Business processing method, business processing apparatus, and computer-readable storage medium
By registering input preprocessing operators in business processing methods and creating and configuring algorithm data streams, the problem of long development cycles in different business scenarios is solved, and operator-based delivery of business and algorithms is realized, improving development efficiency and project delivery efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG DAHUA TECH CO LTD
- Filing Date
- 2022-12-30
- Publication Date
- 2026-07-24
AI Technical Summary
In different business scenarios, existing technologies need to formulate input interface protocols, output interface protocols, and data interaction interface protocols according to actual needs, resulting in long development cycles, large debugging workloads, difficulty in quickly locating and tracing problems, and reduced development efficiency and project delivery efficiency.
By registering input preprocessing operators based on business instructions, creating algorithm data streams, configuring parameters of functional operators, and using input preprocessing operators to distribute the data of the business to be processed to the algorithm data streams for processing, and obtaining the results of output postprocessing operators, the operator-based delivery of business and algorithms is realized.
It enabled customized business operations, improved development efficiency, avoided redundant development, simplified differences in business logic, and improved project delivery efficiency.
Smart Images
Figure CN116107544B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of algorithm integration and scheduling technology, and in particular to a business processing method, a business processing device, and a computer-readable storage medium. Background Technology
[0002] With the rapid development of deep learning technology and the continuous upgrading of computing capabilities in mobile and embedded devices, more and more intelligent algorithms are being implemented in production applications. Application development based on deep learning technology can realize intelligentization and digitalization in numerous scenarios, such as intelligent access control, intelligent buildings, smart transportation, smart healthcare, autonomous driving, and intelligent manufacturing, which is of great significance to accelerating my country's digitalization process.
[0003] Because different business scenarios have different requirements for algorithms, when developing intelligent solutions, it is necessary to formulate input interface protocols, output interface protocols, and data interaction interface protocols for each algorithm used according to actual needs. This results in a long development cycle, a large amount of debugging work, and difficulty in quickly and accurately locating and tracing problems when they occur, which greatly reduces development efficiency and project delivery efficiency. Summary of the Invention
[0004] This application provides a business processing method, a business processing apparatus, and a computer-readable storage medium.
[0005] This application provides a business processing method, the business processing method including:
[0006] Register input preprocessing operators based on business instructions;
[0007] An algorithm data stream is created according to the business instructions, wherein the algorithm data stream includes at least one functional operator and an output post-processing operator;
[0008] The input preprocessing operator is used to configure the configuration parameters of the functional operators in the algorithm data stream;
[0009] The business to be processed is input into the input preprocessing operator, and the processing data of the business to be processed is distributed to the algorithm data stream for business processing through the input preprocessing operator;
[0010] Obtain the business processing result output by the output post-processing operator.
[0011] The data format of the processing data of the service to be processed is a single-frame image, a multi-frame image, a single-frame custom image, an asynchronous multi-frame custom image, or a synchronous multi-frame custom image.
[0012] The business processing method further includes:
[0013] The configuration parameters of the integrated business framework and the integrated input operator are configured using the input preprocessing operator. The configuration parameters of the integrated business framework and the integrated input operator are used to set the algorithm scheme parameters of the algorithm data stream.
[0014] The step of creating an algorithm data stream according to the business instructions includes:
[0015] Several functional operators and output post-processing operators are encapsulated into an algorithm data stream according to the functional order specified by the business instruction;
[0016] Unify the data interaction interface of different functional operators in the algorithm data flow.
[0017] The input preprocessing operator registered based on business instructions includes:
[0018] The function implementation of the business is obtained based on the business instructions;
[0019] Register the function to the input preprocessing operator.
[0020] This application also provides another business processing method, which is applied to an algorithm integration and scheduling framework, the algorithm integration and scheduling framework including an integrated business framework and an integrated scheduling framework; wherein, the business processing method includes:
[0021] The integrated business framework transmits the data to be processed to the input preprocessing operator based on the business instructions, and then transmits the data to the integrated scheduling framework through the input preprocessing operator.
[0022] The integrated scheduling framework inputs the data to be processed into the algorithm data stream, and at least one functional operator in the algorithm data stream performs business processing on the data to be processed, and outputs the business processing result through the output post-processing operator.
[0023] Before the integrated business framework transmits the data to be processed to the input preprocessing operator based on business instructions, the business processing method further includes:
[0024] The integrated business framework and the input preprocessing operator have a unified registration interface through a unified data interaction protocol.
[0025] The integrated business framework uses the registration interface to register the function implementation of the business instructions.
[0026] Before the integrated scheduling framework inputs the data to be processed into the algorithm data stream, the business processing method further includes:
[0027] The integrated scheduling framework initializes the functional operator interfaces in the algorithm data stream;
[0028] The integrated scheduling framework receives the configuration parameters of each functional operator sent by the input preprocessing operator through the functional operator interface, and configures the common function commands and / or custom control commands of each functional operator according to the configuration parameters.
[0029] This application also provides a business processing apparatus, which includes a processor and a memory, wherein the memory stores program data, and the processor is used to execute the program data to implement the business processing method described above.
[0030] This application also provides a computer-readable storage medium for storing program data, which, when executed by a processor, is used to implement the above-described business processing method.
[0031] The beneficial effects of this application are as follows: the business processing device registers input preprocessing operators based on business instructions; creates an algorithm data stream according to the business instructions, wherein the algorithm data stream includes at least one functional operator and an output postprocessing operator; configures the configuration parameters of the functional operators in the algorithm data stream using the input preprocessing operators; inputs the business to be processed into the preprocessing operators, and distributes the processing data of the business to be processed to the algorithm data stream for business processing through the input preprocessing operators; and obtains the business processing results output by the output postprocessing operators. Through the above methods, the business processing device achieves customized business operations by delivering all business processes as operators, processing input preprocessing operations and output postprocessing operations into separate business operators for delivery; and achieves operatorization of both business and algorithms by delivering algorithm modules as separate functional operators. Attached Figure Description
[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:
[0033] Figure 1 This is a flowchart illustrating an embodiment of the business processing method provided in this application;
[0034] Figure 2 This is a flowchart illustrating another embodiment of the business processing method provided in this application;
[0035] Figure 3 This is a schematic diagram of an embodiment of the algorithm integration and scheduling framework provided in this application;
[0036] Figure 4This is a flowchart illustrating the lifecycle of the algorithm integration and scheduling framework provided in this application;
[0037] Figure 5 This is a schematic diagram of the frame sending mode provided in this application;
[0038] Figure 6 This is a schematic diagram of the control command response provided in this application;
[0039] Figure 7 This is a schematic diagram of the structure of an embodiment of the business processing apparatus provided in this application;
[0040] Figure 8 This is a schematic diagram of an embodiment of the computer-readable storage medium provided in this application. Detailed Implementation
[0041] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0042] This application proposes an operator-based algorithm integration method that delivers all business processes as operators. Input preprocessing operations and output postprocessing operations are processed into separate business operators for delivery. Differences in business logic are distinguished by different business processes within the input preprocessing operator, implemented through registration with the integrated business framework, thus achieving customized business operations. Algorithm modules such as target detection algorithms, target recognition algorithms, and target tracking algorithms are processed into separate functional operators for delivery, realizing the operatorization of both business processes and algorithms.
[0043] Please refer to the details. Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the business processing method provided in this application.
[0044] The business processing method of this application is applied to a business processing device, which can be a server or a system consisting of a server and terminal devices working together. Accordingly, the various parts of the business processing device, such as units, sub-units, modules, and sub-modules, can be all located in the server, or they can be located separately in the server and the terminal devices.
[0045] Furthermore, the aforementioned server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules, such as software or software modules used to provide distributed servers, or as a single software program or software module; no specific limitation is made here. In some possible implementations, the business processing method of this application embodiment can be implemented by a processor calling computer-readable instructions stored in memory.
[0046] Specifically, such as Figure 1 As shown, the business processing method in this application embodiment specifically includes the following steps:
[0047] Step S11: Register input preprocessing operators based on business instructions.
[0048] In this embodiment, the service processing device registers input preprocessing operators based on user-input service instructions. These instructions include the service processing details, functions, and processing order specified by the user. Each input preprocessing operator is a module reflecting service differences, primarily responsible for implementing service-related preprocessing operations. The input preprocessing operators use a unified data interaction protocol with the integrated service framework to unify their interface. The functions implemented by each service are registered with the integrated service framework through a unified registration interface, enabling customized service operations.
[0049] Step S12: Create an algorithm data stream according to the business instructions, wherein the algorithm data stream includes at least one functional operator and an output post-processing operator.
[0050] In this embodiment, the service processing device calls several functional operators from the operator library according to service instructions, and assembles them into an algorithm data stream according to the required service processing order. The algorithm data stream also includes an output post-processing operator at its end.
[0051] Among them, functional operators are algorithm modules that implement a certain function. For example, the target detection operator is used to detect the position of the target and output the target's trajectory bounding box information, while the target recognition operator is used to identify the target's attribute information. Different functional operators use a data interaction protocol to unify the data interaction interface, thus achieving interface reuse. The output post-processing operator is a business operator whose main function is to combine, organize, and output the algorithm results output by each functional operator. It, together with the functional operators, forms an algorithm data stream and is registered in the integrated scheduling framework.
[0052] Step S13: Configure the configuration parameters of the functional operators in the algorithm data stream using the input preprocessing operators.
[0053] In this embodiment of the application, after the algorithm data stream is created, the business processing device uses the input preprocessing operator to issue the configuration parameters required when creating each functional operator, which are used to configure the specific control commands of each functional operator. Specifically, these include common function commands such as querying memory information, setting sensitivity, setting alarm information, setting detection area, obtaining data formats supported by the algorithm, setting threshold, setting target filter box, updating rule information, updating image type, and querying rule capability set. In addition, custom control commands implemented by specific operators are also included.
[0054] In addition, the business processing unit can also use the input preprocessing operator to issue configuration parameters of the integrated business framework and the integrated input operator, which are used to configure the algorithm scheme parameters of the algorithm data stream. These parameters mainly include setting the log printing level, obtaining the maximum number of channels supported by the algorithm scheme, obtaining the asynchronous depth, clearing the cache, obtaining the maximum number of channels of the algorithm data stream, binding and unbinding channels, and enabling channels.
[0055] Step S14: Input the business data to be processed into the preprocessing operator, and distribute the processing data of the business data to the algorithm data stream for business processing through the input preprocessing operator.
[0056] In this embodiment of the application, the business processing device inputs the processing data of the business to be processed into the input preprocessing operator, which performs input checks, then distributes the data and sends it into the algorithm data stream. The data is then processed by the various functional operators on the algorithm data stream, and finally the final processing result is obtained in the output postprocessing operator.
[0057] In this application embodiment, the processed data can be divided into single-frame images, multi-frame images, single-frame custom images, asynchronous multi-frame custom images, or synchronous multi-frame custom images, depending on the data format sent.
[0058] Step S15: Obtain the business processing results output by the output post-processing operator.
[0059] In this embodiment, the business processing device registers input preprocessing operators based on business instructions; creates an algorithm data stream according to the business instructions, wherein the algorithm data stream includes at least one functional operator and an output postprocessing operator; configures the configuration parameters of the functional operators in the algorithm data stream using the input preprocessing operators; inputs the business to be processed into the preprocessing operators, and distributes the processing data of the business to be processed to the algorithm data stream for business processing through the input preprocessing operators; and obtains the business processing results output by the output postprocessing operators. Through the above method, the business processing device achieves customized business operations by delivering all business processes as operators, processing input preprocessing operations and output postprocessing operations into separate business operators for delivery; and achieves operatorization of both business and algorithms by delivering algorithm modules as separate functional operators.
[0060] Furthermore, this application proposes an algorithm integration and scheduling framework based on business decoupling, including an integration business module and an integration scheduling module. The integration business module abstracts the common logic of the integrated business and unifies the input process and intelligent analysis results of different integrated businesses. The integration scheduling module distributes input data, schedules the running process, and processes the results of the functional operators. This achieves business decoupling, shields business differences, avoids redundant development, and improves development efficiency.
[0061] Please refer to the details. Figure 2 and Figure 3 , Figure 2 This is a flowchart illustrating another embodiment of the business processing method provided in this application. Figure 3 This is a schematic diagram of an embodiment of the algorithm integration and scheduling framework provided in this application.
[0062] like Figure 3 As shown, an executable program is a program file that runs on a hardware device. It obtains the processing results of the algorithm module by calling the algorithm integration and scheduling framework, and is used to implement specific software functions, such as traffic detection, smoke alarm, intelligent capture and other specific business functions.
[0063] The algorithm integration and scheduling framework is used for algorithm integration and scheduling, and consists of two parts: an integration business framework and an integration scheduling framework. The main function of the integration business framework is to abstract the common logic of integrated business processes, unify the input flow and intelligent analysis results of different integrated business processes, and enable different business operators to use a unified application programming interface. The main function of the integration scheduling framework is to encapsulate the scheduling unit of business operators, provide queue management and other encapsulations for business operators, integrate algorithms with different functions, combine them into an algorithm data stream in a certain order, unify the data interaction interface between different operators, and achieve interface reuse.
[0064] Input preprocessing operators are modules that reflect business differences. Their main function is to implement business-related preprocessing operations. Input preprocessing operators unify their interfaces through a standardized data interaction protocol with the integrated business framework. Each business function is registered with the integrated business framework through a unified registration interface, enabling customized business operations. Functional operators are algorithm modules that implement specific functions. For example, a target detection operator detects the location of a target and outputs its bounding box information; a target recognition operator identifies the target's attribute information. Different functional operators unify their interfaces through a data interaction protocol, achieving interface reuse. Output postprocessing operators are business operators whose main function is to combine, organize, and output the algorithm results from the functional operators. Together with the functional operators, they form an algorithm data stream and are registered with the integrated scheduling framework.
[0065] Specifically, such as Figure 2 As shown, the business processing method in this application embodiment specifically includes the following steps:
[0066] Step S21: The integrated business framework transmits the data to be processed to the input preprocessing operator based on the business instructions, and then transmits the data to the integrated scheduling framework through the input preprocessing operator.
[0067] In this embodiment, the function of the algorithm integration and scheduling framework is to standardize and normalize the data processing flow of the algorithm, according to the time sequence of data flow, such as... Figure 4 As shown, Figure 4 This is a flowchart illustrating the lifecycle of the algorithm integration and scheduling framework provided in this application. The operation of the algorithm integration and scheduling framework is divided into six processes: initialization, creation, data processing, process control, destruction, and deinitialization.
[0068] The following sections will elaborate on each stage according to its life cycle:
[0069] initialization
[0070] The initialization phase includes the following two operations:
[0071] (1) Obtain information such as the compilation time and version of the current code project.
[0072] (2) Initialize the Software Development Kit (SDK).
[0073] create
[0074] The creation phase includes checking creation parameters, initializing algorithm data streams and creating operators, creating and initializing integrated business operators, creating and initializing channel-independent operators, verifying operator header file versions and obtaining algorithm versions, verifying header file versions of the integrated scheduling framework, initializing debugging information, creating and initializing synchronization processing semaphores, creating memory handles, registering result callback functions, registering external printing functions, and registering exception callback functions.
[0075] Specifically, the following steps are included:
[0076] (1) The executable program calls the initialization interface of the integrated business framework to start the initialization and creation of each module.
[0077] (2) The executable program begins to register relevant callback functions, such as registering result callback functions, registering input data release callback functions, registering external print function callback functions, and registering exception callback functions.
[0078] (3) Input preprocessing operators are registered within the integrated business framework.
[0079] (4) The algorithm data stream composed of each functional operator and the output post-processing operator is registered inside the integrated scheduling framework.
[0080] (5) Input the preprocessing operator and send the configuration parameters required for the creation of each functional operator to the integrated scheduling framework.
[0081] (6) The integrated scheduling framework calls the initialization creation interface of each functional operator and passes in the creation configuration parameters.
[0082] (7) Each functional operator completes initialization and creation.
[0083] Data processing
[0084] The main functions of the data processing stage are sending data and obtaining results, primarily involving the data processing flow and related operations during the runtime phase. During runtime, the executable program calls the frame sending interface of the integrated business framework to send data into the input preprocessing operator. The input preprocessing operator performs input checks, then distributes the data and sends it to the integrated scheduling framework.
[0085] Step S22: The integrated scheduling framework inputs the data to be processed into the algorithm data stream, and at least one functional operator in the algorithm data stream performs business processing on the data to be processed, and outputs the business processing result through the output post-processing operator.
[0086] like Figure 4 As shown, the data processing stage also includes: the data being processed by various functional operators on the algorithm data stream, and finally the final processing result is obtained in the output post-processing operator.
[0087] Specifically, the executable program calls the frame sending interface to send the data to be processed into the algorithm data stream, and finally tells the executable program the final business processing result after the data to be processed by each operator through the result callback function.
[0088] Furthermore, the data to be processed varies depending on the data format sent by the executable program, such as... Figure 5 As shown, they can be divided into the following 5 patterns:
[0089] (1) Single frame image
[0090] When the input data is a single-frame image, it means that the frame processes one image at a time. Asynchronous means that sending data and receiving the result are independent of each other; that is, the second data transmission does not need to wait for the result of the first data transmission before transmission.
[0091] (2) Multi-frame images
[0092] When the input data consists of multiple frames of images, it means that the frame processes multiple images at a time, i.e., batch processing. In this case, the functional operators in the algorithm's data stream need to support batch processing operations.
[0093] (3) Single-frame custom image
[0094] When the input data is a single frame of custom data, it means that the framework processes one frame of custom data for an executable program at a time.
[0095] (4) Asynchronous multi-frame custom images
[0096] When the input data is multi-frame custom data, it means that the framework processes multiple frames of executable program custom data each time, and the processing mode is asynchronous.
[0097] (5) Synchronize multiple frames of custom images
[0098] When the input data is multi-frame custom data, it means that the framework processes multiple frames of executable program custom data each time, and the processing mode is synchronous.
[0099] like Figure 4 As shown, the lifecycle of the algorithm integration and scheduling framework also includes:
[0100] Process control
[0101] The main function of the process control phase is to issue specific configuration parameters, which are then distributed to specific operators through the business scheduling framework. The configuration parameters take the form of a series of control commands, which vary depending on the responder, such as... Figure 6 As shown, they are mainly divided into two categories:
[0102] (1) Integrating business framework and integrating input operators
[0103] These control commands mainly include setting the log level, obtaining the maximum number of channels supported by the algorithm scheme, obtaining the asynchronous depth, clearing the cache, obtaining the maximum number of channels for the algorithm data stream, binding and unbinding channels, and enabling channels, etc.
[0104] (2) Specific functional operators
[0105] This type of command includes common function commands such as querying memory information, setting sensitivity, setting alarm information, setting detection area, obtaining data formats supported by the algorithm, setting threshold, setting target filter box, updating rule information, updating image type, and querying rule capability set. In addition, it also includes custom control commands implemented by specific operators.
[0106] destroy
[0107] The main operations in the destruction phase include deleting channels, clearing historical data, deinitializing algorithm data streams, destroying integrated business operators, destroying channel-independent operator handles, and deinitializing synchronization processing semaphores.
[0108] Inverse initialization
[0109] The deinitialization phase involves performing deinitialization operations on the Software Development Kit (SDK).
[0110] In this application embodiment, an algorithm integration and scheduling framework based on business decoupling is proposed, including an integration business module and an integration scheduling module. The integration business module abstracts the common logic of the integrated business and unifies the input process and intelligent analysis results of different integrated businesses. The integration scheduling module distributes input data, schedules the running process, and processes the results of the functional operators. This achieves business decoupling, shields business differences, avoids redundant development, and improves development efficiency.
[0111] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.
[0112] To implement the business processing method of the above embodiments, this application also proposes a business processing apparatus, which can be found in detail below. Figure 7 , Figure 7 This is a schematic diagram of the structure of an embodiment of the business processing apparatus provided in this application.
[0113] The service processing apparatus 500 of this application embodiment includes a memory 51 and a processor 52, wherein the memory 51 and the processor 52 are coupled together.
[0114] The memory 51 is used to store program data, and the processor 52 is used to execute the program data to implement the business processing method described in the above embodiments.
[0115] In this embodiment, processor 52 can also be referred to as a CPU (Central Processing Unit). Processor 52 may be an integrated circuit chip with signal processing capabilities. Processor 52 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor, or processor 52 can be any conventional processor.
[0116] To implement the business processing method of the above embodiments, this application also provides a computer-readable storage medium, such as... Figure 8As shown, the computer-readable storage medium 600 is used to store program data 61, which, when executed by a processor, is used to implement the business processing method as described in the above embodiments.
[0117] This application also provides a computer program product, wherein the computer program product includes a computer program operable to cause a computer to perform the business processing method as described in the embodiments of this application. The computer program product may be a software installation package.
[0118] The business processing methods described in the above embodiments of this application, when implemented as software functional units and sold or used as independent products, can be stored in a device, such as a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0119] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A service processing method, characterized in that, The business processing method includes: Registering an input preprocessing operator based on a business instruction; wherein, the input preprocessing operator is an operator module reflecting business differences, and a unified data interaction protocol is formulated between the input preprocessing operator and the integrated business framework for a unified interface; the function functions implemented by each business are registered into the integrated business framework through a unified registration interface to implement business customization operations; Creating an algorithm data stream according to the business instruction, wherein the algorithm data stream includes at least one functional operator and an output postprocessing operator; wherein, the functional operator is an algorithm module implementing a specific function, and the data interaction interfaces of different functional operators adopt the data interaction protocol for unified interface to achieve interface reuse; the output postprocessing operator is a business operator used to combine, organize, and output the algorithm results output by each functional operator, and the output postprocessing operator and the functional operator form an algorithm data stream and are registered into the integrated scheduling framework; Configuring the configuration parameters of the functional operators in the algorithm data stream by using the input preprocessing operator; Inputting the business to be processed into the input preprocessing operator, and distributing the processing data of the business to be processed to the algorithm data stream through the input preprocessing operator for business processing; Obtaining the business processing result output by the output postprocessing operator; The creating an algorithm data stream according to the business instruction includes: Encapsulating a plurality of functional operators and an output postprocessing operator into an algorithm data stream according to the functional order specified by the business instruction; Unifying the data interaction interfaces of different functional operators in the algorithm data stream; The business processing method further includes: Configuring the configuration parameters of the integrated business framework and the integrated input operator by using the input preprocessing operator, and the configuration parameters of the integrated business framework and the integrated input operator are used to set the algorithm scheme parameters of the algorithm data stream.
2. The business processing method according to claim 1, wherein The data format of the processing data of the business to be processed is a single-frame image, a multi-frame image, a single-frame custom image, an asynchronous multi-frame custom image, or a synchronous multi-frame custom image.
3. The business processing method according to claim 1, wherein The registering an input preprocessing operator based on a business instruction includes: Obtaining the function function implemented by the business based on the business instruction; Registering the function function into the input preprocessing operator.
4. A service processing method, characterized in that The business processing method is applied to an algorithm integration and scheduling framework, and the algorithm integration and scheduling framework includes an integrated business framework and an integrated scheduling framework; wherein, the business processing method includes: The integrated business framework transmits the data to be processed to the input preprocessing operator based on a business instruction, and transmits the data to be processed to the integrated scheduling framework through the input preprocessing operator; wherein, the input preprocessing operator is an operator module reflecting business differences, and a unified data interaction protocol is formulated between the input preprocessing operator and the integrated business framework for a unified interface; the function functions implemented by each business are registered into the integrated business framework through a unified registration interface to implement business customization operations; The integrated scheduling framework inputs the to-be-processed data into the algorithm data stream. At least one functional operator in the algorithm data stream performs business processing on the to-be-processed data, and outputs the business processing result through the output post-processing operator. Among them, the functional operator is an algorithm module that implements a specific function. The data interaction protocol is used to unify the interfaces between different functional operators to achieve interface reuse. The output post-processing operator is a business operator, which is used to combine, organize, and output the algorithm results output by each functional operator. The output post-processing operator and the functional operator form an algorithm data stream and are registered in the integrated scheduling framework. Among them, the algorithm data stream is created according to the service instruction. By encapsulating several functional operators and the output post-processing operator in the function order specified by the service instruction into an algorithm data stream and unifying the data interaction interfaces of different functional operators in the algorithm data stream. The integrated scheduling framework configures the configuration parameters of the integrated service framework and the integrated input operator by using the input preprocessing operator. The configuration parameters of the integrated service framework and the integrated input operator are used to set the algorithm scheme parameters of the algorithm data stream.
5. The service processing method according to claim 4, wherein Before the integrated service framework transfers the to-be-processed data to the input preprocessing operator based on the service instruction, the service processing method further includes: A unified registration interface is unified between the integrated service framework and the input preprocessing operator by formulating a unified data interaction protocol. The integrated service framework uses the registration interface to register the function functions implemented by the service instruction.
6. The service processing method according to claim 4, wherein Before the integrated scheduling framework inputs the to-be-processed data into the algorithm data stream, the service processing method further includes: The integrated scheduling framework initializes the functional operator interfaces in the algorithm data stream. The integrated scheduling framework receives the configuration parameters of each functional operator sent by the input preprocessing operator through the functional operator interfaces, and configures the common function commands and / or custom control commands of each functional operator according to the configuration parameters.
7. A service processing device, characterized in that, The service processing device includes a processor and a memory. Program data is stored in the memory. The processor is used to execute the program data to implement the service processing method according to any one of claims 1 to 6.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program data. When the program data is executed by a processor, it is used to implement the service processing method according to any one of claims 1 to 6.