Artificial intelligence image recognition algorithm upgrading method and device of power transmission monitoring and photographing system, electronic equipment and readable storage medium

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an artificial intelligence image recognition algorithm upgrading method and device of a power transmission monitoring and photographing system, electronic equipment and a readable storage medium. Inputting the description configuration file into the compliance verification algorithm through the power transmission monitoring standardized interface, and performing compliance verification on the algorithm upgrade package; in response to a verification passing result sent by the compliance verification algorithm, an artificial intelligence image recognition algorithm model in the server module is updated based on the artificial intelligence image recognition algorithm model group file; registering a plurality of interface function implementation files in the algorithm service process file to the server process; and the client module enables the artificial intelligence image recognition algorithm model to execute corresponding operation by calling the power transmission monitoring and shooting standardized interface. According to the invention, the same power transmission monitoring and shooting system is applicable to the upgrading operation of artificial intelligence image recognition algorithms of multiple manufacturers.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to an artificial intelligence image recognition algorithm upgrade method, device, electronic device and readable storage medium for a power transmission monitoring system. Background Art

[0002] In the power transmission system, the transmission line is the artery of power transmission, and its safe and stable operation is crucial. However, since the transmission lines are widely distributed in the outdoor environment and exposed to natural conditions for a long time, they are easily interfered by various external factors, especially the attachment or hanging of foreign objects. These hidden dangers may not only cause faults such as short circuits and broken lines, but also pose a threat to the surrounding environment and personnel safety. As an important part of modern power monitoring technology, the transmission line monitoring device realizes all-weather and no-dead-angle monitoring of transmission lines by integrating high-definition cameras, image sensors and advanced image recognition algorithms. These devices can capture the image information of the transmission line in real time, and with the help of image recognition technology, they can automatically identify and classify foreign objects in the image, thereby effectively warning of potential safety hazards.

[0003] With the rapid development of machine learning algorithms, there are many image recognition algorithm developers and they update quickly. The image recognition algorithms in power transmission monitoring systems need to be constantly updated to improve performance. Due to the differences in algorithm library storage and call logic of different image recognition algorithm manufacturers, the compatibility and interoperability between other manufacturers' image recognition algorithms and power transmission line monitoring devices are poor, and power transmission line monitoring devices cannot apply image recognition algorithms from other manufacturers. In addition, since there are many versions and developers of image recognition algorithms, how to avoid the application of unauthorized image recognition algorithms to power transmission line monitoring devices is also a problem that needs to be solved. Summary of the invention

[0004] In order to solve the problems in the related art, the embodiments of the present disclosure provide a method, device, electronic device and readable storage medium for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system, wherein the power transmission monitoring system includes a server module and a client module, wherein the client module includes a software development kit, wherein the software development kit includes a standardized interface for power transmission monitoring and a compliance verification algorithm, wherein the client module communicates with the server module through the standardized interface for power transmission monitoring and wherein the server module includes a server process, wherein the method includes the server module performing the following operations:

[0006] Obtain an algorithm upgrade package, the algorithm upgrade package including an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files;

[0007] Input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package;

[0008] In response to the verification result sent by the compliance verification algorithm, the artificial intelligence image recognition algorithm model in the server module is updated based on the artificial intelligence image recognition algorithm model group file, and a plurality of interface function implementation files in the algorithm service process file are registered to the server process;

[0009] The client module enables the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting;

[0010] in:

[0011] The client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting, including: the client module sends an operation request to the server process of the server module through the standardized interface for power transmission monitoring and shooting, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the standardized interface for power transmission monitoring and shooting.

[0012] According to an embodiment of the present disclosure, the power transmission monitoring and shooting standardized interface includes:

[0013] Algorithm initialization interface, used to perform copyright protection on the algorithm upgrade package and initialize loading of the artificial intelligence image recognition algorithm module;

[0014] A parameter configuration interface, used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result;

[0015] Detecting the recognition result release interface, used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm;

[0016] A video frame detection and recognition processing interface, used to send an image recognition processing request to the server module and obtain the image recognition result sent by the artificial intelligence image recognition algorithm module;

[0017] The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources.

[0018] According to an embodiment of the present disclosure, the interface function implementation file includes:

[0019] An algorithm initialization interface function implementation file, used to respond to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface;

[0020] A parameter configuration interface function implementation file, used to respond to a parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface;

[0021] The detection and identification result release interface function implementation file is used to respond to the result release request sent by the detection and identification result release interface, perform the detection result release operation, and send the identification result release result to the detection and identification result release interface.

[0022] A video frame detection and recognition processing interface function implementation file, which is used to respond to the image recognition processing request, perform pre-processing and post-processing on the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface; the video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function;

[0023] The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file can perform algorithm release operations, and the algorithm release operations include one or more of the following operations: NPU model unloading and removal operations, NPU resource release operations, and system resource release operations. The algorithm release result is sent to the algorithm end release interface.

[0024] According to an embodiment of the present disclosure, the description configuration file includes:

[0025] Upgrade package name, used to identify the algorithm upgrade package;

[0026] An upgrade package publishing platform, used to identify the manufacturer of the algorithm upgrade package;

[0027] The SDK version number of the upgrade package is used to check the compatibility of the algorithm upgrade package;

[0028] The upgrade package MD5 verification code is used to verify the integrity and compliance of the algorithm upgrade package;

[0029] An upgrade package mandatory upgrade code, used to determine whether the algorithm upgrade package needs mandatory upgrade;

[0030] Storage length, used to determine the storage space required when the algorithm upgrade package is running;

[0031] The memory length is used to determine the memory space required when the algorithm upgrade package is running.

[0032] According to an embodiment of the present disclosure, the interface function implementation file also includes a copyright verification function implementation file; the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, including:

[0033] In response to the upgrade package verification request sent by the algorithm initialization interface, the server process calls the copyright verification function implementation file; the upgrade package verification request includes the toolkit version information of the software development toolkit;

[0034] The copyright verification function implementation file obtains the upgrade package SDK version number, the upgrade package MD5 verification code, the toolkit version information, and the IP information and MAC information of the server module, performs copyright verification, and generates a copyright verification result;

[0035] The copyright verification function implementation file returns the copyright verification result to the algorithm initialization interface.

[0036] According to an embodiment of the present disclosure, the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, and also includes:

[0037] In response to the algorithm model loading request sent by the algorithm initialization interface, the server process calls the registered algorithm initialization interface function implementation file, and the algorithm initialization interface function implementation file loads the artificial intelligence image recognition algorithm module into the memory and NPU of the server module, and returns the algorithm model loading result to the algorithm initialization interface after loading is completed;

[0038] In response to the parameter adjustment request sent by the parameter configuration interface, the server process calls the registered parameter configuration interface function implementation file, inputs the parameter set in the parameter adjustment request into the artificial intelligence image algorithm module, tests the artificial intelligence image algorithm module, and returns the parameter adjustment result to the parameter configuration interface;

[0039] In response to the recognition result release request sent by the detection and recognition result release interface, the server process calls the registered detection and recognition result release interface function implementation file, the detection and recognition result release interface function implementation file clears the historical image cache data, releases the memory occupied by the historical image cache data, and returns the historical image cache release result to the recognition result release interface;

[0040] In response to the video frame image sent by the video frame detection and recognition interface, the server process calls the pre-processing function in the registered video frame detection and recognition processing interface function implementation file to generate pre-processing data, the pre-processing function calls the artificial intelligence image recognition algorithm model to process the pre-processing data to generate recognition data, the recognition data is input into the post-processing function to generate post-processing data, and the post-processing data is sent to the video frame detection and recognition interface;

[0041] In response to the algorithm release request issued by the algorithm end release interface, the server process calls the registered algorithm end release interface function implementation file, the algorithm end release interface function implementation file executes the algorithm release operation, obtains the algorithm release result, and sends the algorithm release result to the algorithm end release interface.

[0042] According to an embodiment of the present disclosure, the video frame detection and recognition processing interface further includes:

[0043] The standard foreign body code function is used to generate a hidden danger identification code according to the post-processing data, wherein the hidden danger identification code is an identification code preset according to the type of hidden danger of the transmission line; and return the hidden danger identification code to the video frame detection and identification interface.

[0044] According to an embodiment of the present disclosure, the server module communicates with the client module via the TCP / IP protocol.

[0045] In a second aspect, an embodiment of the present disclosure provides an artificial intelligence image recognition algorithm upgrade device for a power transmission monitoring system, comprising:

[0046] The power transmission monitoring system includes a server module and a client module, the client module includes a software development kit, the software development kit includes a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module through the standardized interface for power transmission monitoring and the server module includes a server process, and the device includes:

[0047] An acquisition module is configured to acquire an algorithm upgrade package, wherein the algorithm upgrade package includes an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, wherein the algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files;

[0048] A verification module is configured to input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package;

[0049] An update module, configured to update the artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file in response to the verification pass result sent by the compliance verification algorithm;

[0050] A registration module, configured to register a plurality of interface function implementation files in the algorithm service process file to the server process;

[0051] A calling module configured so that the client module can call the power transmission monitoring and shooting standardized interface to enable the artificial intelligence image recognition algorithm model to perform corresponding operations;

[0052] Wherein, the client module includes the standardized interface for power transmission monitoring and shooting, and the calling module includes:

[0053] A sending submodule, configured so that the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface;

[0054] A calling submodule, configured so that the server process calls the corresponding interface function implementation file according to the operation request;

[0055] An operation submodule, configured as the interface function implementation file to perform corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request;

[0056] The return submodule is configured so that the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface.

[0057] According to an embodiment of the present disclosure, the power transmission monitoring and shooting standardized interface includes:

[0058] Algorithm initialization interface, used to perform copyright protection on the algorithm upgrade package and initialize loading of the artificial intelligence image recognition algorithm module;

[0059] A parameter configuration interface, used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result;

[0060] Detecting the recognition result release interface, used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm;

[0061] A video frame detection and recognition processing interface, used to send an image recognition processing request to the server module and obtain the image recognition result sent by the artificial intelligence image recognition algorithm module;

[0062] The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources.

[0063] According to an embodiment of the present disclosure, the interface function implementation file includes:

[0064] An algorithm initialization interface function implementation file, used to respond to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface;

[0065] A parameter configuration interface function implementation file, used to respond to a parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface;

[0066] The detection and identification result release interface function implementation file is used to respond to the result release request sent by the detection and identification result release interface, perform the detection result release operation, and send the identification result release result to the detection and identification result release interface.

[0067] A video frame detection and recognition processing interface function implementation file, which is used to respond to the image recognition processing request, perform pre-processing and post-processing on the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface; the video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function;

[0068] The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file can perform algorithm release operations, and the algorithm release operations include one or more of the following operations: NPU model unloading and removal operations, NPU resource release operations, and system resource release operations. The algorithm release result is sent to the algorithm end release interface.

[0069] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer instructions, and wherein the one or more computer instructions are executed by the processor to implement a method as described in any one of the first aspects.

[0070] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, the method described in the first aspect is implemented.

[0071] According to the technical solution provided by the embodiment of the present disclosure, a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system is disclosed, wherein the power transmission monitoring system comprises a server module and a client module, the client module comprises a software development kit, the software development kit comprises a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module via the standardized interface for power transmission monitoring, the server module comprises a server process, and the method comprises the server module performing the following operations: obtaining an algorithm upgrade package, the algorithm upgrade package comprising an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrade package is encrypted and compiled via the software development kit, and the algorithm service process file comprises a plurality of interface function implementation files; the description configuration file is input into the compliance verification algorithm via the standardized interface for power transmission monitoring, and the algorithm upgrade package is compliance verified; in response to the The verification result sent by the compliance verification algorithm is passed, the artificial intelligence image recognition algorithm model in the server module is updated based on the artificial intelligence image recognition algorithm model group file, and multiple interface function implementation files in the algorithm service process file are registered to the server process; the client module enables the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the power transmission monitoring and shooting standardized interface; wherein: the client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the power transmission monitoring and shooting standardized interface, including: the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface.

[0072] The present invention discloses a standardized power transmission monitoring interface in a software development kit and a communication architecture with a separation of a client module and a server module. The standardized power transmission monitoring interface can decouple the algorithm upgrade package from the client module, so that the same power transmission monitoring system can be suitable for upgrade operations of artificial intelligence image recognition algorithms of multiple manufacturers. At the same time, the software development kit also includes a compliance verification algorithm, which obtains the secret key information of the algorithm upgrade package and verifies the algorithm upgrade package through the compliance verification algorithm to prevent the loading of unauthorized algorithm upgrade packages.

[0073] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Other features, objectives and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0075] Figure 1 A flow chart of an artificial intelligence image recognition algorithm upgrade method for a power transmission monitoring system according to an embodiment of the present disclosure is shown.

[0076] Figure 2 A schematic diagram of a standardized interface for power transmission monitoring and filming in a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring and filming system according to an embodiment of the present disclosure is shown.

[0077] Figure 3 A schematic diagram of an interface function implementation file in a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system according to an embodiment of the present disclosure is shown.

[0078] Figure 4 A structural block diagram of an artificial intelligence image recognition algorithm upgrading device for a power transmission monitoring system according to an embodiment of the present disclosure is shown.

[0079] Figure 5 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0080] Figure 6 A schematic diagram showing the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0081] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for the sake of clarity, parts not related to the description of the exemplary embodiments are omitted in the accompanying drawings.

[0082] In the present disclosure, it should be understood that terms such as "include" or "have" are intended to indicate the presence of features, numbers, steps, behaviors, components, parts, or a combination thereof disclosed in the present specification, and are not intended to exclude the possibility that one or more other features, numbers, steps, behaviors, components, parts, or a combination thereof exist or are added.

[0083] It should also be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0084] In the present disclosure, if it involves operations of obtaining user information or user data or displaying user information or user data to others, the operations are all authorized and confirmed by the user, or actively selected by the user.

[0085] In the power transmission system, the transmission line is the artery of power transmission, and its safe and stable operation is crucial. However, since the transmission lines are widely distributed in the outdoor environment and exposed to natural conditions for a long time, they are easily interfered by various external factors, especially the attachment or hanging of foreign objects. These hidden dangers may not only cause faults such as short circuits and broken lines, but also pose a threat to the surrounding environment and personnel safety. As an important part of modern power monitoring technology, the transmission line monitoring device realizes all-weather and no-dead-angle monitoring of transmission lines by integrating high-definition cameras, image sensors and advanced image recognition algorithms. These devices can capture the image information of the transmission line in real time, and with the help of image recognition technology, they can automatically identify and classify foreign objects in the image, thereby effectively warning of potential safety hazards.

[0086] With the rapid development of machine learning algorithms, there are many image recognition algorithm developers and they update quickly. Due to the differences in the storage and calling logic of the algorithm libraries of different image recognition algorithm manufacturers, the compatibility and interoperability between other manufacturers' image recognition algorithms and transmission line monitoring devices are poor, and the transmission line monitoring devices cannot apply other manufacturers' image recognition algorithms. In addition, since there are many versions and developers of image recognition algorithms, how to avoid the application of unauthorized image recognition algorithms to transmission line monitoring devices is also a problem that needs to be solved.

[0087] In order to solve the above technical problems, the present invention discloses an artificial intelligence image recognition algorithm upgrading method for a power transmission monitoring system, the power transmission monitoring system comprises a server module and a client module, the client module comprises a software development kit, the software development kit comprises a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module through the standardized interface for power transmission monitoring and the server module comprises a server process, the method comprises the server module performing the following operations: obtaining an algorithm upgrading package, the algorithm upgrading package comprises an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrading package is encrypted and compiled through the software development kit, and the algorithm service process file comprises a plurality of interface function implementation files; the description configuration file is input into the compliance verification algorithm through the standardized interface for power transmission monitoring and the algorithm upgrading package is compliance verified; in response to the compliance The verification result sent by the regulation verification algorithm is updated based on the artificial intelligence image recognition algorithm model group file in the server module; multiple interface function implementation files in the algorithm service process file are registered to the server process; the client module enables the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the power transmission monitoring and shooting standardized interface; wherein: the client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the power transmission monitoring and shooting standardized interface, including: the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface.

[0088] According to the technical scheme disclosed in the present invention, the power transmission monitoring system is divided into a server module and a client module, a software development kit is set on the client, a power transmission monitoring standardized interface is set in the software development kit, the client sends image processing and other requests to the server module through the power transmission monitoring standardized interface, the server process registers the interface function implementation file in the server module, and the server process receives the image processing and other requests sent by the power transmission monitoring standardized interface. After the request, the corresponding interface function implementation file is called for processing, the interface function implementation file sends the processing result to the power transmission monitoring standardized interface, and the client module receives the processing result through the power transmission monitoring standardized interface. Through the power transmission monitoring standardized interface in the software development kit and the communication architecture separated by the client module and the server module, the power transmission monitoring standardized interface can realize the decoupling of the algorithm upgrade package from the client module, so that the same power transmission monitoring system is applicable to the upgrade operation of the artificial intelligence image recognition algorithm of multiple manufacturers; at the same time, the software development kit also includes a compliance verification algorithm, which obtains the secret key information of the algorithm upgrade package and verifies the algorithm upgrade package through the compliance verification algorithm to prevent the loading of unauthorized algorithm upgrade packages.

[0089] Figure 1 A flowchart of an artificial intelligence image recognition algorithm upgrade method for a power transmission monitoring system according to an embodiment of the present disclosure is shown. The power transmission monitoring system includes a server module and a client module, the client module includes a software development kit, the software development kit includes a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module through the standardized interface for power transmission monitoring, and the server module includes a server process. Figure 1 As shown, the artificial intelligence image recognition algorithm upgrade method includes the following steps S101-S105:

[0090] In step S101, an algorithm upgrade package is obtained, the algorithm upgrade package includes an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file. The algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files. The algorithm interface function implementation file is used to respond to the request and call of the power transmission monitoring and shooting standardized interface, and execute the functions preset by the algorithm interface function implementation file. For example, the algorithm interface function implementation file includes an image pre-processing function, the image pre-processing function is registered in the server process, and the request of the power transmission monitoring and shooting standardized interface is monitored. When the standardized interface response process monitors the image recognition request sent by the power transmission monitoring and shooting standardized interface, the server process calls the image pre-processing function to pre-process the image to be processed in the image recognition request.

[0091] In step S102, the description configuration file is input into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and the algorithm upgrade package is subjected to compliance verification. The software development kit includes a compliance verification algorithm, and the compliance verification algorithm is used to perform compliance verification on the algorithm upgrade package according to the description configuration file. After the server module obtains the algorithm upgrade package through the network or the cloud, it extracts the description configuration file of the algorithm upgrade package, and inputs the description configuration file into the compliance verification algorithm in the software development kit through the power transmission monitoring and shooting standardized interface. The compliance verification algorithm includes hash verification, digital signature verification, metadata verification, etc. The compliance verification algorithm determines whether the algorithm upgrade package is an authorized algorithm upgrade package, and whether the version of the algorithm upgrade package is applicable to the current power transmission monitoring and shooting system, generates a compliance verification result, and sends it to the server module through the power transmission monitoring and shooting standardized interface.

[0092] According to an embodiment of the present disclosure, the description configuration file includes:

[0093] The upgrade package name is used to identify the algorithm upgrade package. The upgrade package name is used to identify whether the file received by the server module is an algorithm upgrade package. If it is determined to be an algorithm upgrade package, a compliance check is performed. If it is determined not to be an algorithm upgrade package, the received file is deleted.

[0094] The upgrade package release platform is used to identify the manufacturer of the algorithm upgrade package. The power transmission monitoring system is a common device in the power industry to ensure the safety of transmission lines. It is used to detect whether there are foreign objects on the transmission lines that affect the safety of power transmission. The core of the power transmission monitoring system is the artificial intelligence image recognition algorithm. In order to improve the foreign object detection capability of the power transmission monitoring system, the power transmission monitoring system needs to have the ability to upgrade the artificial intelligence image recognition algorithm, especially the ability to upgrade the algorithm across algorithm manufacturers. The upgrade package release platform is the algorithm development manufacturer, indicating which manufacturer developed the algorithm upgrade package.

[0095] The upgrade package SDK version number is used to perform compatibility check on the algorithm upgrade package. The algorithm upgrade package is developed using the software development kit (SDK), and the SDK version used when developing the algorithm upgrade package needs to be compatible with the version of the software development kit in the client module when receiving the algorithm upgrade package.

[0096] The upgrade package MD5 verification code is used to verify the integrity and compliance of the algorithm upgrade package.

[0097] The upgrade package mandatory upgrade code is used to determine whether the algorithm upgrade package needs mandatory upgrade. The algorithm upgrade package includes optional upgrade packages and mandatory upgrade packages. When the upgrade package mandatory upgrade code is optional, the user can independently choose whether to update the algorithm upgrade package through the client module. When the upgrade package mandatory upgrade code is mandatory, the subsequent upgrade steps are entered without user selection.

[0098] The storage length is used to determine the storage space required when the algorithm upgrade package is running. The storage length is the storage space required when the algorithm upgrade package is parsed and running. When the storage length is greater than the remaining storage space of the server module, the algorithm upgrade package is stopped from being verified and parsed and the user is notified.

[0099] The memory length is used to determine the memory space required when the algorithm upgrade package is running. When the memory length is less than the available memory space of the server module, the algorithm upgrade package is checked and parsed and the user is notified.

[0100] In step S103, in response to the verification result sent by the compliance verification algorithm, the artificial intelligence image recognition algorithm model in the server module is updated based on the artificial intelligence image recognition algorithm model group file. When the algorithm upgrade package passes the compliance verification, the algorithm upgrade package is parsed, the artificial intelligence image recognition algorithm model group file and the algorithm service process file in the algorithm upgrade package are extracted, and written to the custom location of the server module, completing the update of the artificial intelligence image recognition algorithm model in the server module.

[0101] In step S104, the multiple interface function implementation files are registered to the server process. When the server module is initialized, the server process registers the interface function implementation files in the algorithm service process file to the server process, so that the interface function implementation files are in a listening state and can respond to requests issued by the power transmission monitoring and shooting standardized interface.

[0102] In step S105, the client module causes the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting.

[0103] According to an implementation manner of the present disclosure, the client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting, including: the client module sends an operation request to the server process of the server module through the standardized interface for power transmission monitoring and shooting, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the standardized interface for power transmission monitoring and shooting.

[0104] According to an embodiment of the present disclosure, the client module can obtain monitoring images of the power transmission line and receive user operation instructions, and the power transmission monitoring standardized interface can send the client module's request or the obtained image data to the server module.

[0105] Figure 2 A schematic diagram of a standardized interface for power transmission monitoring in a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system according to an embodiment of the present disclosure is shown. Figure 2 As shown, the transmission monitoring standardized interface includes:

[0106] An algorithm initialization interface is used to perform copyright protection on the algorithm upgrade package and initialize the loading of the artificial intelligence image recognition algorithm module. The algorithm initialization interface includes a copyright protection function, which can verify according to the description configuration file of the algorithm upgrade package to determine whether the algorithm upgrade package is an authorized version of a specific algorithm manufacturer. Furthermore, the copyright protection function can also send the description configuration file of the algorithm upgrade package, the version information of the software development kit, and the version information of the client module to the designated server of the specific algorithm manufacturer to calculate the copyright fee. In addition, the algorithm initialization interface can also receive an algorithm initialization request issued by the client module, send the algorithm initialization request to the server process of the server module, and implement the initialization loading of the artificial intelligence image recognition algorithm module in the server module by the interface function implementation file corresponding to the algorithm initialization interface in the server process.

[0107] A parameter configuration interface is used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result. When the user operates the client module, the detection accuracy parameter, processing speed parameter, resolution parameter, and frame rate parameter of the foreign object need to be adjusted. The parameter configuration interface receives the parameter adjustment request sent by the client module, and the parameter adjustment request includes the parameter category to be adjusted and the adjustment parameter value. The parameter adjustment request is sent to the server process of the server module, and the interface function implementation file corresponding to the parameter configuration interface in the server process implements the parameter configuration of the artificial intelligence image recognition algorithm module in the server module.

[0108] The detection and recognition result release interface is used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm. When performing image recognition, the image recognition algorithm usually requires a large amount of computing resources, such as NPU, memory and cache. After generating the image recognition result, it is necessary to release the above resources to perform other image processing or free up computing resources for other tasks to ensure the stability and performance of the system. After the client module receives the image recognition result of the previous frame, before sending the next frame of image data, it sends a recognition result release request to the detection and recognition result release interface. The detection and recognition result release interface sends the recognition result release request to the server process of the server module. The server process implements the file by calling the corresponding interface function, and executes operations such as closing the opened previous frame image file, cleaning temporary files and cache, releasing NPU resources, and clearing the image data buffer.

[0109] The video frame detection and recognition processing interface is used to send an image recognition processing request to the server module to obtain the image recognition result sent by the artificial intelligence image recognition algorithm module. When the client module needs to process any frame of image, the client module sends the image to be processed to the video frame detection and recognition processing interface, and the video frame detection and recognition processing interface sends the received image to be processed to the server process of the server module. The server process calls the pre-registered corresponding interface function implementation file to pre-process the image to be processed, and then sends it to the artificial intelligence image recognition algorithm model for transmission line foreign body recognition, and finally sends the foreign body recognition result to the video frame detection and recognition processing interface, and finally the video frame detection and recognition processing interface sends it to the client module.

[0110] Furthermore, the video frame detection and identification processing interface also includes: a standard foreign body code function, which is used to generate a hidden danger identification code according to the post-processing data, and the hidden danger identification code is an identification code preset according to the type of hidden danger of the transmission line; and the hidden danger identification code is returned to the video frame detection and identification interface. Since the algorithm upgrade package may be generated by different algorithm manufacturers, the code representation of the image recognition results of different transmission line foreign objects by different algorithm manufacturers may be different, which may cause the client module to be unable to correctly receive the image recognition results. Therefore, according to the application scenario of transmission line monitoring, a standard foreign body code function is set, and the standard foreign body code function includes the types of foreign objects that may be involved in transmission line monitoring and sets a unified standard foreign body code, which can convert the image recognition results output by the artificial intelligence image recognition algorithm into corresponding standard foreign body codes, and then send the corresponding standard foreign body codes to the client module. For example, the image recognition result output by the artificial intelligence image recognition algorithm is "CLASSID_PWR_FILM", which is a plastic film. The standard foreign body code function converts "CLASSID_PWR_FILM" into "10030200" and sends it to the client module through the video frame detection and recognition processing interface. For another example, the image recognition result output by the artificial intelligence image recognition algorithm is "CLASSID_PLASTIC_FILM", which is also a plastic film. The standard foreign body code function also converts "CLASSID_PLASTIC_FILM" into "10030200" and sends it to the client module through the video frame detection and recognition processing interface. The standard foreign body code function in the video frame detection and recognition processing interface can set the image recognition results output by different artificial intelligence image recognition algorithms as standard codes recognized by the client module, and can achieve the technical effect of the power transmission monitoring system obtaining the algorithm upgrade packages provided by different algorithm manufacturers for upgrading, and reduce the workload of algorithm manufacturers to adapt the power transmission monitoring system to the algorithm upgrade packages.

[0111] The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources. When the user no longer needs the power transmission monitoring system to perform the transmission line foreign body recognition function, the client module sends an algorithm release request to the algorithm end release interface, and the algorithm end release interface sends an algorithm release request to the server process of the server module, and the server process calls the corresponding interface function implementation file to release NPU resources and system resources. Releasing NPU resources and system resources includes: cleaning up the interface function implementation files, artificial intelligence image recognition algorithm models in the memory, cleaning up temporary files and cached data, and releasing storage space. Then, the algorithm end release result sent by the server process is returned to the client module.

[0112] According to the implementation manner of the present disclosure, the standardized interface for power transmission monitoring can standardize the client's requests to make them consistent with the format of requests that can be received by the server module and the server process, thereby realizing the upgrade and adaptation of artificial intelligence image recognition algorithms developed by multiple algorithm manufacturers.

[0113] Figure 3 A schematic diagram of an interface function implementation file in a method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system according to an embodiment of the present disclosure is shown. The interface function implementation file is a function file that cooperates with a standardized interface for power transmission monitoring to implement a specific function. Figure 3 As shown, the interface function implementation file includes:

[0114] An algorithm initialization interface function implementation file is used to respond to the algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface. The function pointer of the algorithm initialization interface function implementation file is the same as the function pointer in the algorithm model loading request sent by the algorithm initialization interface. The algorithm initialization interface function implementation file is a callback function, which is registered to the server process when the server is initialized. After registration is completed, the server process listens to the algorithm initialization interface in order to respond to the request sent by the algorithm initialization interface. The loading of the artificial intelligence image recognition algorithm model includes: loading the artificial intelligence image recognition algorithm model into the memory of the server module, and allocating the NPU computing resources in the server module to the artificial intelligence image recognition algorithm model.

[0115] A parameter configuration interface function implementation file is used to respond to the parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface. The parameter configuration interface function implementation file can read the adjustment parameter category and adjustment parameter value in the parameter adjustment request sent by the parameter configuration interface, and verify the parameter adjustment request according to the preset adjustment parameter category and adjustment parameter value threshold. After the parameter adjustment request is verified, the artificial intelligence image recognition algorithm model is adjusted according to the adjustment parameter category and adjustment parameter value. After the artificial intelligence image recognition algorithm model parameter adjustment is completed, the parameter adjustment result is returned to the parameter configuration interface. The parameter configuration interface function implementation file needs to adapt the parameter configuration interface, that is, the function pointer of the parameter configuration interface function implementation file is the same as the function pointer in the parameter adjustment request sent by the parameter configuration interface. The parameter configuration interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the parameter configuration interface to respond to the parameter adjustment request sent by the parameter configuration interface.

[0116] The detection and recognition result release interface function implementation file is used to respond to the result release request sent by the detection and recognition result release interface, perform the detection result release operation, and send the recognition result release result to the detection and recognition result release interface. The detection and recognition result release interface function implementation file can perform the detection result release operation on the server module, clear the processed image data and related cache files, and the occupied NPU resources, so as to improve the processing efficiency of the next frame image. The detection result release operation includes one or more of the following: closing the opened previous frame image file, cleaning temporary files and caches, releasing NPU resources, and clearing image data buffers. The detection and recognition result release interface function implementation file needs to adapt the detection and recognition result release interface, that is, the function pointer of the detection and recognition result release interface function implementation file is the same as the function pointer in the result release request sent by the detection and recognition result release interface, and the detection and recognition result release interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the parameter configuration interface so as to respond to the parameter adjustment request sent by the parameter configuration interface.

[0117] The video frame detection and recognition processing interface function implementation file is used to respond to the image recognition processing request, pre-process the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface. The video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function. Since the artificial intelligence image recognition algorithm modules developed by different algorithm manufacturers have different parameter settings and image data of the training set, the image to be processed needs to be processed according to the requirements of the artificial intelligence image recognition algorithm module before being input into the algorithm module for recognition.

[0118] According to the implementation mode of the present disclosure, the pre-processing function is used to perform one or more of the following operations: quality optimization operation, format unification operation, and image enhancement operation on the image to be processed in the image recognition processing request. When performing the quality optimization operation, the pre-processing function can remove the noise of the image to be processed, enhance the contrast, adjust the brightness, etc., and highlight the effective information in the image to be processed; when performing the format unification operation, the pre-processing function can perform operations such as scaling, cropping, and format conversion on the image to be processed, so that the image to be processed has a unified format and size, which is convenient for the image recognition algorithm module to process; the image enhancement operation includes edge detection and texture enhancement operations on the image to be processed, highlighting the key features in the image to be processed and improving the image recognition efficiency.

[0119] According to the embodiments of the present disclosure, the post-processing function is used to perform image screening and image correction on the initial recognition results output by the artificial intelligence image recognition algorithm to improve the image recognition accuracy. The initial recognition results output by the artificial intelligence image recognition algorithm may not accurately reflect the transmission line foreign matter in the image to be processed due to image light changes, image background complexity, image clarity, and the inclusion of multiple objects in the image. For example, there are multiple detection frames and the confidence of each detection frame is different, that is, there are multiple transmission line foreign matter to be judged. It is necessary to screen the multiple detection frames through the post-processing function, and judge the shape and type of the transmission line foreign matter based on the screening results.

[0120] The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file performs the algorithm release operation, and the algorithm release operation includes one or more of the following operations: releasing NPU resource occupation, releasing memory resource occupation, cleaning up temporary files, and sending the algorithm release result to the algorithm end release interface. The algorithm end release interface function implementation file can release NPU resources and system resources for the server module, including closing all open files and handles, clearing data structures in memory, clearing data structures on NPU, terminating non-essential processes, and sending the release result to the algorithm end release interface. The algorithm end release interface function implementation file needs to adapt the algorithm end release interface, that is, the function pointer of the algorithm end release interface function implementation file is the same as the function pointer in the algorithm release request sent by the algorithm end release interface. The algorithm end release interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the algorithm end release interface to respond to the algorithm release request sent by the algorithm end release interface.

[0121] According to an embodiment of the present disclosure, the description configuration file includes:

[0122] The upgrade package name is used to identify the algorithm upgrade package.

[0123] The upgrade package publishing platform is used to identify the manufacturer of the algorithm upgrade package.

[0124] The SDK version number of the upgrade package is used to perform a compatibility check on the algorithm upgrade package. The client module and the server module communicate through the software development kit, so the SDK version number when the algorithm upgrade package is developed should be compatible with the software development kit. When the SDK version number of the upgrade package does not match the software development kit of the client, the upgrade operation is terminated and the algorithm upgrade package is cleared.

[0125] The MD5 checksum of the upgrade package is used to verify the integrity and compliance of the algorithm upgrade package. Before upgrading the artificial intelligence image recognition algorithm of the power transmission monitoring system, the algorithm upgrade package needs to be checked for integrity and compliance to ensure that the algorithm upgrade package is a complete upgrade package that has been authorized and released, so as to prevent the power transmission monitoring system from working abnormally due to incorrect upgrades.

[0126] The upgrade package mandatory upgrade code is used to determine whether the algorithm upgrade package needs mandatory upgrade. The algorithm upgrade package includes mandatory upgrade packages and optional upgrade packages. The compliance verification algorithm in the software development kit identifies whether the algorithm upgrade package is an algorithm upgrade package that needs mandatory upgrade through the upgrade package mandatory upgrade code.

[0127] The storage length is used to determine the storage space required when the algorithm upgrade package is running.

[0128] The memory length is used to determine the memory space required when the algorithm upgrade package is running.

[0129] According to an embodiment of the present disclosure, the interface function implementation file also includes a copyright verification function implementation file; the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, including:

[0130] In response to the upgrade package verification request sent by the algorithm initialization interface, the server process calls the copyright verification function implementation file; the upgrade package verification request includes the toolkit version information of the software development toolkit.

[0131] The copyright verification function implementation file obtains the upgrade package SDK version number, upgrade package MD5 check code, toolkit version information, and IP information and MAC information of the server module, performs copyright verification, and generates copyright verification results; the copyright verification function implementation file returns the copyright verification results to the algorithm initialization interface. In the application scenario of power transmission line monitoring, the number of power transmission monitoring systems is huge, and the requirements for algorithm upgrade packages vary. Algorithm developers often need to calculate costs based on the number of upgrades to power transmission monitoring systems. The copyright verification function obtains the upgrade package SDK version number and the upgrade package MD5 check code, toolkit version information, and can also obtain relevant information of the server module, such as IP information, MAC information, SN information, etc., and can verify whether the algorithm upgrade package is genuine, and send the copyright verification results to the client module through the algorithm initialization interface. Further, when the algorithm development manufacturer sets up a network authorization server, the copyright verification function implementation file can also send the above information to the algorithm development manufacturer's network authorization server to perform authorization billing management and realize the intellectual property protection of the algorithm development manufacturer.

[0132] According to an embodiment of the present disclosure, the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, and also includes:

[0133] In response to the algorithm model loading request sent by the algorithm initialization interface, the server process calls the registered algorithm initialization interface function implementation file, and the algorithm initialization interface function implementation file loads the artificial intelligence image recognition algorithm module into the memory and NPU of the server module. After loading is completed, the algorithm model loading result is returned to the algorithm initialization interface.

[0134] In response to the parameter adjustment request sent by the parameter configuration interface, the server process calls the registered parameter configuration interface function implementation file, inputs the parameter set in the parameter adjustment request into the artificial intelligence image algorithm module, tests the artificial intelligence image algorithm module, and returns the parameter adjustment result to the parameter configuration interface.

[0135] In response to the recognition result release request sent by the detection and recognition result release interface, the server process calls the registered detection and recognition result release interface function implementation file, and the detection and recognition result release interface function implementation file clears the historical image cache data, releases the memory occupied by the historical image cache data, and returns the historical image cache release result to the recognition result release interface.

[0136] In response to the video frame image sent by the video frame detection and recognition interface, the server process calls the pre-processing function in the registered video frame detection and recognition processing interface function implementation file to generate pre-processing data. The pre-processing function calls the artificial intelligence image recognition algorithm model to process the pre-processing data to generate recognition data. The recognition data is input into the post-processing function to generate post-processing data, and the post-processing data is sent to the video frame detection and recognition interface.

[0137] In response to the algorithm release request sent by the algorithm end release interface, the server process calls the registered algorithm end release interface function implementation file, the algorithm end release interface function implementation file performs the algorithm release operation, obtains the algorithm release result, and sends the algorithm release result to the algorithm end release interface. The algorithm release operation includes NPU model uninstallation and removal operation, NPU resource release operation, and system resource release operation.

[0138] According to an embodiment of the present disclosure, the server module communicates with the client module via the TCP / IP protocol.

[0139] According to the embodiment of the present disclosure, the power transmission monitoring system includes a client module and a server module, which communicate through the TCP / IP protocol of the software development kit. When the server is initialized, the server module registers the interface function file in the upgrade installation package to the server process, and the request sent by the client module is standardized and converted through the standardized interface of the power transmission monitoring in the software development kit and sent to the server process. The server process calls the corresponding interface function file registered to the server process by callback, and the corresponding interface function file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the request content. The technical solution disclosed in the present disclosure can realize the decoupling of the artificial intelligence image recognition algorithm in the client module and the server module, so that the power transmission monitoring system can be compatible with the image recognition algorithms of different algorithm development manufacturers, and improve the compatibility and upgrade convenience of the power transmission monitoring system with the artificial intelligence image recognition algorithm.

[0140] Figure 4 The structural block diagram of the artificial intelligence image recognition algorithm upgrade device of the power transmission monitoring system according to the embodiment of the present disclosure is shown. The power transmission monitoring system includes a server module and a client module, the client module includes a software development kit, the software development kit includes a power transmission monitoring standardized interface and a compliance verification algorithm, the client module communicates with the server module through the power transmission monitoring standardized interface, and the server module includes a server process. Figure 4 As shown, the artificial intelligence image recognition algorithm upgrading device 400 of the power transmission monitoring system includes an acquisition module 410, a verification module 420, an update module 430, a registration module 440 and a calling module 450.

[0141] The acquisition module is configured to acquire an algorithm upgrade package, the algorithm upgrade package includes an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file. The algorithm upgrade package is encrypted and compiled by the software development kit, and the algorithm service process file includes multiple interface function implementation files. The algorithm interface function implementation file is a program code file containing the algorithm logic of the artificial intelligence image recognition algorithm model group file and the interface definition of the power transmission monitoring and shooting standardized interface, which is used to respond to the request and call of the power transmission monitoring and shooting standardized interface, and realize the function preset by the algorithm interface function implementation file. For example, the algorithm interface function implementation file includes an image pre-processing function, which is registered in the server process and monitors the request of the power transmission monitoring and shooting standard interface. When the standardized interface response process monitors the image recognition request sent by the power transmission monitoring and shooting standardized interface, the server process calls the image pre-processing function to pre-process the image to be processed in the image recognition request.

[0142] The verification module is configured to input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package. The software development kit includes a compliance verification algorithm, and the compliance verification algorithm is used to perform compliance verification on the algorithm upgrade package according to the description configuration file. After the server module obtains the algorithm upgrade package through the network or the cloud, it extracts the description configuration file of the algorithm upgrade package, and inputs the description configuration file into the compliance verification algorithm in the software development kit through the power transmission monitoring and shooting standardized interface. The compliance verification algorithm includes hash verification, digital signature verification, metadata verification, etc. The compliance verification algorithm determines whether the algorithm upgrade package is an authorized algorithm upgrade package, and whether the version of the algorithm upgrade package is applicable to the current power transmission monitoring and shooting system, generates a compliance verification result, and sends it to the server module through the power transmission monitoring and shooting standardized interface.

[0143] The update module is configured to update the artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file in response to the verification result sent by the compliance verification algorithm. When the algorithm upgrade package passes the compliance verification, the algorithm upgrade package is parsed, the artificial intelligence image recognition algorithm model group file and the algorithm service process file in the algorithm upgrade package are extracted, and written to the custom location of the server module to complete the update of the artificial intelligence image recognition algorithm model in the server module.

[0144] The registration module is configured to register multiple interface function implementation files in the algorithm service process file to the server process. When the server module is initialized, the server process registers the interface function implementation files in the algorithm service process file to the server process, so that the interface function implementation files are in a listening state and can respond to requests issued by the power transmission monitoring and shooting standardized interface.

[0145] The calling module is configured so that the client module can call the standardized interface for power transmission monitoring and shooting to enable the artificial intelligence image recognition algorithm model to perform corresponding operations.

[0146] Wherein, the client module includes the standardized interface for power transmission monitoring and shooting, and the calling module includes:

[0147] The sending submodule is configured so that the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface.

[0148] A calling submodule, configured so that the server process calls the corresponding interface function implementation file according to the operation request;

[0149] An operation submodule, configured as the interface function implementation file to perform corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request;

[0150] The return submodule is configured so that the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface.

[0151] According to an embodiment of the present disclosure, the power transmission monitoring and shooting standardized interface includes:

[0152] An algorithm initialization interface is used to perform copyright protection on the algorithm upgrade package and initialize the loading of the artificial intelligence image recognition algorithm module. The algorithm initialization interface includes a copyright protection function, which can verify according to the description configuration file of the algorithm upgrade package to determine whether the algorithm upgrade package is an authorized version of a specific algorithm manufacturer. Furthermore, the copyright protection function can also send the description configuration file of the algorithm upgrade package, the version information of the software development kit, and the version information of the client module to the designated server of the specific algorithm manufacturer to calculate the copyright fee. In addition, the algorithm initialization interface can also receive an algorithm initialization request issued by the client module, send the algorithm initialization request to the server process of the server module, and implement the initialization loading of the artificial intelligence image recognition algorithm module in the server module by the interface function implementation file corresponding to the algorithm initialization interface in the server process.

[0153] A parameter configuration interface is used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result. When the user operates the client module, the detection accuracy parameter, processing speed parameter, resolution parameter, and frame rate parameter of the foreign object need to be adjusted. The parameter configuration interface receives the parameter adjustment request sent by the client module, and the parameter adjustment request includes the parameter category to be adjusted and the adjustment parameter value. The parameter adjustment request is sent to the server process of the server module, and the interface function implementation file corresponding to the parameter configuration interface in the server process implements the parameter configuration of the artificial intelligence image recognition algorithm module in the server module.

[0154] The detection and recognition result release interface is used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm. When performing image recognition, the image recognition algorithm usually requires a large amount of computing resources, such as NPU, memory and cache. After generating the image recognition result, it is necessary to release the above resources to perform other image processing or free up computing resources for other tasks to ensure the stability and performance of the system. After the client module receives the image recognition result of the previous frame, before sending the next frame of image data, it sends a recognition result release request to the detection and recognition result release interface. The detection and recognition result release interface sends the recognition result release request to the server process of the server module. The server process implements the file by calling the corresponding interface function, and executes operations such as closing the opened previous frame image file, cleaning temporary files and cache, releasing NPU resources, and clearing the image data buffer.

[0155] The video frame detection and recognition processing interface is used to send an image recognition processing request to the server module to obtain the image recognition result sent by the artificial intelligence image recognition algorithm module. When the client module needs to process any frame of image, the client module sends the image to be processed to the video frame detection and recognition processing interface, and the video frame detection and recognition processing interface sends the received image to be processed to the server process of the server module. The server process calls the pre-registered corresponding interface function implementation file to pre-process the image to be processed, and then sends it to the artificial intelligence image recognition algorithm model for transmission line foreign body recognition, and finally sends the foreign body recognition result to the video frame detection and recognition processing interface, and finally the video frame detection and recognition processing interface sends it to the client module.

[0156] Furthermore, the video frame detection and identification processing interface also includes: a standard foreign body code function, which is used to generate a hidden danger identification code according to the post-processing data, and the hidden danger identification code is an identification code preset according to the type of hidden danger of the transmission line; and the hidden danger identification code is returned to the video frame detection and identification interface. Since the algorithm upgrade package may be generated by different algorithm manufacturers, the code representation of the image recognition results of different transmission line foreign objects by different algorithm manufacturers may be different, which may cause the client module to be unable to correctly receive the image recognition results. Therefore, according to the application scenario of transmission line monitoring, a standard foreign body code function is set, and the standard foreign body code function includes the types of foreign objects that may be involved in transmission line monitoring and sets a unified standard foreign body code, which can convert the image recognition results output by the artificial intelligence image recognition algorithm into corresponding standard foreign body codes, and then send the corresponding standard foreign body codes to the client module. For example, the image recognition result output by the artificial intelligence image recognition algorithm is "CLASSID_PWR_FILM", which is a plastic film. The standard foreign body code function converts "CLASSID_PWR_FILM" into "10030200" and sends it to the client module through the video frame detection and recognition processing interface. For another example, the image recognition result output by the artificial intelligence image recognition algorithm is "CLASSID_PLASTIC_FILM", which is also a plastic film. The standard foreign body code function also converts "CLASSID_PLASTIC_FILM" into "10030200" and sends it to the client module through the video frame detection and recognition processing interface. The standard foreign body code function in the video frame detection and recognition processing interface can set the image recognition results output by different artificial intelligence image recognition algorithms as standard codes recognized by the client module, and can achieve the technical effect of the power transmission monitoring system obtaining the algorithm upgrade packages provided by different algorithm manufacturers for upgrading, and reduce the workload of algorithm manufacturers to adapt the power transmission monitoring system to the algorithm upgrade packages.

[0157] The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources. When the user no longer needs the power transmission monitoring system to perform the transmission line foreign body recognition function, the client module sends an algorithm release request to the algorithm end release interface, and the algorithm end release interface sends an algorithm release request to the server process of the server module, and the server process calls the corresponding interface function implementation file to release NPU resources and system resources. Releasing NPU resources and system resources includes: cleaning up the interface function implementation files, artificial intelligence image recognition algorithm models in the memory, cleaning up temporary files and cached data, and releasing storage space. Then, the algorithm end release result sent by the server process is returned to the client module.

[0158] According to an embodiment of the present disclosure, the interface function implementation file is a function file that cooperates with the standardized interface for power transmission monitoring and shooting to realize a specific function, and the interface function implementation file includes:

[0159] An algorithm initialization interface function implementation file is used to respond to the algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface. The function pointer of the algorithm initialization interface function implementation file is the same as the function pointer in the algorithm model loading request sent by the algorithm initialization interface. The algorithm initialization interface function implementation file is a callback function, which is registered to the server process when the server is initialized. After registration is completed, the server process listens to the algorithm initialization interface in order to respond to the request sent by the algorithm initialization interface. The loading of the artificial intelligence image recognition algorithm model includes: loading the artificial intelligence image recognition algorithm model into the memory of the server module, and allocating the NPU computing resources in the server module to the artificial intelligence image recognition algorithm model.

[0160] A parameter configuration interface function implementation file is used to respond to the parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface. The parameter configuration interface function implementation file can read the adjustment parameter category and adjustment parameter value in the parameter adjustment request sent by the parameter configuration interface, and verify the parameter adjustment request according to the preset adjustment parameter category and adjustment parameter value threshold. After the parameter adjustment request is verified, the artificial intelligence image recognition algorithm model is adjusted according to the adjustment parameter category and adjustment parameter value. After the artificial intelligence image recognition algorithm model parameter adjustment is completed, the parameter adjustment result is returned to the parameter configuration interface. The parameter configuration interface function implementation file needs to adapt the parameter configuration interface, that is, the function pointer of the parameter configuration interface function implementation file is the same as the function pointer in the parameter adjustment request sent by the parameter configuration interface. The parameter configuration interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the parameter configuration interface to respond to the parameter adjustment request sent by the parameter configuration interface.

[0161] The detection and recognition result release interface function implementation file is used to respond to the result release request sent by the detection and recognition result release interface, perform the detection result release operation, and send the recognition result release result to the detection and recognition result release interface. The detection and recognition result release interface function implementation file can perform the detection result release operation on the server module, clear the processed image data and related cache files, and the occupied NPU resources, so as to improve the processing efficiency of the next frame image. The detection result release operation includes one or more of the following: closing the opened previous frame image file, cleaning temporary files and caches, releasing NPU resources, and clearing image data buffers. The detection and recognition result release interface function implementation file needs to adapt the detection and recognition result release interface, that is, the function pointer of the detection and recognition result release interface function implementation file is the same as the function pointer in the result release request sent by the detection and recognition result release interface, and the detection and recognition result release interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the parameter configuration interface so as to respond to the parameter adjustment request sent by the parameter configuration interface.

[0162] The video frame detection and recognition processing interface function implementation file is used to respond to the image recognition processing request, pre-process the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface. The video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function. Since the artificial intelligence image recognition algorithm modules developed by different manufacturers have different parameter settings and image data of the training set, the image to be processed needs to be processed according to the requirements of the artificial intelligence image recognition algorithm module before being input into the algorithm module for recognition.

[0163] According to the implementation mode of the present disclosure, the pre-processing function is used to perform one or more of the following operations: quality optimization operation, format unification operation, and image enhancement operation on the image to be processed in the image recognition processing request. When performing the quality optimization operation, the pre-processing function can remove the noise of the image to be processed, enhance the contrast, adjust the brightness, etc., and highlight the effective information in the image to be processed; when performing the format unification operation, the pre-processing function can perform operations such as scaling, cropping, and format conversion on the image to be processed, so that the image to be processed has a unified format and size, which is convenient for the image recognition algorithm module to process; the image enhancement operation includes edge detection and texture enhancement operations on the image to be processed, highlighting the key features in the image to be processed and improving the image recognition efficiency.

[0164] According to the embodiments of the present disclosure, the post-processing function is used to perform image screening and image correction on the initial recognition results output by the artificial intelligence image recognition algorithm to improve the image recognition accuracy. The initial recognition results output by the artificial intelligence image recognition algorithm may not accurately reflect the transmission line foreign matter in the image to be processed due to image light changes, image background complexity, image clarity, and the inclusion of multiple objects in the image. For example, there are multiple detection frames and the confidence of each detection frame is different, that is, there are multiple transmission line foreign matter to be judged. It is necessary to screen the multiple detection frames through the post-processing function, and judge the shape and type of the transmission line foreign matter based on the screening results.

[0165] The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file performs the algorithm release operation, and the algorithm release operation includes one or more of the following operations: releasing NPU resource occupation, releasing memory resource occupation, cleaning up temporary files, and sending the algorithm release result to the algorithm end release interface. The algorithm end release interface function implementation file can release NPU resources and system resources for the server module, including closing all open files and handles, clearing data structures in memory, clearing data structures on NPU, terminating non-essential processes, and sending the release result to the algorithm end release interface. The algorithm end release interface function implementation file needs to adapt the algorithm end release interface, that is, the function pointer of the algorithm end release interface function implementation file is the same as the function pointer in the algorithm release request sent by the algorithm end release interface. The algorithm end release interface function implementation file is a callback function, which is registered in the server process when the server is initialized. After completing the registration, the server process monitors the algorithm end release interface to respond to the algorithm release request sent by the algorithm end release interface.

[0166] The present disclosure also discloses an electronic device, Figure 5 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0167] like Figure 5 As shown, the electronic device includes a memory and a processor, wherein the memory is used to store computer instructions, and the computer instructions are executed by the processor to implement the method according to an embodiment of the present disclosure.

[0168] A method for upgrading an artificial intelligence image recognition algorithm of a power transmission monitoring system, the power transmission monitoring system comprising a server module and a client module, the client module comprising a software development kit, the software development kit comprising a power transmission monitoring standardized interface and a compliance verification algorithm, the client module communicating with the server module via the power transmission monitoring standardized interface, the server module comprising a server process, the method comprising the server module performing the following operations:

[0169] Obtain an algorithm upgrade package, the algorithm upgrade package including an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files;

[0170] Input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package;

[0171] In response to the verification pass result sent by the compliance verification algorithm, updating the artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file;

[0172] Registering a plurality of interface function implementation files in the algorithm service process file to the server process;

[0173] The client module enables the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting;

[0174] in:

[0175] The client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting, including: the client module sends an operation request to the server process of the server module through the standardized interface for power transmission monitoring and shooting, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the standardized interface for power transmission monitoring and shooting.

[0176] The transmission monitoring and shooting standardized interface includes:

[0177] Algorithm initialization interface, used to perform copyright protection on the algorithm upgrade package and initialize loading of the artificial intelligence image recognition algorithm module;

[0178] A parameter configuration interface, used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result;

[0179] Detecting the recognition result release interface, used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm;

[0180] A video frame detection and recognition processing interface, used to send an image recognition processing request to the server module and obtain the image recognition result sent by the artificial intelligence image recognition algorithm module;

[0181] The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources.

[0182] The interface function implementation file includes:

[0183] An algorithm initialization interface function implementation file, used to respond to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface;

[0184] A parameter configuration interface function implementation file, used to respond to a parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface;

[0185] A detection and identification result release interface function implementation file, which is used to respond to a result release request sent by the detection and identification result release interface, perform a detection result release operation, and send the identification result release result to the detection and identification result release interface;

[0186] A video frame detection and recognition processing interface function implementation file, which is used to respond to the image recognition processing request, perform pre-processing and post-processing on the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface; the video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function;

[0187] The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file can perform algorithm release operations, and the algorithm release operations include one or more of the following operations: NPU model unloading and removal operations, NPU resource release operations, and system resource release operations. The algorithm release result is sent to the algorithm end release interface.

[0188] The description configuration file includes:

[0189] Upgrade package name, used to identify the algorithm upgrade package;

[0190] An upgrade package publishing platform, used to identify the manufacturer of the algorithm upgrade package;

[0191] The SDK version number of the upgrade package is used to check the compatibility of the algorithm upgrade package;

[0192] The upgrade package MD5 verification code is used to verify the integrity and compliance of the algorithm upgrade package;

[0193] An upgrade package mandatory upgrade code, used to determine whether the algorithm upgrade package needs mandatory upgrade;

[0194] Storage length, used to determine the storage space required when the algorithm upgrade package is running;

[0195] The memory length is used to determine the memory space required when the algorithm upgrade package is running.

[0196] The interface function implementation file also includes a copyright verification function implementation file; the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, including:

[0197] In response to the upgrade package verification request sent by the algorithm initialization interface, the server process calls the copyright verification function implementation file; the upgrade package verification request includes the toolkit version information of the software development toolkit;

[0198] The copyright verification function implementation file obtains the upgrade package SDK version number, the upgrade package MD5 verification code, the toolkit version information, and the IP information and MAC information of the server module, performs copyright verification, and generates a copyright verification result;

[0199] The copyright verification function implementation file returns the copyright verification result to the algorithm initialization interface.

[0200] The client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, and also includes:

[0201] In response to the algorithm model loading request sent by the algorithm initialization interface, the server process calls the registered algorithm initialization interface function implementation file, and the algorithm initialization interface function implementation file loads the artificial intelligence image recognition algorithm module into the memory and NPU of the server module, and returns the algorithm model loading result to the algorithm initialization interface after loading is completed;

[0202] In response to the parameter adjustment request sent by the parameter configuration interface, the server process calls the registered parameter configuration interface function implementation file, inputs the parameter set in the parameter adjustment request into the artificial intelligence image algorithm module, tests the artificial intelligence image algorithm module, and returns the parameter adjustment result to the parameter configuration interface;

[0203] In response to the recognition result release request sent by the detection and recognition result release interface, the server process calls the registered detection and recognition result release interface function implementation file, the detection and recognition result release interface function implementation file clears the historical image cache data, releases the memory occupied by the historical image cache data, and returns the historical image cache release result to the recognition result release interface;

[0204] In response to the video frame image sent by the video frame detection and recognition interface, the server process calls the pre-processing function in the registered video frame detection and recognition processing interface function implementation file to generate pre-processing data, the pre-processing function calls the artificial intelligence image recognition algorithm model to process the pre-processing data to generate recognition data, the recognition data is input into the post-processing function to generate post-processing data, and the post-processing data is sent to the video frame detection and recognition interface;

[0205] In response to the algorithm release request issued by the algorithm end release interface, the server process calls the registered algorithm end release interface function implementation file, the algorithm end release interface function implementation file executes the algorithm release operation, obtains the algorithm release result, and sends the algorithm release result to the algorithm end release interface.

[0206] The video frame detection and recognition processing interface also includes:

[0207] The standard foreign body code function is used to generate a hidden danger identification code according to the post-processing data, wherein the hidden danger identification code is an identification code preset according to the type of hidden danger of the transmission line; and return the hidden danger identification code to the video frame detection and identification interface.

[0208] The server module communicates with the client module via TCP / IP protocol.

[0209] Figure 6 A schematic diagram showing the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure is shown.

[0210] like Figure 6 As shown, the computer system includes a processing unit, which can perform the various methods in the above-mentioned embodiments according to the program stored in the read-only memory (ROM) or the program loaded from the storage part into the random access memory (RAM). In the RAM, various programs and data required for the operation of the computer system are also stored. The processing unit, ROM and RAM are connected to each other through a bus. The input / output (I / O) interface is also connected to the bus.

[0211] The following components are connected to the I / O interface: an input part including a keyboard, a mouse, etc.; an output part including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage part including a hard disk, etc.; and a communication part including a network interface card such as a LAN card, a modem, etc. The communication part performs a communication process via a network such as the Internet. The drive is also connected to the I / O interface as needed. Removable media, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive as needed so that the computer program read therefrom is installed into the storage part as needed. Among them, the processing unit can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, an NPU, etc.

[0212] In particular, according to an embodiment of the present disclosure, the method described above can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes a program code for executing the above method. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium.

[0213] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0214] The units or modules involved in the embodiments described in the present disclosure may be implemented by software or programmable hardware. The units or modules described may also be set in a processor, and the names of these units or modules do not constitute limitations on the units or modules themselves in some cases.

[0215] As another aspect, the present disclosure further provides a computer-readable storage medium, which may be a computer-readable storage medium included in the electronic device or computer system in the above embodiment; or a computer-readable storage medium that exists independently and is not assembled into a device. The computer-readable storage medium stores one or more programs, and the programs are used by one or more processors to execute the method described in the present disclosure.

[0216] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present disclosure is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other.

Claims

1. A method for upgrading an artificial intelligence image recognition algorithm for a power transmission monitoring system, characterized in that: The power transmission monitoring system includes a server module and a client module, the client module includes a software development kit, the software development kit includes a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module through the standardized interface for power transmission monitoring and the server module includes a server process, and the method includes the server module performing the following operations: Obtain an algorithm upgrade package, the algorithm upgrade package including an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files; Input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package; In response to the verification pass result sent by the compliance verification algorithm, updating the artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file; Registering a plurality of interface function implementation files in the algorithm service process file to the server process; The client module enables the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting; in: The client module enables the artificial intelligence image recognition algorithm module to perform corresponding operations by calling the standardized interface for power transmission monitoring and shooting, including: the client module sends an operation request to the server process of the server module through the standardized interface for power transmission monitoring and shooting, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the standardized interface for power transmission monitoring and shooting.

2. The method according to claim 1, wherein: The transmission monitoring and shooting standardized interface includes: Algorithm initialization interface, used to perform copyright protection on the algorithm upgrade package and initialize loading of the artificial intelligence image recognition algorithm module; A parameter configuration interface, used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result; Detecting the recognition result release interface, used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm; A video frame detection and recognition processing interface, used to send an image recognition processing request to the server module and obtain the image recognition result sent by the artificial intelligence image recognition algorithm module; The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources.

3. The method according to claim 2, characterized in that The interface function implementation file includes: An algorithm initialization interface function implementation file, used to respond to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface; A parameter configuration interface function implementation file, used to respond to a parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface; A detection and identification result release interface function implementation file, which is used to respond to a result release request sent by the detection and identification result release interface, perform a detection result release operation, and send the identification result release result to the detection and identification result release interface; A video frame detection and recognition processing interface function implementation file, which is used to respond to the image recognition processing request, perform pre-processing and post-processing on the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface; the video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function; The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file can perform algorithm release operations, and the algorithm release operations include one or more of the following operations: NPU model unloading and removal operations, NPU resource release operations, and system resource release operations. The algorithm release result is sent to the algorithm end release interface.

4. The method according to claim 3, characterized in that The description configuration file includes: Upgrade package name, used to identify the algorithm upgrade package; An upgrade package publishing platform, used to identify the manufacturer of the algorithm upgrade package; The SDK version number of the upgrade package is used to check the compatibility of the algorithm upgrade package; The upgrade package MD5 verification code is used to verify the integrity and compliance of the algorithm upgrade package; An upgrade package mandatory upgrade code, used to determine whether the algorithm upgrade package needs mandatory upgrade; Storage length, used to determine the storage space required when the algorithm upgrade package is running; The memory length is used to determine the memory space required when the algorithm upgrade package is running.

5. The method according to claim 4, characterized in that The interface function implementation file also includes a copyright verification function implementation file; the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, including: In response to the upgrade package verification request sent by the algorithm initialization interface, the server process calls the copyright verification function implementation file; the upgrade package verification request includes the toolkit version information of the software development toolkit; The copyright verification function implementation file obtains the upgrade package SDK version number, the upgrade package MD5 verification code, the toolkit version information, and the IP information and MAC information of the server module, performs copyright verification, and generates a copyright verification result; The copyright verification function implementation file returns the copyright verification result to the algorithm initialization interface.

6. The method according to claim 3, characterized in that The client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface, and also includes: In response to the algorithm model loading request sent by the algorithm initialization interface, the server process calls the registered algorithm initialization interface function implementation file, and the algorithm initialization interface function implementation file loads the artificial intelligence image recognition algorithm module into the memory and NPU of the server module, and returns the algorithm model loading result to the algorithm initialization interface after loading is completed; In response to the parameter adjustment request sent by the parameter configuration interface, the server process calls the registered parameter configuration interface function implementation file, inputs the parameter set in the parameter adjustment request into the artificial intelligence image algorithm module, tests the artificial intelligence image algorithm module, and returns the parameter adjustment result to the parameter configuration interface; In response to the recognition result release request sent by the detection and recognition result release interface, the server process calls the registered detection and recognition result release interface function implementation file, the detection and recognition result release interface function implementation file clears the historical image cache data, releases the memory occupied by the historical image cache data, and returns the historical image cache release result to the recognition result release interface; In response to the video frame image sent by the video frame detection and recognition interface, the server process calls the pre-processing function in the registered video frame detection and recognition processing interface function implementation file to generate pre-processing data, the pre-processing function calls the artificial intelligence image recognition algorithm model to process the pre-processing data to generate recognition data, the recognition data is input into the post-processing function to generate post-processing data, and the post-processing data is sent to the video frame detection and recognition interface; In response to the algorithm release request issued by the algorithm end release interface, the server process calls the registered algorithm end release interface function implementation file, the algorithm end release interface function implementation file executes the algorithm release operation, obtains the algorithm release result, and sends the algorithm release result to the algorithm end release interface.

7. The method according to claim 6, characterized in that The video frame detection and recognition processing interface also includes: The standard foreign body code function is used to generate a hidden danger identification code according to the post-processing data, wherein the hidden danger identification code is an identification code preset according to the type of hidden danger of the transmission line; and return the hidden danger identification code to the video frame detection and identification interface.

8. The method according to claim 1, characterized in that The server module communicates with the client module via TCP / IP protocol.

9. An artificial intelligence image recognition algorithm upgrade device for a power transmission monitoring system, characterized in that: The power transmission monitoring system includes a server module and a client module, the client module includes a software development kit, the software development kit includes a standardized interface for power transmission monitoring and a compliance verification algorithm, the client module communicates with the server module through the standardized interface for power transmission monitoring and the server module includes a server process, and the device includes: An acquisition module is configured to acquire an algorithm upgrade package, wherein the algorithm upgrade package includes an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, wherein the algorithm upgrade package is encrypted and compiled through the software development kit, and the algorithm service process file includes a plurality of interface function implementation files; A verification module is configured to input the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance verification on the algorithm upgrade package; An update module, configured to update the artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file in response to the verification pass result sent by the compliance verification algorithm; A registration module, configured to register a plurality of interface function implementation files in the algorithm service process file to the server process; A calling module configured so that the client module can call the power transmission monitoring and shooting standardized interface to enable the artificial intelligence image recognition algorithm model to perform corresponding operations; Wherein, the client module includes the standardized interface for power transmission monitoring and shooting, and the calling module includes: A sending submodule, configured so that the client module sends an operation request to the server process of the server module through the power transmission monitoring and shooting standardized interface; A calling submodule, configured so that the server process calls the corresponding interface function implementation file according to the operation request; An operation submodule, configured as the interface function implementation file to perform corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request; The return submodule is configured so that the interface function implementation file returns the request result to the power transmission monitoring and shooting standardized interface.

10. The device according to claim 9, wherein: The transmission monitoring and shooting standardized interface includes: Algorithm initialization interface, used to perform copyright protection on the algorithm upgrade package and initialize loading of the artificial intelligence image recognition algorithm module; A parameter configuration interface, used to send a parameter configuration request to the server module so that the artificial intelligence image recognition algorithm module optimizes the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result; Detecting the recognition result release interface, used to send a result release request to the server module to release the corresponding resources of the recognition result of the artificial intelligence image recognition algorithm; A video frame detection and recognition processing interface, used to send an image recognition processing request to the server module and obtain the image recognition result sent by the artificial intelligence image recognition algorithm module; The algorithm end release interface is used to send an algorithm release request to the server module to release the algorithm resources of the artificial intelligence image recognition algorithm, and the algorithm resources include NPU resources and system resources.

11. The device according to claim 10, characterized in that The interface function implementation file includes: An algorithm initialization interface function implementation file, used to respond to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send the loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface; A parameter configuration interface function implementation file, used to respond to a parameter adjustment request sent by the parameter configuration interface, send the adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send the parameter adjustment result to the parameter configuration interface; A detection and identification result release interface function implementation file, which is used to respond to a result release request sent by the detection and identification result release interface, perform a detection result release operation, and send the identification result release result to the detection and identification result release interface; A video frame detection and recognition processing interface function implementation file, which is used to respond to the image recognition processing request, perform pre-processing and post-processing on the image to be processed in the image recognition processing request, call the artificial intelligence image recognition algorithm module to perform image recognition, and send the image recognition result to the video frame detection and recognition processing interface; the video frame detection and recognition processing interface function implementation file includes a pre-processing function and a post-processing function; The algorithm end release interface function implementation file is used to respond to the algorithm release request. The algorithm end release interface function implementation file can perform algorithm release operations, and the algorithm release operations include one or more of the following operations: NPU model unloading and removal operations, NPU resource release operations, and system resource release operations. The algorithm release result is sent to the algorithm end release interface.

12. An electronic device, characterized in that: The method comprises a memory and a processor; wherein the memory is used to store computer instructions, wherein the computer instructions are executed by the processor to implement the method steps described in any one of claims 1 to 8.

13. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the method steps described in any one of claims 1 to 8 are implemented.

Images