[go: up one dir, main page]

CN112347916A - Power field operation safety monitoring method and device based on video image analysis - Google Patents

Power field operation safety monitoring method and device based on video image analysis Download PDF

Info

Publication number
CN112347916A
CN112347916A CN202011224459.1A CN202011224459A CN112347916A CN 112347916 A CN112347916 A CN 112347916A CN 202011224459 A CN202011224459 A CN 202011224459A CN 112347916 A CN112347916 A CN 112347916A
Authority
CN
China
Prior art keywords
image
operator
analysis
area
power field
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202011224459.1A
Other languages
Chinese (zh)
Other versions
CN112347916B (en
Inventor
徐海青
陈是同
陶俊
赵云龙
吴小华
毛舒乐
林胜
张天奇
浦正国
杨彬彬
李小威
宋杰
石锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Information and Telecommunication Co Ltd
Anhui Jiyuan Software Co Ltd
Original Assignee
State Grid Information and Telecommunication Co Ltd
Anhui Jiyuan Software Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Information and Telecommunication Co Ltd, Anhui Jiyuan Software Co Ltd filed Critical State Grid Information and Telecommunication Co Ltd
Priority to CN202011224459.1A priority Critical patent/CN112347916B/en
Publication of CN112347916A publication Critical patent/CN112347916A/en
Application granted granted Critical
Publication of CN112347916B publication Critical patent/CN112347916B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/161—Detection; Localisation; Normalisation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biophysics (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Evolutionary Biology (AREA)
  • Human Computer Interaction (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computational Linguistics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)
  • Alarm Systems (AREA)

Abstract

本发明公开了基于视频图像分析的电力现场作业安全监控方法及装置，包括：基于静态图像进行绝对违章状态分析，获取电力现场作业监控视频的静态图像并进行预处理；通过聚类算法获取图像中的作业人员目标区域；获取目标区域内的人体骨架关键点位置，并获取人体分区域图像；结合分区域图像、作业人员目标区域图像以及静态图像整体图像通过卷积神经网络模型判断作业人员行为种类获取违章行为分析结果；本发明通过基于多种图像区域的特征融合，实现了对人体和环境交互特征的融合，同时融合了人体骨架关键点位置特征，提高了行为种类识别判断的准确性，通过关键区域的高效特征表示进行行为种类分析，减少冗余图像信息带来的计算量。

Figure 202011224459

The invention discloses a method and device for safety monitoring of electric field operations based on video image analysis. the target area of the operator; obtain the position of the key points of the human skeleton in the target area, and obtain the sub-regional image of the human body; combine the sub-regional image, the image of the operator's target area and the overall image of the static image to determine the type of operator behavior through the convolutional neural network model Obtain the analysis result of illegal behavior; the present invention realizes the fusion of human body and environment interaction features through feature fusion based on various image regions, and simultaneously integrates the position features of key points of human skeleton, so as to improve the accuracy of behavior type identification and judgment. Efficient feature representation of key regions performs behavioral category analysis and reduces the amount of computation caused by redundant image information.

Figure 202011224459

Description

Power field operation safety monitoring method and device based on video image analysis
Technical Field
The invention relates to the technical field of electric power safety, in particular to a method and a device for monitoring electric power field operation safety based on video image analysis.
Background
In recent years, in order to effectively guarantee the safety of workers on the electric power operation site and the continuity and stability of power supply to users, the national power grid puts forward higher requirements on the safe production and management of electric power. However, due to improper supervision of related supervision departments, management of managers is lost and power operation constructors do not comply or understand the regulations in place, which easily causes large safety accidents.
The behavior problem of finding whether the site operator has potential risks in the operation site in time is mainly solved by adopting manual design and extracting features and then performing behavior recognition and classification, so that the complexity problem of the manually designed features exists, the robustness and the popularization are poor, for behavior recognition by adopting a deep learning method, compared with the traditional behavior analysis based on the manual feature method, a model adopting the deep learning method can automatically obtain meaningful hierarchical feature representation, however, a video segment obtained from the power site is more complex, and how to extract effective features from a video image is still the core work of numerous researchers.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a power field operation safety monitoring method based on video image analysis, which comprises the following steps:
(1) absolute violation state analysis based on static image
Acquiring a static image of a power field operation monitoring video and preprocessing the static image;
acquiring an operator target area in the image through a clustering algorithm;
acquiring the positions of key points of a human body skeleton in a target region through a key point detection model, and acquiring a regional image of a human body;
judging the behavior type of the operator by combining the regional image, the target regional image of the operator and the whole static image through a convolutional neural network model to obtain a violation behavior analysis result;
(2) track violation analysis based on dynamic video
Fitting the relation among the environmental parameters of the construction field area, the work types of workers, the qualification grade of the workers and the predicted value of the danger grade of the construction area based on the training data;
acquiring a monitoring video image of a construction area to be analyzed, performing face recognition on the appearing operators to acquire identity information, and acquiring the work types and qualification grades of the operators based on the face recognition result;
predicting the danger level of the construction area relative to the workers according to the fitting relation based on the environmental parameters of the construction field area, the labor and the types of the workers and the qualification of the workers;
and sending out alarm information when the predicted danger level is greater than a second preset value.
As a further optimization of the scheme, the absolute violation phenomenon comprises live working without wearing insulating gloves, insecure ground wire hanging, on-site standard dressing and typical violation behaviors, and the track violation comprises intrusion into a warning area.
As a further optimization of the above scheme, the obtaining of the target region in the image through the clustering algorithm includes:
(31) acquiring the density of points in the static image according to a preset first calculation method;
(32) taking the point with the maximum density as a first clustering center, and performing density reduction on the point with the first clustering center as the origin and within a preset radius range according to a preset second calculation method;
(33) obtaining a point with the maximum density from all points of the non-clustering centers, judging whether the density value of the point with the maximum density is larger than a preset first threshold value, if so, taking the point with the maximum density from all points of the non-clustering centers as a next clustering center, performing density reduction on the point with the next clustering center as an original point and within a preset radius range according to a preset second calculation method, and repeating the step (33); otherwise, entering a step (34);
(34) and finishing the acquisition of all target areas, and acquiring a clustering area formed by a plurality of clustering centers as a target area.
As a further optimization of the above scheme, the determining the behavior type of the operator through the convolutional neural network model to obtain the analysis result of the violation behavior includes:
inputting the human body regional image and the target region of the operator into a first convolution neural network, extracting features through the first convolution neural network, inputting the extracted features into a feature fusion layer network for feature fusion, and inputting the extracted features into a first classification layer network based on fusion features to obtain an image classification result;
inputting the whole static image into a second convolutional neural network, extracting features through a second convolutional layer network, and inputting the extracted features into a second classification layer network to obtain an image classification result;
and the output results of the first classification layer network and the second classification layer network are input into the classification fusion layer network to obtain the probability of the behavior types of the operators.
As a further optimization of the above scheme, the first convolutional neural network and the second convolutional neural network are trained by using different model parameters as initialization model parameters during training, and the model parameters are corrected by back propagation through calculating a loss function value of an output result after the first classification layer and the second classification layer, respectively.
As a further optimization of the above scheme, the method for acquiring the human body subregion image includes:
training a key point detection model based on the human skeleton key point image data set;
inputting the static image into a key point detection model to obtain key point position detection and classification;
the method comprises the steps of obtaining a region image which contains all key points of a single person and is the smallest in area, and dividing the region image into at least one sub region according to the type of the key points of the human body.
As a further optimization of the above scheme, the method for detecting the key point by the key point detection model comprises:
carrying out feature extraction on an input image through a high-resolution network to obtain a plurality of feature maps with different resolutions;
selecting one input at least two cavity convolution layers with different expansion rates from the feature map output by the high-resolution network to obtain feature maps with different scales, wherein output channels of the cavity convolution layers are 256;
fusing the feature maps of multiple scales to obtain a fused feature map;
and calculating the probability of each point on the image as a key point based on the fusion feature map, and acquiring the point with the maximum probability as the key point.
As further optimization of the scheme, the relation between the environmental parameters of the construction site area, the labor types of workers, the qualification grade of the workers and the risk grade predicted value of the construction area is fitted based on the training data, and a neural network is adopted for fitting.
The invention also provides a power field operation safety monitoring device based on video image analysis, which comprises:
the absolute violation state analysis module is used for carrying out absolute violation state analysis based on the static image and comprises:
the static image preprocessing unit is used for acquiring and preprocessing a static image of the power field operation monitoring video;
the target area extraction unit is used for acquiring the target area of the operator in the image through a clustering algorithm;
the regional image acquisition unit is used for acquiring the positions of key points of the human skeleton in the target region through the key point detection model and acquiring a regional image of the human body;
the behavior type analysis unit is used for judging the behavior type of the operating personnel through the convolutional neural network model by combining the subarea image, the operating personnel target area image and the static image overall image to obtain a violation behavior analysis result;
the track violation analysis module is used for carrying out track violation analysis based on the dynamic video, and comprises the following steps:
the relevant parameter fitting unit is used for fitting the relation among the environmental parameters of the construction site area, the work types of workers, the qualification grade of the workers and the predicted value of the danger grade of the construction area based on the training data;
the system comprises an operator information acquisition unit, a construction area monitoring unit and a monitoring unit, wherein the operator information acquisition unit is used for acquiring a monitoring video image of a construction area to be analyzed, performing face recognition on the appearing operators to acquire identity information, and acquiring the work types and the qualification grades of the operators based on the face recognition result;
and the operator track violation analysis unit is used for predicting the danger level of the construction area relative to the workers according to the fitting relation based on the environmental parameters, the worker types and the worker qualifications of the construction field area, and sending alarm information when the predicted danger level is greater than a second preset value.
As a further optimization of the scheme, the absolute violation phenomenon comprises live working without wearing insulating gloves, insecure ground wire hanging, on-site standard dressing and typical violation behaviors, and the track violation comprises intrusion into a warning area.
The invention discloses a method and a device for monitoring the safety of electric power field operation based on video image analysis, which have the following beneficial effects:
by combining the regional images, the target regional images of the operating personnel and the whole static images, the method realizes the fusion of the interactive characteristics of the human body and the environment and improves the accuracy of the classification judgment of the behavior types of the operating personnel based on the characteristic fusion of various image regions, particularly the combination of the key region of the human body and the whole images containing the environment images and the human body images, and the regional images formed by the key point position information of the human body skeleton are fused, so that the behavior type analysis is realized through the efficient characteristic representation of the key region, and the calculated amount brought by redundant image information is reduced.
Drawings
Fig. 1 is an overall flow chart of a power field operation safety monitoring method based on video image analysis according to an embodiment of the present application;
FIG. 2 is a block diagram of the convolutional neural network model of FIG. 1;
FIG. 3 is a block diagram of a flow chart of a method for judging the behavior type of an operator to obtain an analysis result of violation behaviors by the convolutional neural network model in FIG. 1;
FIG. 4 is a block flow diagram of a method of obtaining the operator target area and the zone images of FIG. 1;
fig. 5 is a block diagram of a power field operation safety monitoring device based on video image analysis according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The method for monitoring the safety of the electric power field operation based on the video image analysis comprises the steps of monitoring an absolute violation state and a track violation behavior, wherein the specific absolute violation state comprises the conditions that live-wire operation does not wear insulating gloves, grounding wires are not firmly hung, the field is normally worn and the track violation behavior comprises the conditions that a warning area breaks into the live-wire operation, and the track violation behavior comprises the conditions that the live-wire operation does not wear the insulating gloves, the grounding wires are not firmly hung and the track.
For monitoring and analyzing the five types of violation behaviors, the method provided by the embodiment of the application comprises the following steps:
(1) absolute violation state analysis based on static image
The method comprises the steps of obtaining static images of power field operation monitoring videos and preprocessing the static images, wherein the optimized power field monitoring video images comprise three-dimensional image data so as to provide more characteristics for subsequent image analysis;
acquiring an operator target area in the image through a clustering algorithm;
acquiring the positions of key points of a human body skeleton in a target region through a key point detection model, and acquiring a regional image of a human body;
judging the behavior type of the operator by combining the regional image, the target regional image of the operator and the whole static image through a convolutional neural network model to obtain a violation behavior analysis result;
in the embodiment, by combining the subarea image, the target area image of the operator and the whole static image, based on the feature fusion of various image areas, especially the combination of the human body key area and the whole image containing the environment image and the human body image, the fusion of the human body and the environment interaction feature is realized, the accuracy of the behavior type classification judgment of the operator is improved, and the subarea image formed by fusing the position information of the human body skeleton key point is realized, so that the behavior type analysis is realized through the efficient feature representation of the key area, and the calculated amount brought by redundant image information is reduced.
(2) Track violation analysis based on dynamic video
Fitting the relation among the environmental parameters of the construction field area, the work types of workers, the qualification grade of the workers and the predicted value of the danger grade of the construction area based on the training data;
acquiring a monitoring video image of a construction area to be analyzed, performing face recognition on the appearing operators to acquire identity information, and acquiring the work types and qualification grades of the operators based on the face recognition result;
predicting the danger level of the construction area relative to the workers according to the fitting relation based on the environmental parameters of the construction field area, the labor and the types of the workers and the qualification of the workers;
and sending out alarm information when the predicted danger level is greater than a second preset value.
In the embodiment, considering that the factors affecting the safety of the operators on the electric power site include subjective factors such as the qualification of the operators, namely, the safety capability condition of the operators, objective factors such as the environmental parameters of the construction site area, and the risk levels of the operators with different safety capabilities in the same working area are different, a method for determining different dangerous areas aiming at different operators is adopted, wherein the qualification of the operators is used for carrying out safety comprehensive capability analysis by combining the personnel information with information fusion such as training records, safety examination information, participation engineering project information, violation records and the like. Wherein the environmental parameters include a safety state of the power device and a safety state of the work tool.
In the embodiment, a convolutional neural network is adopted to fit the relationship among the environmental parameters, the labor types and the qualification grade information of the construction site area and the predicted value of the danger grade of the construction area, the depth characteristic extraction is carried out through a plurality of layers of convolutional layers based on the input characteristic data of the environmental parameters, the labor types and the qualification grade information of the workers serving as the network of the construction site area, the probabilities of different danger grades of the workers entering the construction site are output through softmax on a classification layer, the danger grade with the maximum probability serves as a prediction result, and the workers are considered to enter the warning area when the danger grade exceeds a preset range value and belong to track violation.
In this embodiment, the obtaining of the target region in the image through the clustering algorithm includes the following steps:
(31) acquiring the density of points in a static image according to a preset first calculation method, taking the static image as an example of adopting three-dimensional image data, wherein the first calculation method comprises the following steps:
Figure BDA0002763187060000061
wherein r isax,ray,razRespectively, as a point (x) in the imagei,yi,zi) For the length range value of the central point in the three-dimensional direction, the length range in the three-dimensional direction forms more than one point (x)i,yi,zi) A cuboid spatial range which is a central point;
(32) taking the point with the maximum density as a first clustering center, and performing density reduction on the points with the first clustering center as an origin and within a preset radius range according to a preset second calculation method, wherein the second calculation method comprises the following steps:
Figure BDA0002763187060000062
wherein (x)c,yc,zc) As cluster center coordinates, DcIs a point (x)c,yc,zc) Density value of rbx,rby,rbzThe preferred r is a length range value in the three-dimensional direction with the cluster center as the center, i.e., a range of points to be density-reducedbx,rby,rbzIs respectively rax,ray,raz1.5 times of;
(33) obtaining a point with the maximum density from all points of the non-clustering centers, judging whether the density value of the point with the maximum density is larger than a preset first threshold value, if so, taking the point with the maximum density from all points of the non-clustering centers as a next clustering center, performing density reduction on the point with the next clustering center as an original point and within a preset radius range according to a preset second calculation method, and repeating the step (33); otherwise, entering a step (34);
(34) and finishing the acquisition of all target areas, and acquiring a clustering area formed by a plurality of clustering centers as a target area.
The method comprises the steps of collecting three-dimensional image data and three-dimensional point cloud density subtraction clustering based on depth camera equipment, accurately extracting a clustering region of a single operator in an image, namely a target region, and effectively reducing redundant image data of the target region.
In the embodiment, two convolutional neural network models are adopted, and image feature extraction and behavior type judgment classification are respectively performed on the human body regional image, the operator target region and the static image overall image, and the specific method comprises the following steps:
inputting the human body regional image and the target region of the operator into a first convolution neural network, extracting features through the first convolution neural network, inputting the extracted features into a feature fusion layer network for feature fusion, and inputting the extracted features into a first classification layer network based on fusion features to obtain an image classification result;
inputting the whole static image into a second convolutional neural network, extracting features through a second convolutional layer network, and inputting the extracted features into a second classification layer network to obtain an image classification result;
and inputting the output results of the first classification layer network and the second classification layer network into a classification fusion layer network to obtain the probability of the behavior types of the operators.
In this embodiment, the first convolutional neural network and the second convolutional neural network are trained by using different model parameters as initialization model parameters, and the model parameters are corrected by back propagation through calculating a loss function value of an output result after the first classification layer and the second classification layer, preferably, the initialization parameters of the second convolutional neural network model are the model parameters of the key point detection model in this embodiment, that is, the second convolutional network model is trained by fusing key point feature information helpful for behavior classification determination, so as to improve the behavior classification accuracy of the second convolutional network model, and simultaneously, in the first convolutional network, a human body region image and an operator target region are obtained based on the positions of key points of a human body skeleton, and the operator target region image provides the contour features of the operator, the image analysis range is reduced, and the model training speed and the identification accuracy of the behavior types of the operators are improved.
The method for acquiring the human body subarea image comprises the following steps:
training a key point detection model based on the human skeleton key point image data set;
the method for inputting the static image into the key point detection model to obtain the key point position detection and classification and detecting the key points on the model based on the input image comprises the following steps:
the method comprises the steps of extracting features of an input image through a high-resolution network to obtain a plurality of feature maps with different resolutions, obtaining a plurality of feature maps with different resolutions in parallel from an original input image in the high-resolution network, fusing a plurality of resolution features, fusing different resolution feature information into each output feature image with different resolutions, and fusing the plurality of different resolution feature maps to obtain a final fused feature map through the following steps;
selecting one input at least two cavity convolution layers with different expansion rates from the feature maps output by the high-resolution network to obtain feature maps with different scales, wherein output channels of the cavity convolution layers are 256, and preferably, selecting one input at least two cavity convolution layers with different expansion rates from the feature maps output by the high-resolution network;
fusing the feature maps of multiple scales to obtain a fused feature map;
calculating the probability of each point on the image as a key point based on the fusion feature map, and acquiring the point with the maximum probability as the position of the key point;
acquiring a region image which contains all the single key points and has the minimum area, wherein all the single key points comprise: the method includes the steps that a left eye, a right eye, a left ear, a right ear, a left hand, a right hand and the like are arranged in a region image, the region image is divided into at least one subarea according to the key point category of a human body, for example, all key points belonging to the head are divided into one category as one subarea, and when a plurality of operators are arranged in one image, all key points of a single person are contained, the key points possibly contained in the region image with the minimum area are not the key points of the same person, so that the area size of the subarea of the classified key points is limited, and when the area is smaller than a preset value, the key points contained in the region are judged to be the key points of the non-same person, and it can be understood that the region image containing all key points of the single person and having the.
The embodiment also provides a power field operation safety monitoring device based on video image analysis, which is used for monitoring an absolute violation state and a track violation behavior, wherein the absolute violation state comprises that the live-wire operation does not wear insulating gloves, the grounding wire is not firmly hooked, the field is normally worn and the track violation behavior comprises that an alert area intrudes, and the power field operation safety monitoring device of the embodiment comprises:
the absolute violation state analysis module is used for carrying out absolute violation state analysis based on the static image and comprises:
the static image preprocessing unit is used for acquiring and preprocessing a static image of the power field operation monitoring video;
the target area extraction unit is used for acquiring the target area of the operator in the image through a clustering algorithm;
the regional image acquisition unit is used for acquiring the positions of key points of the human skeleton in the target region through the key point detection model and acquiring a regional image of the human body;
the behavior type analysis unit is used for judging the behavior type of the operating personnel through the convolutional neural network model by combining the subarea image, the operating personnel target area image and the static image overall image to obtain a violation behavior analysis result;
the track violation analysis module is used for carrying out track violation analysis based on the dynamic video, and comprises the following steps:
the relevant parameter fitting unit is used for fitting the relation among the environmental parameters of the construction site area, the work types of workers, the qualification grade of the workers and the predicted value of the danger grade of the construction area based on the training data;
the system comprises an operator information acquisition unit, a construction area monitoring unit and a monitoring unit, wherein the operator information acquisition unit is used for acquiring a monitoring video image of a construction area to be analyzed, performing face recognition on the appearing operators to acquire identity information, and acquiring the work types and the qualification grades of the operators based on the face recognition result;
the operator track violation analysis unit is used for predicting the danger level of the construction area relative to the workers according to the fitting relation based on the environmental parameters of the construction field area, the work types of the workers and the qualification of the workers;
and sending out alarm information when the predicted danger level is greater than a second preset value.
For specific limitations of the electric field operation safety monitoring device, reference may be made to the above limitations of the electric field operation safety monitoring method, which will not be described herein again. All or part of each unit in the electric field operation safety monitoring device can be realized by software, hardware and a combination thereof. The units can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the units.
The present invention is not limited to the above-described embodiments, and those skilled in the art will be able to make various modifications without creative efforts from the above-described conception, and fall within the scope of the present invention.

Claims (10)

1.基于视频图像分析的电力现场作业安全监控方法，其特征在于，包括如下步骤：1. the power field operation safety monitoring method based on video image analysis, is characterized in that, comprises the steps: (1)基于静态图像进行绝对违章状态分析(1) Absolute violation state analysis based on static images 获取电力现场作业监控视频的静态图像并进行预处理；Obtaining static images of monitoring video of power field operations and preprocessing; 通过聚类算法获取图像中的作业人员目标区域；Obtain the target area of the operator in the image through the clustering algorithm; 通过关键点检测模型获取目标区域内的人体骨架关键点位置，并获取人体分区域图像；Obtain the position of the key points of the human skeleton in the target area through the key point detection model, and obtain the sub-regional images of the human body; 结合分区域图像、作业人员目标区域图像以及静态图像整体图像通过卷积神经网络模型判断作业人员行为种类获取违章行为分析结果；Combining the sub-area image, the operator's target area image and the overall image of the static image, the convolutional neural network model is used to determine the type of operator's behavior to obtain the analysis results of illegal behavior; (2)基于动态视频进行轨迹违章分析(2) Trajectory violation analysis based on dynamic video 基于训练数据拟合施工现场区域环境参数、工作人员工种、工作人员资质等级与施工区域危险等级预测值之间的关系；Fitting the relationship between the regional environmental parameters of the construction site, the type of workers, the qualification level of the workers and the predicted value of the risk level of the construction area based on the training data; 获取待分析的施工区域监控视频图像，对出现的作业人员进行人脸识别获取身份信息，基于人脸识别结果获取工作人员工种和工作人员资质等级；Obtain the surveillance video images of the construction area to be analyzed, perform face recognition on the workers that appear to obtain identity information, and obtain the type of workers and the qualification level of the workers based on the results of face recognition; 基于施工现场区域环境参数、工作人员工种、工作人员资质根据拟合关系，预测施工区域相对于工作人员的危险等级；Based on the regional environmental parameters of the construction site, the types of workers, and the qualifications of the workers, according to the fitting relationship, predict the danger level of the construction area relative to the workers; 当预测的危险等级大于第二预设值时发出告警信息。When the predicted danger level is greater than the second preset value, an alarm message is issued. 2.根据权利要求1所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述绝对违章现象包括带电作业未佩戴绝缘手套、接地线挂接不牢靠、现场规范着装、典型违章行为，所述轨迹违章包括警戒区域闯入。2. The method for monitoring power field work safety based on video image analysis according to claim 1, wherein the absolute violations include not wearing insulating gloves during live work, unreliable grounding connection, on-site standard dress, typical Violations, the track violations include intrusion into warning areas. 3.根据权利要求1所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述通过聚类算法获取图像中的目标区域，包括：3. The power field operation safety monitoring method based on video image analysis according to claim 1, wherein the acquisition of the target area in the image by a clustering algorithm comprises: (31)根据预设第一计算方法获取静态图像中点的密度；(31) obtaining the density of the point in the static image according to the preset first calculation method; (32)以密度最大的点作为第一聚类中心，对以第一聚类中心为原点预设半径范围的点根据预设第二计算方法进行密度削减；(32) taking the point with the largest density as the first cluster center, and performing density reduction on the point with the first cluster center as the origin preset radius range according to the preset second calculation method; (33)在非聚类中心的所有点中获取密度最大的点，判断密度最大的点的密度值是否大于预设第一阈值，若是则将所述非聚类中心的所有点中密度最大的点作为下一聚类中心，对以下一聚类中心为原点预设半径范围的点根据预设第二计算方法进行密度削减，重复步骤(33)；否则进入步骤(34)；(33) Obtain the point with the highest density among all the points of the non-cluster center, determine whether the density value of the point with the highest density is greater than the preset first threshold, and if so, select the point with the highest density among all the points of the non-cluster center The point is used as the next cluster center, and the density reduction is performed on the point with the next cluster center as the origin preset radius range according to the preset second calculation method, and step (33) is repeated; otherwise, step (34) is entered; (34)完成所有目标区域的获取，获取多个聚类中心形成的聚类区域作为目标区域。(34) Acquiring all target regions is completed, and a cluster region formed by a plurality of cluster centers is obtained as a target region. 4.根据权利要求3所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述通过卷积神经网络模型判断作业人员行为种类获取违章行为分析结果，包括：4. the power field operation safety monitoring method based on video image analysis according to claim 3, is characterized in that, described by the convolutional neural network model to judge the behavior type of the operator to obtain the analysis result of illegal behavior, comprising: 将人体分区域图像和所述作业人员目标区域输入第一卷积神经网络，经过第一卷积层网络进行特征提取，并将提取的特征输入特征融合层网络进行特征融合，基于融合特征输入第一分类层网络获取图像分类结果；Input the human body sub-region image and the target area of the operator into the first convolutional neural network, perform feature extraction through the first convolutional layer network, and input the extracted features into the feature fusion layer network for feature fusion. A classification layer network obtains image classification results; 将所述静态图像整体输入第二卷积神经网络，经过第二卷积层网络进行特征提取，并将提取的特征输入第二分类层网络获取图像分类结果；Inputting the static image as a whole into the second convolutional neural network, performing feature extraction through the second convolutional layer network, and inputting the extracted features into the second classification layer network to obtain image classification results; 所述第一分类层网络和第二分类层网络输出结果输入分类融合层网络获取作业人员行为种类的概率。The output results of the first classification layer network and the second classification layer network are input into the classification and fusion layer network to obtain the probability of the operator's behavior type. 5.根据权利要求4所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述第一卷积神经网络和第二卷积神经网络在训练时，采用不同模型参数作为初始化模型参数进行训练，分别在第一分类层和第二分类层后通过计算输出结果的损失函数值进行反向传播修正模型参数。5. The power field operation safety monitoring method based on video image analysis according to claim 4, wherein the first convolutional neural network and the second convolutional neural network use different model parameters as initialization during training The model parameters are trained, and after the first classification layer and the second classification layer, respectively, the model parameters are corrected by backpropagation by calculating the loss function value of the output result. 6.根据权利要求4所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述人体分区域图像的获取方法包括：6. The power field operation safety monitoring method based on video image analysis according to claim 4, wherein the acquisition method of the sub-regional images of the human body comprises: 基于人体骨架关键点图像数据集进行关键点检测模型的训练；The keypoint detection model is trained based on the human skeleton keypoint image dataset; 将所述静态图像输入关键点检测模型获取关键点位置检测和分类；Inputting the static image into a keypoint detection model to obtain keypoint position detection and classification; 获取包含单人所有关键点且面积最小的区域图像，并根据人体关键点类别将所述区域图像分为至少一个分区域。Obtain an area image containing all key points of a single person with the smallest area, and divide the area image into at least one sub-area according to the human key point category. 7.根据权利要求6所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述关键点检测模型进行关键点检测的方法为：7. The power field operation safety monitoring method based on video image analysis according to claim 6, is characterized in that, the method that described key point detection model carries out key point detection is: 通过高分辨率网络对输入图像进行特征提取获取多种分辨率不同的特征图；Perform feature extraction on the input image through a high-resolution network to obtain feature maps with different resolutions; 从高分辨率网络输出的所述特征图中选择一个输入至少两个扩张率不同的空洞卷积层获取不同尺度的特征图，所述空洞卷积层的输出通道均为256；From the feature map output by the high-resolution network, select an input at least two dilated convolutional layers with different expansion rates to obtain feature maps of different scales, and the output channels of the dilated convolutional layer are all 256; 将多个尺度的特征图进行融合获取融合特征图；The feature maps of multiple scales are fused to obtain a fusion feature map; 基于融合特征图计算图像上每个点为关键点的概率，获取概率最大点点作为关键点。Based on the fusion feature map, the probability that each point on the image is a key point is calculated, and the point with the highest probability is obtained as a key point. 8.根据权利要求1所述的基于视频图像分析的电力现场作业安全监控方法，其特征在于，所述基于训练数据拟合施工现场区域环境参数、工作人员工种、工作人员资质等级与施工区域危险等级预测值之间的关系，采用神经网络进行拟合。8 . The method for monitoring power field operation safety based on video image analysis according to claim 1 , wherein the fitting of construction site regional environmental parameters, worker types, worker qualification levels and construction area hazards based on training data. 9 . The relationship between the predicted values of the grades is fitted by a neural network. 9.基于视频图像分析的电力现场作业安全监控装置，其特征在于，包括：9. The electric field operation safety monitoring device based on video image analysis is characterized in that, comprising: 绝对违章状态分析模块，用于基于静态图像进行绝对违章状态分析，包括：The absolute violation status analysis module is used for absolute violation status analysis based on static images, including: 静态图像预处理单元，用于获取电力现场作业监控视频的静态图像并进行预处理；The static image preprocessing unit is used to obtain and preprocess the static image of the monitoring video of the power field operation; 目标区域提取单元，用于通过聚类算法获取图像中的作业人员目标区域；The target area extraction unit is used to obtain the target area of the operator in the image through the clustering algorithm; 分区域图像获取单元，用于通过关键点检测模型获取目标区域内的人体骨架关键点位置，并获取人体分区域图像；The sub-region image acquisition unit is used to obtain the position of the key points of the human skeleton in the target area through the key point detection model, and obtain the sub-region images of the human body; 行为种类分析单元，用于结合分区域图像、作业人员目标区域图像以及静态图像整体图像通过卷积神经网络模型判断作业人员行为种类获取违章行为分析结果；The behavior type analysis unit is used to combine the sub-regional image, the image of the target area of the operator and the overall image of the static image to determine the type of behavior of the operator through the convolutional neural network model to obtain the analysis result of illegal behavior; 轨迹违章分析模块，用于基于动态视频进行轨迹违章分析，包括：Track violation analysis module, used for tracking violation analysis based on dynamic video, including: 相关参数拟合单元，用于基于训练数据拟合施工现场区域环境参数、工作人员工种、工作人员资质等级与施工区域危险等级预测值之间的关系；The relevant parameter fitting unit is used to fit the relationship between the regional environmental parameters of the construction site, the type of staff, the qualification level of the staff and the predicted value of the risk level of the construction area based on the training data; 作业人员信息获取单元，用于获取待分析的施工区域监控视频图像，对出现的作业人员进行人脸识别获取身份信息，基于人脸识别结果获取工作人员工种和工作人员资质等级；The operator information acquisition unit is used to acquire the surveillance video image of the construction area to be analyzed, perform face recognition on the operator to obtain identity information, and obtain the type of worker and the qualification level of the worker based on the face recognition result; 作业人员轨迹违章分析单元，用于基于施工现场区域环境参数、工作人员工种、工作人员资质根据拟合关系，预测施工区域相对于工作人员的危险等级，当预测的危险等级大于第二预设值时发出告警信息。The operator trajectory violation analysis unit is used to predict the danger level of the construction area relative to the staff based on the fitting relationship based on the environmental parameters of the construction site area, the type of staff, and the qualifications of the staff. When the predicted danger level is greater than the second preset value A warning message is issued. 10.根据权利要求9所述的基于视频图像分析的电力现场作业安全监控装置，其特征在于，所述绝对违章现象包括带电作业未佩戴绝缘手套、接地线挂接不牢靠、现场规范着装、典型违章行为，所述轨迹违章包括警戒区域闯入。10. The power field operation safety monitoring device based on video image analysis according to claim 9, wherein the absolute violations include not wearing insulating gloves during live work, unreliable ground wire hooking, on-site standard dress, typical Violations, the track violations include intrusion into warning areas.
CN202011224459.1A 2020-11-05 2020-11-05 Video image analysis-based power field operation safety monitoring method and device Active CN112347916B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011224459.1A CN112347916B (en) 2020-11-05 2020-11-05 Video image analysis-based power field operation safety monitoring method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011224459.1A CN112347916B (en) 2020-11-05 2020-11-05 Video image analysis-based power field operation safety monitoring method and device

Publications (2)

Publication Number Publication Date
CN112347916A true CN112347916A (en) 2021-02-09
CN112347916B CN112347916B (en) 2023-11-17

Family

ID=74428579

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011224459.1A Active CN112347916B (en) 2020-11-05 2020-11-05 Video image analysis-based power field operation safety monitoring method and device

Country Status (1)

Country Link
CN (1) CN112347916B (en)

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159633A (en) * 2021-05-11 2021-07-23 河北云酷科技有限公司 Method for evaluating risk level of field operating personnel in industrial field
CN113240695A (en) * 2021-06-02 2021-08-10 四川轻化工大学 Electric power operation personnel wearing identification method based on posture perception
CN113361968A (en) * 2021-07-05 2021-09-07 广东电网有限责任公司佛山供电局 Power grid infrastructure worker safety risk assessment method based on artificial intelligence and big data
CN113516383A (en) * 2021-06-30 2021-10-19 南通四建集团有限公司 Building construction safety control method and device
CN113819953A (en) * 2021-10-19 2021-12-21 山西全安新技术开发有限公司 Violation behavior identification method and intelligent violation-resisting sensor system
CN114092888A (en) * 2021-12-03 2022-02-25 国网四川省电力公司电力科学研究院 An electric shock risk detection system for experimental personnel in an electric power operation laboratory
CN114419356A (en) * 2022-01-26 2022-04-29 中国电力科学研究院有限公司 Detection method, system, equipment and storage medium for densely-arranged power equipment
CN114662864A (en) * 2022-03-03 2022-06-24 国网新疆电力有限公司信息通信公司 Team work intelligent management and control method and system based on artificial intelligence
CN115049979A (en) * 2022-06-14 2022-09-13 中国石油大学(华东) Real-time monitoring method for climbing behaviors of personnel in chemical industry park
CN115171006A (en) * 2022-06-15 2022-10-11 武汉纺织大学 Detection method for automatically identifying personnel entering electric power dangerous area based on deep learning
CN115392407A (en) * 2022-10-28 2022-11-25 中建五局第三建设有限公司 Method, device, equipment and medium for warning danger source based on unsupervised learning
CN115797850A (en) * 2023-02-06 2023-03-14 中国石油大学(华东) Oil field production safety early warning analysis system based on video stream
CN116108397A (en) * 2022-12-22 2023-05-12 福建亿榕信息技术有限公司 A method for identifying illegal behaviors in electric field operations based on multimodal data analysis
CN117911953A (en) * 2024-01-23 2024-04-19 国网内蒙古东部电力有限公司 An artificial intelligence power operation risk supervision and identification system and method
CN118261370A (en) * 2024-03-25 2024-06-28 江苏濠汉信息技术有限公司 Power transmission and distribution line safety management method and device
CN119360438A (en) * 2024-09-29 2025-01-24 北京图安世纪科技股份有限公司 A method and system for identifying and controlling violations of steel enterprise personnel based on AI analysis
CN119580195A (en) * 2025-02-06 2025-03-07 贵州黔程慧通科技发展有限责任公司 Method and system for monitoring high-altitude operation status based on sensor data analysis

Citations (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018086513A1 (en) * 2016-11-08 2018-05-17 杭州海康威视数字技术股份有限公司 Target detection method and device
CN108174165A (en) * 2018-01-17 2018-06-15 重庆览辉信息技术有限公司 Electric power safety operation and O＆M intelligent monitoring system and method
CN108776814A (en) * 2018-06-01 2018-11-09 广东电网有限责任公司 A kind of Electric Power Communication Data resource parallelization clustering method
CN109145789A (en) * 2018-08-09 2019-01-04 炜呈智能电力科技（杭州）有限公司 Power supply system safety work support method and system
CN109218673A (en) * 2018-09-20 2019-01-15 国网江苏省电力公司泰州供电公司 The system and method for power distribution network construction safety coordinated management control is realized based on artificial intelligence
CN110119656A (en) * 2018-02-07 2019-08-13 中国石油化工股份有限公司 Intelligent monitor system and the scene monitoring method violating the regulations of operation field personnel violating the regulations
CN110425005A (en) * 2019-06-21 2019-11-08 中国矿业大学 The monitoring of transportation of belt below mine personnel's human-computer interaction behavior safety and method for early warning
CN110443143A (en) * 2019-07-09 2019-11-12 武汉科技大学 The remote sensing images scene classification method of multiple-limb convolutional neural networks fusion
WO2019218621A1 (en) * 2018-05-18 2019-11-21 北京市商汤科技开发有限公司 Detection method for living being, device, electronic apparatus, and storage medium
CN110610592A (en) * 2019-09-25 2019-12-24 捻果科技(深圳)有限公司 A video analysis and deep learning method for monitoring the safe operation of the apron
CN110826514A (en) * 2019-11-13 2020-02-21 国网青海省电力公司海东供电公司 Construction site violation intelligent identification method based on deep learning
CN110991315A (en) * 2019-11-28 2020-04-10 江苏电力信息技术有限公司 Method for detecting wearing state of safety helmet in real time based on deep learning
CN111027481A (en) * 2019-12-10 2020-04-17 浩云科技股份有限公司 Behavior analysis method and device based on human body key point detection
CN111079658A (en) * 2019-12-19 2020-04-28 夸氪思维(南京)智能技术有限公司 Video-based multi-target continuous behavior analysis method, system and device
CN111339883A (en) * 2020-02-19 2020-06-26 国网浙江省电力有限公司 Identification and detection method of abnormal behavior in substations based on artificial intelligence in complex scenarios
CN111414873A (en) * 2020-03-26 2020-07-14 广州粤建三和软件股份有限公司 Alarm prompting method, device and alarm system based on wearing state of safety helmet
CN111460988A (en) * 2020-03-31 2020-07-28 国网河北省电力有限公司沧州供电分公司 Illegal behavior identification method and device
CN111507317A (en) * 2020-06-30 2020-08-07 之江实验室 Vision-based rotary equipment operation glove wearing detection method and system
CN111582129A (en) * 2020-04-30 2020-08-25 中铁工程装备集团有限公司 Real-time monitoring and alarming method and device for working state of shield machine driver
CN111652128A (en) * 2020-06-02 2020-09-11 浙江大华技术股份有限公司 A safety monitoring method, system and storage device for high-altitude electrical work
CN111832443A (en) * 2020-06-28 2020-10-27 华中科技大学 A Construction Method of Construction Violation Behavior Detection Model and Its Application

Patent Citations (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018086513A1 (en) * 2016-11-08 2018-05-17 杭州海康威视数字技术股份有限公司 Target detection method and device
CN108174165A (en) * 2018-01-17 2018-06-15 重庆览辉信息技术有限公司 Electric power safety operation and O＆M intelligent monitoring system and method
CN110119656A (en) * 2018-02-07 2019-08-13 中国石油化工股份有限公司 Intelligent monitor system and the scene monitoring method violating the regulations of operation field personnel violating the regulations
WO2019218621A1 (en) * 2018-05-18 2019-11-21 北京市商汤科技开发有限公司 Detection method for living being, device, electronic apparatus, and storage medium
CN108776814A (en) * 2018-06-01 2018-11-09 广东电网有限责任公司 A kind of Electric Power Communication Data resource parallelization clustering method
CN109145789A (en) * 2018-08-09 2019-01-04 炜呈智能电力科技（杭州）有限公司 Power supply system safety work support method and system
CN109218673A (en) * 2018-09-20 2019-01-15 国网江苏省电力公司泰州供电公司 The system and method for power distribution network construction safety coordinated management control is realized based on artificial intelligence
CN110425005A (en) * 2019-06-21 2019-11-08 中国矿业大学 The monitoring of transportation of belt below mine personnel's human-computer interaction behavior safety and method for early warning
CN110443143A (en) * 2019-07-09 2019-11-12 武汉科技大学 The remote sensing images scene classification method of multiple-limb convolutional neural networks fusion
CN110610592A (en) * 2019-09-25 2019-12-24 捻果科技(深圳)有限公司 A video analysis and deep learning method for monitoring the safe operation of the apron
CN110826514A (en) * 2019-11-13 2020-02-21 国网青海省电力公司海东供电公司 Construction site violation intelligent identification method based on deep learning
CN110991315A (en) * 2019-11-28 2020-04-10 江苏电力信息技术有限公司 Method for detecting wearing state of safety helmet in real time based on deep learning
CN111027481A (en) * 2019-12-10 2020-04-17 浩云科技股份有限公司 Behavior analysis method and device based on human body key point detection
CN111079658A (en) * 2019-12-19 2020-04-28 夸氪思维(南京)智能技术有限公司 Video-based multi-target continuous behavior analysis method, system and device
CN111339883A (en) * 2020-02-19 2020-06-26 国网浙江省电力有限公司 Identification and detection method of abnormal behavior in substations based on artificial intelligence in complex scenarios
CN111414873A (en) * 2020-03-26 2020-07-14 广州粤建三和软件股份有限公司 Alarm prompting method, device and alarm system based on wearing state of safety helmet
CN111460988A (en) * 2020-03-31 2020-07-28 国网河北省电力有限公司沧州供电分公司 Illegal behavior identification method and device
CN111582129A (en) * 2020-04-30 2020-08-25 中铁工程装备集团有限公司 Real-time monitoring and alarming method and device for working state of shield machine driver
CN111652128A (en) * 2020-06-02 2020-09-11 浙江大华技术股份有限公司 A safety monitoring method, system and storage device for high-altitude electrical work
CN111832443A (en) * 2020-06-28 2020-10-27 华中科技大学 A Construction Method of Construction Violation Behavior Detection Model and Its Application
CN111507317A (en) * 2020-06-30 2020-08-07 之江实验室 Vision-based rotary equipment operation glove wearing detection method and system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ZHENG-PING HU等: "Parallel spatial-temporal convolutional neural networks for anomaly detection and location in crowded scenes", 《JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION》, pages 1 - 11 *
王云峰: "基于深度学习的视频人体动作识别", 《中国优秀硕士学位论文全文数据库 信息科技辑》, vol. 2019, no. 1, pages 138 - 2392 *

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159633A (en) * 2021-05-11 2021-07-23 河北云酷科技有限公司 Method for evaluating risk level of field operating personnel in industrial field
CN113240695A (en) * 2021-06-02 2021-08-10 四川轻化工大学 Electric power operation personnel wearing identification method based on posture perception
CN113516383A (en) * 2021-06-30 2021-10-19 南通四建集团有限公司 Building construction safety control method and device
CN113516383B (en) * 2021-06-30 2024-05-03 南通四建集团有限公司 Building construction safety control method and device
CN113361968A (en) * 2021-07-05 2021-09-07 广东电网有限责任公司佛山供电局 Power grid infrastructure worker safety risk assessment method based on artificial intelligence and big data
CN113361968B (en) * 2021-07-05 2023-01-20 广东电网有限责任公司佛山供电局 Power grid infrastructure worker safety risk assessment method based on artificial intelligence and big data
CN113819953A (en) * 2021-10-19 2021-12-21 山西全安新技术开发有限公司 Violation behavior identification method and intelligent violation-resisting sensor system
CN114092888A (en) * 2021-12-03 2022-02-25 国网四川省电力公司电力科学研究院 An electric shock risk detection system for experimental personnel in an electric power operation laboratory
CN114092888B (en) * 2021-12-03 2023-05-02 国网四川省电力公司电力科学研究院 Electric shock risk detection system for laboratory personnel in electric power operation laboratory
CN114419356A (en) * 2022-01-26 2022-04-29 中国电力科学研究院有限公司 Detection method, system, equipment and storage medium for densely-arranged power equipment
CN114662864A (en) * 2022-03-03 2022-06-24 国网新疆电力有限公司信息通信公司 Team work intelligent management and control method and system based on artificial intelligence
CN115049979A (en) * 2022-06-14 2022-09-13 中国石油大学(华东) Real-time monitoring method for climbing behaviors of personnel in chemical industry park
CN115171006A (en) * 2022-06-15 2022-10-11 武汉纺织大学 Detection method for automatically identifying personnel entering electric power dangerous area based on deep learning
CN115392407A (en) * 2022-10-28 2022-11-25 中建五局第三建设有限公司 Method, device, equipment and medium for warning danger source based on unsupervised learning
CN116108397A (en) * 2022-12-22 2023-05-12 福建亿榕信息技术有限公司 A method for identifying illegal behaviors in electric field operations based on multimodal data analysis
CN116108397B (en) * 2022-12-22 2024-01-09 福建亿榕信息技术有限公司 Electric power field operation violation identification method integrating multi-mode data analysis
CN115797850A (en) * 2023-02-06 2023-03-14 中国石油大学(华东) Oil field production safety early warning analysis system based on video stream
CN117911953A (en) * 2024-01-23 2024-04-19 国网内蒙古东部电力有限公司 An artificial intelligence power operation risk supervision and identification system and method
CN117911953B (en) * 2024-01-23 2025-01-28 国网内蒙古东部电力有限公司 An artificial intelligence power operation risk supervision and identification system and method
CN118261370A (en) * 2024-03-25 2024-06-28 江苏濠汉信息技术有限公司 Power transmission and distribution line safety management method and device
CN119360438A (en) * 2024-09-29 2025-01-24 北京图安世纪科技股份有限公司 A method and system for identifying and controlling violations of steel enterprise personnel based on AI analysis
CN119580195A (en) * 2025-02-06 2025-03-07 贵州黔程慧通科技发展有限责任公司 Method and system for monitoring high-altitude operation status based on sensor data analysis

Also Published As

Publication number Publication date
CN112347916B (en) 2023-11-17

Similar Documents

Publication Publication Date Title
CN112347916A (en) Power field operation safety monitoring method and device based on video image analysis
CN109670441A (en) A kind of realization safety cap wearing knows method for distinguishing, system, terminal and computer readable storage medium
CN116579601B (en) Mine safety production risk monitoring and early warning system and method
CN114782897A (en) Dangerous behavior detection method and system based on machine vision and deep learning
CN112184773A (en) A kind of safety helmet wearing detection method and system based on deep learning
CN116524428A (en) Electric power operation safety risk identification method based on target detection and scene fusion
CN112069988A (en) Gun-ball linkage-based driver safe driving behavior detection method
CN118887767A (en) A forest fire early warning method, system and device based on satellite Internet of Things
CN104077571B (en) A kind of crowd's anomaly detection method that model is serialized using single class
CN111860187A (en) High-precision worn mask identification method and system
CN115880722A (en) Intelligent identification method, system and medium worn by power distribution operating personnel
CN118799821B (en) Construction site risk control method, device, storage medium and electronic equipment
CN113762115B (en) Distribution network operator behavior detection method based on key point detection
CN117911953B (en) An artificial intelligence power operation risk supervision and identification system and method
CN118885941A (en) A fault prediction method based on 360° dynamic image detection system
CN113536842B (en) Electric power operation personnel safety dressing identification method and device
CN113743725A (en) Method and system for detecting ascending qualification of ascending operator based on edge calculation
CN118887608A (en) Method and device for monitoring operation of power transmission line, and electronic equipment
CN117392611A (en) Construction site safety monitoring methods, systems, equipment and storage media
CN117830961A (en) Environment-friendly equipment operation and maintenance behavior analysis method and system based on image analysis
CN117877257A (en) Traffic scene monitoring method and system based on digital twin factory
CN117893968A (en) Intelligent identification model-based electric power operation safety risk monitoring method and device
CN116884513A (en) Method, system and equipment for monitoring dust exposure level of personnel in tunnel construction area
CN114241400A (en) Monitoring method and device of power grid system and computer readable storage medium
CN112883889A (en) Method suitable for detecting violation behaviors in power construction scene

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant