Security management method and device, computer equipment, storage medium and program product

Abstract

The invention relates to a security management method and device, computer equipment, a storage medium and a program product. The method comprises the following steps: acquiring an environment image of a nearby area of a power transmission channel; the environment image comprises an engineering vehicle; inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle; and according to the type information of the engineering vehicle, whether the power transmission channel has the risk of being damaged is determined, and safety management is carried out under the condition that the power transmission channel has the risk of being damaged. According to the embodiment of the invention, the type information of the engineering vehicle included in the environment image can be identified, so that whether the power transmission channel has the risk of being damaged or not can be accurately determined according to the type information of the engineering vehicle, and safety management can be effectively carried out under the condition that the power transmission channel has the risk of being damaged.

Description

Technical Field

[0001] The present application relates to the field of power transmission safety technology, and in particular to a safety management method, device, computer equipment, storage medium and program product. Background Art

[0002] As the amount of electricity consumed by residents continues to increase, the demand for electricity in modern society is also increasing. In order to meet the growing demand for electricity, the distribution of transmission channels has become increasingly dense. Large engineering vehicles (such as excavators, cranes, etc.) often need to operate in areas near transmission channels, which increases the risk of transmission channels being damaged. Therefore, the need for safety management of transmission channels has become increasingly urgent.

[0003] Based on this, how to effectively carry out safety management on power transmission channels has become a technical problem that needs to be urgently solved in the field of power transmission safety technology. Summary of the invention

[0004] Based on this, it is necessary to provide a safety management method, device, computer equipment, storage medium and program product that can effectively perform safety management on power transmission channels in response to the above technical problems.

[0005] In a first aspect, the present application provides a security management method, comprising:

[0006] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0007] Inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle;

[0008] Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

[0009] In one embodiment, determining whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle includes:

[0010] Determining the size information of the engineering vehicle according to the type information of the engineering vehicle;

[0011] Determining the distance between the engineering vehicle and the power transmission channel according to the size information of the engineering vehicle;

[0012] Whether the power transmission channel is at risk of being damaged is determined according to the distance between the engineering vehicle and the power transmission channel.

[0013] In one embodiment, the training process of the target recognition network includes:

[0014] Acquire a vehicle image dataset; the vehicle image dataset includes a plurality of vehicle images and label type information of each vehicle;

[0015] The initial recognition network is trained according to the vehicle image data set to obtain the target recognition network.

[0016] In one embodiment, the training of the initial recognition network according to the vehicle image dataset to obtain the target recognition network includes:

[0017] Determining prediction type information and target decoding results according to the vehicle image dataset and the initial recognition network;

[0018] Determine a target loss according to the annotation type information, the prediction type information and the target decoding result;

[0019] The initial recognition network is trained according to the target loss to obtain the target recognition network.

[0020] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the step of determining the prediction type information and the target decoding result according to the vehicle image dataset and the initial recognition network includes:

[0021] Inputting each of the vehicle images into the encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each of the vehicle images;

[0022] Inputting a target encoded image from the plurality of encoding results into the decoding subnetwork for decoding to obtain the target decoding result; the target decoding result includes a vehicle feature area;

[0023] The target decoding result is input into the classifier for classification to obtain the prediction type information.

[0024] In one embodiment, the initial recognition network further includes a segmentation subnetwork, and each of the vehicle images is input into the encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each of the vehicle images, including:

[0025] Inputting each of the vehicle images into the segmentation subnetwork to perform image block segmentation to obtain a plurality of image blocks corresponding to each of the vehicle images;

[0026] The multiple image blocks corresponding to each of the vehicle images are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each of the vehicle images.

[0027] In one of the embodiments, determining the target loss according to the annotation type information, the prediction type information and the target decoding result includes:

[0028] Determining a contrast loss according to the prediction type information and the target decoding result;

[0029] The target loss is determined according to the contrast loss, the prediction type information and the annotation type information.

[0030] In a second aspect, the present application also provides a security management device, including:

[0031] An acquisition module, used to acquire an environmental image of an area near the power transmission channel; the environmental image includes engineering vehicles;

[0032] A target recognition module, used for inputting the environment image into a preset target recognition network for target recognition, and obtaining the type information of the engineering vehicle;

[0033] The safety management module is used to determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and to perform safety management when the power transmission channel is at risk of being damaged.

[0034] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0035] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0036] Inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle;

[0037] Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0039] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0040] Inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle;

[0041] Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

[0042] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0043] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0044] Inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle;

[0045] Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

[0046] The above-mentioned safety management method, device, computer equipment, storage medium and program product obtain an environmental image of the vicinity of the power transmission channel; the environmental image includes engineering vehicles; the environmental image is input into a preset target recognition network for target recognition to obtain type information of the engineering vehicles; based on the type information of the engineering vehicles, it is determined whether the power transmission channel is at risk of being damaged, and safety management is performed when there is a risk of damage to the power transmission channel. The embodiment of the present application can identify the type information of the engineering vehicles included in the environmental image, so that it can accurately determine whether the power transmission channel is at risk of being damaged based on the type information of the engineering vehicles, and thus, safety management can be effectively performed when there is a risk of damage to the power transmission channel. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0048] FIG1 is an application environment diagram of a security management method in one embodiment;

[0049] FIG2 is a schematic diagram of a flow chart of a security management method in one embodiment;

[0050] FIG3 is a schematic diagram showing the principle of determining the distance between an engineering vehicle and a power transmission channel based on pinhole imaging in one embodiment;

[0051] FIG4 is a schematic diagram of a flow chart of a training step in an embodiment;

[0052] FIG5 is a schematic diagram of a flow chart of steps for training a target recognition network in one embodiment;

[0053] FIG6 is a schematic flow chart of a step of determining prediction type information in one embodiment;

[0054] FIG7 is a schematic flow chart of a security management method in another embodiment;

[0055] FIG8 is a structural block diagram of a security management device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0058] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0059] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0060] With the continuous increase in residential electricity consumption, the demand for electricity in modern society is also increasing. In order to meet the growing demand for electricity, the power transmission system continues to expand, and the distribution of transmission channels has become increasingly dense. The dense transmission channel network covers cities, villages and various terrains, among which the terrain may include but is not limited to various complex terrain environments such as mountains, rivers, and forests. These transmission channels often need to cross various obstacles, such as roads, buildings, rivers, etc., and are therefore easily affected and damaged by external factors. In particular, in some engineering construction, road construction and other activities, large engineering vehicles (such as excavators, cranes, etc.) often need to operate in areas near transmission channels, which increases the risk of transmission channels being damaged. If not effectively monitored and managed, it is possible to cause serious damage to transmission lines and facilities, and even cause accidents. Therefore, the need for safe management of transmission channels has become increasingly urgent.

[0061] However, traditional technology can only determine whether there are engineering vehicles in the image of the transmission channel, but cannot measure the distance between the engineering vehicle and the transmission channel, and thus cannot accurately determine whether there is a risk of damage to the transmission channel, thus resulting in invalid alarms or missed reports. Based on this, how to effectively manage the safety of the transmission channel has become a technical problem that needs to be solved in the field of power transmission safety technology.

[0062] After the above introduction to the background technology of the security management method provided by the embodiment of the present application, the implementation environment involved in the security management method provided by the embodiment of the present application will be briefly described below. The security management method provided by the embodiment of the present application can be applied to a computer device as shown in Figure 1. The computer device can be a terminal, or it can also be a server, and the computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a security management method is implemented. The display unit of the computer device is used to form a visually visible picture, and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0063] Those skilled in the art will understand that the structure shown in FIG. 1 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. A specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0064] In one embodiment, as shown in FIG. 2 , a security management method is provided, which is described by taking the method applied to the computer device in FIG. 1 as an example, and includes the following steps:

[0065] S201, acquiring an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles.

[0066] Among them, the vicinity of the power transmission channel refers to a preset area located around the power transmission channel. Of course, the embodiment of the present application does not limit the scope and shape of the preset area. The environmental image refers to the image of the scenery located in the vicinity of the power transmission channel. It should be noted that the environmental image includes engineering vehicles. In the embodiment of the present application, optionally, the computer device can obtain the environmental image of the vicinity of the power transmission channel in real time; or, the computer device can also obtain the environmental image of the vicinity of the power transmission channel at regular intervals. Of course, the embodiment of the present application does not limit the specific implementation method of obtaining the environmental image.

[0067] S202, inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle.

[0068] In the embodiment of the present application, optionally, the computer device can directly input the environment image into a preset target recognition network for target recognition to obtain the type information of the engineering vehicle; or, the computer device can also first perform image preprocessing on the environment image, and then input the environment image after image preprocessing into a preset target recognition network for target recognition to obtain the type information of the engineering vehicle. Of course, the embodiment of the present application does not limit the specific implementation method of target recognition. Among them, the type information of the engineering vehicle refers to the specific model of the engineering vehicle, and the preset target recognition network may include but is not limited to the TransFG (A Transformer Architecture for Fine-Grained Recognition, a Transformer architecture for fine-grained target detection) network and other fine-grained target detection networks, wherein Transformer is a network model for feature extraction based on a self-attention mechanism. Fine-grained target detection is a branch of target detection technology, which mainly focuses on identifying different subcategories of the same category of targets. In the implementation of the present application, considering that the distance measurement technology based on a monocular camera requires the actual size information of the engineering vehicle as a known condition, fine-grained target detection technology can be used to distinguish different types of engineering vehicles to obtain the type information and size information of the engineering vehicle.

[0069] S203: Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

[0070] In an embodiment of the present application, optionally, the computer device may directly determine whether there is a risk of damage to the transmission channel based on the type information of the engineering vehicle, and perform safety management when there is a risk of damage to the transmission channel. Alternatively, the computer device may first determine the distance between the engineering vehicle and the transmission channel based on the type information of the engineering vehicle, thereby determining whether there is a risk of damage to the transmission channel based on the distance between the engineering vehicle and the transmission channel, and perform safety management when there is a risk of damage to the transmission channel. Of course, the embodiment of the present application does not limit the specific implementation method of safety management. Among them, the safety management method may include but is not limited to various warning methods such as voice and interface prompts.

[0071] In the above safety management method, an environmental image of the vicinity of the power transmission channel is obtained; the environmental image includes engineering vehicles; the environmental image is input into a preset target recognition network for target recognition to obtain type information of the engineering vehicles; based on the type information of the engineering vehicles, it is determined whether the power transmission channel is at risk of being damaged, and safety management is performed when there is a risk of damage to the power transmission channel. The embodiment of the present application can identify the type information of the engineering vehicles included in the environmental image, so that it can accurately determine whether the power transmission channel is at risk of being damaged based on the type information of the engineering vehicles, and thus, it can effectively perform safety management when there is a risk of damage to the power transmission channel.

[0072] In one embodiment, a method for determining whether a power transmission channel is at risk of being damaged is provided, that is, the “determining whether a power transmission channel is at risk of being damaged according to the type information of the engineering vehicle” in the above S203 includes:

[0073] According to the type information of the engineering vehicle, the size information of the engineering vehicle is determined.

[0074] According to the size information of the engineering vehicle, determine the distance between the engineering vehicle and the transmission channel.

[0075] Determine whether the transmission channel is at risk of being damaged based on the distance between the engineering vehicle and the transmission channel.

[0076] In the embodiment of the present application, the computer device can determine the size information of the engineering vehicle according to the type information of the engineering vehicle. The size information of the engineering vehicle may include but is not limited to the height information of the engineering vehicle. Thus, the computer device can determine the distance between the engineering vehicle and the power transmission channel according to the size information of the engineering vehicle.

[0077] Exemplarily, the computer device can determine the distance between the engineering vehicle and the power transmission channel based on the triangle similarity relationship of pinhole imaging, as shown in FIG3 . FIG3 is a schematic diagram of the principle of determining the distance between the engineering vehicle and the power transmission channel based on pinhole imaging in one embodiment. Specifically, the height h of the image of the engineering vehicle (i.e., the target to be measured in FIG3 ) in the image at the imaging surface can be determined according to the following formula (1) and known imaging information.

[0078]

[0079] Among them, the known imaging information may include n, N, H sensor , n is the number of pixels occupied by the engineering vehicle in the image at the imaging surface, N is the vertical resolution of the image at the imaging surface (i.e. the total number of pixels in the vertical direction of the image), H sensor H is the vertical dimension of the image sensor used for imaging. sensor is the inherent information of the image sensor.

[0080] Afterwards, in combination with FIG3 , the computer device can determine the distance D between the engineering vehicle and the power transmission channel using the following formula (2) based on the height h of the image formed by the engineering vehicle at the imaging surface, the size information of the engineering vehicle (such as the actual height information H of the engineering vehicle) and the known imaging information.

[0081]

[0082] The known imaging information may also include f, where f is the focal length of the imaging camera, D is the distance between the engineering vehicle and the power transmission channel, H is the actual height of the engineering vehicle, and h is the height of the image of the engineering vehicle in the image.

[0083] Thus, the computer device can determine whether the engineering vehicle will damage the power transmission channel based on the distance between the engineering vehicle and the power transmission channel, so as to determine whether there is a risk of the power transmission channel being damaged.

[0084] In this embodiment, the size information of the engineering vehicle can be accurately determined based on the type information of the engineering vehicle, and the distance between the engineering vehicle and the transmission channel can be accurately determined based on the accurate size information of the engineering vehicle. Thus, based on the distance between the engineering vehicle and the transmission channel, it can be accurately determined whether the transmission channel is at risk of being damaged. In this way, safety management can be effectively performed when there is a risk of the transmission channel being damaged.

[0085] In one embodiment, a training method of a target recognition network is provided, that is, the training process of the target recognition network, as shown in FIG4 , including:

[0086] S301, obtaining a vehicle image dataset; the vehicle image dataset includes a plurality of vehicle images and label type information of each vehicle.

[0087] In an embodiment of the present application, a computer device can acquire multiple candidate vehicle images in the transmission channel area through a monocular camera installed on the transmission channel, and screen the multiple candidate vehicle images to obtain multiple vehicle images including engineering vehicles. Thus, the computer device can annotate the multiple vehicle images to obtain the annotation type information of each vehicle. Exemplarily, the annotation type information can be (x, y, c), where (x, y) represents the center point coordinates of the engineering vehicle, and c represents the subdivision category of the engineering vehicle, such as the model of the excavator, the model of the crane, etc. In this way, a vehicle image dataset can be constructed based on multiple vehicle images and the annotation type information of each vehicle.

[0088] S302, training an initial recognition network according to the vehicle image data set to obtain a target recognition network.

[0089] In the embodiment of the present application, optionally, the computer device can directly train the initial recognition network according to the vehicle image data set to obtain the target recognition network. Alternatively, the computer device can also first determine the prediction type information and the target decoding result according to the vehicle image data set and the initial recognition network, and then train the initial recognition network according to the vehicle image data set, the prediction type information and the target decoding result to obtain the target recognition network. Of course, the embodiment of the present application does not limit the specific implementation method of training the target recognition network.

[0090] In this embodiment, a vehicle image dataset can be obtained, wherein the vehicle image dataset includes multiple vehicle images and label type information of each vehicle, so that the initial recognition network can be accurately trained according to the vehicle image dataset to obtain an accurate target recognition network.

[0091] In one embodiment, a method for implementing a training target recognition network is provided, namely, the “training an initial recognition network according to a vehicle image data set to obtain a target recognition network” in the above S302, as shown in FIG5, including:

[0092] S401, determining prediction type information and target decoding results according to the vehicle image data set and the initial recognition network.

[0093] Among them, the initial recognition network refers to an untrained fine-grained target detection network, the predicted type information refers to the vehicle type information predicted by the initial recognition network, and the target decoding result refers to the encoding result of the image block that can reflect the intra-class difference and has recognition, that is, the encoding result of the characteristic area (such as the vehicle logo, body structure, etc.) that can characterize the difference between different vehicles. In the embodiment of the present application, optionally, the computer device can simultaneously determine the predicted type information and the target decoding result based on the vehicle image data set and the initial recognition network. Alternatively, the computer device can also first determine the target decoding result based on the vehicle image data set and the initial recognition network, and then determine the predicted type information based on the target decoding result and the initial recognition network. Of course, the embodiment of the present application does not limit the specific implementation method for determining the predicted type information and the target decoding result.

[0094] S402, determining a target loss according to the annotation type information, the prediction type information and the target decoding result.

[0095] In the embodiment of the present application, optionally, the computer device can directly determine the target loss based on the annotation type information, the prediction type information and the target decoding result. Alternatively, the computer device can also first determine the contrast loss based on the prediction type information and the target decoding result, and then determine the target loss based on the contrast loss, the prediction type information and the annotation type information. Of course, the embodiment of the present application does not limit the specific implementation method for determining the target loss.

[0096] S403, training the initial recognition network according to the target loss to obtain a target recognition network.

[0097] In an embodiment of the present application, the computer device can adjust the initial model parameters of the initial recognition network according to the target loss to obtain intermediate model parameters, and use the intermediate model parameters as new initial model parameters, thereby using the new initial model parameters to determine the new target loss, and then adjust the new initial model parameters according to the new target loss, and iterate until the new target loss tends to converge. At this time, it means that the recognition effect of the initial recognition network is better, and the new initial model parameters at this time can be determined as the target model parameters, and the target recognition network is obtained according to the target model parameters.

[0098] In this embodiment, the predicted type information and the target decoding result can be accurately determined based on the vehicle image data set and the initial recognition network, so that the target loss can be accurately determined based on the labeled type information, the predicted type information and the target decoding result, and the initial recognition network can be trained based on the target loss to obtain an accurate target recognition network. In this way, it can be ensured that the target recognition network has the ability to distinguish the detailed characteristics of various types of engineering vehicles, so as to provide accurate type information of engineering vehicles for subsequent safety management.

[0099] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier. Based on this, an implementation method for determining the prediction type information and the target decoding result is provided, that is, the “determining the prediction type information and the target decoding result according to the vehicle image data set and the initial recognition network” in S401, as shown in FIG6, including:

[0100] S501, input each vehicle image into the encoding sub-network for feature encoding, and obtain multiple encoding results corresponding to each vehicle image.

[0101] In the embodiment of the present application, since the initial recognition network includes a coding subnetwork, the computer device can input each vehicle image into the coding subnetwork for feature coding to obtain multiple coding results corresponding to each vehicle image.

[0102] In one embodiment, the initial recognition network further includes a segmentation subnetwork, and S501 includes:

[0103] Each vehicle image is input into the segmentation subnetwork for image block segmentation to obtain multiple image blocks corresponding to each vehicle image.

[0104] The multiple image blocks corresponding to each vehicle image are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image.

[0105] In the embodiment of the present application, the computer device can input each vehicle image into the segmentation subnetwork to perform image block segmentation, and obtain multiple image blocks corresponding to each vehicle image. In addition, in order to retain the local adjacent structure between the image blocks, the adjacent image blocks can include a preset overlapping portion during segmentation. For example, assuming that the resolution of the input vehicle image is H W, the size of the image block is P P, the step size of the sliding window is S, then the input vehicle image can be segmented into N image blocks using the following formula (3):

[0106] (3)

[0107] Thus, the size of the space between adjacent image blocks is (PS) The overlapping area of ​​P. Therefore, the computer device can input multiple image blocks corresponding to each vehicle image into the encoding subnetwork for feature encoding according to VIT (Vision Transformer, a Transformer model that uses images as input and is used for image classification). For example, each image block can be converted into a vector by linear mapping, and the position information can be embedded, and then encoded by the encoder of the Transformer to obtain multiple encoding results corresponding to each vehicle image.

[0108] S502, inputting a target encoded image from the plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area.

[0109] In the embodiment of the present application, since the embodiment of the present application emphasizes the subtle differences between different subclasses (more subdivided vehicle types), based on this, the computer device can delete the encoding results of the image blocks that are not distinctive, and determine the encoding results of the image blocks that can reflect the differences within the class and are recognizable as the target encoded image. Thus, the computer device can input the target encoded image from the multiple encoding results into the decoding subnetwork for decoding, that is, the target encoded image can be input into the decoder of the Transformer for decoding to obtain the target decoding result. Among them, the target decoding result includes vehicle feature areas that can characterize the differences between different vehicles.

[0110] S503: Input the target decoding result into a classifier for classification to obtain prediction type information.

[0111] In an embodiment of the present application, the computer device may input the target decoding result into a classifier for classification to obtain predicted type information of each vehicle image.

[0112] In this embodiment, each vehicle image can be input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image. Then, the target encoded image among the multiple encoding results is input into the decoding subnetwork for decoding to obtain the target decoding result, which can accurately determine the encoding result of the feature area that can characterize the difference between different vehicles. Therefore, the target decoding result is input into the classifier for classification, and the prediction type information can be accurately obtained.

[0113] In one embodiment, a method for determining a target loss is provided, namely, the step of “determining a target loss according to the annotation type information, the prediction type information and the target decoding result” in S402, including:

[0114] According to the prediction type information and the target decoding result, the contrast loss is determined.

[0115] The target loss is determined based on the contrast loss, the prediction type information, and the annotation type information.

[0116] In the embodiment of the present application, in order to minimize the similarity of the classification basis between the vehicle image data with different annotated type information and maximize the similarity of the classification basis between the vehicle image data with the same annotated type information, the computer device can determine the contrast loss according to the predicted type information and the target decoding result using the following formula (4): .

[0117]

[0118] Where B represents the batch size of the vehicle image dataset; Represents the classification basis of different vehicle image data, that is, the target decoding result, which is a set of feature vectors; Represents the prediction type information of different vehicle image data; Sim is The dot product of is used to calculate the similarity of the predicted type information; is a constant used to filter samples of different categories with large similarity, thereby improving the model's ability to distinguish similar categories.

[0119] Thus, the computer device can determine the target loss using the following formula (5) based on the contrast loss, prediction type information and annotation type information: .

[0120]

[0121] in, For prediction type information The cross entropy loss between the annotation type information c.

[0122] In this embodiment, the contrast loss can be determined according to the prediction type information and the target decoding result, so that the target loss can be accurately determined according to the contrast loss, the prediction type information and the annotation type information.

[0123] In an optional embodiment, as shown in FIG. 7 , a security management method is provided, which is applied to a computer device, and includes:

[0124] S601, obtaining a vehicle image dataset; the vehicle image dataset includes a plurality of vehicle images and annotation type information of each vehicle;

[0125] S602, inputting each vehicle image into a segmentation subnetwork to perform image block segmentation to obtain a plurality of image blocks corresponding to each vehicle image;

[0126] S603, inputting multiple image blocks corresponding to each vehicle image into the encoding subnetwork for feature encoding, and obtaining multiple encoding results corresponding to each vehicle image;

[0127] S604, inputting a target encoded image from the plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area;

[0128] S605, inputting the target decoding result into a classifier for classification to obtain prediction type information;

[0129] S606, determining a contrast loss according to the prediction type information and the target decoding result;

[0130] S607, determining a target loss according to the comparison loss, the prediction type information, and the annotation type information;

[0131] S608, training the initial recognition network according to the target loss to obtain a target recognition network;

[0132] S609, acquiring an environmental image of the vicinity of the power transmission channel; the environmental image includes engineering vehicles;

[0133] S610, inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle;

[0134] S611, determining size information of the engineering vehicle according to the type information of the engineering vehicle;

[0135] S612, determining the distance between the engineering vehicle and the power transmission channel according to the size information of the engineering vehicle;

[0136] S613, determining whether the transmission channel is at risk of being damaged based on the distance between the engineering vehicle and the transmission channel, and performing safety management when the transmission channel is at risk of being damaged.

[0137] In the above-mentioned safety management method, the type information of the engineering vehicles included in the environmental image can be identified, so that it can be accurately determined whether the transmission channel is at risk of being damaged based on the type information of the engineering vehicles. In this way, safety management can be effectively performed when the transmission channel is at risk of being damaged. The embodiment of the present application combines fine-grained target detection technology with monocular distance measurement technology, analyzes the position and size of the engineering vehicle in the vehicle image through fine-grained detection results, and measures the distance between the engineering vehicle and the transmission channel based on the position and size of the engineering vehicle. In this way, it is possible to more accurately determine whether the engineering vehicle has a potential risk of damage, reduce the occurrence of false alarms and missed alarms, reduce the risk of damage caused by external factors, improve the safety management level of the transmission channel, and ensure the safe and stable operation of the power transmission system.

[0138] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0139] Based on the same inventive concept, the embodiment of the present application also provides a security management device for implementing the security management method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more security management device embodiments provided below can refer to the limitations on the security management method above, and will not be repeated here.

[0140] In an exemplary embodiment, as shown in FIG8 , a security management device is provided, including: an acquisition module 31, a target identification module 32 and a security management module 33, wherein:

[0141] The acquisition module 31 is used to acquire the environmental image of the vicinity of the power transmission channel; the environmental image includes engineering vehicles.

[0142] The target recognition module 32 is used to input the environment image into a preset target recognition network for target recognition to obtain the type information of the engineering vehicle.

[0143] The safety management module 33 is used to determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and to perform safety management when the power transmission channel is at risk of being damaged.

[0144] In one embodiment, the security management module 33 includes:

[0145] A first determining unit, used to determine the size information of the engineering vehicle according to the type information of the engineering vehicle;

[0146] A second determining unit, used to determine the distance between the engineering vehicle and the power transmission channel according to the size information of the engineering vehicle;

[0147] The third determination unit is used to determine whether the power transmission channel has a risk of being damaged according to the distance between the engineering vehicle and the power transmission channel.

[0148] In one embodiment, the security management device further includes:

[0149] A vehicle image data set acquisition module is used to acquire a vehicle image data set; the vehicle image data set includes multiple vehicle images and annotation type information of each vehicle;

[0150] The training module is used to train the initial recognition network according to the vehicle image data set to obtain the target recognition network.

[0151] In one embodiment, the training module includes:

[0152] A fourth determination unit, configured to determine prediction type information and a target decoding result according to the vehicle image data set and the initial recognition network;

[0153] A target loss determination unit, used to determine the target loss according to the annotation type information, the prediction type information and the target decoding result;

[0154] The training unit is used to train the initial recognition network according to the target loss to obtain the target recognition network.

[0155] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the fourth determination unit includes:

[0156] The encoding subunit is used to input each vehicle image into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image;

[0157] A decoding subunit, used for inputting a target encoded image from a plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area;

[0158] The classification subunit is used to input the target decoding result into the classifier for classification to obtain the prediction type information.

[0159] In one embodiment, the initial recognition network further includes a segmentation subnetwork and an encoding subunit, which are specifically used to:

[0160] Input each vehicle image into the segmentation subnetwork for image block segmentation to obtain multiple image blocks corresponding to each vehicle image;

[0161] The multiple image blocks corresponding to each vehicle image are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image.

[0162] In one embodiment, the target loss determination unit includes:

[0163] A contrast loss determination subunit, used to determine the contrast loss according to the prediction type information and the target decoding result;

[0164] The target loss determination subunit is used to determine the target loss according to the contrast loss, the prediction type information and the labeling type information.

[0165] Each module in the above security management device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0166] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG1. ​​The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a security management method is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0167] Those skilled in the art will understand that the structure shown in FIG. 1 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0168] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0169] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0170] Input the environment image into the preset target recognition network for target recognition to obtain the type information of the engineering vehicle;

[0171] Determine whether the transmission channel is at risk of being damaged based on the type information of the engineering vehicle, and perform safety management if the transmission channel is at risk of being damaged.

[0172] In one embodiment, it is determined whether there is a risk of damage to the power transmission channel according to the type information of the engineering vehicle, and the processor further implements the following steps when executing the computer program:

[0173] Determine the size information of the engineering vehicle according to the type information of the engineering vehicle;

[0174] Determine the distance between the engineering vehicle and the power transmission channel based on the size information of the engineering vehicle;

[0175] Determine whether the transmission channel is at risk of being damaged based on the distance between the engineering vehicle and the transmission channel.

[0176] In one embodiment, during the training process of the target recognition network, the processor further implements the following steps when executing the computer program:

[0177] Acquire a vehicle image dataset; the vehicle image dataset includes multiple vehicle images and annotation type information of each vehicle;

[0178] The initial recognition network is trained according to the vehicle image dataset to obtain the target recognition network.

[0179] In one embodiment, the initial recognition network is trained according to the vehicle image data set to obtain a target recognition network, and the processor further implements the following steps when executing the computer program:

[0180] Determine the prediction type information and target decoding result according to the vehicle image dataset and the initial recognition network;

[0181] Determine the target loss according to the annotation type information, the prediction type information and the target decoding result;

[0182] The initial recognition network is trained according to the target loss to obtain the target recognition network.

[0183] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the prediction type information and the target decoding result are determined according to the vehicle image data set and the initial recognition network. When the processor executes the computer program, the following steps are also implemented:

[0184] Input each vehicle image into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image;

[0185] Inputting a target encoded image from the plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area;

[0186] The target decoding result is input into the classifier for classification to obtain the predicted type information.

[0187] In one embodiment, the initial recognition network further includes a segmentation subnetwork, which inputs each vehicle image into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image, and the processor further implements the following steps when executing the computer program:

[0188] Input each vehicle image into the segmentation subnetwork for image block segmentation to obtain multiple image blocks corresponding to each vehicle image;

[0189] The multiple image blocks corresponding to each vehicle image are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image.

[0190] In one embodiment, the target loss is determined according to the annotation type information, the prediction type information and the target decoding result, and the processor further implements the following steps when executing the computer program:

[0191] Determine the contrast loss based on the prediction type information and the target decoding result;

[0192] The target loss is determined based on the contrast loss, the prediction type information, and the annotation type information.

[0193] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0194] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0195] Input the environment image into the preset target recognition network for target recognition to obtain the type information of the engineering vehicle;

[0196] Determine whether the transmission channel is at risk of being damaged based on the type information of the engineering vehicle, and perform safety management if the transmission channel is at risk of being damaged.

[0197] In one embodiment, it is determined whether there is a risk of damage to the power transmission channel according to the type information of the engineering vehicle, and when the computer program is executed by the processor, the following steps are also implemented:

[0198] Determine the size information of the engineering vehicle according to the type information of the engineering vehicle;

[0199] Determine the distance between the engineering vehicle and the power transmission channel based on the size information of the engineering vehicle;

[0200] Determine whether the transmission channel is at risk of being damaged based on the distance between the engineering vehicle and the transmission channel.

[0201] In one embodiment, during the training process of the target recognition network, the computer program further implements the following steps when executed by the processor:

[0202] Acquire a vehicle image dataset; the vehicle image dataset includes multiple vehicle images and annotation type information of each vehicle;

[0203] The initial recognition network is trained according to the vehicle image dataset to obtain the target recognition network.

[0204] In one embodiment, the initial recognition network is trained according to the vehicle image data set to obtain a target recognition network, and the computer program, when executed by the processor, further implements the following steps:

[0205] Determine the prediction type information and target decoding result according to the vehicle image dataset and the initial recognition network;

[0206] Determine the target loss according to the annotation type information, the prediction type information and the target decoding result;

[0207] The initial recognition network is trained according to the target loss to obtain the target recognition network.

[0208] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the prediction type information and the target decoding result are determined according to the vehicle image data set and the initial recognition network. When the computer program is executed by the processor, the following steps are also implemented:

[0209] Input each vehicle image into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image;

[0210] Inputting a target encoded image from the plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area;

[0211] The target decoding result is input into the classifier for classification to obtain the predicted type information.

[0212] In one embodiment, the initial recognition network further includes a segmentation subnetwork, which inputs each vehicle image into an encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each vehicle image, and the computer program further implements the following steps when executed by the processor:

[0213] Input each vehicle image into the segmentation subnetwork for image block segmentation to obtain multiple image blocks corresponding to each vehicle image;

[0214] The multiple image blocks corresponding to each vehicle image are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image.

[0215] In one embodiment, the target loss is determined according to the annotation type information, the prediction type information and the target decoding result, and the computer program further implements the following steps when executed by the processor:

[0216] Determine the contrast loss based on the prediction type information and the target decoding result;

[0217] The target loss is determined based on the contrast loss, the prediction type information, and the annotation type information.

[0218] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0219] Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles;

[0220] Input the environment image into the preset target recognition network for target recognition to obtain the type information of the engineering vehicle;

[0221] Determine whether the transmission channel is at risk of being damaged based on the type information of the engineering vehicle, and perform safety management if the transmission channel is at risk of being damaged.

[0222] In one embodiment, it is determined whether there is a risk of damage to the power transmission channel according to the type information of the engineering vehicle, and when the computer program is executed by the processor, the following steps are also implemented:

[0223] Determine the size information of the engineering vehicle according to the type information of the engineering vehicle;

[0224] Determine the distance between the engineering vehicle and the power transmission channel based on the size information of the engineering vehicle;

[0225] Determine whether the transmission channel is at risk of being damaged based on the distance between the engineering vehicle and the transmission channel.

[0226] In one embodiment, during the training process of the target recognition network, the computer program further implements the following steps when executed by the processor:

[0227] Acquire a vehicle image dataset; the vehicle image dataset includes multiple vehicle images and annotation type information of each vehicle;

[0228] The initial recognition network is trained according to the vehicle image dataset to obtain the target recognition network.

[0229] In one embodiment, the initial recognition network is trained according to the vehicle image data set to obtain a target recognition network, and the computer program, when executed by the processor, further implements the following steps:

[0230] Determine the prediction type information and target decoding result according to the vehicle image dataset and the initial recognition network;

[0231] Determine the target loss according to the annotation type information, the prediction type information and the target decoding result;

[0232] The initial recognition network is trained according to the target loss to obtain the target recognition network.

[0233] In one embodiment, the initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the prediction type information and the target decoding result are determined according to the vehicle image data set and the initial recognition network. When the computer program is executed by the processor, the following steps are also implemented:

[0234] Input each vehicle image into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image;

[0235] Inputting a target encoded image from the plurality of encoding results into a decoding subnetwork for decoding to obtain a target decoding result; the target decoding result includes a vehicle feature area;

[0236] The target decoding result is input into the classifier for classification to obtain the predicted type information.

[0237] In one embodiment, the initial recognition network further includes a segmentation subnetwork, which inputs each vehicle image into an encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each vehicle image, and the computer program further implements the following steps when executed by the processor:

[0238] Input each vehicle image into the segmentation subnetwork for image block segmentation to obtain multiple image blocks corresponding to each vehicle image;

[0239] The multiple image blocks corresponding to each vehicle image are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each vehicle image.

[0240] In one embodiment, the target loss is determined according to the annotation type information, the prediction type information and the target decoding result, and the computer program further implements the following steps when executed by the processor:

[0241] Determine the contrast loss based on the prediction type information and the target decoding result;

[0242] The target loss is determined based on the contrast loss, the prediction type information, and the annotation type information.

[0243] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0244] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0245] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A safety management method, characterized in that: The method comprises: Acquire an environmental image of an area near a power transmission channel; the environmental image includes engineering vehicles; Inputting the environment image into a preset target recognition network for target recognition to obtain type information of the engineering vehicle; Determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and perform safety management if the power transmission channel is at risk of being damaged.

2. The method according to claim 1, characterized in that The determining, according to the type information of the engineering vehicle, whether the power transmission channel is at risk of being damaged includes: Determining the size information of the engineering vehicle according to the type information of the engineering vehicle; Determining the distance between the engineering vehicle and the power transmission channel according to the size information of the engineering vehicle; Whether the power transmission channel is at risk of being damaged is determined according to the distance between the engineering vehicle and the power transmission channel.

3. The method according to claim 1 or 2, characterized in that: The training process of the target recognition network includes: Acquire a vehicle image dataset; the vehicle image dataset includes a plurality of vehicle images and label type information of each vehicle; The initial recognition network is trained according to the vehicle image data set to obtain the target recognition network.

4. The method according to claim 3, characterized in that: The training of the initial recognition network according to the vehicle image data set to obtain the target recognition network includes: Determining prediction type information and target decoding results according to the vehicle image dataset and the initial recognition network; Determine a target loss according to the annotation type information, the prediction type information and the target decoding result; The initial recognition network is trained according to the target loss to obtain the target recognition network.

5. The method according to claim 4, characterized in that The initial recognition network includes an encoding subnetwork, a decoding subnetwork and a classifier, and the determining of prediction type information and target decoding results according to the vehicle image dataset and the initial recognition network includes: Inputting each of the vehicle images into the encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each of the vehicle images; Inputting a target encoded image from the plurality of encoding results into the decoding subnetwork for decoding to obtain the target decoding result; the target decoding result includes a vehicle feature area; The target decoding result is input into the classifier for classification to obtain the prediction type information.

6. The method according to claim 5, characterized in that The initial recognition network further includes a segmentation subnetwork, and each of the vehicle images is input into the encoding subnetwork for feature encoding to obtain a plurality of encoding results corresponding to each of the vehicle images, including: Inputting each of the vehicle images into the segmentation subnetwork to perform image block segmentation to obtain a plurality of image blocks corresponding to each of the vehicle images; The multiple image blocks corresponding to each of the vehicle images are input into the encoding subnetwork for feature encoding to obtain multiple encoding results corresponding to each of the vehicle images.

7. The method according to claim 4, characterized in that The determining the target loss according to the annotation type information, the prediction type information and the target decoding result includes: Determining a contrast loss according to the prediction type information and the target decoding result; The target loss is determined according to the contrast loss, the prediction type information and the annotation type information.

8. A security management device, characterized in that: The device comprises: An acquisition module, used to acquire an environmental image of an area near the power transmission channel; the environmental image includes engineering vehicles; A target recognition module, used for inputting the environment image into a preset target recognition network for target recognition, and obtaining the type information of the engineering vehicle; The safety management module is used to determine whether the power transmission channel is at risk of being damaged according to the type information of the engineering vehicle, and to perform safety management when the power transmission channel is at risk of being damaged.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images