Super-computing cluster intelligent video analysis method based on edge computing and related device

Abstract

The embodiment of the invention discloses a super-computing cluster intelligent video analysis method based on edge computing and a related device, and the method comprises the steps: obtaining target video analysis task parameters which comprise a target task type and a target task object obtained by video monitoring equipment; determining target task complexity according to the target task type and the target task object; when the target task complexity is greater than a preset threshold value, determining n supercomputing modules according to the target task complexity; splitting the target task type and the target task object to obtain task configuration parameters of n + 1 subtasks, and determining an execution sequence of the n + 1 subtasks; the first subtask objects corresponding to the n + 1 subtasks form a target task object; and performing video analysis on the target task object according to the execution sequence and the task configuration parameters of the n + 1 sub-tasks to obtain a target video analysis result. According to the embodiment of the invention, the video analysis efficiency of the security and protection monitoring system can be improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, the field of Internet of Things technology, or the field of artificial intelligence technology, and specifically to a supercomputing cluster intelligent video analysis method and related devices based on edge computing. Background Art

[0002] The security monitoring system on the market is usually composed of cameras, transmission networks (such as Ethernet or wireless networks), centralized servers and background management systems. The operation mode of the security monitoring system is mainly to collect video streams through cameras, and then transmit the video data to a remote data center or cloud platform for processing and analysis.

[0003] At present, although the existing security monitoring systems have met the basic security needs to a certain extent, they also have significant limitations, resulting in low video analysis efficiency. Therefore, the problem of how to improve the video analysis efficiency of security monitoring systems needs to be solved urgently. Summary of the invention

[0004] The embodiments of the present application provide a supercomputing cluster intelligent video analysis method and related devices based on edge computing, which can improve the video analysis efficiency of the security monitoring system.

[0005] In a first aspect, an embodiment of the present application provides a supercomputing cluster intelligent video analysis method based on edge computing, which is applied to an edge computing device. The supercomputing cluster intelligent video analysis system includes the edge computing device, m supercomputing modules and a video monitoring device, where m is an integer greater than 1; the method includes:

[0006] Acquire target video analysis task parameters, wherein the target video analysis task parameters include a target task type and a target task object acquired by the video surveillance device;

[0007] Determine the complexity of the target task according to the target task type and the target task object;

[0008] When the complexity of the target task is greater than a preset threshold, n supercomputing modules are determined according to the complexity of the target task, where n is a positive integer less than or equal to m;

[0009] The target task type and the target task object are split to obtain task configuration parameters of n+1 subtasks, and the execution order of the n+1 subtasks is determined. The task configuration parameters of each subtask are: first task content, first subtask object, and first device identifier; the first subtask object corresponding to the n+1 subtask constitutes the target task object;

[0010] Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n+1 subtasks to obtain a target video analysis result.

[0011] In a second aspect, an embodiment of the present application provides a supercomputer cluster intelligent video analysis device based on edge computing. This device is applied to an edge computing device. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1; the device includes: an acquisition unit, a determination unit, a splitting unit, and an analysis unit, where,

[0012] The acquisition unit is configured to acquire target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object acquired by the video monitoring device;

[0013] The determination unit is configured to determine the target task complexity according to the target task type and the target task object; when the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m;

[0014] The splitting unit is configured to split the target task type and the target task object to obtain task configuration parameters of n+1 subtasks, and determine the execution order of the n+1 subtasks. The task configuration parameters of each subtask are: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n+1 subtasks constitute the target task object;

[0015] The analysis unit is configured to perform video analysis on the target task object according to the execution order and the task configuration parameters of the n+1 subtasks to obtain a target video analysis result.

[0016] In a third aspect, an embodiment of the present application provides an edge computing device, including a processor, a memory, a communication interface, and one or more programs. Among them, the above one or more programs are stored in the above memory and are configured to be executed by the above processor. The above programs include instructions for performing the steps in the first aspect of the embodiment of the present application.

[0017] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium stores a computer program for electronic data exchange. The computer program causes a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application.

[0018] Fifth aspect, an embodiment of the present application provides a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application. The computer program product may be a software installation package.

[0019] Implementing the embodiments of the present application has the following beneficial effects:

[0020] It can be seen that the supercomputer cluster intelligent video analysis method and related devices described in the embodiments of the present application. The method is applied to edge computing devices. The supercomputer cluster intelligent video analysis system includes edge computing devices, m supercomputer modules, and video surveillance devices, where m is an integer greater than 1. Obtain target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object obtained by the video surveillance device. Determine the target task complexity according to the target task type and the target task object. When the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m. Split the target task type and the target task object to obtain task configuration parameters for n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameters for each subtask are: the first task content, the first subtask object, and the first device identifier. The first subtask objects corresponding to the n + 1 subtasks constitute the target task object. Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain a target video analysis result. It is possible to perform intelligent task splitting based on the task type, task object, and performance of the supercomputer module to ensure video analysis efficiency. Thus, it is possible to ensure video processing real-time performance and contribute to improving the video analysis efficiency of the security surveillance system. Description of the Drawings

[0021] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0022] Figure 1 It is a flowchart of a supercomputer cluster intelligent video analysis method based on edge computing provided by an embodiment of the present application;

[0023] Figure 2 It is a system architecture diagram of a supercomputer cluster intelligent video analysis system provided by an embodiment of the present application;

[0024] Figure 3It is a schematic diagram of an application scenario of a supercomputer cluster intelligent video analysis method based on edge computing provided by an embodiment of the present application;

[0025] Figure 4 It is a schematic diagram of the system architecture of another supercomputer cluster intelligent video analysis system provided by an embodiment of the present application;

[0026] Figure 5 It is a schematic diagram of the structure of an edge computing device provided by an embodiment of the present application;

[0027] Figure 6 It is a block diagram of the functional units of a supercomputer cluster intelligent video analysis device based on edge computing provided by an embodiment of the present application. Detailed implementation manners

[0028] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0029] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0030] Referring to "embodiment" herein means that a specific feature, structure or characteristic described in connection with the embodiment may be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0031] In the embodiments of the present application, the electronic devices involved can be devices with communication capabilities. Such electronic devices can include various wearable devices with wireless communication functions (such as smart glasses, smart bracelets, Internet of Things devices (such as smart refrigerators, smart washing machines, smart TVs), smart watches, etc.), handheld devices, smart home devices, in-vehicle devices (such as in-vehicle cameras, dash cams, in-vehicle speakers, etc.), computing devices or other processing devices connected to a wireless modem, as well as various forms of user equipment (UE), mobile stations (MS), terminal devices, and so on.

[0032] In the embodiments of the present application, the edge computing device mainly operates based on the concept of distributed computing. Specifically, it uses devices located at the edge of the network to execute computationally intensive tasks to relieve the burden on the centralized data center and improve application performance. The edge computing device can perform real-time analysis and decision-making on locally collected data by running specific applications and services. The edge computing device can include at least one of the following: edge computing boxes, edge servers, mobile supercomputing boxes, etc., which are not limited herein.

[0033] In the embodiments of the present application, the supercomputing cluster intelligent video analysis based on edge computing can achieve fast preprocessing and feature extraction of video data by clustering and managing a large number of edge computing nodes, reducing the data transmission burden and latency. At the same time, it performs accurate image recognition, behavior analysis, event detection, etc. with the help of advanced artificial intelligence algorithms (such as deep learning models, large language models, etc.).

[0034] In the embodiments of the present application, the supercomputing module can include an edge computing device, or an electronic device, or a supercomputing cluster.

[0035] The supercomputing cluster intelligent video analysis based on edge computing in the embodiments of the present application can also be applied to large-scale video surveillance networks that require cross-regional collaboration, support unified cloud management and resource allocation, improve the overall monitoring efficiency and service quality, that is, not only can it improve the speed and accuracy of video data analysis, but also enhance the scalability and flexibility of the system to meet diverse application requirements.

[0036] The following will introduce the embodiments of the present application in detail.

[0037] Please refer to Figure 1 , Figure 1It is a schematic flowchart of a supercomputer cluster intelligent video analysis method based on edge computing provided by an embodiment of the present application. This method is applied to an edge computing device. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. The supercomputer cluster intelligent video analysis method based on edge computing includes:

[0038] 101. Obtain target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object obtained by the video monitoring device.

[0039] In an embodiment of the present application, the target video analysis task parameters may include a target task type and a target task object obtained by the video monitoring device. The target task type may represent at least one of the following: task process, task content, task purpose, task difficulty, task priority, resources required for the task, etc., which are not limited herein.

[0040] Among them, the target task type may include one or more task types.

[0041] Among them, the target task object may include one or more video monitoring images. For example, the target task object may include a video segment within a preset time period. The preset time period may be set in advance or be the system default.

[0042] In specific implementation, as Figure 2 shown, the supercomputer cluster intelligent video analysis system may include an edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. Among them, the edge computing device may include m supercomputer modules, or the m supercomputer modules may also be other electronic devices independent of the edge computing device. For example, an Internet of Things or at least one supercomputer cluster may be formed among the edge computing device, m supercomputer modules, and the video monitoring device, and they are communicatively connected to each other.

[0043] In specific implementation, as Figure 3 shown, the edge computing device may receive a video analysis instruction from an electronic device, and in response to the video analysis instruction, obtain the target video analysis task parameters.

[0044] In specific implementation, as Figure 4As shown in the figure, the supercomputer cluster intelligent video analysis system may include an edge computing device, m supercomputer modules, a video monitoring device, and a cloud server, where m is an integer greater than 1. Among them, the edge computing device may include m supercomputer modules, or the m supercomputer modules may also be other electronic devices independent of the edge computing device. For example, an Internet of Things or at least one supercomputer cluster may be formed among the edge computing device, the m supercomputer modules, the video monitoring device, and the cloud server, and the edge computing device, the m supercomputer modules, the video monitoring device, and the cloud server are communicatively connected.

[0045] 102. Determine the target task complexity according to the target task type and the target task object.

[0046] Among them, in specific implementation, the target task complexity may be determined according to the target task type and the target task object. To a certain extent, the target task complexity reflects the video processing difficulty or the size of the video processing workload.

[0047] Optionally, the determining the target task complexity according to the target task type and the target task object includes:

[0048] Determine the target task process corresponding to the target task type;

[0049] Determine the target algorithm corresponding to the target task process;

[0050] Determine the reference complexity of the target algorithm;

[0051] Determine the first memory size and the first quality evaluation value of the target task object;

[0052] Determine the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity.

[0053] In specific implementation, the mapping relationship between the preset task type and the task process may be stored in advance. Furthermore, based on this mapping relationship, the target task process corresponding to the target task type may be determined. The mapping relationship between the preset task process and the algorithm may also be stored in advance. Furthermore, based on this mapping relationship, the target algorithm corresponding to the target task process may be determined. Among them, different algorithms may implement different functions. The algorithms may include at least one of the following: image enhancement algorithm, face recognition algorithm, behavior analysis algorithm, license plate recognition algorithm, etc., which are not limited herein.

[0054] Among them, the target algorithm may include at least one algorithm, that is, it may include one or more algorithms.

[0055] Correspondingly, the mapping relationship between the preset algorithm and the complexity can also be pre-stored. Furthermore, based on this mapping relationship, the reference complexity of the target algorithm can be determined. Of course, the first memory size and the first quality evaluation value of the target task object can also be determined. To a certain extent, the first memory size and the first quality evaluation value of the target task object reflect the size of the data processing workload. Furthermore, the target task complexity can be determined according to the first memory size, the first quality evaluation value, and the reference complexity. In this way, on the one hand, the algorithm corresponding to the task type can be determined and the complexity corresponding to the algorithm can be initially determined. On the other hand, the complexity can be dynamically adjusted based on the size of the data processing workload, so that the final complexity is deeply related to the task and the size of the data processing workload of the task object. Thus, the accuracy of the selection of the supercomputing module can be guaranteed, and it is also helpful to improve the video analysis efficiency of the security monitoring system.

[0056] Among them, the first quality evaluation value reflects the quality of the target task object. For example, for any image of the target task object, features can be extracted from the image to obtain multiple features, the total number of features can be determined, the area size of the image can be determined, and based on the total number of features / area size = feature point distribution index, the mean value of the feature point distribution indexes of all images in the target task object can be determined to obtain the first mean value. According to the preset mapping relationship between the mean value and the image quality evaluation value, the first image quality evaluation value corresponding to the first mean value can be determined based on this mapping relationship.

[0057] Optionally, the above step of determining the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity can be implemented in the following manner:

[0058] Determine the first adjustment parameter corresponding to the first memory size;

[0059] Determine the second adjustment parameter corresponding to the first quality evaluation value;

[0060] Adjust the reference complexity according to the first adjustment parameter and the second adjustment parameter to obtain the target task complexity.

[0061] In specific implementation, the mapping relationship between the preset memory size and the adjustment parameter can be stored in advance. Furthermore, based on this mapping relationship, the first adjustment parameter corresponding to the first memory size can be determined. Also, the mapping relationship between the preset quality evaluation value and the adjustment parameter can be stored in advance. Furthermore, based on this mapping relationship, the second adjustment parameter corresponding to the first quality evaluation value can be determined. Finally, the reference complexity can be adjusted according to the first adjustment parameter and the second adjustment parameter to obtain the target task complexity. The target task complexity = (1 + the first adjustment parameter) * (1 + the second adjustment parameter) * the reference complexity. In this way, the complexity can be dynamically adjusted based on the size of the data processing workload, so that the final complexity is deeply related to the task and the size of the data processing workload of the task object. Thus, the accuracy of the selection of the supercomputing module can be ensured, and it also helps to improve the video analysis efficiency of the security monitoring system.

[0062] Among them, the value ranges of the first adjustment parameter and the second adjustment parameter can be set in advance or defaulted by the system. For example, the first adjustment parameter can be greater than 0, and the second adjustment parameter can be between 0 and 1.

[0063] 103. When the target task complexity is greater than the preset threshold, determine n supercomputing modules according to the target task complexity, where n is a positive integer less than or equal to m.

[0064] Among them, the preset threshold can be set in advance or defaulted by the system. n is a positive integer less than or equal to m.

[0065] In specific implementation, when the target task complexity is greater than the preset threshold, it means that the edge computing device cannot complete the target task type alone, and the step of determining n supercomputing modules according to the target task complexity is executed; on the contrary, when the target task complexity is less than or equal to the preset threshold, it means that the edge computing device can complete the target task type alone, and then n supercomputing modules can be determined according to the target task complexity.

[0066] Optionally, the above step 103 of determining n supercomputing modules according to the target task complexity can be implemented in the following manner:

[0067] Determine k combinations of supercomputing modules corresponding to the target task complexity, and each combination of supercomputing modules includes at least one supercomputing module; k is a positive integer;

[0068] Filter the k combinations of supercomputing modules according to the target task type to obtain the target combination of supercomputing modules;

[0069] Determine the n supercomputing modules according to the target combination of supercomputing modules.

[0070] Among them, the mapping relationship between the preset task complexity and the supercomputer module combinations can be stored in advance. Furthermore, based on this mapping relationship, k supercomputer module combinations corresponding to the target task complexity can be determined. Each supercomputer module combination includes at least one supercomputer module, and k is a positive integer. Of course, the k supercomputer module combinations can also be filtered according to the target task type to obtain the target supercomputer module combination, that is, the target supercomputer module combination can successfully complete the tasks corresponding to the target task type. Finally, n supercomputer modules can be determined according to the target supercomputer module combination, and the n supercomputer modules are the supercomputer modules in the target supercomputer module combination. In this way, the corresponding supercomputer module combination can be adapted based on the task complexity, and the supercomputer module combination corresponding to the target task type can be selected, thereby ensuring the task execution efficiency and helping to improve the video analysis efficiency of the security monitoring system.

[0071] Optionally, the above step of filtering the k supercomputer module combinations according to the target task type and the target task duration parameter to obtain the target supercomputer module combination can be implemented as follows:

[0072] Filter the k supercomputer module combinations according to the target task type to obtain x supercomputer module combinations, where x is a positive integer less than or equal to k;

[0073] Estimate the task duration of each supercomputer module combination in the x supercomputer module combinations for executing the tasks of the target task type and the target task object to obtain x task durations;

[0074] Select the minimum task duration among the x task durations, and obtain the supercomputer module combination corresponding to the minimum task duration to obtain the target supercomputer module combination.

[0075] In the embodiments of the present application, the k supercomputer module combinations can be filtered according to the target task type to obtain x supercomputer module combinations, where x is a positive integer less than or equal to k; each supercomputer module combination in the x supercomputer module combinations can complete the tasks corresponding to the target task type.

[0076] Next, the task duration of each supercomputer module combination in the x supercomputer module combinations for executing the tasks of the target task type and the target task object can be estimated to obtain x task durations, then the minimum task duration among the x task durations can be selected, and the supercomputer module combination corresponding to the minimum task duration can be obtained to obtain the target supercomputer module combination. In this way, the video analysis efficiency can be ensured, and thus the real-time performance of video processing can be ensured, which helps to improve the video analysis efficiency of the security monitoring system.

[0077] 104. Split the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameters of each subtask include: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object.

[0078] Among them, the first task content may include at least one of the following: task process, task purpose, resources required for the task, etc., which are not limited here.

[0079] Among them, the first subtask object may include one or more video frames. For example, the first subtask object may include a video clip.

[0080] Among them, the first device identifier can be used to identify the supercomputing module, and the first device identifier may include at least one of the following: IP address, MAC address, supercomputing module number, etc., which are not limited here.

[0081] In specific implementation, the target task type and the target task object can be split to obtain the task configuration parameters of n + 1 subtasks, and the execution order of the n + 1 subtasks can be determined. The task configuration parameters of each subtask include: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object, that is, the first subtask objects corresponding to the n + 1 subtasks can form the entire target task object. In this way, intelligent task splitting can be performed based on the task type, task object, and the performance of the supercomputing module, thereby improving the video analysis efficiency of the security monitoring system.

[0082] Optionally, in step 104 above, splitting the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks and determining the execution order of the n + 1 subtasks can be implemented as follows:

[0083] Determine the available computing resources of the edge computing device to obtain the first available computing resources;

[0084] Determine the available computing resources of the n supercomputing modules to obtain n second available computing resources;

[0085] Determine n + 1 resource scheduling values according to the first available computing resources and the n second available computing resources;

[0086] Set the edge computing device to the highest priority;

[0087] Determine the priorities of the n supercomputing modules according to the n + 1 resource scheduling values to obtain n priorities. The larger the resource scheduling value, the higher the priority;

[0088] Determine the execution order according to the n priorities and the highest priority;

[0089] According to the time sequence of the target task object and the n+1 resource scheduling values, split the target task type and the target task object to obtain n+1 subtasks and the task configuration parameters of the n+1 subtasks.

[0090] In specific implementation, the available computing resources can be understood as idle resources or computable resources that can be scheduled. Specifically, the available computing resources of the edge computing device can be determined to obtain the first available computing resources, and the available computing resources of each of the n supercomputer modules can also be determined to obtain n second available computing resources. Then, according to the first available computing resources and the n second available computing resources, n+1 resource scheduling values are determined. Specifically, the total available computing resources can be calculated from the first available computing resources and the n second available computing resources, and then the ratio between the first available computing resources and the total available computing resources is determined. This ratio corresponds to a resource scheduling value, and the ratio between each of the n second available computing resources and the total available computing resources is obtained to get n ratios, and each ratio corresponds to a resource scheduling value. The resource scheduling value can be used to represent the amount of computing resources that need to be scheduled.

[0091] Next, the edge computing device can be set as the highest priority, and then the priorities of the n supercomputer modules are determined according to the n+1 resource scheduling values to obtain n priorities. The larger the resource scheduling value, the higher the priority. Based on determining the execution order according to the n priorities and the highest priority, and according to the time sequence of the target task object and the n+1 resource scheduling values, the target task type and the target task object are split to obtain n+1 subtasks and the task configuration parameters of the n+1 subtasks. That is, the video clip can be split into n+1 video sub-clips (the first subtask object), each video sub-clip corresponds to a subtask and a first device identifier, and each subtask corresponds to a resource scheduling value. Furthermore, intelligent task splitting can be performed based on the task type, the task object, and the performance of the supercomputer module, which can ensure the video analysis efficiency. Thus, the real-time nature of video processing can be ensured, which helps to improve the video analysis efficiency of the security monitoring system.

[0092] 105. Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n+1 subtasks to obtain a target video analysis result.

[0093] In specific implementation, video analysis can be performed on the target task object according to the execution order and the task configuration parameters of the n+1 subtasks. Specifically, the edge computing device and n supercomputer modules can, in accordance with the execution order, have each device among the edge computing device and the n supercomputer modules complete the first task content and the first subtask object, thereby obtaining the target video analysis result. Since intelligent task splitting is performed based on the task type, task object, and the performance of the supercomputer modules, the video analysis efficiency can be ensured. Thus, the real-time nature of video processing can be guaranteed, which helps to improve the video analysis efficiency of the security monitoring system.

[0094] In specific implementation, the video of the camera can be pulled for real-time video processing, and through a variety of algorithms, intelligent monitoring in complex scenarios can be completed. By deploying computing resources close to the data source, instant data analysis and processing are achieved, significantly reducing the need to transmit data to the cloud, thereby reducing latency and enhancing the response speed of the system.

[0095] In specific implementation, data processing can be performed locally or close to the data source. For example, the Linux operating system can be adopted to further enhance the security of the system. Its built-in security mechanisms and flexible configuration options enable the system to effectively prevent malicious attacks and data leakage.

[0096] For example, in public places with a large flow of people, it becomes particularly important to accurately count the number of pedestrians and identify abnormal behaviors; in the field of logistics and warehousing, timely detecting and alarming the presence of foreign objects is also crucial for ensuring production safety. By introducing advanced computer vision technologies and deep learning algorithms, real-time analysis of video content is achieved, enabling the identification of specific targets or behavior patterns, which greatly improves the intelligence level and accuracy of the system.

[0097] In specific implementation, a modular design concept can be adopted, enabling convenient and rapid replacement and upgrade between components. It allows users to flexibly configure different module combinations according to actual needs to meet diverse application requirements. The advantage of this is that, according to the development of technology and changes in market demand, the hardware or software part can be updated at any time to ensure that the system is always in the best state. For example, when a new and more efficient processor comes out, users only need to replace the corresponding hardware module to enjoy the benefits brought by performance improvement without having to rebuild the entire system architecture. At the same time, the modular design also simplifies the assembly process, reduces the manufacturing difficulty, and improves the production efficiency.

[0098] The embodiments of the present application can not only solve the problems of high bandwidth occupancy, high latency, and inefficient target detection in the security monitoring system in the related art, but also provide a new development direction for future intelligent security. By introducing edge computing technology, a more efficient, secure, and flexible video processing solution is achieved, which has broad application prospects and market potential. These improvements not only enhance the user experience but also provide great convenience for the deployment and maintenance in practical applications.

[0099] In the embodiments of the present application, the intelligent video analysis system of the supercomputer cluster based on edge computing can be composed of multiple groups of edge supercomputer modules. Each supercomputer module is jointly composed of a core board module, an expansion board module, a cooling system, and a network interface. Multiple modules are managed by a cluster management system and work together collaboratively.

[0100] In specific implementation, the intelligent video analysis system of the supercomputer cluster based on edge computing is responsible for coordinating the work among various supercomputer modules, including task allocation, data synchronization, and fault detection, etc., to ensure the stable operation of the system. According to the current task load situation, it intelligently adjusts the working states of each module, adopts advanced power management and energy-saving technologies, reduces the overall energy consumption while ensuring high performance, so as to achieve the best performance and energy efficiency ratio, and realizes dynamic resource scheduling. In addition, it supports easily adding new supercomputer modules to the existing cluster without interrupting the service, greatly enhancing the adaptability and expandability of the system.

[0101] In the embodiments of the present application, the intelligent video analysis system of the supercomputer cluster based on edge computing can adopt a compact design to make the device easy to carry and quickly deploy, and the modular structure also simplifies the later maintenance and upgrade work.

[0102] For example, in the embodiments of the present application, the video stream is first pulled into the edge computing device and temporarily stored in the temporary storage space provided by the expansion board. Subsequently, the core board module starts a preset target detection algorithm (such as smoke detection, pedestrian statistics, fall detection, abnormal aggregation, intrusion into dangerous areas, etc.) to analyze the video content. After the detection is completed, the processing information is generated and saved in the log, and at the same time, the existing alarm information is sent to the customer. Throughout the process, the intelligent video analysis system of the supercomputer cluster based on edge computing can continuously monitor the state of the video stream and adjust the algorithm parameters or switch different processing modes according to the actual situation.

[0103] It can be seen that the intelligent video analysis method of the supercomputer cluster based on edge computing described in the embodiments of the present application is applied to an edge computing device. The intelligent video analysis system of the supercomputer cluster includes an edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. Obtain target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object obtained by the video monitoring device. Determine the target task complexity according to the target task type and the target task object. When the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m. Split the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameters of each subtask are: the first task content, the first subtask object, and the first device identifier. The first subtask objects corresponding to the n + 1 subtasks constitute the target task object. Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain the target video analysis result. It is possible to perform intelligent task splitting based on the task type, task object, and performance of the supercomputer module to ensure the video analysis efficiency. Therefore, the real-time performance of video processing can be guaranteed, which helps to improve the video analysis efficiency of the security monitoring system.

[0104] Consistent with the above embodiments, please refer to Figure 5 , Figure 5 is a schematic structural diagram of an edge computing device provided by an embodiment of the present application. As shown in the figure, the edge computing device includes a processor, a memory, a communication interface, and one or more programs. The above one or more programs are stored in the above memory and are configured to be executed by the above processor. In the embodiments of the present application, the edge computing device is applied to an intelligent video analysis system of a supercomputer cluster. The intelligent video analysis system of the supercomputer cluster includes the edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. The above program includes instructions for performing the following steps:

[0105] Obtain target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object obtained by the video monitoring device;

[0106] Determine the target task complexity according to the target task type and the target task object;

[0107] When the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m;

[0108] Split the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameters of each subtask are: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object;

[0109] Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain the target video analysis result.

[0110] Optionally, in terms of determining the target task complexity according to the target task type and the target task object, the above program includes instructions for performing the following steps:

[0111] Determine the target task process corresponding to the target task type;

[0112] Determine the target algorithm corresponding to the target task process;

[0113] Determine the reference complexity corresponding to the target algorithm;

[0114] Determine the first memory size and the first quality evaluation value of the target task object;

[0115] Determine the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity.

[0116] Optionally, in terms of determining the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity, the above program includes instructions for performing the following steps:

[0117] Determine the first adjustment parameter corresponding to the first memory size;

[0118] Determine the second adjustment parameter corresponding to the first quality evaluation value;

[0119] Adjust the reference complexity according to the first adjustment parameter and the second adjustment parameter to obtain the target task complexity.

[0120] Optionally, in terms of determining n supercomputer modules according to the target task complexity, the above program includes instructions for performing the following steps:

[0121] Determine k combinations of supercomputer modules corresponding to the target task complexity. Each combination of supercomputer modules includes at least one supercomputer module; k is a positive integer;

[0122] Filter the k combinations of supercomputer modules according to the target task type to obtain the target combination of supercomputer modules;

[0123] Determine the n supercomputer modules according to the determined target supercomputer module combination.

[0124] Optionally, in the aspect of filtering the k supercomputer module combinations according to the target task type and the target task duration parameter to obtain a target supercomputer module combination, the above program includes instructions for performing the following steps:

[0125] Filter the k supercomputer module combinations according to the target task type to obtain x supercomputer module combinations, where x is a positive integer less than or equal to k;

[0126] Estimate the task durations of each of the x supercomputer module combinations for executing the target task type and the target task object to obtain x task durations;

[0127] Select the minimum task duration among the x task durations, and obtain the supercomputer module combination corresponding to the minimum task duration to obtain the target supercomputer module combination.

[0128] Optionally, in the aspect of splitting the target task type and the target task object to obtain task configuration parameters of n + 1 subtasks and determining the execution order of the n + 1 subtasks, the above program includes instructions for performing the following steps:

[0129] Determine the available computing resources of the edge computing device to obtain the first available computing resources;

[0130] Determine the available computing resources of the n supercomputer modules to obtain n second available computing resources;

[0131] Determine n + 1 resource scheduling values according to the first available computing resources and the n second available computing resources;

[0132] Set the edge computing device to the highest priority;

[0133] Determine the priorities of the n supercomputer modules according to the n + 1 resource scheduling values to obtain n priorities, where the larger the resource scheduling value, the higher the priority;

[0134] Determine the execution order according to the n priorities and the highest priority;

[0135] Split the target task type and the target task object according to the time sequence of the target task object and the n + 1 resource scheduling values to obtain n + 1 subtasks and the task configuration parameters of the n + 1 subtasks.

[0136] It can be seen that for the edge computing device described in the embodiments of the present application, the edge computing device is applied to a supercomputer cluster intelligent video analysis system. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video surveillance device, where m is an integer greater than 1. The target video analysis task parameters are obtained. The target video analysis task parameters include the target task type and the target task object obtained by the video surveillance device. The target task complexity is determined according to the target task type and the target task object. When the target task complexity is greater than a preset threshold, n supercomputer modules are determined according to the target task complexity, where n is a positive integer less than or equal to m. The target task type and the target task object are split to obtain the task configuration parameters of n + 1 subtasks, and the execution order of the n + 1 subtasks is determined. The task configuration parameters of each subtask are: the first task content, the first subtask object, and the first device identifier. The first subtask objects corresponding to the n + 1 subtasks constitute the target task object. Video analysis is performed on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain the target video analysis result. Intelligent task splitting can be performed based on the task type, the task object, and the performance of the supercomputer modules to ensure the video analysis efficiency. Therefore, the real-time nature of video processing can be ensured, which helps to improve the video analysis efficiency of the security surveillance system.

[0137] Figure 6 It is a functional unit composition block diagram of a supercomputer cluster intelligent video analysis device 600 based on edge computing involved in the embodiments of the present application. The supercomputer cluster intelligent video analysis device 600 based on edge computing is applied to an edge computing device. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video surveillance device, where m is an integer greater than 1. The supercomputer cluster intelligent video analysis device 600 based on edge computing includes: an acquisition unit 601, a determination unit 602, a splitting unit 603, and an analysis unit 604. Among them,

[0138] The acquisition unit 601 is configured to acquire target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object acquired by the video surveillance device;

[0139] The determination unit 602 is configured to determine a target task complexity according to the target task type and the target task object; when the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m;

[0140] The splitting unit 603 is configured to split the target task type and the target task object, obtain the task configuration parameters of n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameter of each subtask includes: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object.

[0141] The analysis unit 604 is configured to perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks, and obtain a target video analysis result.

[0142] Optionally, in terms of determining the target task complexity according to the target task type and the target task object, the determining unit 602 is specifically configured to:

[0143] Determine the target task process corresponding to the target task type;

[0144] Determine the target algorithm corresponding to the target task process;

[0145] Determine the reference complexity corresponding to the target algorithm;

[0146] Determine the first memory size and the first quality evaluation value of the target task object;

[0147] Determine the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity.

[0148] Optionally, in terms of determining n supercomputer modules according to the target task complexity, the determining unit 602 is specifically configured to:

[0149] Determine k combinations of supercomputer modules corresponding to the target task complexity, and each combination of supercomputer modules includes at least one supercomputer module; k is a positive integer;

[0150] Filter the k combinations of supercomputer modules according to the target task type to obtain a target combination of supercomputer modules;

[0151] Determine the n supercomputer modules according to the target combination of supercomputer modules.

[0152] Optionally, in terms of filtering the k combinations of supercomputer modules according to the target task type and the target task duration parameter to obtain a target combination of supercomputer modules, the determining unit 602 is specifically configured to:

[0153] Screen the k combinations of supercomputer modules according to the target task type to obtain x combinations of supercomputer modules, where x is a positive integer less than or equal to k;

[0154] Estimate the task duration of each of the x combinations of supercomputer modules for the target task type and the target task object, obtaining x task durations;

[0155] Select the minimum task duration among the x task durations, and obtain the combination of supercomputer modules corresponding to the minimum task duration, obtaining the target combination of supercomputer modules.

[0156] Optionally, in terms of splitting the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks and determining the execution order of the n + 1 subtasks, the splitting unit 603 is specifically configured to:

[0157] Determine the available computing resources of the edge computing device to obtain the first available computing resources;

[0158] Determine the available computing resources of the n supercomputer modules to obtain n second available computing resources;

[0159] Determine n + 1 resource scheduling values according to the first available computing resources and the n second available computing resources;

[0160] Set the edge computing device to the highest priority;

[0161] Determine the priorities of the n supercomputer modules according to the n + 1 resource scheduling values, obtaining n priorities, the larger the resource scheduling value, the higher the priority;

[0162] Determine the execution order according to the n priorities and the highest priority;

[0163] Split the target task type and the target task object according to the time sequence of the target task object and the n + 1 resource scheduling values, obtaining n + 1 subtasks and the task configuration parameters of the n + 1 subtasks.

[0164] It can be seen that the edge-computing-based supercomputer cluster intelligent video analysis device described in the embodiments of the present application is applied to an edge computing device. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. The target video analysis task parameters are obtained, and the target video analysis task parameters include the target task type and the target task object obtained by the video monitoring device. The target task complexity is determined according to the target task type and the target task object. When the target task complexity is greater than a preset threshold, n supercomputer modules are determined according to the target task complexity, where n is a positive integer less than or equal to m. The target task type and the target task object are split to obtain the task configuration parameters of n + 1 subtasks, and the execution order of the n + 1 subtasks is determined. The task configuration parameter of each subtask includes: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object. Video analysis is performed on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain the target video analysis result. Intelligent task splitting can be performed based on the task type, task object, and the performance of the supercomputer module to ensure the video analysis efficiency. Therefore, the real-time performance of video processing can be ensured, which helps to improve the video analysis efficiency of the security monitoring system.

[0165] It can be understood that the functions of the program modules of the edge-computing-based supercomputer cluster intelligent video analysis device in this embodiment can be specifically implemented according to the methods in the above method embodiments. The specific implementation process can refer to the relevant descriptions in the above method embodiments and will not be elaborated here.

[0166] The embodiments of the present application further provide a computer storage medium, where the computer storage medium stores a computer program for electronic data exchange, and the computer program enables a computer to execute some or all of the steps of any of the methods recorded in the above method embodiments.

[0167] The embodiments of the present application further provide a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute some or all of the steps of any of the methods recorded in the above method embodiments. The computer program product can be a software installation package.

[0168] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0169] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not described in detail in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0170] In several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the above division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in an electrical or other form.

[0171] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0172] In addition, each functional unit in the various embodiments of the present application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0173] If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on such an understanding, the technical solution of the present 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 for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the above methods in the various embodiments of the present application. The aforementioned memory includes: various media such as USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0174] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable memory. The memory may include: a flash drive, a read-only memory (abbreviation: ROM), a random access memory (abbreviation: RAM), a magnetic disk, an optical disk, etc.

[0175] The above embodiments of the present application have been introduced in detail. Specific examples are used in this article to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. An intelligent video analysis method for a supercomputer cluster based on edge computing, characterized in that This method is applied to an edge computing device. The supercomputer cluster intelligent video analysis system includes the edge computing device, m supercomputer modules, and a video monitoring device, where m is an integer greater than 1. The method includes: Obtain target video analysis task parameters, where the target video analysis task parameters include a target task type and a target task object obtained by the video monitoring device; Determine the target task complexity according to the target task type and the target task object; When the target task complexity is greater than a preset threshold, determine n supercomputer modules according to the target task complexity, where n is a positive integer less than or equal to m; Split the target task type and the target task object to obtain task configuration parameters for n + 1 subtasks, and determine the execution order of the n + 1 subtasks. The task configuration parameters for each subtask are: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object; Perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain a target video analysis result.

2. The method according to claim 1, wherein The determining the target task complexity according to the target task type and the target task object includes: Determine a target task process corresponding to the target task type; Determine a target algorithm corresponding to the target task process; Determine a reference complexity corresponding to the target algorithm; Determine the first memory size and the first quality evaluation value of the target task object; Determine the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity.

3. The method according to claim 2, wherein The determining the target task complexity according to the first memory size, the first quality evaluation value, and the reference complexity includes: Determine a first adjustment parameter corresponding to the first memory size; Determine a second adjustment parameter corresponding to the first quality evaluation value; Adjust the reference complexity according to the first adjustment parameter and the second adjustment parameter to obtain the target task complexity.

4. The method according to any one of claims 1 to 3, characterized in that, The determining n supercomputer modules according to the target task complexity includes: Determine k combinations of supercomputer modules corresponding to the target task complexity, where each combination of supercomputer modules includes at least one supercomputer module; k is a positive integer; Filter the k combinations of supercomputer modules according to the target task type to obtain a target combination of supercomputer modules; Determine the n supercomputer modules according to the target combination of supercomputer modules.

5. The method according to claim 4, characterized in that The filtering the k combinations of supercomputer modules according to the target task type and the target task duration parameter to obtain a target combination of supercomputer modules includes: Screen the k combinations of supercomputer modules according to the target task type to obtain x combinations of supercomputer modules, where x is a positive integer less than or equal to k; Estimate the task duration for each of the x combinations of supercomputer modules to execute the target task type and the target task object to obtain x task durations; Select the minimum duration among the x task durations, and obtain the combination of supercomputing modules corresponding to the minimum duration to obtain the target combination of supercomputing modules.

6. The method according to any one of claims 1-3, characterized in that, The splitting of the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks and determine the execution order of the n + 1 subtasks includes: Determine the available computing resources of the edge computing device to obtain the first available computing resources; Determine the available computing resources of the n supercomputing modules to obtain n second available computing resources; Determine n + 1 resource scheduling values according to the first available computing resources and the n second available computing resources; Set the edge computing device to the highest priority; Determine the priorities of the n supercomputing modules according to the n + 1 resource scheduling values to obtain n priorities. The larger the resource scheduling value, the higher the priority; Determine the execution order according to the n priorities and the highest priority; Split the target task type and the target task object according to the time sequence of the target task object and the n + 1 resource scheduling values to obtain n + 1 subtasks and the task configuration parameters of the n + 1 subtasks.

7. An intelligent video analysis device for a supercomputer cluster based on edge computing, characterized in that, The device is applied to an edge computing device. The supercomputing cluster intelligent video analysis system includes the edge computing device, m supercomputing modules and a video monitoring device, where m is an integer greater than 1. The device includes: an acquisition unit, a determination unit, a splitting unit and an analysis unit, where, The acquisition unit is used to acquire target video analysis task parameters, and the target video analysis task parameters include a target task type and a target task object acquired by the video monitoring device; The determination unit is used to determine the target task complexity according to the target task type and the target task object; when the target task complexity is greater than a preset threshold, determine n supercomputing modules, where n is a positive integer less than or equal to m; The splitting unit is used to split the target task type and the target task object to obtain the task configuration parameters of n + 1 subtasks and determine the execution order of the n + 1 subtasks. The task configuration parameters of each subtask are: the first task content, the first subtask object, and the first device identifier; the first subtask objects corresponding to the n + 1 subtasks constitute the target task object; The analysis unit is used to perform video analysis on the target task object according to the execution order and the task configuration parameters of the n + 1 subtasks to obtain a target video analysis result.

8. The device according to claim 7, wherein In terms of determining the target task complexity according to the target task type and the target task object, the determination unit specifically is used to: Determine the target task process corresponding to the target task type; Determine the target algorithm corresponding to the target task process; Determine the reference complexity of the target algorithm; Determine the first memory size and the first quality evaluation value of the target task object; Determine the target task complexity according to the first memory size, the first quality evaluation value and the reference complexity.

9. An edge computing device, characterized in that, Comprising a processor, a memory, a communication interface, and one or more programs, the one or more programs being stored in the memory and configured to be executed by the processor, the programs comprising instructions for performing the steps in the method according to any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program being executed by a processor to implement the method according to any one of claims 1 to 6.

Images