image acquisition device at the edge, wherein the image acquisition device is deployed in the area to be monitored, and the image acquisition device communicates with a cloud server, such as... Figure 5 The diagram shows another edge-cloud collaborative video surveillance device, which includes: Acquisition module 50 is used to continuously acquire video streams of the area to be monitored; The judgment module 52 is used to determine whether the images contained in the video stream meet the preset triggering conditions; The sending module 54 is used to send a trigger event generated when the judgment result of the judgment module is yes to the cloud server, so that the cloud server can perform quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition, generate an alarm level according to the quantitative analysis result, and generate a graded alarm signal according to the preset alarm rules.

[0062] The edge-cloud collaborative video surveillance device provided in this embodiment of the invention has the same technical features as the edge-cloud collaborative video surveillance method provided in the above embodiments, so it can also solve the same technical problems and achieve the same technical effects.

[0063] Furthermore, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above method.

[0064] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the above-described method.

[0065] Furthermore, embodiments of the present invention also provide a schematic diagram of the structure of an electronic device, such as... Figure 6 The diagram shows the structure of the electronic device, which includes a processor 61 and a memory 60. The memory 60 stores computer-executable instructions that can be executed by the processor 61, and the processor 61 executes the computer-executable instructions to implement the above-described method.

[0066] exist Figure 6 In the illustrated embodiment, the electronic device further includes a bus 62 and a communication interface 63, wherein the processor 61, the communication interface 63, and the memory 60 are connected via the bus 62.

[0067] The memory 60 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 63 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 62 may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 62 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0068] Processor 61 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 61 or by instructions in software form. Processor 61 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in the memory, and the processor 61 reads the information in the memory and uses its hardware to complete the aforementioned method.

[0069] The computer program product of the video surveillance method and system based on edge-cloud collaboration provided in this invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0070] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0071] Furthermore, in the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0072] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion 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.) to execute all or part of the steps of the methods described in the various embodiments of this invention. 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.

[0073] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0074] Finally, it should be noted that the above embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A video surveillance method based on edge-cloud collaboration, characterized in that, Applied to cloud servers; The cloud server communicates with at least one image acquisition device at the edge; the method includes: The image acquisition device at the edge receives a trigger event reported by the image acquisition device; wherein the trigger event is reported by the image acquisition device when it determines that the image in the acquired video stream meets the preset trigger conditions. Quantitative analysis of the frame image corresponding to the triggering event is performed based on the triggering conditions; Alarm levels are generated based on quantitative analysis results, and graded alarm signals are generated according to preset alarm rules.

2. The method according to claim 1, characterized in that, The reporting content carried in the triggering event includes a frame image that meets the triggering conditions; The triggering condition includes at least one of the following: The pixel change rate between adjacent frame images exceeds a preset threshold; The image acquisition device detects a human-shaped target, and the confidence level of the human-shaped target exceeds a preset confidence threshold. The image acquisition device detects that the original human-shaped target has disappeared, and the duration of the disappearance reaches a preset duration threshold.

3. The method according to claim 2, characterized in that, The quantitative analysis includes personnel identity association analysis; The step of quantitatively analyzing the frame image corresponding to the triggering event based on the triggering conditions includes: Extract the human-shaped region contained in the frame image, perform feature extraction on the human-shaped target in the human-shaped region, and generate the feature vector corresponding to the human-shaped target; Calculate the similarity between the feature vector and each known feature vector in the preset personnel database; If there exists a known feature vector with a similarity greater than a preset first similarity threshold, then the person identity corresponding to the known feature vector is used to determine the person identity of the humanoid target.

4. The method according to claim 3, characterized in that, The method further includes: If there is no known feature vector with a similarity greater than the first similarity threshold, the humanoid target is identified as an unknown person and the unknown person is recorded.

5. The method according to claim 1, characterized in that, The quantitative analysis also includes cross-camera tracking analysis; The step of quantitatively analyzing the frame image corresponding to the triggering event based on the triggering conditions further includes: If there are multiple image acquisition devices, then extract the appearance feature vectors detected by each of the image acquisition devices; Calculate the similarity of the appearance feature vectors extracted by different image acquisition devices within the same time window; When the similarity is greater than a preset second similarity threshold, the appearance feature vectors extracted by different image acquisition devices are determined to be the appearance feature vectors of the same person. The occurrence records of the appearance feature vector of the same person across different image acquisition devices are correlated, and the trajectory of the person is tracked based on the correlation results.

6. The method according to claim 3, characterized in that, The steps of generating alarm levels based on quantitative analysis results and generating graded alarm signals according to preset alarm rules include: If the humanoid target is identified as a person and is detected to be in a preset target area, a first alarm signal is generated at preset time intervals; the first alarm signal is used to indicate that the humanoid target is in the preset target area. If the identity of the humanoid target is determined, and the humanoid target is detected to disappear from the target area, the duration of the disappearance is recorded. If the duration of the disappearance reaches a preset duration threshold, a second alarm message is generated. The second alarm message is used to indicate that the humanoid target has disappeared from the target area. After generating the second alarm information, the duration of the disappearance is continuously recorded. If the duration of the disappearance reaches a preset second duration threshold, a third alarm information is generated, and a reminder message is sent to the user terminal based on the third alarm information.

7. The method according to claim 1, characterized in that, The method further includes: The graded alarm signals are aggregated according to a preset statistical period, and multi-dimensional quantitative indicators are calculated based on the graded alarm signals within the preset statistical period. Generate a statistical report containing the quantitative indicators across multiple dimensions.

8. A video surveillance method based on edge-cloud collaboration, characterized in that, The method involves communicating with at least one image acquisition device at the edge, the image acquisition device being deployed in the area to be monitored, and the image acquisition device communicating with a cloud server. Continuously collect video streams of the area to be monitored; Determine whether the images contained in the video stream meet the preset triggering conditions; If so, a trigger event generated when the trigger condition is met is sent to the cloud server, so that the cloud server performs quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition, generates an alarm level according to the quantitative analysis result, and generates a graded alarm signal according to the preset alarm rules.

9. The method according to claim 8, characterized in that, The step of determining whether the images contained in the video stream meet the preset triggering conditions includes: If it is determined that the image contained in the video stream meets one of the following triggering conditions, then the image contained in the video stream is determined to meet the preset triggering condition: The pixel change rate between adjacent frame images exceeds a preset threshold; The image acquisition device detects a human-shaped target, and the confidence level of the human-shaped target exceeds a preset confidence threshold. The image acquisition device detects that the original human-shaped target has disappeared, and the duration of the disappearance reaches a preset duration threshold.

10. A video surveillance system based on edge-cloud collaboration, characterized in that, The system includes: The edge terminal includes at least one image acquisition device deployed in the area to be monitored for continuously acquiring video streams of the area to be monitored; and determining whether the images contained in the video stream meet preset trigger conditions. If so, a trigger event generated when the trigger conditions are met is sent to the cloud server. The cloud server establishes communication with the image acquisition device at the edge, and is used to receive the trigger event, perform quantitative analysis on the frame image corresponding to the trigger event based on the trigger condition, generate an alarm level according to the quantitative analysis result, and generate a graded alarm signal according to the preset alarm rules.

Images