Intelligent solution operation method, device, electronic device and storage medium

Abstract

The present application relates to a method, device, electronic device and storage medium for running an intelligent solution, wherein the method for running an intelligent solution includes: obtaining each operator and each edge information in the intelligent solution by parsing the topological structure of the intelligent solution, wherein the edge information is the connection relationship between each operator in the intelligent solution; performing deduplication processing on each operator to obtain a deduplicated operator set, wherein each operator in the deduplicated operator set is identified as each target operator; loading each target operator; performing deduplication and integration processing on each edge information to obtain target edge information; each target operator and target edge information constitutes a target topological structure; running the intelligent solution according to the target topological structure to obtain an operation result. Through the present application, the problem of low computing resource utilization is solved and the computing resource utilization is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to intelligent solution operation methods, devices, electronic devices and storage media. Background Art

[0002] With the continuous advancement of science and technology, intelligent systems are attracting more and more attention. In the field of video surveillance, the deployment and application of intelligent solutions have changed from single functional modules to integrated systems that can meet the diverse and complex needs of users. In this context, intelligent solutions are delivered to users in the form of a single module, allowing users to select and deploy different modules according to their specific needs. However, with the continuous improvement of user needs, the same device often needs to deploy multiple intelligent solutions to achieve more advanced functions and better user experience. Then, the intelligent solution is delivered to the user in the form of a single module. When the user needs to dynamically turn it on or off, the module needs to be loaded or unloaded at the granularity of the entire intelligent solution. In addition, on embedded devices, due to limited resources, it is often difficult to turn on multiple intelligent solutions at the same time, or there is performance loss. At the intelligent solution level, optimization is currently mainly carried out through the following two methods:

[0003] 1. Time-sharing reuse of multiple schemes: Assuming that only one scheme is enabled at a time, when three schemes need to be run at the same time, each intelligent scheme can only be allocated 1 / 3 of the processing time and cannot run in other time periods.

[0004] 2. Reduce the resolution or bit rate of the video frame to increase the processing speed of a single operator.

[0005] However, in the currently adopted solution 1, the running time allocated to each intelligent solution is short, which affects the accuracy of operator processing; in solution 2, the resolution or bit rate of the video frame is reduced, which affects the accuracy of operator processing, resulting in false alarms. In addition, the intelligent solution is composed of operators, and is currently loaded at the granularity of the intelligent solution. The common operators in multiple intelligent solutions will be repeatedly loaded into the memory, resulting in a waste of memory. In addition, if the same frame is input into two intelligent solutions respectively, the common operators in the two intelligent solutions will also be repeatedly run, which will also cause a waste of processor resources.

[0006] There is currently no effective solution to the problem of low computing resource utilization in related technologies. Summary of the invention

[0007] In this embodiment, a smart solution operation method, device, electronic device and storage medium are provided to solve the problem of low computing resource utilization in related technologies.

[0008] In a first aspect, a smart solution operation method is provided in this embodiment, including:

[0009] Analyze the topological structure of the intelligent solution to obtain information about each operator and each edge in the intelligent solution, where the edge information is the connection relationship between the operators in the intelligent solution;

[0010] De-duplication processing is performed on each operator to obtain a de-duplication set of operators, and each operator in the de-duplication set of operators is identified as each target operator;

[0011] Load each target operator;

[0012] De-duplicate and integrate each side information to obtain the target side information;

[0013] Each target operator and target edge information constitute the target topology structure;

[0014] Run the intelligent solution according to the target topology and obtain the operation results.

[0015] In some of the embodiments, the topology structure of the intelligent solution also includes configuration information of each operator, and loading each target operator includes:

[0016] According to the different configuration information required by each operator in the intelligent solution, the corresponding configuration information is loaded into each target operator.

[0017] In some of the embodiments, the topological structure of the intelligent solution also includes a label of the intelligent solution, and each target operator corresponding to each configuration information is distinguished by marking the label of the intelligent solution.

[0018] In some of the embodiments, each side information is deduplicated and integrated to obtain target side information, including:

[0019] Compare the same operator nodes in each edge information, retain the operators with different configuration information, and obtain the deduplicated edge information;

[0020] Perform topological sorting on the deduplicated edge information to obtain the target edge information.

[0021] In some of the embodiments, the topological structure of the intelligent solution further includes a label of the intelligent solution; and the same operators in each side information are compared, and operators with different configuration information are retained to obtain the side information after deduplication, including:

[0022] Compare operators of the same node in each edge information and retain operators with different configuration information;

[0023] The retained operators are marked with labels of the intelligent solutions corresponding to the retained operators to obtain the deduplicated side information.

[0024] In some of the embodiments, the intelligent solution is run according to the target topology structure to obtain the running results, including:

[0025] When running an intelligent solution according to the target topology, when multiple different operator operation results are imported into the same operator node in the intelligent solution, different intermediate results are calculated in the operator node according to the imported different operator operation results;

[0026] Different intermediate results are transmitted to corresponding downstream nodes according to the labels of their respective marked intelligent solutions until the operation is completed to obtain the operation results; wherein the operation results include the labels of the intelligent solutions.

[0027] In some of the embodiments, in response to a new request for a new smart solution, idle CPU time is used to reparse and load new operators, generate a new topology corresponding to the new smart solution, and run the new smart solution according to the new topology;

[0028] In response to the deletion request of the intelligent solution, the idle CPU time is used to unload the unnecessary target operators and target edge information, generate a new topology structure, and run the deleted intelligent solution according to the new topology result.

[0029] In the second aspect, in this embodiment, an intelligent solution operation device is provided, including: a parsing module, a loading module, an integration module and an operation module, wherein:

[0030] A parsing module is used to parse the topological structure of the intelligent solution and obtain information about each operator and each edge in the intelligent solution, wherein the edge information is the connection relationship between the operators in the intelligent solution;

[0031] A loading module is used to perform deduplication processing on each operator to obtain a deduplication set of operators; each operator in the deduplication set of operators is identified as a target operator; and each target operator is loaded;

[0032] The integration module is used to perform deduplication and integration processing on each edge information to obtain the target edge information; each target operator and the target edge information form a target topological structure;

[0033] The running module is used to run the intelligent solution according to the target topology structure and obtain the running result.

[0034] In a third aspect, an electronic device is provided in this embodiment, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intelligent solution operation method described in the first aspect is implemented.

[0035] In a fourth aspect, a storage medium is provided in this embodiment, on which a computer program is stored, and when the program is executed by a processor, the intelligent solution operation method described in the first aspect is implemented.

[0036] Compared with the related art, the intelligent solution operation method provided in this embodiment obtains each operator and each edge information in the intelligent solution by parsing the topological structure of the intelligent solution, wherein the edge information is the connection relationship between each operator in the intelligent solution; deduplicates each operator to obtain a deduplicated operator set, and each operator in the deduplicated operator set is identified as a target operator; loads each target operator; deduplicates and integrates each edge information to obtain target edge information; each target operator and target edge information constitutes a target topological structure; runs the intelligent solution according to the target topological structure to obtain an operation result, thereby solving the problem of low computing resource utilization and improving computing resource utilization.

[0037] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 It is a hardware structure block diagram of the terminal of the intelligent solution operation method of this embodiment.

[0040] Figure 2 It is a flow chart of the intelligent solution operation method of this embodiment.

[0041] Figure 3 It is a schematic diagram of smart solution 1 of the smart solution operation method in this embodiment.

[0042] Figure 4 It is a schematic diagram of smart solution 2 of the smart solution operation method in this embodiment.

[0043] Figure 5 This is an example diagram of the side information after the smart solution 1 and the smart solution 2 are integrated by the smart solution operation method in this embodiment.

[0044] Figure 6 It is a flow chart of another intelligent solution operation method of this embodiment.

[0045] Figure 7 It is a structural block diagram of the intelligent solution running device of this embodiment. DETAILED DESCRIPTION

[0046] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0047] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with general skills in the technical field to which this application belongs. The words "one", "a", "the", "these" and the like in this application do not indicate a quantitative limitation, and they may be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone. Generally, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0048] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 is a hardware structure diagram of a terminal of the intelligent solution operation method of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0049] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the intelligent solution operation method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0050] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0051] In this embodiment, a smart solution operation method is provided. Figure 2 is a flow chart of the intelligent solution operation method of this embodiment. Figure 2 As shown, the process includes the following steps:

[0052] Step S201 , analyzing the topological structure of the smart solution to obtain each operator and each edge information in the smart solution, wherein the edge information is the connection relationship between each operator in the smart solution.

[0053] Specifically, operators provide a single algorithm function, such as target detection, target tracking, etc. Operators may need to rely on some machine learning or deep learning models, such as YOLO or ViT networks used for target detection, or they may not need to use models and rely on internal logic to achieve their functions. Each operator has its own input and output requirements. The same operator can be run with different configurations (models are also considered as a configuration) to achieve different functions. Intelligent solutions are composed of multiple operators, providing a complete solution, such as passenger flow statistics, behavior analysis, etc. Multiple operators form a directed acyclic graph (DAG), in which operators are nodes and edge information represents the flow of data. The input of the entire graph is video frames, and the output is the running result. The topological structure of the graph composed of operators is represented by a file, which can be in a format such as protobuf or json, recorded as a graph file. The graph file records the type and configuration information of each operator, as well as the relationship between operators (edge ​​information). Modules are code units that can be dynamically loaded and run when the program is running. They are generally dynamic link libraries (.so format files under Linux and .dll format files under Windows), and can also be dynamically unloaded. At present, the intelligent solution is delivered as a module with the entire intelligent solution. With the improvement of user demand, the same device often needs to deploy multiple intelligent solutions, and dynamically open or close according to user demand. At this time, it is necessary to load or unload with the entire intelligent solution as the granularity. The intelligent solution itself is composed of operators. When the entire intelligent solution is loaded or unloaded, the same operator will be repeatedly loaded into the memory, resulting in a waste of memory. In addition, the common operators in different intelligent solutions will also be repeatedly run, resulting in a waste of computing resources. Therefore, in this embodiment, considering that there are often common operators or models between multiple intelligent solutions, each operator in the intelligent solution is compiled into a separate module for direct delivery. For example, target detection is the first operator of the intelligent solution, which can be used by a large number of intelligent solutions. The intelligent solution operation method in this embodiment can be a single intelligent solution or multiple intelligent solutions running simultaneously. Before running, a scheduler module is first loaded. When each intelligent solution is delivered, the graph file of the intelligent solution itself, that is, the topological structure of the intelligent solution itself, will be delivered. When the user runs a single intelligent solution, the graph file of the intelligent solution is parsed by the scheduler module, and the scheduler module directly uses the parsed graph file to run the operator once.

[0054] When the user selects multiple intelligent solutions to run simultaneously, the scheduler module parses the graph file of each intelligent solution to obtain the operators and edge information used in each intelligent solution. The edge information refers to the connection relationship between the operators in an intelligent solution. The edge information can be used to determine the operator running order of the intelligent solution, thereby determining the operator type required by each intelligent solution and the relationship diagram between the operators in each intelligent solution. For example, Figure 3 FIG. 1 is a schematic diagram of the intelligent solution 1 of the intelligent solution operation method in this embodiment. Figure 3 As shown in the figure, in smart solution 1, the operators used are: A, B1, C, D, E, and their edge information is recorded as: [1,A,B1], [1,B1,D], [1,B1,C], [1,D,E], [1,C,E], where 1 represents the number (label) of smart solution 1, which is used to distinguish that the above operators belong to smart solution 1, and [A,B1] represents the edge information from operator A to operator B1. Figure 4 Schematic diagram of intelligent solution 2 of the intelligent solution operation method in this embodiment, such as Figure 4 As shown in the figure, in smart solution 2, the operators used are: A, B2, D, E, and their edge information is recorded as: [2, A, B2], [2, B2, D], [2, D, E], where 2 represents the number (label) of smart solution 2, which is used to distinguish that the above operators belong to smart solution 2, and [A, B2] represents the edge information from operator A to operator B2. Smart solution 1 and smart solution 2 are parsed by the scheduler module, and the operators in the two smart solutions include: A, B1, C, D, E; A, B2, D, E, and the edge information in the two smart solutions includes: [1, A, B1], [1, B1, D], [1, B1, C], [1, D, E], [1, C, E]; [2, A, B2], [2, B2, D], [2, D, E].

[0055] Step S202, performing deduplication processing on each operator to obtain a deduplicated operator set, each operator in the deduplicated operator set is identified as a target operator; and each target operator is loaded.

[0056] Specifically, according to the operators obtained by parsing in step S201, if there are operators of the same type, such as operator A in both smart solution 1 and smart solution 2, different configuration numbers and solution numbers are ignored, and each operator is deduplicated to obtain a deduplicated operator set. Figure 3 and Figure 4 After deduplication of operators in the table, an operator set (A, B, C, D, E) is obtained. Based on the deduplication process, the operator types required for each intelligent solution are determined, that is, the target operators: A, B, C, D, E. The scheduler loads these target operators A, B, C, D, E.

[0057] Step S203, deduplicate and integrate each edge information to obtain target edge information; each target operator and target edge information form a target topological structure.

[0058] Specifically, according to the side information obtained by parsing, the same side information exists in the side information of different intelligent solutions, for example, the side information [1, D, E] in intelligent solution 1 and the side information [2, D, E] in intelligent solution 2 are the same. In this embodiment, in order to improve the utilization of computing resources, the same operators and the same side information are merged and deduplicated to avoid repeated loading, thereby improving the utilization of computing resources. Therefore, according to the above method, each side information is deduplicated, and after deduplication, the side information of the two intelligent solutions is integrated to obtain the target side information. Figure 5 : is an example diagram of the side information after the intelligent solution operation method in this embodiment integrates the intelligent solution 1 and the intelligent solution 2. Figure 5 As shown in the figure, for the edge information in smart solution 1 and the edge information in smart solution 2, the target edge information is obtained: [A, B1], [A, B2], [B1, D], [B2, D], [B1, C], [D, E], [C, E]. Since the common operators will be shared, a number label of the smart solution to which the operator belongs is added to each operator node, as shown in the figure. Figure 5 The A operator in the diagram will be used by both Smart Solution 1 and Smart Solution 2, and (1,2) will be marked on the A operator. The B1 operator will only be used by Smart Solution 1, and only 1 will be marked on the B1 operator to distinguish the operator calls on each operator node. The target operator loaded by the integrated scheduler module and the target edge information after deduplication and integration are obtained to obtain the graph file, that is, the target topology structure.

[0059] Step S204, running the intelligent solution according to the target topology structure to obtain the running result.

[0060] Specifically, the scheduler module runs each operator according to the generated new graph file. The scheduler module is responsible for ensuring that data flows correctly and timely between operators. The scheduler module will pass the input frame to the first operator node in the graph and perform calculations. When the predecessor operators of an operator are all calculated, the scheduler will schedule the operator and use the output of the upstream operator as the input of the current operator. When all operators in the new graph file are calculated, the running results of the current frame are output.

[0061] Through the above steps S201 to S204, the topological structure of the intelligent solution is parsed to obtain each operator and each edge information in the intelligent solution, wherein the edge information is the connection relationship between each operator in the intelligent solution; each operator is deduplicated to obtain a deduplicated operator set, and each operator in the deduplicated operator set is identified as each target operator; each target operator is loaded; each edge information is deduplicated and integrated to obtain target edge information; each target operator and target edge information constitutes a target topological structure; the intelligent solution is run according to the target topological structure to obtain the running result. Compared with the existing related technology that directly treats the entire intelligent solution as a module and delivers it by reducing the video frame resolution or bit rate, this embodiment parses the topological structure of each intelligent solution, extracts all operators and edge information, deduplicates both operators and edge information, obtains a new topological structure, and reduces memory usage by not repeatedly loading the same operator and edge information, improves the utilization rate of operators, and improves the utilization rate of computing resources.

[0062] In some of the embodiments, the topology structure of the intelligent solution also includes configuration information of each operator, and loading each target operator includes:

[0063] According to the different configuration information required by each operator in the intelligent solution, the corresponding configuration information is loaded into each target operator.

[0064] Specifically, the topological structure of the intelligent scheme also includes the configuration information of each operator, wherein the operator and its configuration information can be independent of each other, and one operator can be configured with multiple configuration information. In this embodiment, the operator type required for the operation of each intelligent scheme is obtained by deduplicating the operators, wherein the same operator has different configuration information in different intelligent schemes, and executes different operation programs. According to the configuration information required by each operator in each intelligent scheme, the corresponding configuration information is loaded into the deduplicated operator. Figure 3 and Figure 4 As shown, Figure 3 Case 1 Neutralization Figure 4 The B operator is used in Solution 2. At this time, it is only necessary to load the B operator once, and then configure the corresponding configuration information for the loaded B operator according to the configuration information required by the B operator in Solution 1, and record the B operator configured with the configuration information of Solution 1 as the B1 operator; similarly, according to the configuration information required by the B operator in Solution 2, configure the corresponding configuration information for the loaded B operator, and record the B operator configured with the configuration information of Solution 2 as the B2 operator. Among them, marking the B operator as B1 operator and B2 operator can effectively distinguish the different configuration information in the B operator, making it convenient for the intelligent solution to call and use it according to its own label during operation. In this process, the same operator can be loaded only once, and the operation requirements of different intelligent solutions can be met by loading different configuration information for the same operator.

[0065] In another embodiment, the topological structure of the intelligent solution further includes a label of the intelligent solution, and each target operator corresponding to each configuration information is distinguished by marking the label of the intelligent solution.

[0066] Specifically, when multiple smart solutions need to run at the same time, a label is set for each smart solution to distinguish different smart solutions. When loading the target operator, the corresponding operator is loaded according to different configuration information. The configuration information contains the label of the smart solution. The smart solution label corresponding to the configuration information is marked on the corresponding target operator to distinguish different target operators, making it easier for the scheduler module to call and use the operator according to the label, ensuring the smooth operation of the operator in the smart solution. Figure 5 As shown, the B operator is marked, the B operator used in the smart solution 1 is marked as B1, and the B operator in the smart solution 2 is marked as B2.

[0067] In some embodiments, each side information is deduplicated and integrated to obtain target side information, including:

[0068] The same operator nodes in each edge information are compared, and operators with different configuration information are retained to obtain the deduplicated edge information; the deduplicated edge information is topologically sorted to obtain the target edge information.

[0069] Specifically, in order to reduce memory usage, when multiple intelligent solutions are running at the same time, all the edge information obtained by parsing is sorted. Among them, the same operator nodes in different intelligent solutions are compared to see whether the configuration information of the called operators is the same. Figure 3 and Figure 4 As shown in the figure, the operator nodes where the A operator is located all call the A operator with the same configuration, so one A operator is removed and only one A operator is retained in the operator node; in the B operator node, scheme 1 and scheme 2 use B operators with different configuration information, so both B1 operator and B2 operator are retained. Similarly, the same processing is performed on other operator nodes, and finally the following is obtained: Figure 5 The operators shown in Figure 3 and Figure 4 The relationship between operators, i.e., edge information, is integrated, such as Figure 3 and Figure 4 There is an edge from operator A to operator B in both, but the two B operators have different configurations, so we retain two different edges from operator A to operator B [A, B1] and [A, B2], and finally get Figure 5 The target side information is shown.

[0070] In another embodiment, the topological structure of the intelligent solution further includes a label of the intelligent solution; comparing the same operators in each side information, retaining operators with different configuration information, and obtaining the side information after deduplication, including:

[0071] The operators of the same node in each edge information are compared, and the operators with different configuration information are retained; the retained operators are marked with labels of the intelligent solutions corresponding to the retained operators, and the deduplicated edge information is obtained.

[0072] Specifically, in the above embodiment, after operators with different configuration information are retained through comparison of the same operator nodes, in order to distinguish the intelligent solutions corresponding to the operators of the same type retained, the operators corresponding to the different configuration information are also marked with the labels of the corresponding intelligent solutions according to the labels of different intelligent solutions attached to the different configuration information, such as Figure 5 The B operator in smart solution 1 is marked as B1, and the B operator in smart solution 2 is marked as B2, so as to obtain the deduplicated side information [A, B1], [A, B2], [B1, D], [B2, D], [B1, C], [D, E], [C, E].

[0073] In some embodiments, the intelligent solution is run according to the target topology structure to obtain the running results, including:

[0074] When running an intelligent solution according to a target topology, when multiple different operator calculation results in the intelligent solution are imported into the same operator node, different intermediate results are calculated in the operator node according to the different imported operator calculation results; the different intermediate results are transmitted to the corresponding downstream nodes according to the labels of the intelligent solutions they are marked with, until the operation is completed to obtain the operation result; wherein, the operation result includes the label of the intelligent solution.

[0075] Specifically, when running each intelligent solution according to the newly generated target topology, all intermediate results are also marked with labels corresponding to the intelligent solution to distinguish the intelligent solution to which the intermediate result belongs. Figure 5When the intermediate results calculated by the B1 operator and the B2 operator are imported into the D operator node, the D operator calculates the intermediate results of the intelligent solution 1 at the D operator node based on the intermediate results calculated by the imported B1 operator, and marks the label 1 of the solution 1 on the intermediate results, and passes the marked intermediate results downstream to the next operator node marked with the solution 1 label 1, and the operator of the next operator node calculates the intermediate results again; similarly, the D operator calculates the intermediate results of the intelligent solution 2 at the D operator node based on the intermediate results calculated by the imported B2 operator, and marks the label 2 of the solution 2 on the intermediate results, and passes the marked intermediate results downstream to the next operator node marked with the solution 2 label 2, and the operator of the next operator node calculates the intermediate results again. This process is repeated until all operators are processed and the operation results of each intelligent solution are obtained, wherein the operation results contain the labels of the intelligent solutions. In this process, if the results calculated by the same operator need to flow to different operators, the intermediate results calculated by the operator need to be copied and passed to different downstream nodes respectively, such as Figure 5 As shown in the figure, the same A operator is used in both Scheme 1 and Scheme 2. After the A operator calculates the intermediate result, it needs to copy the intermediate result to obtain the intermediate results of two A operators, and then pass the intermediate results of the two A operators to the B1 operator and the B2 operator respectively. By marking the corresponding smart scheme label on each intermediate result, when each smart scheme is running, the intermediate results calculated by each smart scheme will not flow to the operator nodes of other smart schemes, ensuring the accuracy of the operation of the smart scheme.

[0076] In another embodiment, in response to a request to add a new intelligent solution, the idle CPU time is used to re-parse and load new operators, generate a new topology corresponding to the new intelligent solution, and run the new intelligent solution according to the new topology; in response to a request to delete an intelligent solution, the idle CPU time is used to unload unnecessary target operators and target edge information, generate a new topology, and run the deleted intelligent solution according to the new topology result.

[0077] Specifically, when a user needs to add a new intelligent solution, the scheduler module uses the available CPU time to parse the topological structure in the new intelligent solution, obtains the operator type and edge information required by the new intelligent solution, compares the obtained operator type and edge information with the topological structure of each intelligent solution being run, and automatically loads the required operators according to the intelligent solution running method described in the above embodiment, regenerates the new topological structure, and the scheduler module runs the operators in each intelligent solution according to the new topological result to obtain the running results of each intelligent solution. Among them, when automatically loading, since the operators and edge information required by the newly added intelligent solution may already exist in the intelligent solution that is already running, when the new intelligent solution is added, it is only necessary to add the operators that did not exist before, without loading all the operators in the new intelligent solution, which improves the flexibility of operation, reduces memory usage, reduces the reuse of operators, and improves the utilization rate of operators.

[0078] Similarly, when the user needs to delete a running smart solution, the scheduler module uses the idle CPU time to find only the operators and side information of the smart solution that needs to be deleted in the topology of the running smart solution, that is, the target operator and target side information that are no longer needed. The smart solution that needs to be deleted can be deleted by deleting these operators and side information. It is not necessary to delete all operators and side information in the smart solution, which improves the flexibility of deleting smart solutions. The scheduler module runs the corresponding operators according to the topology of the deleted smart solution and obtains the corresponding operation results.

[0079] In this embodiment, a smart solution operation method is also provided. Figure 6 is a flow chart of another intelligent solution operation method of this embodiment. Figure 6 As shown, the process includes the following steps:

[0080] Step S601, parsing the topological structure of the smart solution to obtain each operator in the smart solution, configuration information of each operator, each edge information and a label of the smart solution, wherein the edge information is the connection relationship between each operator in the smart solution;

[0081] Step S602, performing deduplication processing on each operator to obtain a deduplicated operator set, and each operator in the deduplicated operator set is identified as a target operator;

[0082] Step S603, according to the different configuration information required by each operator in the intelligent solution, the corresponding configuration information is loaded into each target operator; wherein each target operator is distinguished by a label marking the intelligent solution;

[0083] Step S604, comparing the same operator nodes in each edge information, retaining operators with different configuration information, marking the retained operators with labels of the intelligent solutions corresponding to the retained operators, topologically sorting the edge information corresponding to the retained operators, and obtaining target edge information;

[0084] Step S605, each target operator and target edge information form a target topology structure; run the intelligent solution according to the target topology structure to obtain the running result;

[0085] Step S606, in response to the new intelligent solution's request, using the idle CPU time to reparse and load new operators, generate a new topology structure corresponding to the new intelligent solution, and run the new intelligent solution according to the new topology structure;

[0086] Step S607, in response to the deletion request of the intelligent solution, unload unnecessary target operators and target edge information by using the idle CPU time, generate a new topology structure, and run the deleted intelligent solution according to the new topology result.

[0087] Through the above steps S601 to S607, compared with the current method of directly treating the entire intelligent solution as a module and delivering it by reducing the video frame resolution or bit rate, firstly, this embodiment extracts all operators, operator configuration information, side information and labels of each intelligent solution by parsing the topological structure of each operator in each intelligent solution, deduplicating all operators and then loading the operators required to run the intelligent solution, so that the scheduler only loads the operators needed and does not repeatedly load the same operators, thereby solving the problem of repeated loading into the memory, causing memory waste, and also improving the utilization rate of the operators. Secondly, this embodiment deduplicates and merges all side information, removes duplicate side information, obtains integrated side information, marks the labels of each intelligent solution on the operators in the integrated side information, obtains a new topological structure, and runs each operator according to the new topological structure. Among them, by subdividing the delivery granularity to the operator level, the reused operators and repeated edge information are deduplicated, the use of duplicate operators is reduced, and thus the memory usage is reduced. By integrating the operators and edge information after deduplication, a new topological structure is obtained to realize the simultaneous operation of multiple intelligent solutions, thereby improving the utilization rate of operators. Finally, when the user needs to add or delete an intelligent solution, it is only necessary to add or uninstall the operators and edge information involved in the target intelligent solution, without adding or deleting all the operators and edge information in the target intelligent solution, and without restarting the process, which improves the flexibility of intelligent solution delivery.

[0088] In this embodiment, an intelligent solution operation device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0089] Figure 7 is a structural block diagram of the intelligent solution operation device of this embodiment, such as Figure 7 As shown, the device 70 includes: a parsing module 71, a loading module 72, an integration module 73 and an operation module 74, wherein:

[0090] The parsing module 71 is used to parse the topological structure of the intelligent solution to obtain information about each operator and each edge in the intelligent solution, wherein the edge information is the connection relationship between the operators in the intelligent solution;

[0091] The loading module 72 is used to perform deduplication processing on each operator to obtain a deduplication set of operators; each operator in the deduplication set of operators is identified as a target operator; and each target operator is loaded;

[0092] The integration module 73 is used to perform deduplication and integration processing on each edge information to obtain the target edge information; each target operator and the target edge information form a target topological structure;

[0093] The operation module 74 is used to run the intelligent solution according to the target topology structure to obtain the operation result.

[0094] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0095] In this embodiment, an electronic device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0096] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0097] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0098] S1, analyzing the topological structure of the intelligent solution to obtain information about each operator and each edge in the intelligent solution, wherein the edge information is the connection relationship between the operators in the intelligent solution;

[0099] S2, performing deduplication processing on each operator to obtain a deduplicated operator set, and each operator in the deduplicated operator set is identified as a target operator;

[0100] S3, load each target operator;

[0101] S4, deduplicate and integrate each edge information to obtain target edge information; each target operator and target edge information form a target topological structure;

[0102] S5, running the intelligent solution according to the target topology structure to obtain the running result.

[0103] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0104] In addition, in combination with the intelligent solution operation method provided in the above embodiments, a storage medium can also be provided in this embodiment to implement the intelligent solution operation method. The storage medium stores a computer program; when the computer program is executed by the processor, any intelligent solution operation method in the above embodiments is implemented.

[0105] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0106] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0107] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0108] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0109] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A method for operating an intelligent solution, characterized in that: include: Analyze the topological structure of each intelligent solution to obtain each operator and each edge information in each intelligent solution, wherein the edge information is the connection relationship between each operator in each intelligent solution; Performing deduplication processing on the operators to obtain a deduplicated operator set, wherein the operators in the deduplicated operator set are identified as target operators; Loading the target operators; wherein the topological structure of the intelligent solution also includes configuration information of each operator; Compare the same operator nodes in the edge information, retain operators with different configuration information, and obtain duplicated edge information; Topologically sort the deduplicated edge information to obtain the target edge information; The target operators and the target edge information form a target topological structure; The intelligent solutions are run according to the target topology structure to obtain running results.

2. The intelligent solution operation method according to claim 1, characterized in that: The topological structure of each intelligent solution also includes configuration information of each operator, and the loading of each target operator includes: According to the different configuration information required by each operator in each intelligent solution, the corresponding configuration information is loaded into each target operator.

3. The intelligent solution operation method according to claim 2, characterized in that: The topological structure of each intelligent solution also includes a label of each intelligent solution, and each target operator corresponding to each configuration information is distinguished by marking the label of each intelligent solution.

4. The intelligent solution operation method according to claim 1, characterized in that: The topological structure of each intelligent solution also includes a label of each intelligent solution; comparing the same operators in each side information, retaining operators with different configuration information, and obtaining the side information after deduplication, includes: Compare operators of the same node in each edge information, and retain operators with different configuration information; The retained operators are marked with labels of intelligent solutions corresponding to the retained operators to obtain the deduplicated side information.

5. The intelligent solution operation method according to claim 3, characterized in that: The step of running each intelligent solution according to the target topology structure to obtain an operation result includes: When running the intelligent schemes according to the target topology structure, when multiple different operator operation results in the intelligent schemes are imported into the same operator node, different intermediate results are calculated in the operator node according to the imported different operator operation results; The different intermediate results are transmitted to the corresponding downstream nodes according to the labels of the smart solutions marked respectively, until the operation is completed to obtain the operation result; wherein the operation result includes the labels of the smart solutions.

6. The intelligent solution operation method according to claim 1, characterized in that: The method further comprises: In response to a new request for a new intelligent solution, using idle CPU time, re-parse and load new operators, generate a new topology structure corresponding to the new intelligent solution, and run the new intelligent solution according to the new topology structure; In response to the deletion request of the intelligent solution, the idle CPU time is used to unload the unnecessary target operators and target edge information, generate a new topology structure, and run the deleted intelligent solution according to the new topology result.

7. An intelligent solution operation device, characterized in that: It includes: parsing module, loading module, integration module and running module, among which, The parsing module is used to parse the topological structure of each intelligent solution to obtain each operator and each edge information in each intelligent solution, wherein the edge information is the connection relationship between each operator in each intelligent solution; The loading module is used to perform deduplication processing on the operators to obtain a deduplication set of operators; each operator in the deduplication set of operators is identified as a target operator; and each target operator is loaded; wherein the topological structure of the intelligent solution also includes configuration information of each operator; The integration module is used to compare the same operator nodes in the edge information, retain operators with different configuration information, and obtain the edge information after deduplication; topologically sort the edge information after deduplication to obtain the target edge information; the target operators and the target edge information form a target topological structure; The operation module is used to operate each intelligent solution according to the target topology structure to obtain an operation result.

8. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the smart solution running method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent solution operation method described in any one of claims 1 to 6 are implemented.

Images