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

Abstract

The application provides an abnormality detection method and device, electronic equipment and storage medium, and relates to the technical field of monitoring. The method comprises the following steps: receiving a target image obtained by a target pan-tilt camera shooting a target monitoring point; when it is judged that the target monitoring point is initially abnormal based on the target image, each pan-tilt camera in a pan-tilt camera group is dispatched to shoot the target monitoring point respectively, and a group of images are obtained, each pan-tilt camera in the pan-tilt camera group being located at a different angle position of the target monitoring point; each image in the group of images is subjected to image recognition respectively, and a group of recognition results are obtained; and whether the target monitoring point is abnormal is determined according to the group of recognition results. The technical scheme provided by the application achieves the effect of improving the abnormality detection efficiency.

Description

Technical Field

[0001] This application relates to the field of monitoring technology, specifically to an anomaly detection method, device, electronic device, and storage medium. Background Technology

[0002] In the field of power system automation monitoring technology, substation inspection mainly relies on periodic scanning by fixed cameras or single pan-tilt cameras. Image acquisition is completed through preset programs or manual remote control, and then anomaly identification is performed on the single-view images to determine whether there are any abnormalities in the substation (such as a certain power equipment or component). The methods in related technologies have the problem of single-view limitations, which may lead to false detections due to obstruction, changes in lighting, or the target size being too small. Some related technologies also adopt the method of manually adjusting the angle of other cameras to obtain images from multiple angles. However, this method cannot obtain multi-view data in real time, delaying the anomaly confirmation time. It can be seen that the anomaly detection methods in related technologies have the technical problem of low efficiency. Summary of the Invention

[0003] To address the aforementioned technical problems, this application provides an anomaly detection method, apparatus, electronic device, and storage medium.

[0004] In a first aspect, this application provides an anomaly detection method, comprising: receiving a target image uploaded by a target PTZ camera, wherein the target image is obtained by the target PTZ camera capturing a target monitoring point; if the target monitoring point is determined to be initially abnormal based on the target image, scheduling each PTZ camera in the PTZ camera group to capture images of the target monitoring point, thereby obtaining a set of images, wherein the set of images includes images captured by each PTZ camera in the PTZ camera group capturing images of the target monitoring point, and each PTZ camera in the PTZ camera group being located at different angular positions of the target monitoring point; performing image recognition on each image in the set of images to obtain a set of recognition results, wherein each recognition result in the set of recognition results corresponds to one image in the set of images; and determining whether an anomaly exists at the target monitoring point based on the set of recognition results.

[0005] By adopting the above technical solution, the target image uploaded by the target PTZ camera is first received and it is determined whether the target monitoring point is initially abnormal. If it is initially abnormal, a group of PTZ cameras at different angles are scheduled to take a set of images. The set of images is then identified to obtain a set of identification results. Finally, the identification results are used to determine whether there is an abnormality at the target monitoring point. This avoids the limitation of single-view limitation, can acquire multi-view data in real time, improves the efficiency of anomaly detection, and reduces false detections.

[0006] Optionally, each pan-tilt camera in the pan-tilt camera group is scheduled to take pictures of the target monitoring point to obtain a set of images, including: identifying the pan-tilt camera group that meets the preset conditions; scheduling each pan-tilt camera in the pan-tilt camera group to take pictures of the target monitoring point simultaneously and uploading the pictures; and acquiring the pictures uploaded by each pan-tilt camera in the pan-tilt camera group to obtain a set of images.

[0007] By adopting the above technical solution, the system first receives target images uploaded by the target PTZ camera. Based on this, when the target monitoring point is determined to be initially abnormal, the PTZ camera group that meets the preset conditions can accurately select suitable cameras to assist in shooting. These cameras are scheduled to simultaneously shoot and upload images of the target monitoring point, enabling real-time acquisition of multi-view data. This avoids delays in anomaly confirmation time caused by manual intervention. Finally, the images uploaded by each camera are obtained to form a set of images, providing a multi-angle and more comprehensive data foundation for subsequent image recognition and accurate judgment of whether there are anomalies at the target monitoring point, thus improving the accuracy and efficiency of anomaly detection. Through automated scheduling and synchronous shooting, the multi-view image acquisition is fully automated, reducing manual intervention and improving the system's operating efficiency and intelligence level. The synchronously captured multi-view images can more comprehensively reflect the state of the target monitoring point, providing richer information for subsequent image recognition and anomaly judgment, thereby improving the accuracy of anomaly detection.

[0008] Optionally, a pan-tilt camera group that meets preset conditions is determined, including one of the following: forming a pan-tilt camera group from each pan-tilt camera in the pan-tilt camera cluster whose distance from the target pan-tilt camera is less than a preset distance threshold, wherein the pan-tilt camera group includes the target pan-tilt camera; or determining candidate pan-tilt cameras from the pan-tilt camera cluster according to a preset pan-tilt configuration table, and forming a pan-tilt camera group from the candidate pan-tilt cameras, wherein the pan-tilt configuration table records the monitoring points covered by each pan-tilt camera in the pan-tilt camera group, and the monitoring range of each pan-tilt camera in the candidate pan-tilt cameras covers the target monitoring point.

[0009] By adopting the above technical solution, a pan-tilt camera group that meets the preset conditions can be determined from the pan-tilt camera group. The pan-tilt camera group can be flexibly assembled according to the distance of the target pan-tilt camera or the pan-tilt configuration table, so that the pan-tilt camera group can shoot the target monitoring point from different angles, avoid the limitations of single-view, reduce false detection or missed detection caused by obstruction, changes in lighting or the target size being too small, improve the accuracy and efficiency of anomaly detection, and at the same time realize the acquisition of multi-view data in real time and timely confirmation of anomalies.

[0010] Optionally, each pan-tilt camera in the pan-tilt camera group is scheduled to capture images of the target monitoring point to obtain a set of images, including: determining the target information of the target monitoring point based on the target images, wherein the target information includes target spatial coordinate information and target size information; sending the target information to each pan-tilt camera in the pan-tilt camera group so that each pan-tilt camera in the pan-tilt camera group can adjust its attitude and focus based on the target information, and synchronously capture images of the target monitoring point and upload the captured images; receiving the images uploaded by each pan-tilt camera in the pan-tilt camera group to obtain a set of images.

[0011] By adopting the above technical solution, after receiving the target image uploaded by the target pan-tilt camera and determining that the target monitoring point is initially abnormal, the target spatial coordinates and size information of the target monitoring point are determined based on the target image. This information is then sent to each pan-tilt camera in the pan-tilt camera group, allowing each camera to adjust its posture and focal length accordingly, synchronously capture images of the target monitoring point, and upload them. This improves the accuracy and relevance of the captured images and avoids blurry images or failure to cover the target monitoring point due to unsuitable camera posture and focal length. Finally, the images uploaded by each pan-tilt camera are received to obtain a set of images for subsequent identification, which helps to comprehensively and accurately identify whether there is anomaly at the target monitoring point from multiple angles, solving the problem of low efficiency in anomaly detection methods in related technologies.

[0012] Optionally, the target information of the target monitoring point is determined based on the target image, including: determining the pixel coordinates and abnormal pixel size of the target monitoring point based on the target image; obtaining target spatial coordinate information based on the target pose data, target field of view data and pixel coordinates of the target pan-tilt camera; and obtaining target size information based on the target pose data, target field of view data and abnormal pixel size of the target pan-tilt camera.

[0013] By adopting the above technical solution, the pixel coordinates and abnormal pixel size of the target monitoring point can be determined using the target image. Then, by combining the target pose data and target field of view data of the target PTZ camera, the target spatial coordinate information and target size information of the target monitoring point can be accurately obtained. This provides an accurate basis for each PTZ camera in the subsequent PTZ camera group to adjust its attitude and focus based on the target information and to synchronously shoot the target monitoring point. It avoids the problem of poor shooting effect due to inaccurate target information, improves the accuracy of image recognition, and thus more efficiently determines whether there are abnormalities in the target monitoring point.

[0014] Optionally, determine whether there is an anomaly at the target monitoring point based on a set of identification results, including: determining that there is an anomaly at the target monitoring point when the proportion of the anomaly identification results contained in a set of identification results is greater than or equal to a preset proportion threshold.

[0015] By adopting the above technical solution, image recognition is performed on a set of images to obtain a set of recognition results. Based on the comparison between the proportion of abnormal recognition results in a set of recognition results and the preset proportion threshold, it is determined whether there is an anomaly at the target monitoring point. This can overcome the limitations of a single viewpoint, avoid false detection or missed detection caused by occlusion, changes in lighting, or the target size being too small, and improve the accuracy and efficiency of anomaly detection.

[0016] Optionally, determining whether a target monitoring point is abnormal based on a set of identification results includes: calculating a comprehensive score by weighted summation of the confidence values ​​corresponding to each identification result in the set of identification results; and determining that the target monitoring point is abnormal when the comprehensive score is greater than a preset score threshold.

[0017] By adopting the above technical solution, a comprehensive score is obtained by weighted summation of the confidence values ​​corresponding to each identification result in a set of identification results. This can more scientifically and accurately assess the anomalies of the target monitoring point. When the comprehensive score is greater than the preset score threshold, it is determined that there is an anomaly in the target monitoring point. This can improve the accuracy and reliability of anomaly detection, avoid false detection or missed detection caused by single-view detection, and improve the efficiency of anomaly detection.

[0018] Optionally, after determining that there is an anomaly at the target monitoring point, the above method may also include: issuing an alarm message.

[0019] By adopting the above technical solution, when it is determined that there is indeed an anomaly at the target monitoring point, the system will trigger an alarm mechanism and send alarm information to the designated object (such as maintenance personnel, central control system, etc.). These alarm information can be conveyed through various channels, such as SMS, email, in-application notifications, or directly displaying warnings in the monitoring center.

[0020] In a second aspect of this application, an anomaly detection device is also provided, comprising: a receiving module for receiving a target image uploaded by a target PTZ camera, wherein the target image is obtained by the target PTZ camera capturing a target monitoring point; a scheduling module for scheduling each PTZ camera in a PTZ camera group to capture images of the target monitoring point separately, thereby obtaining a set of images, wherein the set of images includes images captured by each PTZ camera in the PTZ camera group, and each PTZ camera in the PTZ camera group is located at a different angular position of the target monitoring point; an identification module for performing image identification on each image in the set of images, thereby obtaining a set of identification results, wherein each identification result in the set of identification results corresponds to one image in the set of images; and a determination module for determining whether an anomaly exists at the target monitoring point based on the set of identification results.

[0021] In a third aspect of this application, an electronic device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the program to implement the method steps of any of the above claims.

[0022] In a fourth aspect of this application, a computer-readable storage medium is also provided, which stores instructions that, when executed, perform the method steps of any of the above claims.

[0023] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: 1. It avoids the limitations of a single-view perspective, can acquire multi-view data in real time, improves anomaly detection efficiency, and reduces false detections; 2. Through automated scheduling and synchronous shooting, the acquisition of multi-view images is fully automated, reducing manual intervention and improving the system's operating efficiency and intelligence level. The synchronously captured multi-view images can more comprehensively reflect the status of the target monitoring point, providing richer information for subsequent image recognition and anomaly judgment, thereby improving the accuracy of anomaly detection. 3. Based on the target image, the target spatial coordinate information and target size information of the target monitoring point are determined. This information is then sent to each pan-tilt camera in the pan-tilt camera group, allowing each camera to adjust its posture and focal length accordingly, synchronously capture images of the target monitoring point, and upload the images. This improves the accuracy and relevance of the captured images and avoids blurry images or failure to cover the target monitoring point due to unsuitable camera posture and focal length. Attached Figure Description

[0024] Figure 1 This is a flowchart of an anomaly detection method provided in an embodiment of this application; Figure 2 This is an architecture diagram of a multi-device collaborative active monitoring and decision-making system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the spatial coordinate transformation algorithm provided in the embodiments of this application; Figure 4 This is a schematic diagram of the multi-perspective decision-making process provided in the embodiments of this application; Figure 5 This is a structural block diagram of an anomaly detection device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application.

[0025] Explanation of reference numerals in the attached figures: 600 - Electronic device; 601 - Processor; 602 - Communication bus; 603 - User interface; 604 - Network interface; 605 - Memory. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0027] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0028] In the description of the embodiments of this application, the term "multiple" means two or more. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0029] This application provides an anomaly detection method, referring to... Figure 1 , Figure 1 This is a flowchart of an anomaly detection method provided in an embodiment of this application. The method includes: Step S101: Receive the target image uploaded by the target PTZ camera, wherein the target image is obtained by the target PTZ camera capturing the target monitoring point; Step S102: If the target monitoring point is determined to be initially abnormal based on the target image, each pan-tilt camera in the pan-tilt camera group is scheduled to take pictures of the target monitoring point to obtain a set of images. The set of images includes the images obtained by each pan-tilt camera in the pan-tilt camera group taking pictures of the target monitoring point. Each pan-tilt camera in the pan-tilt camera group is located at a different angle position of the target monitoring point. Step S103: Perform image recognition on each image in the set of images to obtain a set of recognition results, wherein each recognition result in the set of recognition results corresponds to one image in the set of images; Step S104: Determine whether there is an anomaly at the target monitoring point based on a set of identification results.

[0030] Through the above steps, the target image uploaded by the target PTZ camera is first received and it is determined whether the target monitoring point is initially abnormal. If it is initially abnormal, a group of PTZ cameras at different angles are scheduled to take a set of images. The set of images is then identified to obtain a set of identification results. Finally, the identification results are used to determine whether there is an abnormality at the target monitoring point. This avoids the limitation of single-view limitation, can acquire multi-view data in real time, improves the efficiency of anomaly detection, and reduces false detections.

[0031] The execution entity of the above steps can be a server, the cloud, or a central processing node. This embodiment provides an anomaly detection method for power system automation monitoring, such as in substations. First, a target image captured by a target PTZ camera is received from the target monitoring point. A preliminary judgment is made on the target image, for example, based on analysis, to preliminarily determine whether an anomaly might exist at the target monitoring point. If an anomaly is preliminarily determined, a group (or multiple) of PTZ cameras, located at different angles of the target monitoring point, are further dispatched to capture images of the same target monitoring point, resulting in a set of images. This set of images contains images from different perspectives of the target monitoring point. Image recognition processing is performed on each of these images to obtain a set of recognition results. The combined set of recognition results is then used to finally determine whether an actual anomaly exists at the target monitoring point. In other words, a single PTZ camera (target PTZ camera) performs routine shooting of the monitoring point, and image recognition technology is used to determine if a preliminary anomaly exists. When a preliminary anomaly is detected, a group of PTZ cameras (PTZ camera group) pre-deployed at different spatial angles are automatically dispatched to simultaneously capture images of the target monitoring point, acquiring a multi-view image dataset. After independently analyzing each set of view images, the recognition results from multiple perspectives are combined for joint decision-making to ultimately determine whether an anomaly exists. The aforementioned PTZ camera group may or may not include the target PTZ camera. In related technologies, due to the reliance on a single camera or the angle limitations of a fixed camera, false detections are easily caused by factors such as occlusion and changes in lighting. Furthermore, manual adjustment of the camera angle cannot achieve real-time acquisition of multi-view data, resulting in low detection efficiency. This embodiment uses a PTZ camera group to capture images of the target monitoring point from different angles, enabling more comprehensive capture of the target's feature information. Even when some views are obstructed or lighting conditions are poor, images from other angles may still provide sufficient information for anomaly detection, effectively reducing the probability of false detection. After detecting an initial anomaly, the PTZ camera group can be quickly scheduled for multi-angle shooting, and image recognition and analysis can be performed in real time without manual adjustment of the camera angle, greatly shortening the anomaly confirmation time and improving the system's response speed. This embodiment reduces manual intervention and improves the system's intelligence level through automated scheduling and multi-view collaborative analysis. The system can automatically judge initial anomalies and trigger further multi-view detection, realizing an automated anomaly detection process and improving the intelligence level of power system automated monitoring. Taking the abnormal detection of power equipment in a substation as an example, a group of PTZ cameras is deployed throughout the substation. The PTZ camera group includes the aforementioned target PTZ cameras and PTZ camera sets. The aforementioned target monitoring points can be power equipment in the substation, a local area of ​​the power equipment, or components in the equipment, etc.

[0032] In an optional embodiment, each pan-tilt camera in the pan-tilt camera group is scheduled to capture images of the target monitoring point to obtain a set of images, including: identifying the pan-tilt camera group that meets preset conditions; scheduling each pan-tilt camera in the pan-tilt camera group to simultaneously capture images of the target monitoring point and upload the captured images; and acquiring the images uploaded by each pan-tilt camera in the pan-tilt camera group to obtain a set of images.

[0033] In the above embodiments, the target image uploaded by the target PTZ camera is first received. Based on this, if the target monitoring point is determined to be initially abnormal, the PTZ camera group that meets the preset conditions can accurately select a suitable camera to assist in shooting. These cameras are scheduled to synchronously shoot and upload images of the target monitoring point, enabling real-time acquisition of multi-view data. This avoids delays in anomaly confirmation time due to manual intervention. Finally, the images uploaded by each camera are acquired to form a set of images, providing a multi-angle and more comprehensive data foundation for subsequent image recognition and accurate judgment of whether the target monitoring point has anomalies, thus improving the accuracy and efficiency of anomaly detection. Through automated scheduling and synchronous shooting, the multi-view image acquisition is fully automated, reducing manual intervention and improving the system's operating efficiency and intelligence level. The synchronously captured multi-view images can more comprehensively reflect the state of the target monitoring point, providing richer information for subsequent image recognition and anomaly judgment, thereby improving the accuracy of anomaly detection.

[0034] A PTZ camera group is formed by selecting PTZ cameras that meet preset conditions. These preset conditions can be determined based on the geographical location of the monitoring point, equipment layout, and anomaly type. For example, the preset conditions can be that the target monitoring point is within the field of view coverage, or that the distance to the target monitoring point is less than a preset distance threshold. In other words, the most suitable PTZ camera group for multi-angle shooting of the target monitoring point is selected. The above preset conditions can be flexibly set according to the actual application scenario to ensure that the selected PTZ camera group can cover the target monitoring point from different angles. Each PTZ camera in the PTZ camera group is scheduled to shoot the target monitoring point simultaneously and upload the captured images to the server. Synchronous shooting ensures the time consistency of multi-view images and avoids information mismatch caused by differences in shooting time. In this way, the server can obtain a set of images, which includes a set of images of the target monitoring point from different perspectives. Not all available PTZ cameras are suitable for capturing images of a specific target monitoring point. By selecting cameras that meet preset conditions, resources can be utilized more effectively, avoiding unnecessary energy consumption and data redundancy. Synchronous shooting ensures that images are acquired at the same time, reducing data inconsistencies caused by environmental changes (such as changes in lighting) and helping to improve the accuracy of subsequent analysis. Synchronous operation speeds up the image acquisition process, shortening the time required from initial anomaly detection to comprehensive assessment, and improving the system's response speed. In this embodiment, by dynamically selecting PTZ camera groups that meet preset conditions, the system can flexibly adjust the combination of PTZ cameras participating in shooting according to different monitoring needs and scenarios, improving the system's adaptability and flexibility.

[0035] In an optional embodiment, determining a PTZ camera group that meets preset conditions includes one of the following: forming a PTZ camera group from each PTZ camera in the PTZ camera group whose distance from the target PTZ camera is less than a preset distance threshold, wherein the PTZ camera group includes the target PTZ camera; determining candidate PTZ cameras from the PTZ camera group according to a preset PTZ configuration table, and forming a PTZ camera group from the candidate PTZ cameras, wherein the PTZ configuration table records the monitoring points covered by each PTZ camera in the PTZ camera group, and the monitoring range of each PTZ camera in the candidate PTZ cameras covers the target monitoring point.

[0036] In the above embodiments, a pan-tilt camera group that meets preset conditions can be determined from the pan-tilt camera group. The pan-tilt camera group can be flexibly assembled according to the distance of the target pan-tilt camera or the pan-tilt configuration table, so that the pan-tilt camera group can shoot the target monitoring point from different angles, avoid the limitation of single view, reduce false detection or missed detection caused by occlusion, changes in lighting or the target size being too small, improve the accuracy and efficiency of anomaly detection, and at the same time realize the acquisition of multi-view data in real time to promptly confirm anomalies.

[0037] This embodiment provides two different methods for determining the PTZ camera group. The first is a distance-based method, which selects all PTZ cameras whose distance from the target PTZ camera is less than a certain preset threshold to form the PTZ camera group. This method utilizes the spatial distribution relationship of the PTZ cameras and filters out PTZ cameras that can effectively cover the target monitoring point by distance, so as to ensure that the same target monitoring point can be photographed from different angles. For example, all PTZ cameras within a radius of 50 meters (or other distances) are suitable for scenarios with compact physical space. The second is a configuration table-based method, which selects candidate PTZ cameras whose monitoring range covers the target monitoring point according to a pre-set PTZ configuration table to form the PTZ camera group. That is, the selected cameras must be able to effectively cover the target monitoring point. The PTZ configuration table records the mapping relationship between each camera and the monitoring point. This method is suitable for scenarios with complex logical associations, such as multi-view coverage of distributed devices. Both distance-based and configuration table-based selection methods allow the system to flexibly adjust according to actual conditions, improving its ability to cope with different scenarios. This embodiment uses one of the above two methods to determine the PTZ camera group, which can ensure that data of the target monitoring point is obtained from multiple advantageous positions, reducing detection errors caused by viewing angle limitations. By reasonably selecting the PTZ camera group, unnecessary PTZ camera participation in shooting is avoided, resource utilization is optimized, and system operating costs are reduced.

[0038] In an optional embodiment, each gimbal camera in the gimbal camera group is scheduled to capture images of the target monitoring point to obtain a set of images. This includes: determining the target information of the target monitoring point based on the target images, wherein the target information includes target spatial coordinate information and target size information; sending the target information to each gimbal camera in the gimbal camera group so that each gimbal camera in the gimbal camera group can adjust its attitude and focal length based on the target information, and synchronously capture images of the target monitoring point and upload the captured images; and receiving the images uploaded by each gimbal camera in the gimbal camera group to obtain a set of images.

[0039] In the above embodiments, after receiving the target image uploaded by the target PTZ camera and determining that the target monitoring point is initially abnormal, the target spatial coordinates and target size information of the target monitoring point are determined based on the target image. This information is then sent to each PTZ camera in the PTZ camera group, allowing each camera to adjust its posture and focal length accordingly, synchronously capture images of the target monitoring point, and upload them. This improves the accuracy and relevance of the captured images and avoids blurry images or failure to cover the target monitoring point due to unsuitable camera posture and focal length. Finally, the images uploaded by each PTZ camera are received to obtain a set of images for subsequent identification, which helps to comprehensively and accurately identify whether there is an anomaly at the target monitoring point from multiple angles, solving the problem of low efficiency in anomaly detection methods in related technologies.

[0040] First, the target image uploaded by the target PTZ camera is analyzed to extract the target spatial coordinates and size information of the target monitoring point. The target spatial coordinates are used to determine the specific location of the target monitoring point in space, while the target size information reflects the size and specifications of the target monitoring point. This information is the basis for subsequent operations. The acquired target information is sent to each PTZ camera in the PTZ camera group. Based on this information, each PTZ camera precisely adjusts its own posture (such as angle and direction) and focal length. By adjusting the posture, it is ensured that the camera can be aimed at the target monitoring point from a suitable angle. By adjusting the focal length, it is ensured that the captured image clearly and completely presents the details of the target monitoring point. Subsequently, each PTZ camera synchronously captures images of the target monitoring point and uploads the captured images. The server receives the images uploaded by each PTZ camera in the PTZ camera group, obtaining a set of images for subsequent image recognition and anomaly detection. This embodiment analyzes the target image and determines the target information, enabling each pan-tilt camera in the pan-tilt camera group to precisely adjust its posture and focal length based on the target's spatial coordinates and size information. This ensures clear and complete images, improving image quality. Related technologies require manual adjustment of camera angles and focal lengths one by one (e.g., remote control of PTZ parameters), which is time-consuming and error-prone. This embodiment achieves millisecond-level adaptive parameter adjustment by automatically sending target information. Fixed-focal-length cameras struggle to capture small, distant targets (e.g., insulator defects). This embodiment improves resolution by dynamically adjusting the focal length and automatically optimizing the magnification based on the target size information. Through this embodiment, each pan-tilt camera in the pan-tilt camera group adjusts its posture and focal length based on the target information before capturing images, resulting in clearer images with more suitable perspectives. This allows for a more complete presentation of the details of the target monitoring point, providing high-quality data for subsequent image recognition and improving the accuracy of anomaly detection. Furthermore, the ability to flexibly adjust the pan-tilt camera group's shooting parameters based on the spatial coordinates and size information of different target monitoring points makes it suitable for various complex monitoring scenarios and different types of target monitoring points, enhancing the adaptability and versatility of the entire anomaly detection system.

[0041] In an optional embodiment, determining the target information of the target monitoring point based on the target image includes: determining the pixel coordinates and abnormal pixel size of the target monitoring point based on the target image; obtaining target spatial coordinate information based on the target pose data, target field of view data and pixel coordinates of the target pan-tilt camera; and obtaining target size information based on the target pose data, target field of view data and abnormal pixel size of the target pan-tilt camera.

[0042] In the above embodiments, the pixel coordinates and abnormal pixel size of the target monitoring point can be determined using the target image. Then, by combining the target pose data and target field of view data of the target PTZ camera, the target spatial coordinate information and target size information of the target monitoring point can be accurately obtained. This provides an accurate basis for each PTZ camera in the subsequent PTZ camera group to adjust its posture and focal length based on the target information and to synchronously shoot the target monitoring point. This avoids the problem of poor shooting effect due to inaccurate target information, improves the accuracy of image recognition, and thus more efficiently determines whether there is an anomaly at the target monitoring point.

[0043] The target image captured by the target pan-tilt camera is analyzed and processed. Image recognition and analysis algorithms are used to determine the pixel coordinates of the target monitoring point in the image, i.e., its position on the image plane. Simultaneously, the abnormal pixel size occupied by the target monitoring point in the image is calculated; this pixel-level data forms the basis for subsequent calculations. Then, combining the target pose data of the target pan-tilt camera (including the camera's position and attitude information in space), the target field of view data (the angle of the spatial range that the camera can observe), and the determined pixel coordinates, the pixel coordinates on the image plane are transformed to the actual spatial coordinate system using geometric optics and spatial coordinate transformation principles. This yields the target spatial coordinate information of the target monitoring point, determining its position in the true image plane. The specific location in real space; based on the target pose data, target field of view data, and abnormal pixel size of the target gimbal camera, a correspondence model between image pixels and actual size is established to convert the pixel size of the target monitoring point in the image into the actual physical size, thereby obtaining the target size information and clarifying the actual size and specifications of the target monitoring point; in practical applications, when calculating the target spatial coordinates and target size information, depth information, such as the z-value of the target image in the camera coordinate system, is also required. That is, the world coordinate system position of the target monitoring point is calculated by using the pixel coordinates of the target monitoring point, depth information, target gimbal camera pose data, and field of view data; the physical size information of the target can be calculated based on the size of the abnormal pixel area and the camera magnification. This embodiment combines the pixel information of the target image with the pose and field of view data of the gimbal camera to accurately calculate the spatial coordinates and size information of the target, providing more accurate input for subsequent anomaly detection. It realizes automatic extraction and calculation of target information, reduces manual intervention, improves the system's operating efficiency and intelligence level, and the accurate target information can guide the gimbal camera to adjust its posture and focal length, ensuring that it captures multi-view images that are more in line with the actual situation of the target monitoring point, improving the effectiveness and quality of the images, enhancing the reliability of anomaly detection, and improving the accuracy of anomaly detection.

[0044] In an optional embodiment, determining whether a target monitoring point is abnormal based on a set of identification results includes: determining that the target monitoring point is abnormal when the proportion of abnormal identification results included in a set of identification results is greater than or equal to a preset proportion threshold.

[0045] In the above embodiments, image recognition is performed on a set of images to obtain a set of recognition results. The presence of anomalies at the target monitoring point is determined by comparing the proportion of abnormal recognition results in the set of recognition results with a preset proportion threshold. This can overcome the limitations of a single viewpoint, avoid false detections or missed detections caused by occlusion, changes in lighting, or small target size, and improve the accuracy and efficiency of anomaly detection.

[0046] After performing image recognition on a set of images captured by the pan-tilt camera group, a set of recognition results is obtained. The number of recognition results judged as abnormal is counted, and their proportion in the set is calculated. When this proportion is greater than or equal to a preset threshold, based on the majority judgment principle, the target monitoring point is considered abnormal; otherwise, it is considered normal. This method quantitatively analyzes the recognition results of multi-view images to determine the state of the target monitoring point using a unified standard. This embodiment, by integrating the recognition results of multi-view images and using a preset threshold as the judgment basis, can more comprehensively and objectively assess the state of the target monitoring point, reducing errors caused by single-view misjudgment or subjective judgment, and improving the accuracy of anomaly judgment. The preset threshold can be flexibly adjusted according to different application scenarios to adapt to various specific needs and conditions, improving the system's flexibility and adaptability. For example, the preset threshold can be 50% (or 60%, or others).

[0047] In an optional embodiment, determining whether a target monitoring point is abnormal based on a set of identification results includes: calculating a comprehensive score by weighted summation of the confidence values ​​corresponding to each identification result in the set of identification results; and determining that the target monitoring point is abnormal when the comprehensive score is greater than a preset score threshold.

[0048] In the above embodiments, a comprehensive score is obtained by weighted summation of the confidence values ​​corresponding to each identification result in a set of identification results. This can more scientifically and accurately assess the anomalies of the target monitoring point. When the comprehensive score is greater than the preset score threshold, it is determined that there is an anomaly in the target monitoring point. This can improve the accuracy and reliability of anomaly detection, avoid false detection or missed detection caused by single-view detection, and improve the efficiency of anomaly detection.

[0049] After performing image recognition on a set of images captured by a gimbal camera array, each recognition result corresponds to a confidence value, which reflects the reliability of the recognition result. Different gimbal cameras will have different confidence values ​​for their image recognition results due to factors such as shooting angle and image quality. This embodiment assigns different weights to the confidence values ​​of each recognition result, considering the differences in the importance of each result, and then sums these weighted confidence values ​​to obtain a comprehensive score. When the comprehensive score is greater than a preset scoring threshold, the target monitoring point is determined to be abnormal; otherwise, it is considered that the target monitoring point is not abnormal. This method quantifies and integrates the reliability of multi-view image recognition results, forming a unified judgment criterion. Assuming a PTZ camera group comprises 3 (or other) PTZ cameras with confidence values ​​C1, C2, and C3, and corresponding weighting coefficients K1, K2, and K3, in practical applications, the weighting coefficients of each PTZ camera can be set based on the distance between each PTZ camera and the target monitoring point. Alternatively, the weighting coefficients can be set based on the proportion of the target monitoring point in the images captured by each PTZ camera. The larger the proportion of the target monitoring point in the image, the larger the corresponding weighting coefficient. For example, K1 = 0.45, K2 = 0.35, and K3 = 0.2. A comprehensive score is obtained by weighted summation of the confidence values. For example, the preset score threshold is 0.5 (or other values). When the comprehensive score exceeds this preset score threshold, the target monitoring point is considered to be abnormal. Related technologies rely on only some simple rules to judge anomalies, which cannot fully reflect the comprehensive situation of multi-view image recognition results. However, this embodiment fully considers the reliability differences of image recognition results from different perspectives by weighted summation of the confidence values ​​of the recognition results, making the judgment results more in line with the actual situation, reducing the possibility of misjudgment, greatly improving the accuracy of anomaly judgment, and overcoming the limitations of a single judgment standard.

[0050] In an optional embodiment, after determining that there is an anomaly at the target monitoring point, the method further includes issuing an alarm message.

[0051] In the above embodiments, once it is determined that there is indeed an anomaly at the target monitoring point, the system will trigger an alarm mechanism and send alarm information to the designated object (such as maintenance personnel, central control system, etc.). These alarm information can be conveyed through various channels, such as SMS, email, in-application notifications, or directly displaying warnings in the monitoring center.

[0052] Upon detecting an anomaly at a target monitoring point, the system triggers an alarm mechanism, sending alarm messages to relevant personnel or systems. Alarm messages can take various forms, such as sound, light, SMS, or pop-up notifications. Their purpose is to promptly notify relevant personnel of the anomaly at the target monitoring point, enabling staff to respond quickly and take appropriate measures. Automatically issuing alarm messages upon detecting anomalies further enhances the system's automation level, reduces manual intervention, and improves operational efficiency. Automatic alarm issuance ensures that anomalies are detected promptly, allowing relevant personnel or systems to quickly take measures to minimize the impact of the anomaly on system operation.

[0053] Obviously, the embodiments described above are only some embodiments of this application, and not all embodiments. The present application will be specifically described below with reference to specific embodiments.

[0054] This application provides a method and system for multi-device collaborative proactive monitoring and decision-making. Figure 2 This is an architecture diagram of a multi-device collaborative active monitoring and decision-making system provided in an embodiment of this application. The system will be described below.

[0055] 1. Collaborative architecture (substation PTZ group with mutual master-slave relationships) Target gimbal (e.g., gimbal A that is found to be abnormal): When an abnormality is triggered, the target's three-dimensional spatial coordinates (converted from pixel coordinates and depth information) and target size (based on the abnormal pixel size and the field of view (FOV) of the camera at the current magnification) are generated. Other gimbals (such as gimbals B / C / D...): Receive the target's three-dimensional spatial coordinates and target size information from the server, and automatically adjust the gimbal's attitude and focal length through spatial coordinate system transformation algorithms (such as homogeneous matrix transformation); Multi-view synchronous acquisition module: Each gimbal in other gimbals captures the target based on a unified timestamp and uploads the data to the edge computing node (or server) via network protocols (such as RTSP / ONVIF); Distributed voting decision module: Employs a majority-rule strategy (e.g., an alarm is triggered if ≥50% of devices identify anomalies) and combines a confidence-weighted algorithm to optimize the voting results.

[0056] 2. Working principle Anomaly Triggering Phase: When PTZ A detects a suspected anomaly (such as equipment overheating or component detachment), it calculates the target's three-dimensional spatial coordinates and target size using the camera pose and field of view (FOV). Collaborative scheduling phase: The system calls the spatial location database of neighboring gimbals, plans the shortest turning path, and sends the target coordinates and size to the subordinate gimbals; Multi-view data fusion: Each PTZ camera simultaneously captures images and performs local recognition, then uploads the results to a central decision-making node (or server) for voting; Alarm decision-making stage: If the voting results meet the preset threshold, a tiered alarm is triggered (e.g., a level 1 alarm requires manual review, while a level 2 alarm is directly pushed to the operation and maintenance platform).

[0057] The aforementioned edge computing nodes and central decision-making nodes can reside on the same server. Figure 2 The image recognition algorithm, spatial coordinate transformation algorithm, and gimbal scheduling control can all be executed by a single server.

[0058] Figure 3 This is a schematic diagram of the spatial coordinate transformation algorithm provided in an embodiment of this application, wherein, Ow-XwYwZw: World coordinate system, unit: meters; Oc-XcYcZc: Camera coordinate system, with the optical center at the origin, in meters (m). o-xy: Image coordinate system, with the optical center at the midpoint of the image, in mm; uv: pixel coordinate system, unit is pixel; P: A point in the world coordinate system; p: Projection in the image coordinate system; f: Camera focal length, equal to the distance between o and Oc; In addition, shifting the world coordinate system up the Z-axis by a height of h and rotating it around the X-axis by θ can transform it to the camera coordinate system.

[0059] The coordinate transformation is explained below: (1) Camera matrix calculation 1) Intrinsic parameter matrix K: Where W and H are the image width and height, (c x ,c y ) = (W / 2, H / 2), (c x ,c y ) represents the center point of the image, f x f y θ represents the horizontal focal length and vertical focal length of the camera, respectively. h θ v These represent the horizontal and vertical viewing angles, respectively. 2) Extrinsic parameter matrix [R|t]: The rotation matrix R is determined by three-degree-of-freedom rotation quantities (such as Euler angles or rotation vectors); The translation vector t = -R·C, where C is the three-dimensional coordinate of the camera in the world coordinate system; 3) Projection matrix P: P = K[R|t]; (2) Calculation of three-dimensional coordinates and radius 1) Back projection of the center point: Find the center pixel coordinates (u, v) of the rectangle, with the corresponding depth value z. depth (z value in camera coordinate system).

[0060] Three-dimensional points projected back onto the camera coordinate system: 2) World coordinate system transformation: Radius calculation: Calculate the pixel-to-space ratio: horizontal ratio and vertical ratio; Horizontal scale: z depth / f x Vertical scale: z depth / f y ; Take the maximum value of the rectangle's width and height (in pixels), max(w,h), where w and h refer to the width and height of the detected defect or anomaly point. Calculate the spatial radius based on the proportions. The spatial radius corresponds to the aforementioned target size information. 3. Detailed Implementation Methods (1) Hardware deployment: 1) Multiple cameras supporting PTZ (pan-tilt-zoom) are installed inside the substation; 2) Deploy target detection models (such as YOLO) and spatial coordinate transformation services on edge servers.

[0062] (2) Workflow: Figure 4 This is a schematic diagram of a multi-perspective decision-making process provided in an embodiment of this application, including the following steps: Step 1: During routine inspection, PTZ A detects a crack on the surface of an insulator (confidence level 65%), triggering an abnormal event; Step 2: The system calculates the three-dimensional coordinates and size of the insulator based on the pose parameters (pitch angle, azimuth angle, focal length) of PTZ A, and broadcasts it to PTZ B / C / D; Step 3: The gimbals B / C / D adjust their own posture through inverse kinematics calculations to complete focusing and shooting within 3 seconds; Step 4: Each PTZ unit performs local identification (B / C are identified as cracks with confidence levels of 72% and 68% respectively; D fails to be identified due to angle issues). The decision module executes the decision logic and finally determines whether there is an anomaly. For example, if the decision module determines that the number of valid votes is 3 / 4, an alarm is triggered.

[0063] Spatial coordinate mapping: The camera intrinsic and extrinsic parameter matrix is ​​established using the Zhang Zhengyou calibration method (or simplified to based on the camera field of view without considering distortion), and pixel coordinates are converted to world coordinates by combining gimbal PTZ data and depth estimation algorithm.

[0064] Voting weight design: The confidence weight is dynamically adjusted based on the distance between the gimbal and the target and the lighting conditions (e.g., weight of close-range devices +20%).

[0065] The embodiments of this application solve the following technical problems in the related art: solve the problem of misjudgment caused by environmental interference in single-view detection; eliminate the problem of collaborative response delay when multiple devices run independently; and improve the confidence and robustness of anomaly identification through multi-view data fusion.

[0066] Compared with the prior art, the embodiments of this application have at least the following technical effects: Improved accuracy: Multi-view cross-validation reduces the false alarm rate by ≥30% (simulation experimental data), which is especially suitable for complex occlusion scenarios; Efficiency optimization: Collaborative response time ≤ 2 seconds (actual value), an 80% improvement in efficiency compared to manual scheduling; Resource conservation: Dynamic scheduling reduces redundant shooting attempts and extends the lifespan of the gimbal mechanism; Highly scalable: Supports elastic scaling of PTZ group size and is compatible with different brands of equipment (through standardized protocol adaptation layer).

[0067] This application also provides an anomaly detection device, such as... Figure 5 As shown, Figure 5 This is a structural block diagram of an anomaly detection device provided in an embodiment of this application. The device includes: The receiving module 51 is used to receive the target image uploaded by the target pan-tilt camera, wherein the target image is obtained by the target pan-tilt camera capturing the target monitoring point; The scheduling module 52 is used to schedule each pan-tilt camera in the pan-tilt camera group to take pictures of the target monitoring point when the target monitoring point is determined to be in a preliminary abnormality based on the target image, and to obtain a set of images. The set of images includes the images obtained by each pan-tilt camera in the pan-tilt camera group taking pictures of the target monitoring point. Each pan-tilt camera in the pan-tilt camera group is located at a different angle position of the target monitoring point. The recognition module 53 is used to perform image recognition on each image in a set of images to obtain a set of recognition results, wherein each recognition result in the set of recognition results corresponds to one image in the set of images; The determination module 54 is used to determine whether there is an anomaly at the target monitoring point based on a set of identification results.

[0068] In an optional embodiment, the scheduling module 52 includes: a first determining unit, used to determine a PTZ camera group that meets preset conditions; a scheduling unit, used to schedule each PTZ camera in the PTZ camera group to synchronously capture images of the target monitoring point and upload the captured images; and an acquiring unit, used to acquire the images uploaded by each PTZ camera in the PTZ camera group to obtain a set of images.

[0069] In an optional embodiment, the first determining unit includes one of the following: a first forming subunit, configured to form a gimbal camera group by grouping each gimbal camera in the gimbal camera group whose distance from the target gimbal camera is less than a preset distance threshold, wherein the gimbal camera group includes the target gimbal camera; and a second forming subunit, configured to determine candidate gimbal cameras from the gimbal camera group according to a preset gimbal configuration table, and form the candidate gimbal cameras into a gimbal camera group, wherein the gimbal configuration table records the monitoring points covered by each gimbal camera in the gimbal camera group, and the monitoring range of each gimbal camera in the candidate gimbal camera group covers the target monitoring point.

[0070] In an optional embodiment, the scheduling module 52 includes: a second determining unit, configured to determine target information of the target monitoring point based on the target image, wherein the target information includes target spatial coordinate information and target size information; a sending unit, configured to send the target information to each gimbal camera in the gimbal camera group, so that each gimbal camera in the gimbal camera group can adjust its attitude and focal length based on the target information, and synchronously capture images of the target monitoring point and upload the captured images; and a receiving unit, configured to receive the images uploaded by each gimbal camera in the gimbal camera group to obtain a set of images.

[0071] In an optional embodiment, the second determining unit includes: a determining subunit, configured to determine the pixel coordinates and abnormal pixel size of the target monitoring point based on the target image; and an obtaining subunit, configured to obtain target spatial coordinate information based on the target pose data, target field of view data and pixel coordinates of the target PTZ camera, and to obtain target size information based on the target pose data, target field of view data and abnormal pixel size of the target PTZ camera.

[0072] In an optional embodiment, the determination module 54 includes a third determination unit, used to determine that there is an anomaly at the target monitoring point when the proportion of abnormal identification results contained in a set of identification results is greater than or equal to a preset proportion threshold.

[0073] In an optional embodiment, the determination module 54 includes: a calculation unit, used to perform a weighted summation of the confidence values ​​corresponding to each identification result in a set of identification results to obtain a comprehensive score; and a fourth determination unit, used to determine that there is an anomaly at the target monitoring point when the comprehensive score is greater than a preset score threshold.

[0074] In an optional embodiment, the above-mentioned device further includes an alarm module, which is used to issue an alarm message after determining that there is an anomaly at the target monitoring point.

[0075] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided above belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0076] This application also provides a computer-readable storage medium storing instructions that, when executed, perform the steps of any of the methods described above.

[0077] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0078] This application also discloses an electronic device. For example... Figure 6 As shown, Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 600 may include: at least one processor 601, at least one communication bus 602, a user interface 603, at least one network interface 604, and a memory 605.

[0079] The communication bus 602 is used to enable communication between these components.

[0080] The user interface 603 may include a display screen and a camera. Optionally, the user interface 603 may also include a standard wired interface and a wireless interface.

[0081] The network interface 604 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0082] The processor 601 may include one or more processing cores. The processor 601 connects to various parts of the electronic device (such as a server) using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 605, and by calling data stored in memory 605. Optionally, the processor 601 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 601 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 601.

[0083] The memory 605 may include random access memory (RAM) or read-only memory. Optionally, the memory 605 may include a non-transitory computer-readable storage medium. The memory 605 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 605 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 605 may also be at least one storage device located remotely from the aforementioned processor 601. (Refer to...) Figure 6 The memory 605, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for an anomaly detection method.

[0084] exist Figure 6In the illustrated electronic device 600, the user interface 603 is mainly used to provide an input interface for the user and to acquire user input data; while the processor 601 can be used to call an application program of an anomaly detection method stored in the memory 605. When executed by one or more processors 601, the electronic device 600 performs one or more of the methods described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0085] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0086] In the various embodiments provided in this application, it should be understood that the disclosed apparatus or system can be implemented in other ways. For example, the apparatus or system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] 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.

[0089] 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 device (CMD). 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 memory 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 of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0090] The above description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the disclosure herein.

[0091] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art that are not described in this disclosure.

Claims

1. An anomaly detection method, characterized in that, include: Receive target images uploaded by the target PTZ camera, wherein the target images are obtained by the target PTZ camera capturing images of the target monitoring point; If the target monitoring point is determined to be initially abnormal based on the target image, each pan-tilt camera in the pan-tilt camera group is scheduled to take pictures of the target monitoring point to obtain a set of images. The set of images includes the images obtained by each pan-tilt camera in the pan-tilt camera group taking pictures of the target monitoring point. Each pan-tilt camera in the pan-tilt camera group is located at a different angle position of the target monitoring point. Image recognition is performed on each image in the set of images to obtain a set of recognition results, wherein each recognition result in the set of recognition results corresponds to one image in the set of images; Based on the set of identification results, determine whether the target monitoring point has any anomalies; The process involves scheduling each gimbal camera in the gimbal camera group to capture images of the target monitoring point, resulting in a set of images. This includes: determining target information for the target monitoring point based on the target images, where the target information includes target spatial coordinates and target size; sending the target information to each gimbal camera in the gimbal camera group so that each camera can adjust its attitude and focus based on the target information, and simultaneously capture images of the target monitoring point and upload the captured images; and receiving the images uploaded by each gimbal camera in the gimbal camera group to obtain the set of images. Determining target information of the target monitoring point based on the target image includes: determining the pixel coordinates and abnormal pixel size of the target monitoring point based on the target image; obtaining the target spatial coordinate information based on the target pose data, target field of view data and the pixel coordinates of the target PTZ camera; and obtaining the target size information based on the target pose data, target field of view data and the abnormal pixel size of the target PTZ camera. Determining whether the target monitoring point is abnormal based on the set of identification results includes: calculating a comprehensive score by weighted summation of the confidence values ​​corresponding to each identification result in the set of identification results; wherein, in the weighted summation calculation, the weight coefficient of each identification result is determined based on at least one of the following: the distance between the PTZ camera corresponding to each identification result and the target monitoring point; the proportion of the target monitoring point in the image captured by the PTZ camera corresponding to each identification result, the larger the proportion of the target monitoring point, the larger the corresponding weight coefficient; when the comprehensive score is greater than a preset score threshold, it is determined that the target monitoring point is abnormal.

2. The method according to claim 1, characterized in that, Each pan-tilt camera in the pan-tilt camera group is scheduled to capture images of the target monitoring point, resulting in a set of images, including: Identify the gimbal camera group that meets the preset conditions; The system schedules each pan-tilt camera in the pan-tilt camera group to simultaneously capture images of the target monitoring point and upload the captured images. The images uploaded by each gimbal camera in the gimbal camera group are obtained to obtain the group of images.

3. The method according to claim 2, characterized in that, The gimbal camera group that meets the preset conditions includes one of the following: The gimbal camera group consists of each gimbal camera in the gimbal camera group whose distance from the target gimbal camera is less than a preset distance threshold, wherein the gimbal camera group includes the target gimbal camera. Candidate PTZ cameras are determined from the PTZ camera group according to the preset PTZ configuration table, and the candidate PTZ cameras are combined into the PTZ camera group. The PTZ configuration table records the monitoring points covered by each PTZ camera in the PTZ camera group, and the monitoring range of each PTZ camera in the candidate PTZ cameras covers the target monitoring point.

4. The method according to claim 1, characterized in that, Determining whether the target monitoring point has any anomalies based on the set of identification results includes: If the proportion of abnormal identification results in the set of identification results is greater than or equal to a preset proportion threshold, it is determined that the target monitoring point is abnormal.

5. An anomaly detection device, characterized in that, include: The receiving module is used to receive target images uploaded by the target PTZ camera, wherein the target images are obtained by the target PTZ camera capturing images of the target monitoring point; The scheduling module is used to schedule each pan-tilt camera in the pan-tilt camera group to take pictures of the target monitoring point when the target monitoring point is determined to be in a preliminary abnormality based on the target image, thereby obtaining a set of images. The set of images includes the images obtained by each pan-tilt camera in the pan-tilt camera group taking pictures of the target monitoring point, and each pan-tilt camera in the pan-tilt camera group is located at a different angle position of the target monitoring point. The recognition module is used to perform image recognition on each image in the set of images to obtain a set of recognition results, wherein each recognition result in the set of recognition results corresponds to one image in the set of images; The determination module is used to determine whether there is an anomaly at the target monitoring point based on the set of identification results; The scheduling module includes: a second determining unit, used to determine target information of the target monitoring point based on the target image, wherein the target information includes target spatial coordinate information and target size information; a sending unit, used to send the target information to each gimbal camera in the gimbal camera group, so that each gimbal camera in the gimbal camera group can adjust its attitude and focal length based on the target information, and synchronously capture images of the target monitoring point and upload the captured images; and a receiving unit, used to receive the images uploaded by each gimbal camera in the gimbal camera group to obtain a set of images. The second determining unit includes: a determining subunit, used to determine the pixel coordinates and abnormal pixel size of the target monitoring point based on the target image; and an obtaining subunit, used to obtain target spatial coordinate information based on the target pose data, target field of view data and pixel coordinates of the target PTZ camera, and to obtain target size information based on the target pose data, target field of view data and abnormal pixel size of the target PTZ camera. The determining module includes: a calculation unit, used to perform a weighted summation calculation on the confidence values ​​corresponding to each identification result in a set of identification results to obtain a comprehensive score; wherein, in the weighted summation calculation, the weight coefficient of each identification result is determined according to at least one of the following: the distance between the PTZ camera corresponding to each identification result and the target monitoring point; the proportion of the target monitoring point in the image captured by the PTZ camera corresponding to each identification result, the larger the proportion of the target monitoring point, the larger the corresponding weight coefficient; and a fourth determining unit, used to determine that the target monitoring point is abnormal when the comprehensive score is greater than a preset score threshold.

6. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 4.

Images