WO2020001427A1

WO2020001427A1 - Analysis task execution method, apparatus and system, and electronic device

Info

Publication number: WO2020001427A1
Application number: PCT/CN2019/092756
Authority: WO
Inventors: 冯仁光
Original assignee: 杭州海康威视数字技术股份有限公司
Priority date: 2018-06-28
Filing date: 2019-06-25
Publication date: 2020-01-02
Also published as: CN110659125A

Abstract

Embodiments of the present application provide an analysis task execution method, apparatus and system, and an electronic device. The method comprises: determining sub-tasks included in an analysis task to be executed; obtaining algorithm components for executing the sub-tasks as algorithm components to be assembled; assembling, according to a preset arrangement rule, the algorithm components to be assembled to obtain a target analysis algorithm; and executing, by using the target analysis algorithm, the analysis task to be executed. Only algorithm components providing different functions need to be stored, and by means of algorithm arrangement, assembled into an analysis algorithm capable of executing the analysis task to be executed, and it is not required to store one analysis algorithm for each analysis task, thereby effectively reducing the consumption of storage resources caused to achieve good adaptability of an analysis device.

Description

Method, device, system and electronic equipment for analyzing task execution

This application claims priority from a Chinese patent application filed on June 28, 2018 with the Chinese Patent Office under the application number 201810688355.2 and entitled "A Method, Device, System, and Electronic Equipment for Analyzing Task Execution", the entire contents of which are incorporated by reference. Incorporated in this application.

Technical field

The present application relates to the field of data analysis technology, and in particular, to a method, a device, a system, and an electronic device for performing an analysis task.

Background technique

The data collected by the equipment often requires intelligent analysis to obtain the information the user needs. Specifically, an analysis task may be generated for the collected data, and the analysis device performs the analysis task to obtain information required by the user.

In the related art, the analysis device may use an analysis algorithm written in advance to perform an analysis task. For different analysis tasks, different analysis algorithms may be required to perform, for example, the analysis algorithm required for the analysis task of face recognition in a video is different from the analysis algorithm required for the analysis task of vehicle tracking in the video. Therefore, the analysis tasks that the analysis device can perform are limited to the types of analysis algorithms written in the local area. If the analysis algorithms used to perform certain analysis tasks are not written in the local area, the analysis device may not be able to perform the analysis tasks. In order to increase the types of analysis tasks that can be performed by the analysis device, analysis algorithms can be written in the analysis device for various analysis tasks in advance, but it will occupy a large amount of storage resources.

Summary of the invention

The purpose of the embodiments of the present application is to provide a method, a device, a system, and an electronic device for performing an analysis task, so as to achieve a better adaptability of the analysis device on the premise that less storage resources are occupied. The specific technical solutions are as follows:

In a first aspect of the embodiments of the present application, a method for performing an analysis task is provided. The method includes:

Determine the subtasks to be included in the analysis task to be performed;

Acquiring an algorithm component for performing the sub-task as an algorithm component to be assembled;

Assemble the algorithm component to be assembled according to a preset arrangement rule to obtain a target analysis algorithm;

And using the target analysis algorithm to perform the analysis task to be performed.

With reference to the first aspect, in a first possible implementation manner, the assembly of the algorithm components according to a preset arrangement rule to obtain a target analysis algorithm includes:

According to the order of the preset priorities of the algorithm components to be assembled from high to low, the algorithm components to be assembled are assembled in series from front to back to obtain a target analysis algorithm. The preset priorities of the following three types of algorithm components are High to low: Algorithm components used to provide object detection functions, algorithm components used to provide video tracking functions, and algorithm components used to provide intelligent application functions.

With reference to the first aspect, in a second possible implementation manner, before the determining a subtask included in an analysis task to be performed, the method further includes:

Determining whether an analysis algorithm for performing an analysis task to be performed is stored locally;

If an analysis algorithm for executing the to-be-executed analysis task is stored locally, using the analysis algorithm saved for executing the to-be-executed analysis task locally to execute the to-be-executed analysis task;

If the analysis algorithm for executing the to-be-executed analysis task is not stored locally, the step of determining a sub-task included in the to-be-executed analysis task is performed.

With reference to the first aspect, in a third possible implementation manner, before using the target analysis algorithm to perform an analysis task to be performed, the method further includes:

Hot-switch the currently used analysis algorithm to the target analysis algorithm.

In a second aspect provided in the embodiments of the present application, an analysis task execution device is provided, where the device includes:

A task analysis module for determining sub-tasks included in the analysis task to be performed;

A component acquisition module, configured to acquire an algorithm component for performing the sub-task as an algorithm component to be assembled;

An algorithm assembly module is configured to assemble the assembly algorithm components according to a preset arrangement rule to obtain a target analysis algorithm;

An execution module is configured to use the target analysis algorithm to execute the analysis task to be performed.

With reference to the second aspect, in a first possible implementation manner, the algorithm assembly module is specifically configured to move the algorithm component to be assembled from front to bottom according to a preset priority of the algorithm component to be assembled from high to low. It is assembled in series to obtain the target analysis algorithm. Among them, the preset priorities of the following three types of algorithm components are from high to low: algorithm components for providing target detection functions, algorithm components for providing video tracking functions, and intelligence Algorithmic components for application functions.

With reference to the second aspect, in a second possible implementation manner, the task analysis module is further configured to determine whether a sub-task included in the analysis task to be executed is stored locally for performing the analysis to be executed before the determination. Task analysis algorithm;

If the analysis algorithm for executing the to-be-executed analysis task is not stored locally, executing the step of determining a sub-task included in the to-be-executed analysis task;

The execution module is further configured to execute the analysis task to be executed by using the analysis algorithm stored locally to execute the analysis task to be executed if the analysis algorithm used to execute the analysis task to be executed is stored locally.

With reference to the second aspect, in a third possible implementation manner, the execution module is further configured to hot-switch the currently used analysis algorithm before performing the analysis task to be performed using the target analysis algorithm. An algorithm for the target analysis.

In a third aspect of the embodiments of the present application, an analysis task execution system is provided, and the system includes:

A management platform, configured to obtain an analysis task of a device connected to the analysis task execution system as a to-be-executed analysis task;

An analysis server, configured to execute the to-be-executed analysis task according to any one of the foregoing analysis task execution methods;

A storage server storing an algorithm component in the storage server, the storage server being configured to provide the analysis server with an algorithm component for performing the analysis task to be performed.

With reference to the third aspect, in a first possible implementation manner, the algorithm components stored in the storage server are classified according to at least one of the following classification basis: supported platform architecture, bit width, The type of data processed, the type of target processed, and the type of subtask processed.

In a fourth aspect of the embodiments of the present application, an electronic device is provided, including:

Memory for storing computer programs;

The processor is configured to implement any of the foregoing analysis task execution methods when executing a program stored in the memory.

In a fifth aspect of the embodiments of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the analysis described in any of the foregoing is implemented. Task execution method.

The analysis task execution method, device, system and electronic equipment provided in the embodiments of the present application may only need to store algorithm components that provide different functions, and arrange them into an analysis algorithm capable of performing the analysis task to be performed through algorithm orchestration. This analysis task stores an analysis algorithm, which effectively reduces the consumption of storage resources in order to achieve a better adaptability of the analysis equipment. Of course, the implementation of any product or method of this application does not necessarily need to achieve all the advantages described above at the same time.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings in the following description are only Some embodiments of the application, for those of ordinary skill in the art, can obtain other drawings according to these drawings without paying creative labor.

FIG. 1 is a schematic flowchart of an analysis task execution method according to an embodiment of the present application; FIG.

FIG. 2 is another schematic flowchart of an analysis task execution method according to an embodiment of the present application; FIG.

FIG. 3 is another schematic flowchart of an analysis task execution method according to an embodiment of the present application; FIG.

FIG. 4 is a schematic structural diagram of an analysis task execution device according to an embodiment of the present application; FIG.

5a is a schematic diagram of a framework of an analysis task execution system provided by an embodiment of the present application;

5b is another schematic diagram of an analysis task execution system according to an embodiment of the present application;

FIG. 5c is a schematic diagram of a switching method of a target analysis algorithm according to an embodiment of the present application; FIG.

FIG. 6 is a schematic structural diagram of an analysis task execution electronic device according to an embodiment of the present application.

detailed description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

Referring to FIG. 1, FIG. 1 is a schematic flowchart of an analysis task execution algorithm according to an embodiment of the present application, which may include:

S101. Determine a subtask included in an analysis task to be performed.

The analysis task to be executed may include only one subtask, or may include multiple subtasks at the same time. Each subtask corresponds to a link in the process of the analysis task to be executed. For example, one task is to analyze the human body in the video stream screen. Behavior, the specific process may be to first identify the human body in the video picture and track the identified human body, and perform human behavior analysis based on the human body information obtained by the tracking, that is, the task may include a human body identification subtask and a human body tracking subtask. , And behavior analysis subtasks.

S102. Acquire an algorithm component for executing a subtask as an algorithm component to be assembled.

For each sub-task included in the analysis task to be executed, an algorithm component for performing the sub-task may be obtained. The algorithm component used to execute a subtask refers to an algorithm component that has a function of executing the subtask. In this embodiment, the algorithm component cannot run alone. Only in the assembled analysis algorithm, the algorithm component can be used to execute the subtask.

In an optional embodiment, the algorithm component is stored in the algorithm warehouse, and the algorithm component used to execute the subtask is obtained. The algorithm component may be obtained from the algorithm warehouse to execute the subtask. In other embodiments, the algorithm component may also be stored locally, which is not limited in this embodiment of the present application.

S103: Assemble the algorithm component to be assembled according to a preset arrangement rule to obtain a target analysis algorithm.

The preset orchestration rules can be configured according to the actual needs of the user. In a preferred embodiment, the components to be assembled are connected in series from front to back according to the preset priority of the algorithm components to be assembled. Assemble to get the target analysis algorithm.

Further, the algorithm components to be assembled can be pre-divided into three categories according to different functions, including: algorithm components for providing target detection functions (such as algorithm components for providing human detection functions), and algorithm components for providing video tracking functions. Algorithm components that provide intelligent application functions (such as algorithm components that provide human behavior analysis functions). Among these three types of algorithm components, the preset priority of the algorithm component used to provide the target detection function is higher than the algorithm used to provide the tracking function. The preset priority of the component. The preset priority of the algorithm component used to provide the tracking function is higher than the preset priority of the algorithm component used to provide the intelligent application function.

Exemplarily, when the algorithm components to be assembled are assembled, if all three types of algorithm components exist, the algorithm component used to provide the target detection function is installed in the first link, and the algorithm component used to provide the video tracking function is installed Installed in the second link. If there is only one algorithm component for providing intelligent application functions, install the algorithm component for providing intelligent application functions in the third link. If there are multiple algorithms for providing intelligent application functions Components, these algorithm components used to provide intelligent application functions are installed in the third and subsequent links.

S104. Use a target analysis algorithm to perform a task to be performed for analysis.

Because the target analysis algorithm is assembled by the algorithm components used to perform the sub-tasks of the task to be analyzed, the target analysis algorithm can perform each sub-task in the analysis task to be performed, and the target analysis algorithm can be used to perform the analysis to be performed task.

In this embodiment, only the algorithm components that provide different functions need to be stored, and the algorithm can be assembled into an analysis algorithm that can perform the analysis task to be performed. It is no longer necessary to store an analysis algorithm for each analysis task, which effectively reduces the Realize the better adaptability of analysis equipment, and the consumption of storage resources. On the other hand, it can also enable more types of analysis tasks to be performed with a limited storage space.

Referring to FIG. 2, FIG. 2 is another schematic flowchart of an analysis task execution method according to an embodiment of the present application, which may include:

S201. It is determined whether an analysis algorithm for performing an analysis task to be executed is locally stored. If an analysis algorithm for performing an analysis task to be executed is stored locally, S202 is performed. S203.

This embodiment can be applied to an electronic device with intelligent analysis capabilities, for example, it can be applied to an intelligent analysis server, or it can be applied to a smart camera equipped with an intelligent processing chip. Taking the application of this embodiment to an intelligent analysis server as an example, this step may be determining whether an analysis algorithm for performing an analysis task to be performed exists in a memory of the intelligent server, where the memory may include an internal memory and an external memory.

S202: Use the analysis algorithm saved locally to perform the analysis task to be executed to execute the analysis task to be executed.

When an analysis algorithm for performing an analysis task to be performed is stored locally, it may be considered that it is not necessary to reassemble the analysis algorithm using an algorithm component. By directly using the analysis algorithm saved locally to perform the analysis task to be performed, the analysis task to be performed can already be performed. Of course, in other embodiments, for the actual needs of the user, when the analysis algorithm used to perform the analysis task to be performed is stored locally, the analysis algorithm may be assembled by using the algorithm component, and the analysis algorithm obtained by the assembly is used to perform the analysis to be performed. Task, this embodiment does not limit this.

S203: Determine a subtask included in the analysis task to be performed.

This step is the same as S101. For details, refer to the foregoing description of S101, and details are not described herein again.

S204. Obtain an algorithm component for executing a subtask as an algorithm component to be assembled.

This step is the same as S102. For details, refer to the foregoing description of S102, and details are not described herein again.

S205: Assemble the algorithm components to be assembled according to a preset orchestration rule to obtain a target analysis algorithm.

This step is the same as S103. For details, refer to the foregoing description of S103, and details are not described herein again.

S206. Use the target analysis algorithm to perform the analysis task to be performed.

This step is the same as S104. For details, refer to the foregoing description of S104, and details are not described herein again.

In this embodiment, when an analysis algorithm for executing an analysis task to be executed is already stored locally, the analysis algorithm for executing an analysis task to be executed locally may be directly used to execute the analysis task to be executed, avoiding problems caused by the algorithm. Resources are wasted due to unnecessary assembly of components.

Referring to FIG. 3, FIG. 3 is another schematic flowchart of an analysis task execution method according to an embodiment of the present application, which may include:

S301. Determine a subtask included in an analysis task to be performed.

S302. Acquire an algorithm component for performing a subtask as an algorithm component to be assembled.

S303: Assemble the algorithm components to be assembled according to a preset arrangement rule to obtain a target analysis algorithm.

S304. Hot-switch the currently used analysis algorithm to the target analysis algorithm.

Among them, hot-swap refers to the fact that it is not necessary to restart the device, and it can switch the currently used analysis algorithm to the target analysis algorithm.

S305. Use a target analysis algorithm to perform a task to be performed for analysis.

Since the target analysis algorithm is assembled by the algorithm component to be assembled in this embodiment, there is no need to modify any global variables based on the current analysis algorithm, so hot switching can be used to achieve faster analysis algorithm switching. Because hot-swap does not require restarting the device, it can effectively shorten the interval required to perform different types of to-be-analyzed tasks.

Referring to FIG. 4, FIG. 4 is a schematic structural diagram of an analysis task execution apparatus according to an embodiment of the present application, which may include:

A task analysis module 401, configured to determine a subtask included in an analysis task to be performed;

A component acquisition module 402, configured to acquire an algorithm component for performing a sub-task as an algorithm component to be assembled;

An algorithm assembly module 403 is configured to assemble an algorithm component according to a preset orchestration rule to obtain a target analysis algorithm;

The execution module 404 is configured to perform an analysis task to be performed by using a target analysis algorithm.

Further, the algorithm assembly module 403 is specifically configured to assemble the algorithm components to be assembled in series from front to back according to the preset priority of the algorithm components to be assembled, to obtain a target analysis algorithm. Among them, the following three types of algorithms The preset priorities of the components are from high to low: the algorithm component used to provide the target detection function, the algorithm component used to provide the video tracking function, and the algorithm component used to provide the intelligent application function.

Further, the task analysis module 401 is further configured to determine, before determining the subtask included in the analysis task to be executed, whether an analysis algorithm for performing the analysis task to be executed is stored locally;

If the analysis algorithm used to perform the analysis task to be performed is not stored locally, perform the step of determining a subtask included in the analysis task to be performed;

The execution module 404 is further configured to execute an analysis task to be executed if the analysis algorithm for executing the analysis task to be executed is stored locally, and use the analysis algorithm stored to execute the analysis task to be executed.

Further, the execution module 404 is further configured to hot-swap the currently used analysis algorithm to the target analysis algorithm before using the target analysis algorithm to perform the analysis task to be performed.

Referring to FIG. 5a, FIG. 5a shows a framework diagram of an analysis task execution system provided by an embodiment of the present application, which may include:

A management platform 501, configured to obtain an analysis task of a device connected to the analysis task execution system as a to-be-executed analysis task;

An analysis server 502 is configured to execute a to-be-executed analysis task according to any one of the foregoing analysis task execution methods;

The storage server 503 stores an algorithm component in the storage server, and the storage server is configured to provide the analysis server with an algorithm component for performing an analysis task to be performed.

The analysis server 502 may be a virtual server, or a server as a physical device. The storage server can be a single storage device or a storage cluster composed of multiple storage devices.

In an optional embodiment, as shown in FIG. 5b, the storage server 503 is a storage cluster composed of multiple storage devices 5031. These storage devices can utilize a load balancer 504 to achieve storage load balancing.

Further, the algorithm components stored in the storage server are classified according to at least one of the following classification basis: supported platform architecture, bit width, type of data processed, type of target processed, processed Subtask type. In other embodiments, classification may also be performed according to the manufacturer of the algorithm component, the writing standard, and the like, which is not limited in this embodiment. It is understandable that classifying the algorithm components helps the storage server to better manage these algorithm components.

Further, the analysis server 502 may obtain the algorithm component directly from the storage server 503, or the management platform 501 may obtain the algorithm component from the storage server 503, and then dispatch the obtained algorithm component to the analysis server 502.

The following optional embodiments will describe the workflow of the analysis system:

The management platform 501 may obtain the analysis task of the device accessed by the analysis system, assign the analysis task to the analysis server 502, and register an intelligent analysis service for data interaction with the analysis server 502 to implement the analysis Management of analysis server 502.

The management platform 501 obtains, from the storage server 503, an algorithm component for executing each subtask included in the analysis task, and dispatches the obtained algorithm component to the analysis server 502. Further, if the management platform 501 is connected to multiple analysis servers 502, the management platform may obtain the operating performance of the multiple analysis servers 502, and allocate the analysis task to the analysis server 502 with the highest analysis performance to improve the analysis. Task execution efficiency. Further, before the management platform 501 dispatches algorithm components to the analysis server 502, the management platform 501 may configure parameter settings of these algorithm components. For example, the management platform 501 may receive configuration parameters input by a user for analysis tasks, and according to these The configuration parameters configure the algorithm components, so that the target analysis algorithm assembled by these algorithm components can better complete the analysis task.

After obtaining the algorithm components, the analysis server 502 composes these algorithm components to obtain a target analysis algorithm according to a preset arrangement rule, and uses the target analysis algorithm to perform an analysis task. There may be only one target analysis algorithm in the analysis server 502, or there may be multiple target analysis algorithms. The analysis server 502 with multiple target analysis algorithms can hot-swap the currently used analysis algorithm to the target analysis algorithm corresponding to the analysis task according to the different analysis tasks. The switching method can be shown in Figure 5c to achieve the use of a The analysis server 502 can complete a variety of analysis tasks.

Further, in addition to the above-mentioned scheduling algorithm components, one or more of the following functions may be implemented between the management platform 501 and the analysis server 502:

Algorithm update: The management platform 501 may update the target analysis algorithm regularly or according to the demand control analysis server 502. It can be understood that after the analysis server 502 composes the algorithm components into a target analysis algorithm, the developer may optimize the algorithm components to improve the performance of these algorithm components. In this case, update the target analysis in the analysis server 502. Algorithms can improve the performance of the analysis server when performing analysis tasks. The analysis server 502 may periodically or be controlled by the management platform 501, and feed back the version of the target analysis algorithm to the management platform 501, so that the management platform 501 can better manage the target analysis algorithm in the analysis server 502. The version may be determined by the version of the algorithm component used to assemble the target analysis algorithm, or may be determined by the date when the target analysis algorithm is assembled.

Logout and keep-alive: The management platform 501 maintains intelligent analysis services, which consumes a certain amount of system resources. After an analysis server 502 completes all analysis tasks assigned to the analysis server 502, if the computing power provided by other analysis servers 502 is sufficient After completing the analysis task generated by the device connected to the analysis system, the analysis task may not be assigned to the analysis server 502, and the management platform 501 may cancel the intelligent analysis service corresponding to the analysis server 502. If the analysis task is not assigned to the analysis server 502 for a short period of time, the analysis task may also be assigned to the analysis server 502 in the future. The management platform 501 can set the intelligent analysis service to a keep-alive state to facilitate reuse. Wake up quickly without the need to re-register for a new smart analytics service.

An embodiment of the present application further provides an electronic device, as shown in FIG. 6, including:

The memory 601 is configured to store a computer program;

The processor 602 is configured to execute the following steps when executing a program stored in the memory 601:

Determine the subtasks to be included in the analysis task to be performed;

Obtaining an algorithm component for performing a sub-task as an algorithm component to be assembled;

Assemble the algorithm components to be assembled according to preset orchestration rules to assemble the target analysis algorithm;

Use target analysis algorithms to perform pending analysis tasks.

Further, the algorithm components are assembled according to preset orchestration rules to obtain a target analysis algorithm, including:

According to the order of the preset priorities of the algorithm components to be assembled, the algorithm components to be assembled are assembled in series from front to back to obtain the target analysis algorithm. Among them, the preset priorities of the following three types of algorithm components are from high to low: An algorithm component for providing a target detection function, an algorithm component for providing a video tracking function, and an algorithm component for providing an intelligent application function.

Further, before determining the subtasks included in the analysis task to be performed, the method further includes:

If the analysis algorithm used to execute the analysis task to be executed is stored locally, the analysis algorithm used to execute the analysis task to be executed locally is used to execute the analysis task to be executed;

If the analysis algorithm used to perform the analysis task to be performed is not stored locally, the step of determining a subtask included in the analysis task to be performed is performed.

Further, before using the target analysis algorithm to perform the analysis task to be performed, the method further includes:

Hot-swap the currently used analysis algorithm to the target analysis algorithm.

The memory mentioned in the above electronic device may include random access memory (Random Access Memory, RAM), and may also include non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far from the foregoing processor.

The above processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc .; it may also be a digital signal processor (Digital Signal Processing, DSP), special integration Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In still another embodiment provided by the present application, a computer-readable storage medium is also provided. The computer-readable storage medium stores instructions, and when the computer-readable storage medium is run on a computer, the computer executes any of the foregoing embodiments. Analyze task execution methods.

In still another embodiment provided by the present application, a computer program product containing instructions is also provided. When the computer program product is run on a computer, the computer is caused to execute any analysis task execution method in the foregoing embodiment.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the embodiments of the present application are wholly or partially generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be from a website site, a computer, a server, or a data center. Transmission via wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) to another website site, computer, server, or data center. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, a data center, or the like that includes one or more available medium integration. The available medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (Solid State Disk (SSD)), and the like.

It should be noted that in this article, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that these entities or operations There is any such actual relationship or order among them. Moreover, the terms "including", "comprising", or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also those that are not explicitly listed Or other elements inherent to such a process, method, article, or device. Without more restrictions, the elements defined by the sentence "including a ..." do not exclude the existence of other identical elements in the process, method, article, or equipment that includes the elements.

Each embodiment in this specification is described in a related manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the device, system, electronic device, computer-readable storage medium, and computer program product, since they are basically similar to the method embodiment, the description is relatively simple. For the related parts, see the description of the method embodiment Just fine.

The above descriptions are merely preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any modification, equivalent replacement, and improvement made within the spirit and principle of this application are included in the protection scope of this application.

Claims

An analysis task execution method, characterized in that the method includes:

Determine the subtasks to be included in the analysis task to be performed;

Acquiring an algorithm component for performing the sub-task as an algorithm component to be assembled;

Assemble the algorithm component to be assembled according to a preset arrangement rule to obtain a target analysis algorithm;

And using the target analysis algorithm to perform the analysis task to be performed.
The method according to claim 1, wherein the assembly of the algorithm components according to preset orchestration rules to obtain a target analysis algorithm comprises:

According to the order of the preset priorities of the algorithm components to be assembled from high to low, the algorithm components to be assembled are assembled in series from front to back to obtain a target analysis algorithm. The preset priorities of the following three types of algorithm components are High to low: Algorithm components used to provide object detection functions, algorithm components used to provide video tracking functions, and algorithm components used to provide intelligent application functions.
The method according to claim 1, wherein before the determining a subtask included in the analysis task to be performed, the method further comprises:

Determining whether an analysis algorithm for performing an analysis task to be performed is stored locally;

If an analysis algorithm for executing the to-be-executed analysis task is stored locally, using the analysis algorithm saved for executing the to-be-executed analysis task locally to execute the to-be-executed analysis task;

If the analysis algorithm for executing the to-be-executed analysis task is not stored locally, the step of determining a sub-task included in the to-be-executed analysis task is performed.
The method according to claim 1, wherein before the using the target analysis algorithm to perform an analysis task to be performed, the method further comprises:

Hot-switch the currently used analysis algorithm to the target analysis algorithm.
An analysis task execution device is characterized in that the device includes:

A task analysis module for determining sub-tasks included in the analysis task to be performed;

A component acquisition module, configured to acquire an algorithm component for performing the sub-task as an algorithm component to be assembled;

An algorithm assembly module is configured to assemble the assembly algorithm components according to a preset arrangement rule to obtain a target analysis algorithm;

An execution module is configured to use the target analysis algorithm to execute the analysis task to be performed.
The device according to claim 5, wherein the algorithm assembly module is specifically configured to serially connect the algorithm components to be assembled from front to back according to a preset priority of the algorithm components to be assembled. Assemble to obtain the target analysis algorithm. Among them, the preset priorities of the following three types of algorithm components are from high to low: algorithm components for providing target detection functions, algorithm components for providing video tracking functions, and intelligent application functions. Algorithm components.
The apparatus according to claim 5, wherein the task analysis module is further configured to determine whether a sub-task included in the analysis task to be executed is stored locally before the determination of the sub-task included in the analysis task to be executed. Analysis algorithm

If the analysis algorithm for executing the to-be-executed analysis task is not stored locally, executing the step of determining a sub-task included in the to-be-executed analysis task;

The execution module is further configured to execute the analysis task to be executed by using the analysis algorithm stored locally to execute the analysis task to be executed if the analysis algorithm used to execute the analysis task to be executed is stored locally.
The device according to claim 5, wherein the execution module is further configured to hot-switch the currently used analysis algorithm to all the analysis algorithms before performing the analysis task to be performed using the target analysis algorithm. Describe the target analysis algorithm.
An analysis task execution system is characterized in that the system includes:

A management platform, configured to obtain an analysis task of a device connected to the analysis task execution system as a to-be-executed analysis task;

An analysis server, configured to execute the analysis task to be performed according to the method steps of any one of claims 1-4;

A storage server storing an algorithm component in the storage server, the storage server being configured to provide the analysis server with an algorithm component for performing the analysis task to be performed.
The system according to claim 9, wherein the algorithm components stored in the storage server are classified according to at least one of the following classification basis: supported platform architecture, bit width, processed Data type, type of target processed, type of subtask processed.
An electronic device, comprising:

Memory for storing computer programs;

The processor is configured to implement the method steps according to any one of claims 1-4 when executing a program stored in the memory.
A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method steps according to any one of claims 1-4 are implemented.