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

Abstract

The present disclosure relates to the technical field of artificial intelligence, in particular to an artificial intelligence image recognition algorithm upgrading method and device for a power transmission monitoring system, an electronic device and a readable storage medium, the method comprising: obtaining an algorithm upgrading package; inputting a description configuration file into a compliance verification algorithm through a power transmission monitoring standardized interface to perform compliance verification on the algorithm upgrading package; in response to a verification pass result sent by the compliance verification algorithm, updating an artificial intelligence image recognition algorithm model in a server module based on a model group file of the artificial intelligence image recognition algorithm; registering a plurality of interface function implementation files in an algorithm service process file to the server process; and enabling the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the power transmission monitoring standardized interface through a client module. The present disclosure can realize upgrading operations of artificial intelligence image recognition algorithms of multiple manufacturers in the same power transmission monitoring system.

Description

TECHNICAL FIELD

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

[0002] In a power transmission system, the power transmission line is the artery of power transmission, and its safe and stable operation is crucial. However, as the power transmission line is widely distributed in outdoor environments and is exposed to natural conditions for a long time, it is easily disturbed by various external factors, especially the attachment or suspension of foreign objects. These hidden dangers not only may cause short circuits, broken lines and other faults, but also may pose a threat to the surrounding environment and personnel safety. The power transmission line monitoring and shooting device, as an important part of modern power monitoring technology, realizes all-weather and dead-angle-free monitoring of the power transmission line by integrating high-definition cameras, image sensors and advanced image recognition algorithms. These devices can capture image information of the power transmission line in real time, and automatically identify and classify foreign objects in the image with the help of image recognition technology, thereby effectively warning potential safety hazards.

[0003] With the rapid development of machine learning algorithms, there are many developers of image recognition algorithms and the update speed is fast. The image recognition algorithm in the power transmission monitoring and shooting system needs to be updated continuously to improve performance. Because the algorithm library storage and calling logic of different image recognition algorithm manufacturers are different, the compatibility and interoperability between the image recognition algorithm of other manufacturers and the power transmission line monitoring and shooting device are poor, and the power transmission line monitoring and shooting device has the problem that it cannot apply the image recognition algorithm of other manufacturers. In addition, because there are many versions and developers of image recognition algorithms, how to avoid applying unauthorized image recognition algorithms to the power transmission line monitoring and shooting device is also a problem to be solved. SUMMARY

[0004] To solve the problems in the related art, the present disclosure provides an artificial intelligence image recognition algorithm upgrading method and device for a power transmission monitoring and shooting system, an electronic device and a readable storage medium.

[0005] In a first aspect, an artificial intelligence image recognition algorithm upgrading method for a power transmission monitoring and shooting system is provided in the embodiments of the present disclosure. The power transmission monitoring and shooting system includes a server module and a client module. The client module includes a software development kit, and the software development kit includes a power transmission monitoring and shooting standardized interface and a compliance verification algorithm. The client module communicates with the server module through the power transmission monitoring and shooting standardized interface. The server module includes a server process. The method includes the following operations performed by the server module:

[0006] An algorithm upgrade package is acquired, 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 being encrypted and compiled by the software development kit, the algorithm service process file including a plurality of interface function implementation files;

[0007] The description configuration file is input to the compliance verification algorithm through the power transmission monitoring and photographing standardized interface, and the algorithm upgrade package is subjected to compliance verification;

[0008] In response to a verification pass 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, and a plurality of interface function implementation files in the algorithm service process file are registered to the server process;

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

[0010] Among them:

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

[0012] According to embodiments of the present disclosure, the power transmission monitoring and photographing standardized interface includes:

[0013] An algorithm initialization interface is configured to protect the copyright of the algorithm upgrade package and initialize and load the artificial intelligence image recognition algorithm model;

[0014] A parameter configuration interface is configured to send a parameter configuration request to the server module, so that the artificial intelligence image recognition algorithm model optimizes an image recognition result, and the parameter configuration request includes a parameter set for adjusting the image recognition result;

[0015] A detection recognition result release interface is configured to send a result release request to the server module, so as to release corresponding resources of a recognition result of the artificial intelligence image recognition algorithm;

[0016] The video frame detection and recognition processing interface is configured to send an image recognition processing request to the server module, and acquire the image recognition result sent by the artificial intelligence image recognition algorithm module.

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

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

[0019] The algorithm initialization interface function implementation file is configured to, in response to an algorithm model loading request sent by the algorithm initialization interface, load the artificial intelligence image recognition algorithm model, and send a loading result of the artificial intelligence image recognition algorithm model to the algorithm initialization interface.

[0020] The parameter configuration interface function implementation file is configured to, in response to a parameter adjustment request sent by the parameter configuration interface, send an adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send a parameter adjustment result to the parameter configuration interface.

[0021] The detection and recognition result release interface function implementation file is configured to, in response to a result release request sent by the detection and recognition result release interface, perform a detection result release operation, and send a recognition result release result to the detection and recognition result release interface.

[0022] The video frame detection and recognition processing interface function implementation file is configured to, in response to the image recognition processing request, perform pre-processing and post-processing on a to-be-processed image 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.

[0023] The algorithm end release interface function implementation file is configured to respond to the algorithm release request, and the algorithm end release interface function implementation file can perform an algorithm release operation, the algorithm release operation including one or more of the following operations: an NPU model uninstallation removal operation, an NPU resource release operation, and a system resource release operation, and the algorithm end release interface function implementation file sends an algorithm release result to the algorithm end release interface.

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

[0025] The upgrade package name is configured to identify the algorithm upgrade package.

[0026] The upgrade package release platform is configured to identify a manufacturer of the algorithm upgrade package.

[0027] An SDK version number of the upgrade package, used for compatibility checking of the algorithm upgrade package;

[0028] An MD5 check code of the upgrade package, used for checking integrity and compliance of the algorithm upgrade package;

[0029] An upgrade forced upgrade code of the upgrade package, used for judging whether the algorithm upgrade package needs to be forced to upgrade;

[0030] A storage length, used for determining a storage space required when the algorithm upgrade package runs;

[0031] A memory length, used for determining a memory space required when the algorithm upgrade package runs.

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

[0033] In response to an upgrade package check request sent by the algorithm initialization interface, the server process calls the copyright check function implementation file; the upgrade package check request comprises SDK version information of the software development kit;

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

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

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

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

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

[0041] 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.

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

[0043] A standard foreign matter code function is configured to generate a hidden danger identification code according to the post-processing data, the hidden danger identification code being an identification code preset according to a power transmission line hidden danger type, 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 and the client module communicate through a TCP / IP protocol.

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

[0046] 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 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, the server module comprises a server process, and the device comprises:

[0047] An acquisition module is configured to acquire an algorithm upgrading package, the algorithm upgrading package comprising an artificial intelligence image recognition algorithm model group file, an algorithm service process file and a description configuration file, the algorithm upgrading package being encrypted and compiled through the software development kit, and the algorithm service process file comprising 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 standardized interface and perform compliance verification on the algorithm upgrading package;

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

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

[0051] A calling module is configured to cause the artificial intelligence image recognition algorithm model to perform corresponding operations through the power transmission monitoring standardized interface by the client module;

[0052] The client module comprises the power transmission monitoring standardized interface, the calling module, and the calling module comprises:

[0053] A sending submodule is configured to send an operation request to the server process of the server module through the power transmission monitoring standardized interface by the client module;

[0054] A calling submodule is configured to call the corresponding interface function implementation file according to the operation request by the server process;

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

[0056] The return submodule is configured to return a request result to the power transmission monitoring standard interface by the interface function implementation file.

[0057] According to an embodiment of the present disclosure, the power transmission monitoring standard interface comprises:

[0058] An algorithm initialization interface is configured to protect the copyright of the algorithm upgrade package, and initialize and load the artificial intelligence image recognition algorithm module.

[0059] A parameter configuration interface is configured 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 comprises a parameter set for adjusting the image recognition result.

[0060] A detection recognition result release interface is configured to send a result release request to the server module, and release corresponding resources of the recognition result of the artificial intelligence image recognition algorithm.

[0061] A video frame detection and recognition processing interface is configured to send an image recognition processing request to the server module, and acquire the image recognition result sent by the artificial intelligence image recognition algorithm module.

[0062] An algorithm end release interface is configured to send an algorithm release request to the server module, and release algorithm resources of the artificial intelligence image recognition algorithm, wherein the algorithm resources comprise NPU resources and system resources.

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

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

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

[0066] A detection recognition result release interface function implementation file is configured to respond to a result release request sent by the detection recognition result release interface, execute a detection result release operation, and send a recognition result release result to the detection recognition result release interface.

[0067] The video frame detection and recognition processing interface function implementation file is configured to respond to the image recognition processing request, pre-process and post-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 comprises a pre-processing function and a post-processing function.

[0068] The algorithm end release interface function implementation file is configured to respond to the algorithm release request, and the algorithm end release interface function implementation file is capable of performing an algorithm release operation, the algorithm release operation comprising one or more of the following operations: an NPU model unloading removal operation, an NPU resource release operation, and a system resource release operation, and sending an algorithm release result to the algorithm end release interface.

[0069] In a third aspect, an electronic device is provided, including a memory and a processor, wherein the memory is configured to store one or more computer instructions, and the one or more computer instructions are executed by the processor to implement the method of any one of the first aspect.

[0070] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium has stored thereon computer instructions, which are executed by a processor to implement the method of the first aspect.

[0071] According to the technical scheme provided by the embodiment of the present disclosure, a kind of artificial intelligence image recognition algorithm upgrading method of power transmission monitoring and shooting system is disclosed, the power transmission monitoring and shooting system includes server module and client module, the client module includes software development kit, the software development kit includes power transmission monitoring and shooting standard interface and compliance verification algorithm, the client module communicates with the server module by the power transmission monitoring and shooting standard interface, the server module includes server process, the method includes the following operations performed by the server module: obtain algorithm upgrade package, the algorithm upgrade package includes artificial intelligence image recognition algorithm model group file, algorithm service process file, description configuration file, the algorithm upgrade package is encrypted and compiled by the software development kit, the algorithm service process file includes multiple interface function implementation files;The description configuration file is input into the compliance verification algorithm by the power transmission monitoring and shooting standard interface, and the compliance of the algorithm upgrade package is verified;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 the multiple interface function implementation files in the algorithm service process file are registered to the server process;The client module makes the artificial intelligence image recognition algorithm model execute corresponding operation by calling the power transmission monitoring and shooting standard interface;Wherein: the client module makes the artificial intelligence image recognition algorithm module execute corresponding operation by calling the power transmission monitoring and shooting standard interface, including: the client module sends operation request to the server process of the server module by the power transmission monitoring and shooting standard interface, the server process calls corresponding interface function implementation file according to the operation request, the interface function implementation file performs corresponding operation on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns request result to the power transmission monitoring and shooting standard interface.

[0072] The present disclosure can decouple the algorithm upgrade package and the client module through the power transmission monitoring and shooting standard interface in the software development kit and the communication architecture of client module-server module separation, so that the same power transmission monitoring and shooting system can be applied to the upgrading operation of artificial intelligence image recognition algorithms of multiple manufacturers.

[0073] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0074] Other features, objects, and advantages of the present disclosure will become more apparent from the following detailed description when read in conjunction with the accompanying drawings. In the drawings:

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

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

[0077] Figure 3 A schematic diagram of an interface function implementation file in an artificial intelligence image recognition algorithm upgrading method 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 of 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 structural schematic diagram 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 as to make those skilled in the art readily implement them. Also, portions irrelevant to the description of the exemplary embodiments have been omitted in the accompanying drawings for the sake of clarity.

[0082] In the present disclosure, it should be understood that terms such as “include” or “have” are intended to indicate that there are features, numbers, steps, actions, components, parts or combinations thereof disclosed in the specification, and do not exclude the possibility that one or more other features, numbers, steps, actions, components, parts or combinations thereof exist or are added.

[0083] It is further noted that the embodiments in the present disclosure and the features in the embodiments can be combined with each other without conflict. The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0084] In the present disclosure, if the operation of acquiring user information or user data or the operation of showing user information or user data to others is involved, the operation is an operation authorized, confirmed by the user, or actively selected by the user.

[0085] In a power transmission system, the power transmission line as the artery of electric energy transmission, its safe and stable operation is of great importance. However, due to the wide distribution of the power transmission line in the outdoor environment, long-term exposure to natural conditions, it is easy to be disturbed by various external factors, especially the attachment or suspension of foreign matters. These hidden dangers not only may cause short circuit, broken line and other faults, but also may pose a threat to the surrounding environment and personnel safety. The power transmission line monitoring device as an important part of modern power monitoring technology, through the integration of high-definition camera, image sensor and advanced image recognition algorithm, realizes all-weather and dead angle-free monitoring of the power transmission line. These devices can capture the image information of the power transmission line in real time, and with the help of image recognition technology, automatically identify and classify the foreign matters in the image, so as to effectively warn the potential safety hazards.

[0086] With the rapid development of machine learning algorithms, there are many developers of image recognition algorithms and the update speed is fast. Due to the different storage and calling logic of algorithm library of different image recognition algorithm manufacturers, the compatibility and interoperability between the image recognition algorithms of other manufacturers and the power transmission line monitoring device are poor, and the power transmission line monitoring device has the problem of being unable to apply the image recognition algorithms of other manufacturers. In addition, due to the large number of versions and development manufacturers of image recognition algorithms, how to avoid applying unauthorized image recognition algorithms to the power transmission line monitoring device is also a problem to be solved.

[0087] In order to solve the above technical problems, the application discloses an artificial intelligence image recognition algorithm upgrading method of a power transmission monitoring and shooting system, the power transmission monitoring and shooting system comprises a server module and a client module, the client module comprises a software development kit, the software development kit comprises a power transmission monitoring and shooting standardized interface and a compliance verification algorithm, the client module communicates with the server module through the power transmission monitoring and shooting standardized interface, the server module comprises a server process, and the method comprises the following operations performed by the server module: acquiring 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, the algorithm service process file comprises a plurality of interface function implementation files; inputting the description configuration file into the compliance verification algorithm through the power transmission monitoring and shooting standardized interface, and performing compliance verification on the algorithm upgrading package; in response to a verification passing result sent by the compliance verification algorithm, updating an artificial intelligence image recognition algorithm model in the server module based on the artificial intelligence image recognition algorithm model group file; registering the plurality of interface function implementation files in the algorithm service process file to the server process; the client module makes the artificial intelligence image recognition algorithm model perform corresponding operations by calling the power transmission monitoring and shooting standardized interface; wherein: the client module makes the artificial intelligence image recognition algorithm model perform corresponding operations by calling the power transmission monitoring and shooting standardized interface, which comprises 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 a request result to the power transmission monitoring and shooting standardized interface.

[0088] According to the technical solution of the present disclosure, the power transmission monitoring and shooting system is divided into a server module and a client module, a software development kit is set in the client, a power transmission monitoring and shooting 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 and shooting standardized interface, an interface function implementation file is registered in the server process of the server module, after the server process receives the image processing and other requests sent by the power transmission monitoring and shooting standardized interface, the corresponding interface function implementation file is called for processing, the processing result is sent to the power transmission monitoring and shooting standardized interface by the interface function implementation file, and the client module receives the processing result through the power transmission monitoring and shooting standardized interface. Through the power transmission monitoring and shooting standardized interface in the software development kit and the communication architecture of the separation of the client module and the server module, the power transmission monitoring and shooting standardized interface can decouple the algorithm upgrade package and the client module, so that the same power transmission monitoring and shooting system is applicable to the upgrade operation of the artificial intelligence image recognition algorithm of multiple manufacturers; meanwhile, the software development kit also includes a compliance verification algorithm, the secret key information of the algorithm upgrade package is obtained, and the algorithm upgrade package is verified through the compliance verification algorithm, so as to prevent the loading of unauthorized algorithm upgrade packages.

[0089] Figure 1 A flowchart of an artificial intelligence image recognition algorithm upgrade method of a power transmission monitoring and shooting system according to an embodiment of the present disclosure is shown. The power transmission monitoring and shooting 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 and shooting standardized interface and a compliance verification algorithm, the client module communicates with the server module through the power transmission monitoring and shooting standardized interface, and the server module includes a server process. As shown in Figure 1 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 of the power transmission monitoring and shooting standardized interface and calling, and executes the preset function of 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 listened to. When the standardized interface responds to the process and listens to 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 photographing standardized interface, and the algorithm upgrade package is subjected to compliance verification. The software development kit includes a compliance verification algorithm, which is used to verify the compliance of the algorithm upgrade package according to the description configuration file. After the server module obtains the algorithm upgrade package through a network or cloud, the description configuration file of the algorithm upgrade package is extracted, and the compliance verification algorithm in the software development kit is input into the compliance verification algorithm through the power transmission monitoring and photographing 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 suitable for the current power transmission monitoring and photographing system, generates a compliance verification result, and sends the compliance verification result to the server module through the power transmission monitoring and photographing standardized interface.

[0092] According to the embodiments 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, compliance verification is performed. If it is determined to be a non-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 and photographing system is a common device in the power industry to ensure the safety of power transmission lines, and is used to detect whether there are foreign matters affecting power transmission safety on the power transmission line. The core of the power transmission monitoring and photographing system is an artificial intelligence image recognition algorithm. In order to improve the foreign matter detection capability of the power transmission monitoring and photographing system, the power transmission monitoring and photographing system needs to have the capability of upgrading the artificial intelligence image recognition algorithm, especially the capability of upgrading the algorithm across algorithm manufacturers. The upgrade package release platform, i.e., the algorithm development manufacturer, indicates which manufacturer developed the algorithm upgrade package.

[0095] The upgrade package SDK version number is used to check the compatibility of the algorithm upgrade package. The algorithm upgrade package is developed by using the software development kit (SDK), and the SDK version used during development of the algorithm upgrade package needs to be adapted to the version of the software development kit in the client module to receive the algorithm upgrade package.

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

[0097] The upgrade package forces the upgrade code, which is used to determine whether the algorithm upgrade package needs to be forced to upgrade. The algorithm upgrade package includes optional upgrade packages and mandatory upgrade packages. When the upgrade package force upgrade code is optional, the user can autonomously choose whether to update the algorithm upgrade package through the client module. When the upgrade package force upgrade code is mandatory, the subsequent upgrade step is 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 size 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 checksum is stopped and the algorithm upgrade package is 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 checksum is stopped and the algorithm upgrade package is parsed and the user is notified.

[0100] In step S103, in response to the verification pass 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 plurality of interface function implementation files are registered to the server process. When the server module is initialized, the server process registers the interface function implementation file in the algorithm service process file to the server process, so that the interface function implementation file is in a listening state and can respond to the request issued by the power transmission monitoring standard interface.

[0102] In step S105, the client module calls the power transmission monitoring standard interface to make the artificial intelligence image recognition algorithm model perform corresponding operations.

[0103] According to the embodiments of the present disclosure, the client module calls the power transmission monitoring standard interface to make the artificial intelligence image recognition algorithm module perform corresponding operations, including: the client module sends an operation request to a server process of the server module through the power transmission monitoring standard interface, the server process calls a 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 a request result to the power transmission monitoring standard interface.

[0104] According to the embodiments of the present disclosure, the client module can obtain the monitoring image of the power transmission line and receive the operation instruction of the user, and the power transmission monitoring standard interface can send the request of the client module or the obtained image data to the server module.

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

[0106] An algorithm initialization interface is configured to protect the copyright of the algorithm upgrading package and initialize and load the artificial intelligence image recognition algorithm module. The algorithm initialization interface includes a copyright protection function, which can verify the algorithm upgrading package according to a description configuration file of the algorithm upgrading package to determine whether the algorithm upgrading package is an authorized version of a specific algorithm manufacturer. Further, the copyright protection function can also send the description configuration file of the algorithm upgrading package, version information of the software development kit, and version information of the client module to a 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 sent by the client module, send the algorithm initialization request to a server process of the server module, and implement the initialization and loading of the artificial intelligence image recognition algorithm module in the server module by an interface function implementation file corresponding to the algorithm initialization interface in the server process.

[0107] A parameter configuration interface is configured 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 parameter set for adjusting the image recognition result. During the operation of the client module, the user needs to adjust the detection accuracy parameter, processing speed parameter, resolution parameter, and frame rate parameter of the foreign matter. The parameter configuration interface receives a parameter adjustment request sent by the client module, the parameter adjustment request includes the parameter category to be adjusted and the adjusted parameter value, and sends the parameter adjustment request to the server process of the server module. 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] A detection recognition result release interface is configured 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. The image recognition algorithm usually needs a large amount of computing resources such as NPU, memory, and cache during image recognition. After generating the image recognition result, the above-mentioned resources need to be released to perform other image processing or free up computing resources for other tasks, so as to ensure the stability and performance of the system. After the client module receives the image recognition result of the previous frame, the detection recognition result release interface sends a recognition result release request to the detection recognition result release interface before sending the next frame of image data. The detection recognition result release interface sends the recognition result release request to the server process of the server module. The server process executes the operations of closing the opened previous frame image file, cleaning the temporary file and cache, releasing the NPU resource, and clearing the image data buffer by calling the corresponding interface function implementation file.

[0109] A video frame detection and recognition processing interface is configured 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. 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 matter recognition. Finally, the foreign matter recognition result is sent to the video frame detection and recognition processing interface, and then sent to the client module.

[0110] Further, the video frame detection and recognition processing interface further comprises a standard foreign matter code function, configured to generate a hidden danger recognition code according to the post-processing data, the hidden danger recognition code being a recognition code preset according to a power line hidden danger type; and return the hidden danger recognition code to the video frame detection and recognition interface. Since the algorithm upgrade package can be generated by different algorithm manufacturers, the code representation of the image recognition result of different algorithm manufacturers for different power line foreign matters can be different, which can cause the client module to fail to correctly receive the image recognition result. Therefore, according to the application scene of the power line monitoring and shooting, the standard foreign matter code function is set, the standard foreign matter code function includes the foreign matter types that can be involved in the power line monitoring and shooting and sets a unified standard foreign matter code, the image recognition result output by the artificial intelligence image recognition algorithm can be converted into the corresponding standard foreign matter code, and the corresponding standard foreign matter code is sent to the client module. For example, the image recognition result output by the artificial intelligence image recognition algorithm is “CLASSID_PWR_FILM”, that is, plastic film, the standard foreign matter code function converts “CLASSID_PWR_FILM” into “10030200” and sends 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 plastic film, and the standard foreign matter code function also converts “CLASSID_PLASTIC_FILM” into “10030200” and sends to the client module through the video frame detection and recognition processing interface. The standard foreign matter code function in the video frame detection and recognition processing interface can set the image recognition result output by different artificial intelligence image recognition algorithms as a standard code recognized by the client module, can realize the technical effect that the power monitoring and shooting system upgrades the algorithm upgrade package provided by different algorithm manufacturers, and reduces the workload of the algorithm manufacturer in adapting the algorithm upgrade package to the power monitoring and shooting system.

[0111] The algorithm end release interface is configured to send an algorithm release request to the server module to release algorithm resources of the artificial intelligence image recognition algorithm, the algorithm resources including NPU resources and system resources. When the user no longer needs the power monitoring and shooting system to perform the power line foreign matter recognition function, the client module sends an algorithm release request to the algorithm end release interface, the algorithm end release interface sends an algorithm release request to a server process of the server module, and the server process calls a corresponding interface function to realize file implementation to release NPU resources and system resources. Releasing the NPU resources and the system resources includes cleaning up the interface function implementation file, the artificial intelligence image recognition algorithm model, cleaning up temporary files and cache data, and releasing storage space. Then, an algorithm end release result sent by the server process is returned to the client module.

[0112] According to the embodiments of the present disclosure, the power transmission monitoring and photographing standardized interface can standardize the request of the client, so that the request is consistent with the format of the request that can be received by the server module and the server process, and the upgrading and adaptation of the artificial intelligence image recognition algorithm developed by multiple algorithm manufacturers are realized.

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

[0114] The algorithm initialization interface function implementation file is used for responding to an algorithm model loading request sent by the algorithm initialization interface, loading the artificial intelligence image recognition algorithm model, and sending a 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 that in the algorithm model loading request sent by the algorithm initialization interface, and the algorithm initialization interface function implementation file is a callback function and is registered in the server process at server initialization. After the registration is completed, the server process listens to the algorithm initialization interface, so as 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 NPU computing resources in the server module to the artificial intelligence image recognition algorithm model.

[0115] The parameter configuration interface function implementation file is configured to respond to a parameter adjustment request sent by the parameter configuration interface, send an adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model, and send a parameter adjustment result to the parameter configuration interface. The parameter configuration interface function implementation file can read an adjustment parameter category and an adjustment parameter value in the parameter adjustment request sent by the parameter configuration interface, verify the parameter adjustment request according to a preset adjustment parameter category and adjustment parameter value threshold, and adjust the artificial intelligence image recognition algorithm model according to the adjustment parameter category and the adjustment parameter value after the parameter adjustment request is verified. After the parameter adjustment of the artificial intelligence image recognition algorithm model is completed, the parameter adjustment result is returned to the parameter configuration interface. The parameter configuration interface function implementation file needs to be adapted to the parameter configuration interface, that is, a function pointer of the parameter configuration interface function implementation file is the same as a function pointer in the parameter adjustment request sent by the parameter configuration interface, the parameter configuration interface function implementation file is a callback function, and the parameter configuration interface function implementation file is registered in a server process when the server is initialized. After the registration is completed, the server process listens to the parameter configuration interface, so as to respond to the parameter adjustment request sent by the parameter configuration interface.

[0116] The detection recognition result release interface function implementation file is configured to respond to a result release request sent by the detection recognition result release interface, execute a detection result release operation, and send a recognition result release result to the detection recognition result release interface. The detection recognition result release interface function implementation file can perform a detection result release operation on a server module, clear processed image data and related cache files, and occupied NPU resources, and improve the processing efficiency of a next frame of image. The detection result release operation includes one or more of the following: closing an opened last frame of image file, cleaning temporary files and cache, releasing NPU resources, and clearing an image data buffer. The detection recognition result release interface function implementation file needs to be adapted to the detection recognition result release interface, that is, a function pointer of the detection recognition result release interface function implementation file is the same as a function pointer in the result release request sent by the detection recognition result release interface, the detection recognition result release interface function implementation file is a callback function, and the detection recognition result release interface function implementation file is registered in a server process when the server is initialized. After the registration is completed, the server process listens to 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 for responding to the image recognition processing request, pre-processing the image to be processed in the image recognition processing request, calling the artificial intelligence image recognition algorithm module for image recognition, and sending 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. Different artificial intelligence image recognition algorithm modules developed by different algorithm manufacturers need to process the image to be processed according to the requirements of the artificial intelligence image recognition algorithm module before inputting the algorithm module for recognition due to different parameter settings and image data of training sets.

[0118] According to the embodiments of the present disclosure, the pre-processing function is used for one or more of the optimized quality operation, the format uniform operation, and the image enhancement operation on the image to be processed in the image recognition processing request. When performing the optimized quality operation, the pre-processing function can remove noise, enhance contrast, adjust brightness, and the like of the image to be processed, and highlight effective information in the image to be processed; when performing the format uniform operation, the pre-processing function can perform scaling, cropping, format conversion, and the like on the image to be processed, so that the image to be processed has a uniform format and size, and is convenient for processing of the image recognition algorithm module; and the image enhancement operation includes edge detection and texture enhancement operation on the image to be processed, highlights key features in the image to be processed, and improves image recognition efficiency.

[0119] According to the embodiments of the present disclosure, the post-processing function is used for image screening and image correction on the initial recognition result output by the artificial intelligence image recognition algorithm, and improves the image recognition accuracy. The initial recognition result output by the artificial intelligence image recognition algorithm may be caused by image light variation, image background complexity, image definition, and multiple objects in the image, so that the initial recognition result cannot accurately reflect the power line foreign matter in the image to be processed. For example, there are multiple detection boxes and different confidence levels of each detection box, that is, there are multiple power line foreign matters to be judged, and the post-processing function is needed to screen the multiple detection boxes, and according to the screening result, the shape and type of the power line foreign matter are judged.

[0120] An algorithm end release interface function implementation file is configured to respond to the algorithm release request, and the algorithm end release interface function implementation file performs an algorithm release operation, which includes one or more of the following operations: releasing NPU resource occupation, releasing memory resource occupation, cleaning up temporary files, and sending an algorithm release result to the algorithm end release interface. The algorithm end release interface function implementation file can release NPU resources and system resources of the server module, including closing all opened files and handles, clearing data structures in the memory, clearing data structures on the NPU, terminating unnecessary processes, and sending the release result to the algorithm end release interface. The algorithm end release interface function implementation file needs to adapt to 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, and the algorithm end release interface function implementation file is a callback function registered in the server process during server initialization. After the registration is completed, the server process listens to 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] An upgrade package name is used to identify the algorithm upgrade package.

[0123] An upgrade package release platform is used to identify the manufacturer of the algorithm upgrade package.

[0124] An upgrade package SDK version number is used to perform compatibility checking on the algorithm upgrade package. The client module and the server module communicate through the software development kit, so the SDK version number during development of the algorithm upgrade package should be adapted to the software development kit. When the upgrade package SDK version number does not match the software development kit of the client, the upgrade operation is terminated and the algorithm upgrade package is cleared.

[0125] An upgrade package MD5 check code is used to check the integrity and compliance of the algorithm upgrade package. The artificial intelligence image recognition algorithm of the power transmission monitoring and photographing system needs to perform integrity checking and compliance checking on the algorithm upgrade package before upgrading, to ensure that the algorithm upgrade package is an authorized and complete upgrade package, and to prevent the power transmission monitoring and photographing system from working abnormally due to incorrect upgrading.

[0126] An upgrade package forced upgrade code is used to determine whether the algorithm upgrade package needs to be forced to upgrade. The algorithm upgrade package includes a forced upgrade package and a selectable upgrade package, and the compliance checking algorithm in the software development kit identifies whether the algorithm upgrade package is an algorithm upgrade package that needs to be forced to upgrade through the upgrade package forced upgrade code.

[0127] Storage length, used to determine the memory space required when the algorithm upgrade package runs.

[0128] Memory length, used to determine the memory space required when the algorithm upgrade package runs.

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

[0130] In response to an 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 comprises SDK version information of the software development kit.

[0131] The copyright verification function implementation file acquires the SDK version number of the upgrade package, the upgrade package MD5 check code, the SDK version information, and IP information and MAC information of the server module, performs copyright verification, and generates a copyright verification result; and the copyright verification function implementation file returns the copyright verification result to the algorithm initialization interface. In the application scenario of power transmission line monitoring and photographing, the number of power transmission monitoring and photographing systems is large, and the requirements for algorithm upgrade packages are different. Algorithm development manufacturers often need to calculate fees according to the number of upgrades of the power transmission monitoring and photographing systems. The copyright verification function can acquire the SDK version number of the upgrade package, the upgrade package MD5 check code, and the SDK version information, and can also acquire related information of the server module, such as IP information, MAC information, and SN information, to verify whether the algorithm upgrade package is a genuine product. The copyright verification result is sent to the client module through the algorithm initialization interface. Further, when the algorithm development manufacturer sets a network authorization server, the copyright verification function implementation file can also send the above information to the network authorization server of the algorithm development manufacturer to perform authorization billing management and realize 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 a server process of the server module through the power transmission monitoring and photographing standardized interface, the server process calls the corresponding interface function implementation file according to the operation request, the interface function implementation file performs a corresponding operation on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns a request result to the power transmission monitoring and photographing standardized interface, further comprising:

[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, 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.

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

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

[0137] 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 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 removal operation, NPU resource release operation, and system resource release operation.

[0138] According to the embodiments of the present disclosure, the server module and the client module communicate through the TCP / IP protocol.

[0139] According to an embodiment of the present disclosure, the power transmission monitoring and photographing system comprises a client module and a server module, which communicate through a software development kit TCP / IP protocol, the server module registers interface function files in an upgrade installation package into a server process when the server is initialized, sends a request sent by the client module to the server process after standardizing conversion through a power transmission monitoring and photographing standardized interface in the software development kit, the server process calls corresponding interface function files registered into the server process through a callback mode, and the corresponding interface function files perform corresponding operations on the artificial intelligence image recognition algorithm model according to the request content. The technical solution of the present disclosure can decouple the artificial intelligence image recognition algorithm in the client module and the server module, so that the power transmission monitoring and photographing system can be compatible with image recognition algorithms of different algorithm development manufacturers, and the compatibility and upgrade convenience of the power transmission monitoring and photographing system for the artificial intelligence image recognition algorithm are improved.

[0140] Figure 4 A structural block diagram of an artificial intelligence image recognition algorithm upgrade device of a power transmission monitoring and photographing system according to an embodiment of the present disclosure is shown. The power transmission monitoring and photographing system comprises a server module and a client module, the client module comprises a software development kit, the software development kit comprises a power transmission monitoring and photographing standardized interface and a compliance verification algorithm, the client module communicates with the server module through the power transmission monitoring and photographing standardized interface, and the server module comprises a server process. As shown in Figure 4 The artificial intelligence image recognition algorithm upgrade device 400 of the power transmission monitoring and photographing system comprises 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 comprises 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 comprises a plurality of interface function implementation files. The algorithm interface function implementation file is a program code file comprising algorithm logic of the artificial intelligence image recognition algorithm model group file and interface definitions of the power transmission monitoring and photographing standardized interface, is used for responding to a request and calling of the power transmission monitoring and photographing standardized interface, and realizes a function preset in the algorithm interface function implementation file. For example, the algorithm interface function implementation file comprises an image pre-processing function, the image pre-processing function is registered in the server process, listens to a request of the power transmission monitoring and photographing standardized interface, and when the image recognition request sent by the power transmission monitoring and photographing standardized interface is listened to by the standardized interface response process, the server process calls the image pre-processing function to pre-process an image to be processed in the image recognition request.

[0142] The check module is configured to input the description configuration file into the compliance check algorithm through the power transmission monitoring and shooting standardized interface, and perform compliance check on the algorithm upgrade package. The software development kit includes a compliance check algorithm for performing compliance check 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, the description configuration file of the algorithm upgrade package is extracted, and the compliance check algorithm is input into the software development kit through the power transmission monitoring and shooting standardized interface. The compliance check algorithm includes hash check, digital signature check, metadata check, etc. The compliance check algorithm judges whether the algorithm upgrade package is an authorized algorithm upgrade package, and whether the version of the algorithm upgrade package is suitable for the current power transmission monitoring and shooting system. The compliance check result is generated and sent 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 pass result sent by the compliance check algorithm. When the algorithm upgrade package passes the compliance check, 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, thereby updating the artificial intelligence image recognition algorithm model in the server module.

[0144] The registration module is configured to register a plurality of 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 sent by the power transmission monitoring and shooting standardized interface.

[0145] The calling module is configured to cause the artificial intelligence image recognition algorithm model to perform corresponding operations by the client module calling the power transmission monitoring and shooting standardized interface.

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

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

[0148] The calling submodule is configured to call the corresponding interface function implementation file according to the operation request by the server process;

[0149] An operation submodule is configured to cause the interface function implementation file to perform a corresponding operation on the artificial intelligence image recognition algorithm model according to the operation request.

[0150] A return submodule is configured to cause the interface function implementation file to return a request result to the power transmission monitoring and photographing standardized interface.

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

[0152] An algorithm initialization interface is configured to protect the copyright of the algorithm upgrade package and initialize and load the artificial intelligence image recognition algorithm module. The algorithm initialization interface comprises a copyright protection function, which can verify whether the algorithm upgrade package is an authorized version of a specific algorithm manufacturer according to a description configuration file of the algorithm upgrade package. Further, the copyright protection function can send the description configuration file of the algorithm upgrade package, version information of the software development kit, and version information of the client module to a designated server of the specific algorithm manufacturer to calculate copyright fees. In addition, the algorithm initialization interface can receive an algorithm initialization request sent by the client module, send the algorithm initialization request to a server process of the server module, and cause an interface function implementation file corresponding to the algorithm initialization interface in the server process to implement the initialization and loading of the artificial intelligence image recognition algorithm module in the server module.

[0153] A parameter configuration interface is configured to send a parameter configuration request to the server module to cause the artificial intelligence image recognition algorithm module to optimize the image recognition result. The parameter configuration request comprises a parameter set for adjusting the image recognition result. During the operation of the client module, a user needs to adjust the detection precision parameter, processing speed parameter, resolution parameter, and frame rate parameter of a foreign object. The parameter configuration interface receives a parameter adjustment request sent by the client module, the parameter adjustment request comprising a parameter category to be adjusted and an adjusted parameter value, and sends the parameter adjustment request to a server process of the server module. An 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 recognition result release interface is configured to send a result release request to the server module to release 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, cache and the like. After the image recognition result is generated, the above-mentioned resources need to be released to perform other image processing or free up computing resources for other tasks, so as to ensure the stability and performance of the system. After the client module receives the image recognition result of the previous frame, the detection recognition result release interface sends a recognition result release request to the detection recognition result release interface before sending the next frame of image data. The detection recognition result release interface sends the recognition result release request to the server process of the server module. The server process implements the file of the corresponding interface function by calling, and performs operations such as closing the opened previous frame of image file, cleaning the temporary file and cache, releasing the NPU resource, and clearing the image data buffer.

[0155] The video frame detection recognition processing interface is configured 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 recognition processing interface. The video frame detection 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 the image to the artificial intelligence image recognition algorithm model for power line foreign matter recognition. Finally, the foreign matter recognition result is sent to the video frame detection recognition processing interface, and then sent to the client module.

[0156] Further, the video frame detection and recognition processing interface further comprises a standard foreign matter code function, configured to generate a hidden danger recognition code according to the post-processing data, the hidden danger recognition code being a recognition code preset according to a power line hidden danger type; and return the hidden danger recognition code to the video frame detection and recognition interface. Since the algorithm upgrade package can be generated by different algorithm manufacturers, the code representation of the image recognition result of different algorithm manufacturers for different power line foreign matters can be different, which can cause the client module to fail to correctly receive the image recognition result. Therefore, according to the application scene of the power line monitoring and shooting, the standard foreign matter code function is set, the standard foreign matter code function includes the foreign matter types that can be involved in the power line monitoring and shooting and sets a unified standard foreign matter code, the image recognition result output by the artificial intelligence image recognition algorithm can be converted into the corresponding standard foreign matter code, and the corresponding standard foreign matter code is sent to the client module. For example, the image recognition result output by the artificial intelligence image recognition algorithm is “CLASSID_PWR_FILM”, that is, plastic film, the standard foreign matter code function converts “CLASSID_PWR_FILM” into “10030200” and sends 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 plastic film, and the standard foreign matter code function also converts “CLASSID_PLASTIC_FILM” into “10030200” and sends to the client module through the video frame detection and recognition processing interface. The standard foreign matter code function in the video frame detection and recognition processing interface can set the image recognition result output by different artificial intelligence image recognition algorithms as a standard code recognized by the client module, can realize the technical effect that the power monitoring and shooting system upgrades the algorithm upgrade package provided by different algorithm manufacturers, and reduces the workload of the algorithm manufacturer in adapting the algorithm upgrade package to the power monitoring and shooting system.

[0157] An algorithm end release interface is configured to send an algorithm release request to the server module to release algorithm resources of the artificial intelligence image recognition algorithm, the algorithm resources including NPU resources and system resources. When the user no longer needs the power monitoring and shooting system to perform the power line foreign matter recognition function, the client module sends an algorithm release request to the algorithm end release interface, the algorithm end release interface sends an algorithm release request to a server process of the server module, and the server process calls a corresponding interface function to release NPU resources and system resources. Releasing the NPU resources and the system resources includes cleaning the interface function implementation file, the artificial intelligence image recognition algorithm model, cleaning temporary files and cache data, and releasing the storage space in the memory. 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 for implementing specific functions in cooperation with the power transmission monitoring standard interface, and the interface function implementation file includes:

[0159] The algorithm initialization interface function implementation file is configured to load the artificial intelligence image recognition algorithm model in response to an algorithm model loading request sent by the algorithm initialization interface, and send a 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 that in the algorithm model loading request sent by the algorithm initialization interface, and the algorithm initialization interface function implementation file is a callback function registered in a server process during server initialization. After the registration is completed, the server process listens to the algorithm initialization interface 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 NPU computing resources in the server module to the artificial intelligence image recognition algorithm model.

[0160] The parameter configuration interface function implementation file is configured to send an adjustment parameter item in the parameter adjustment request to the artificial intelligence image recognition algorithm model in response to a parameter adjustment request sent by the parameter configuration interface, and send a parameter adjustment result to the parameter configuration interface. The parameter configuration interface function implementation file can read an adjustment parameter category and an adjustment parameter value in the parameter adjustment request sent by the parameter configuration interface, verify the parameter adjustment request according to preset adjustment parameter category and adjustment parameter value thresholds, and adjust the artificial intelligence image recognition algorithm model according to the adjustment parameter category and the adjustment parameter value after the parameter adjustment request is verified. After the parameter adjustment of the artificial intelligence image recognition algorithm model is completed, a parameter adjustment result is returned to the parameter configuration interface. The parameter configuration interface function implementation file needs to adapt to the parameter configuration interface, that is, the function pointer of the parameter configuration interface function implementation file is the same as that in the parameter adjustment request sent by the parameter configuration interface, and the parameter configuration interface function implementation file is a callback function registered in a server process during server initialization. After the registration is completed, the server process listens to the parameter configuration interface to respond to the parameter adjustment request sent by the parameter configuration interface.

[0161] The detection recognition result release interface function implementation file is configured to respond to a result release request sent by the detection recognition result release interface, execute a detection result release operation, and send a recognition result release result to the detection recognition result release interface. The detection recognition result release interface function implementation file can perform a detection result release operation on the server module, clear processed image data and related cache files, and NPU resources occupied, thereby improving the processing efficiency of the next frame of image. The detection result release operation includes one or more of the following: closing an opened last frame of image file, cleaning temporary files and cache, releasing NPU resources, and clearing an image data buffer. The detection recognition result release interface function implementation file needs to adapt to the detection recognition result release interface, that is, the function pointer of the detection recognition result release interface function implementation file is the same as the function pointer in the result release request sent by the detection recognition result release interface. The detection recognition result release interface function implementation file is a callback function and is registered in the server process during server initialization. After registration, the server process listens to the parameter configuration interface to respond to a parameter adjustment request sent by the parameter configuration interface.

[0162] The video frame detection recognition processing interface function implementation file is configured 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 recognition processing interface. The video frame detection recognition processing interface function implementation file includes a pre-processing function and a post-processing function. Different manufacturers develop artificial intelligence image recognition algorithm modules, and 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 due to different parameter settings and image data of training sets.

[0163] According to an embodiment of the present disclosure, the pre-processing function is configured to perform one or more of an optimized quality operation, a format uniform operation, and an image enhancement operation on the image to be processed in the image recognition processing request. When performing the optimized quality operation, the pre-processing function can remove noise from the image to be processed, enhance contrast, adjust brightness, and the like, thereby highlighting effective information in the image to be processed. When performing the format uniform operation, the pre-processing function can perform scaling, cropping, format conversion, and the like on the image to be processed, so that the image to be processed has a uniform format and size, thereby facilitating processing by the image recognition algorithm module. The image enhancement operation includes edge detection and texture enhancement operations on the image to be processed, thereby highlighting key features in the image to be processed and improving image recognition efficiency.

[0164] According to an embodiment of the present disclosure, the post-processing function is used for image screening and image correction on the initial recognition result output by the artificial intelligence image recognition algorithm, so as to improve the image recognition accuracy. The initial recognition result output by the artificial intelligence image recognition algorithm may be caused by image light variation, image background complexity, image definition, and multiple objects contained in the image, so that the initial recognition result cannot accurately reflect the foreign matter of the power transmission line in the image to be processed. For example, there are multiple detection boxes and the confidence of each detection box is different, that is, there are multiple power transmission line foreign matters to be judged. The post-processing function is needed to screen multiple detection boxes, and according to the screening result, the shape and type of the power transmission line foreign matter are judged.

[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 an 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 an algorithm release result to the algorithm end release interface. The algorithm end release interface function implementation file can release NPU resources and system resources of the server module, including closing all opened files and handles, clearing data structures in the memory, clearing data structures on the NPU, terminating unnecessary processes, and sending the release result to the algorithm end release interface. The algorithm end release interface function implementation file needs to adapt to 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 the registration is completed, the server process listens to the algorithm end release interface, so as 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] As Figure 5 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 the embodiment of the present disclosure.

[0168] An artificial intelligence image recognition algorithm upgrading method 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 through the power transmission monitoring standardized interface, the server module comprising a server process, the method comprising the server module performing the following operations:

[0169] obtaining an algorithm upgrading package, the algorithm upgrading package comprising an artificial intelligence image recognition algorithm model group file, an algorithm service process file, and a description configuration file, the algorithm upgrading package being encrypted and compiled through the software development kit, the algorithm service process file comprising a plurality of interface function implementation files;

[0170] inputting the description configuration file into the compliance verification algorithm through the power transmission monitoring standardized interface, and performing compliance verification on the algorithm upgrading package;

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

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

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

[0174] wherein:

[0175] the client module causing the artificial intelligence image recognition algorithm model to perform corresponding operations by calling the power transmission monitoring standardized interface comprises: the client module sending an operation request to a server process of the server module through the power transmission monitoring standardized interface, the server process calling a corresponding interface function implementation file according to the operation request, the interface function implementation file performing a corresponding operation on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returning a request result to the power transmission monitoring standardized interface.

[0176] the power transmission monitoring standardized interface comprises:

[0177] an algorithm initialization interface, used for copyright protection of the algorithm upgrading package and initialization loading of the artificial intelligence image recognition algorithm model;

[0178] A parameter configuration interface is configured 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 parameter set for adjusting the image recognition result;

[0179] A detection recognition result release interface is configured to send a result release request to the server module, so as to release the corresponding resource of the recognition result of the artificial intelligence image recognition algorithm;

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

[0181] An algorithm end release interface is configured to send an algorithm release request to the server module, so as to release the algorithm resource of the artificial intelligence image recognition algorithm, wherein the algorithm resource includes NPU resource and system resource.

[0182] The interface function implementation file includes:

[0183] An algorithm initialization interface function implementation file is configured 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 is configured 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 a parameter adjustment result to the parameter configuration interface;

[0185] A detection recognition result release interface function implementation file is configured to respond to a result release request sent by the detection recognition result release interface, execute a detection result release operation, and send a recognition result release result to the detection recognition result release interface;

[0186] A video frame detection recognition processing interface function implementation file is configured to respond to the image recognition processing request, pre-process and post-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 recognition processing interface; the video frame detection recognition processing interface function implementation file includes a pre-processing function and a post-processing function;

[0187] An algorithm end release interface function implementation file is used to respond to the algorithm release request, and the algorithm end release interface function implementation file can perform an algorithm release operation, which includes one or more of the following operations: an NPU model uninstallation removal operation, an NPU resource release operation, and a system resource release operation, and the algorithm end release interface function implementation file sends an algorithm release result to the algorithm end release interface.

[0188] The description configuration file includes:

[0189] An upgrade package name is used to identify the algorithm upgrade package.

[0190] An upgrade package release platform is used to identify the manufacturer of the algorithm upgrade package.

[0191] An upgrade package SDK version number is used to perform compatibility checking on the algorithm upgrade package.

[0192] An upgrade package MD5 check code is used to check the integrity and compliance of the algorithm upgrade package.

[0193] An upgrade package forced upgrade code is used to determine whether the algorithm upgrade package needs to be forced to upgrade.

[0194] A storage length is used to determine the storage space required when the algorithm upgrade package is running.

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

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

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

[0198] The copyright verification function implementation file acquires the upgrade package SDK version number, the upgrade package MD5 check code, the toolkit version information, and 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 a server process of the server module through the power transmission monitoring and photographing standardized interface, the server process calls a corresponding interface function implementation file according to the operation request, the interface function implementation file performs a corresponding operation on the artificial intelligence image recognition algorithm model according to the operation request, and the interface function implementation file returns a request result to the power transmission monitoring and photographing standardized interface.

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

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

[0205] 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.

[0206] The video frame detection and identification processing interface further includes:

[0207] a standard foreign matter code function for generating a hazard identification code according to the post-processing data, the hazard identification code being an identification code preset according to a power transmission line hazard type; and returning the hazard identification code to the video frame detection and recognition interface.

[0208] The server module and the client module communicate through a TCP / IP protocol.

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

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

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

[0212] In particular, the method described above can be implemented as a computer software program according to the embodiments of the present disclosure. For example, the embodiments of the present disclosure include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program containing program code for executing the above method. In such embodiments, the computer program can be downloaded and installed from a network through the communication portion, and / or installed from a removable medium.

[0213] The flow and block diagrams in the drawings represent possible architectural, functional, and operational architectures of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block can represent a module, a segment, or a portion of code that comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustrations, and combinations thereof, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.

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

[0215] As another aspect, the present disclosure also provides a computer readable storage medium, which can be the computer readable storage medium included in the electronic device or the computer system in the above embodiments, or can exist separately from the device and not be assembled into the device. The computer readable storage medium stores one or more programs for execution by one or more processors to perform the methods described in the present disclosure.

[0216] The above description is merely preferred embodiments of the present disclosure and a description of principles of applied technologies. It should be understood by those skilled in the art that the scope of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and also includes other technical solutions formed by the combinations of the above technical features or equivalent features without departing from the inventive concept. For example, the above technical features can be replaced with the technical features disclosed in the present disclosure (but not limited to) having similar functions to form technical solutions.

Claims

1. A method for upgrading an artificial intelligence image recognition algorithm in 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 (SDK), which 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. The server module includes a server process. The method includes the server module performing the following operations: Obtain the algorithm upgrade package, which 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 using the software development kit. The algorithm service process file includes multiple interface function implementation files. The description configuration file is input into the compliance verification algorithm through the power transmission monitoring standardization interface to perform compliance verification on the algorithm upgrade package; In response to the verification pass 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; Register multiple interface function implementation files in the algorithm service process file to the server process; The client module invokes the standardized interface for power transmission monitoring to enable the artificial intelligence image recognition algorithm model to perform corresponding operations. in: The client module invokes the standardized power transmission monitoring interface to enable the artificial intelligence image recognition algorithm module to perform corresponding operations, including: the client module sending an operation request to the server process of the server module through the standardized power transmission monitoring interface; the server process invoking the corresponding interface function implementation file according to the operation request; the interface function implementation file performing corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request; and the interface function implementation file returning the request result to the standardized power transmission monitoring interface.

2. The method according to claim 1, wherein, The standardized interface for power transmission monitoring includes: An algorithm initialization interface is used to protect the copyright of the algorithm upgrade package and to initialize and load the artificial intelligence image recognition algorithm module. 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 can optimize the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result; 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; The video frame detection and recognition processing interface is 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, including NPU resources and system resources.

3. The method according to claim 2, characterized in that, The interface function implementation file includes: The 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 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 items 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 implementation file for the detection and recognition result release interface function 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 video frame detection and recognition processing interface function implementation file is used to respond to the image recognition processing request, perform preprocessing and postprocessing 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 preprocessing functions and postprocessing functions; 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 the algorithm release operation, which includes one or more of the following operations: NPU model unloading and removal operation, NPU resource release operation, system resource release operation, and send the algorithm release result to the algorithm end release interface.

4. The method according to claim 3, characterized in that, The description configuration file includes: The upgrade package name is used to identify the algorithm upgrade package. An upgrade package distribution platform is used to identify the creators of the algorithm upgrade packages; The upgrade package SDK version number is used to perform compatibility checks on the algorithm upgrade package. The upgrade package's MD5 checksum is used to verify the integrity and compliance of the algorithm upgrade package; The upgrade package forced upgrade code is used to determine whether the algorithm upgrade package needs to be forcibly upgraded; Storage length, used to determine the storage space required for the algorithm upgrade package to run; Memory length, used to determine the memory space required for the algorithm upgrade package to run.

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 standardized power transmission monitoring 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 standardized power transmission monitoring interface, including: In response to the upgrade package verification request sent by the algorithm initialization interface, the server process calls the copyright verification function to implement the file; the upgrade package verification request includes the toolkit version information of the software development kit; The copyright verification function retrieves the upgrade package SDK version number, upgrade package MD5 checksum, tool package version information, and the server module's IP and MAC information, 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 standardized power transmission monitoring 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. The interface function implementation file returns the request result to the standardized power transmission monitoring interface. The system 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. 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. 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 detection and 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 preprocessing function in the file of the registered video frame detection and recognition processing interface function to generate preprocessing data. The preprocessing function calls the artificial intelligence image recognition algorithm model to process the preprocessing data and generate recognition data. The recognition data is then input into the postprocessing function to generate postprocessing data, and the postprocessing 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: A standard foreign object code function is used to generate a hazard identification code based on the post-processed data. The hazard identification code is a preset identification code based on the type of hazard in the power transmission line. The hazard identification code is then returned to the video frame detection and identification interface.

8. The method according to claim 1, characterized in that, The server module and the client module communicate via the 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 (SDK), which 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. The server module includes a server process. The device includes: The acquisition module is configured to acquire an algorithm upgrade package, which 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 using the software development kit, and the algorithm service process file includes multiple interface function implementation files. The verification module is configured to input the description configuration file into the compliance verification algorithm through the power transmission monitoring standardization interface, and perform compliance verification on the algorithm upgrade package; 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 pass result sent by the compliance verification algorithm. The registration module is configured to register multiple interface function implementation files from the algorithm service process file to the server process. The calling module is configured such that the client module invokes the standardized interface for power transmission monitoring to cause the artificial intelligence image recognition algorithm model to perform corresponding operations. The client module includes the standardized interface for power transmission monitoring, and the calling module includes: 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 standardized interface; The calling submodule is configured so that the server process calls the corresponding interface function implementation file according to the operation request; The operation submodule is configured to perform corresponding operations on the artificial intelligence image recognition algorithm model according to the operation request, as described in the interface function implementation file. The return submodule is configured to return the request result to the standardized power transmission monitoring interface as the interface function implementation file.

10. The apparatus according to claim 9, wherein, The standardized interface for power transmission monitoring includes: An algorithm initialization interface is used to protect the copyright of the algorithm upgrade package and to initialize and load the artificial intelligence image recognition algorithm module. 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 can optimize the image recognition result; the parameter configuration request includes a set of parameters for adjusting the image recognition result; 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; The video frame detection and recognition processing interface is 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, including NPU resources and system resources.

11. The apparatus according to claim 10, characterized in that, The interface function implementation file includes: The 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 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 items 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 implementation file for the detection and recognition result release interface function 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 video frame detection and recognition processing interface function implementation file is used to respond to the image recognition processing request, perform preprocessing and postprocessing 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 preprocessing functions and postprocessing functions; 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 the algorithm release operation, which includes one or more of the following operations: NPU model unloading and removal operation, NPU resource release operation, system resource release operation, and send the algorithm release result to the algorithm end release interface.

12. An electronic device, characterized in that, It includes 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 steps of the method according to any one of claims 1-8.

13. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by a processor, the computer instructions implement the steps of the method described in any one of claims 1-8.

Images