CN118038119A - Safety belt wearing detection method and system in high-altitude power operation scene - Google Patents
Safety belt wearing detection method and system in high-altitude power operation scene Download PDFInfo
- Publication number
- CN118038119A CN118038119A CN202311835286.0A CN202311835286A CN118038119A CN 118038119 A CN118038119 A CN 118038119A CN 202311835286 A CN202311835286 A CN 202311835286A CN 118038119 A CN118038119 A CN 118038119A
- Authority
- CN
- China
- Prior art keywords
- safety belt
- wearing
- power operation
- operation scene
- detection
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 82
- 230000006870 function Effects 0.000 claims abstract description 31
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000012544 monitoring process Methods 0.000 claims abstract description 13
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000004590 computer program Methods 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 7
- 230000009467 reduction Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 230000009471 action Effects 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 238000011897 real-time detection Methods 0.000 claims description 2
- 238000002203 pretreatment Methods 0.000 claims 1
- 230000007246 mechanism Effects 0.000 abstract description 6
- 230000000694 effects Effects 0.000 abstract description 4
- 239000000284 extract Substances 0.000 abstract 1
- 238000013135 deep learning Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000026676 system process Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for detecting the wearing of a safety belt in a high-altitude power operation scene, which relate to the technical field of target detection and comprise the steps of collecting a data set in the power operation scene and preprocessing; optimizing a loss function and constructing a safety belt wearing detection network structure; training by using the preprocessed data set and the optimized loss function to obtain a safety belt wearing detection model; and detecting images acquired on site in real time, and detecting the wearing of the safety belt in the high-altitude power operation scene. The invention introduces a data enhancement and attention mechanism, remarkably improves the learning ability of the network, and more effectively extracts key information from the monitoring image. The detection effect on a long-distance small target is improved, and the applicability of the system in a complex high-altitude power operation scene is enhanced. The labor cost is greatly reduced, and convenience are provided for subsequent high-altitude power operation through an automatic safety belt detection system.
Description
Technical Field
The invention relates to the technical field of target detection, in particular to a method and a system for detecting wearing of a safety belt in a high-altitude power operation scene.
Background
In recent years, with the increasing maturity of computer vision computation, especially the rapid development of neural network technology, deep learning technology is widely applied in various production environments. The idea of deep learning was originally derived from the study of artificial neural networks by western math and computer scientists. The artificial neural network is an algorithm model imitating the behavior characteristics of the animal neural network, and the distributed parallel information processing is realized by adjusting the interconnection relation among a large number of internal nodes, so that the purpose of processing information is achieved. The introduction of deep learning not only has great influence in life, but also is gradually applied to high-altitude power operation scenes.
In the high-altitude power operation field, dangerous factors such as special electrified equipment and the like exist, the environment is complex, and safety accidents are easy to occur. If the personnel which do not meet the power working requirements enter the construction site, serious casualties can be caused. To reduce such potential hazards, it is desirable to monitor the belt wear of the operator during routine power operations. However, due to insufficient safety awareness of the worker, the safety belt is not normally worn or worn easily. The monitoring personnel monitor the safety belt constantly, so that the labor intensity is high, the intelligent level is relatively low, and the efficiency of manually detecting whether to wear the safety belt is low.
Disclosure of Invention
The invention is provided in view of the problems of the existing safety belt wearing detection method in the high-altitude power operation scene. A YOLOX target detection model is introduced in the high-altitude power operation scene, so that a more effective means is provided for accurate detection of the safety belt. However, in the case that the target to be detected is far away or has shielding, the detection effect may be less than ideal, which is not consistent with the actual high-altitude power operation scene. Therefore, the invention aims to provide a method and a system for detecting the wearing of a safety belt in a high-altitude power operation scene.
In order to solve the technical problems, the invention provides the following technical scheme:
In a first aspect, the present invention provides a method for detecting belt wear in an overhead power operation scenario, including collecting a data set in the power operation scenario, and preprocessing the data set; optimizing a loss function of a safety belt wearing detection network applicable to an overhead power operation scene and constructing a safety belt wearing detection network structure; the detection network is trained by using the preprocessed data set and the optimized loss function, and a safety belt wearing detection model is obtained; and detecting the images acquired on site in real time by using the trained safety belt wearing detection model, and detecting the wearing of the safety belt in the high-altitude power operation scene.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the preprocessing includes the steps of collecting relevant video and images from an overhead power operation site; storing the acquired video frame by frame as images, and integrating the images with the acquired images for manual screening; manually labeling the screened data set, and dividing the data set into three labels of standard wearing, non-standard wearing and non-wearing; and expanding the marked data set by adopting a data enhancement technology.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the loss function comprises wear detection loss and normalized regular loss, wherein the wear detection loss is used for focusing on classification performance of a model on the wearing condition of the safety belt, and is defined as follows:
Where N is the number of samples, C is the number of categories, y i,j is a binary label, indicating whether sample i belongs to category j, Representing the model predictive probability of the ith belonging to the jth class, w j represents the weight of the jth class.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the normalized regularization loss is used for normalizing model parameters, so that overfitting is avoided, and the related definition formula is as follows:
Where α represents the weight of the regularization term, K represents the total number of model parameters, and θ κ represents the kth model parameter; the final loss function combines the wear detection loss and the normalized canonical loss together, and simultaneously considers classification performance and model complexity to prevent overfitting, and is defined as follows:
Loss＝Losswear+Lossnormal
Where Loss wear represents wear detection Loss and Loss normal represents normalized regular Loss.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the construction of the safety belt wearing detection network structure comprises the following steps of performing feature extraction operation on an input image, performing feature extraction through a 3x3 convolution layer, and enabling each layer to be followed by a batch normalization and activation function to promote stable convergence of a network and learn more complex features; performing feature enhancement, up-sampling an input feature image, performing channel dimension reduction through a 1x1 convolution layer, obtaining a feature image with wider context through a down-sampling path, and fusing the feature images of the two paths through dimension reduction of the 1x1 convolution layer; extracting multi-scale features from an input image, introducing multi-scale output layers on different levels, upsampling a low-resolution feature map, fusing the upsampled feature map with a high-resolution feature map from a backbone network through feature fusion operation, and performing scale matching operation to ensure that the feature map has consistent scales; processing the input feature map through global averaging pooling, learning the weight relation of each channel through linear transformation, applying softmax operation to linear transformation results to obtain the attention weight of each channel, and applying the attention weight to the original feature map; the feature is extracted through a convolution layer, the feature map is converted into a vector with a fixed size through global average pooling, a full-connection layer is used for mapping output of the global average pooling to information of a predicted target class and a boundary box, and an activation function is used for obtaining probability of the target class and coordinates of the boundary box.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the training comprises the following steps that the preprocessed image data is used, and parameters of a neural network are iteratively optimized through a back propagation algorithm, so that the parameters can accurately detect the wearing state of the safety belt in an aerial power operation scene; verifying the model obtained through training by using a verification set, and performing optimization of model parameters according to a verification result so as to improve the generalization capability of the model on unseen data; and storing the trained safety belt wearing detection model, and preparing to be deployed into an actual system for monitoring the wearing state of the safety belt in a real-time high-altitude power operation scene.
As a preferable scheme of the method for detecting the wearing of the safety belt in the high-altitude power operation scene, the invention comprises the following steps: the real-time detection field acquired image comprises the following steps of inputting real-time video stream or image to detect the wearing condition of the safety belt of the high-altitude power operation personnel in time; judging whether the wearing situation of the safety belt is unworn or not according to the detected wearing situation of the safety belt; when the system judges that the system is not worn or is not worn normally, the system can trigger an acoustic alarm, immediately send an alarm signal to the surrounding environment in an acoustic reminding mode, attract the attention of related personnel, send an alarm notification, send alarm information to appointed personnel or a monitoring center through a communication means, and timely take necessary safety measures and actions.
In a second aspect, the present invention provides a seat belt wear detection system in an overhead power operation scenario, comprising: the acquisition module is used for acquiring a data set in an electric power operation scene and preprocessing the data set; the optimizing module is used for optimizing a loss function of the safety belt wearing detection network applicable to the high-altitude power operation scene and constructing a safety belt wearing detection network structure; the training module is used for training the target detection network by using the preprocessed data set and the optimized loss function to obtain a safety belt wearing detection model; the detection module is used for detecting the image acquired on site in real time by using the trained safety belt wearing detection model and carrying out safety belt wearing detection under the high-altitude power operation scene.
In a third aspect, the present invention provides a computer device comprising a memory and a processor, the memory storing a computer program, wherein: and the processor executes the computer program to realize the safety belt wearing detection method in the high-altitude power operation scene.
In a fourth aspect, the present invention provides a computer-readable storage medium having stored thereon a computer program, wherein: and the computer program, when executed by the processor, realizes the step of the safety belt wearing detection method in the high-altitude power operation scene.
The invention has the beneficial effects that a data enhancement and attention mechanism is introduced, the learning capacity of the network is obviously improved, and key information is more effectively extracted from the monitoring image. Through ingenious improvement of the network structure, the detection effect of a long-distance small target is successfully improved, and the applicability of the system in a complex high-altitude power operation scene is enhanced. The labor cost is greatly reduced, and great convenience and convenience are provided for subsequent high-altitude power operation through an automatic safety belt detection system.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a seat belt wear detection method in a high-altitude power operation scenario.
Fig. 2 is a block diagram of a seat belt wear detection method in an overhead power operation scenario.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
Example 1
Referring to fig. 1, a first embodiment of the present invention provides a method for detecting belt wear in a high-altitude power operation scenario, including:
s1: and collecting a data set in the electric power operation scene, and preprocessing the data set.
Specifically, video and image acquisition: shooting is performed at an overhead power job site using appropriate equipment (e.g., high definition cameras) to obtain relevant video and images. The acquired data is ensured to have enough resolution and definition so that accurate results can be obtained in the subsequent processing steps.
The video is stored as an image frame by frame: the acquired video is split into individual images frame by frame, ensuring that each frame is captured.
Manual screening: and manually screening the acquired images, and judging whether the images are subjected to normal wearing, non-normal wearing and non-wearing according to preset standards. Manual screening may be performed by a professional or automated screening may be performed using computer vision techniques.
Manual labeling: and (3) manually labeling the screened data set, and adding corresponding labels (normative worn, non-normative worn and non-worn) for each image. The accuracy of labeling is critical to training a machine learning model, ensuring that the labels are consistent with the image content.
Data set augmentation (data enhancement): the annotated dataset is augmented using data enhancement techniques such as rotation, flipping, scaling, etc. The data enhancement helps to improve the generalization capability of the model, so that the model is more robust under different scenes and conditions.
S2: optimizing a loss function of a safety belt wearing detection network and constructing a safety belt wearing detection network structure, wherein the loss function is suitable for a high-altitude power operation scene.
In particular, the loss function includes wear detection loss and normalized canonical loss,
Wear detection loss is used for focusing on classification performance of the model on the wearing condition of the safety belt, and is defined as follows:
Where N is the number of samples, C is the number of categories, y i,j is a binary label, indicating whether sample i belongs to category j, Representing the model predictive probability of the ith belonging to the jth class, w j represents the weight of the jth class.
Normalized canonical losses are used to normalize model parameters to avoid overfitting, and the relevant definitions are as follows:
Where α represents the weight of the regularization term, K represents the total number of model parameters, and θ κ represents the kth model parameter; the final loss function combines the wear detection loss and the normalized canonical loss together, and simultaneously considers classification performance and model complexity to prevent overfitting, and is defined as follows:
Loss＝Losswear+Lossnormal
Where Loss wear represents wear detection Loss and Loss normal represents normalized regular Loss.
Carrying out feature extraction operation on an input image, carrying out feature extraction through a 3x3 convolution layer, and enabling each layer to be followed by a batch normalization and activation function to promote stable convergence of a network and learn more complex features;
Performing feature enhancement, up-sampling an input feature image, performing channel dimension reduction through a 1x1 convolution layer, obtaining a feature image with wider context through a down-sampling path, and fusing the feature images of the two paths through dimension reduction of the 1x1 convolution layer;
extracting multi-scale features from an input image, introducing multi-scale output layers on different levels, upsampling a low-resolution feature map, fusing the upsampled feature map with a high-resolution feature map from a backbone network through feature fusion operation, and performing scale matching operation to ensure that the feature map has consistent scales;
Processing the input feature map through global averaging pooling, learning the weight relation of each channel through linear transformation, applying softmax operation to linear transformation results to obtain the attention weight of each channel, and applying the attention weight to the original feature map;
extracting features by a convolution layer, converting the feature map into a vector with a fixed size by global average pooling, mapping the output of the global average pooling to information of a predicted target class and a boundary box by a full connection layer, and obtaining the probability of the target class and the coordinates of the boundary box by using an activation function
S3: the target detection network is trained by using the preprocessed data set and the optimized loss function, and a belt wearing detection model is obtained.
Specifically, the image data is preprocessed: preprocessing the acquired image data, including scaling, normalization, denoising and the like, so that the data input into the neural network have a consistent format and quality.
Building a neural network model: a neural network model suitable for belt wearing detection is designed, and a Convolutional Neural Network (CNN) and other structures can be selected to be used. The structures of the input layer, the hidden layer and the output layer are determined, and an activation function and a loss function are selected.
Taking the preprocessed image data as input, and iteratively optimizing parameters of the neural network through a back propagation algorithm. Training is performed using a training set, and the network weights and biases are adjusted in a manner that minimizes the loss function.
And verifying the model obtained through training by using a verification set, and evaluating the performance of the model. And (3) optimizing the model parameters according to the verification result, and improving the generalization capability of the model by means of adjusting the learning rate, increasing regularization and the like. And saving the trained belt wearing detection model as a file for subsequent deployment.
And integrating the stored model into an actual system for real-time monitoring of the wearing state of the safety belt in the high-altitude power operation scene. Ensuring that the model is compatible with the interfaces of the actual system and performing the necessary tests and verifications.
And monitoring the wearing state of the safety belt in an actual system, and acquiring image data of an overhead power operation scene in real time. And deducing the image data by using the deployed model to obtain a prediction result of the wearing state of the safety belt. And carrying out real-time monitoring and feedback according to the prediction result, and ensuring the safety of operators.
S4: and detecting the images acquired on site in real time by using the trained safety belt wearing detection model, and detecting the wearing of the safety belt in the high-altitude power operation scene.
Specifically, the safety belt wearing condition of high altitude electric power operation personnel in real time: the safety belt wearing detection network structure has the capability of real-time monitoring, and timely detects the wearing state of the safety belt of the high-altitude electric power operation personnel by inputting real-time video stream or image.
Based on the detected wearing condition of the safety belt, judgment is carried out to distinguish whether the safety belt is not worn or is not worn normally.
When the system judges that the system is not worn or is not worn normally, the system can trigger an acoustic alarm, and immediately send an alarm signal to the surrounding environment in an acoustic reminding mode so as to attract the attention of related personnel. And sending alarm notification to related personnel, and simultaneously, the system also has the function of sending alarm notification to related personnel, and sending alarm information to appointed personnel or a monitoring center through a communication means so as to take necessary security measures and actions in time. This helps to improve the safety and emergency response capability of the workplace.
Further, this embodiment also provides a safety belt wearing detection system under high altitude power operation scene, includes: collecting a data set in an electric power operation scene, and preprocessing the data set; optimizing a loss function of a safety belt wearing detection network applicable to an overhead power operation scene and constructing a safety belt wearing detection network structure; the detection network is trained by using the preprocessed data set and the optimized loss function, and a safety belt wearing detection model is obtained; and detecting the images acquired on site in real time by using the trained safety belt wearing detection model, and detecting the wearing of the safety belt in the high-altitude power operation scene.
The embodiment also provides a computer device, which is suitable for the situation of the safety belt wearing detection method in the high-altitude power operation scene, and comprises the following steps: a memory and a processor; the memory is used for storing computer executable instructions and the processor is used for executing the computer executable instructions to implement all or part of the steps of the method according to the embodiments of the present invention as set forth in the embodiments above.
The present embodiment also provides a storage medium having stored thereon a computer program which, when executed by a processor, performs the method of any of the alternative implementations of the above embodiments. The storage medium may be implemented by any type of volatile or non-volatile Memory device or combination thereof, such as static random access Memory (Static Random Access Memory, SRAM), electrically erasable Programmable Read-Only Memory (ELECTRICALLY ERASABLE PROGRAMMABLE READ-Only Memory, EEPROM), erasable Programmable Read-Only Memory (Erasable Programmable Read OnlyMemory, EPROM), programmable Read-Only Memory (PROM), read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk, or optical disk.
The storage medium according to the present embodiment belongs to the same inventive concept as the data storage method according to the above embodiment, and technical details not described in detail in the present embodiment can be seen in the above embodiment, and the present embodiment has the same advantageous effects as the above embodiment.
From the above, the method introduces a data enhancement and attention mechanism, and increases the adaptability of the model to different scenes and changes by expanding training data. The training set is facilitated to be expanded, more diversified data is provided, and therefore the generalization capability of the network is enhanced. Through the attention mechanism, the network can pay attention to important areas in the image more specifically, so that the extraction efficiency of the key information is improved. The network can focus on the key area, and the interference of redundant information is reduced, so that key features are more effectively learned. Through ingenious improvement of the network structure, including adding deeper layers and adjusting the size of convolution kernels, the sensitivity and accuracy to long-distance small targets are improved. The applicability of the system in a complex high-altitude power operation scene is enhanced, the wearing state of the safety belt can be detected more accurately when the system processes images in the complex high-altitude power operation scene, and the applicability and practicality of the system are improved. The automated belt detection system is capable of completing the monitoring task without human intervention, thereby reducing labor costs and labor requirements. The high-altitude power operation system provides convenience and convenience for high-altitude power operation, can greatly improve safety in real-time monitoring, and simultaneously lightens the burden of manual operation.
Example 2
Referring to fig. 2, a second embodiment of the present invention provides a method for detecting belt wear in a high-altitude power operation scene, and in order to verify the beneficial effects of the present invention, scientific demonstration is performed through economic benefit calculation and simulation experiments.
Table 1 comparison table of test results
From the above, the method introduces a data enhancement and attention mechanism, and increases the adaptability of the model to different scenes and changes by expanding training data. The training set is facilitated to be expanded, more diversified data is provided, and therefore the generalization capability of the network is enhanced. Through the attention mechanism, the network can pay attention to important areas in the image more specifically, so that the extraction efficiency of the key information is improved. The network can focus on the key area, and the interference of redundant information is reduced, so that key features are more effectively learned. Through ingenious improvement of the network structure, including adding deeper layers and adjusting the size of convolution kernels, the sensitivity and accuracy to long-distance small targets are improved. The applicability of the system in a complex high-altitude power operation scene is enhanced, the wearing state of the safety belt can be detected more accurately when the system processes images in the complex high-altitude power operation scene, and the applicability and practicality of the system are improved. The automated belt detection system is capable of completing the monitoring task without human intervention, thereby reducing labor costs and labor requirements. The high-altitude power operation system provides convenience and convenience for high-altitude power operation, can greatly improve safety in real-time monitoring, and simultaneously lightens the burden of manual operation.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (10)
1. A method for detecting wearing of a safety belt in a high-altitude power operation scene is characterized by comprising the following steps of: comprising the steps of (a) a step of,
Collecting a data set in an electric power operation scene, and preprocessing the data set;
Optimizing a loss function of a safety belt wearing detection network applicable to an overhead power operation scene and constructing a safety belt wearing detection network structure;
the detection network is trained by using the preprocessed data set and the optimized loss function, and a safety belt wearing detection model is obtained;
And detecting the images acquired on site in real time by using the trained safety belt wearing detection model, and detecting the wearing of the safety belt in the high-altitude power operation scene.
2. The seat belt wear detection method in an overhead power operation scenario according to claim 1, wherein: the pre-treatment comprises the steps of,
Collecting relevant videos and images from an overhead power operation site;
Storing the acquired video frame by frame as images, and integrating the images with the acquired images for manual screening;
Manually labeling the screened data set, and dividing the data set into three labels of standard wearing, non-standard wearing and non-wearing;
And expanding the marked data set by adopting a data enhancement technology.
3. The seat belt wear detection method in the high-altitude power operation scene according to claim 2, wherein: the loss function includes wear detection loss and normalized canonical loss,
The wear detection loss is used for focusing on classification performance of the model on the wearing condition of the safety belt, and is defined as follows:
Where N is the number of samples, C is the number of categories, y i,j is a binary label, indicating whether sample i belongs to category j, Representing the model predictive probability of the ith belonging to the jth class, w j represents the weight of the jth class.
4. The seatbelt wear detection method in an overhead power operation scenario according to claim 3, wherein: the normalized regularization loss is used for normalizing model parameters, so that overfitting is avoided, and the related definition formula is as follows:
Where α represents the weight of the regularization term, K represents the total number of model parameters, and θ κ represents the kth model parameter;
the final loss function combines the wear detection loss and the normalized canonical loss together, and simultaneously considers classification performance and model complexity to prevent overfitting, and is defined as follows:
Loss＝Losswear+Lossnormal
Where Loss wear represents wear detection Loss and Loss normal represents normalized regular Loss.
5. The method for detecting the wearing of a safety belt in an overhead power operation scene according to claim 4, wherein: the construction of the seatbelt wear detection network structure includes the steps of,
Carrying out feature extraction operation on an input image, carrying out feature extraction through a 3x3 convolution layer, and enabling each layer to be followed by a batch normalization and activation function to promote stable convergence of a network and learn more complex features;
Performing feature enhancement, up-sampling an input feature image, performing channel dimension reduction through a 1x1 convolution layer, obtaining a feature image with wider context through a down-sampling path, and fusing the feature images of the two paths through dimension reduction of the 1x1 convolution layer;
extracting multi-scale features from an input image, introducing multi-scale output layers on different levels, upsampling a low-resolution feature map, fusing the upsampled feature map with a high-resolution feature map from a backbone network through feature fusion operation, and performing scale matching operation to ensure that the feature map has consistent scales;
Processing the input feature map through global averaging pooling, learning the weight relation of each channel through linear transformation, applying softmax operation to linear transformation results to obtain the attention weight of each channel, and applying the attention weight to the original feature map;
The feature is extracted through a convolution layer, the feature map is converted into a vector with a fixed size through global average pooling, a full-connection layer is used for mapping output of the global average pooling to information of a predicted target class and a boundary box, and an activation function is used for obtaining probability of the target class and coordinates of the boundary box.
6. The method for detecting the wearing of a safety belt in an overhead power operation scene according to claim 5, wherein: the training comprises the steps of,
The preprocessed image data is used, and parameters of the neural network are iteratively optimized through a back propagation algorithm, so that the parameters can accurately detect the wearing state of the safety belt in the high-altitude power operation scene;
Verifying the model obtained through training by using a verification set, and performing optimization of model parameters according to a verification result so as to improve the generalization capability of the model on unseen data;
and storing the trained safety belt wearing detection model, and preparing to be deployed into an actual system for monitoring the wearing state of the safety belt in a real-time high-altitude power operation scene.
7. The method for detecting the wearing of a safety belt in an overhead power operation scene according to claim 6, wherein: the real-time detection of the live acquired image comprises the steps of,
The method comprises the steps of inputting real-time video stream or image according to the wearing condition of the safety belt of the high-altitude power operator in real time, and detecting the wearing state of the safety belt of the high-altitude power operator in time;
Judging whether the wearing situation of the safety belt is unworn or not according to the detected wearing situation of the safety belt;
When the system judges that the system is not worn or is not worn normally, the system can trigger an acoustic alarm, immediately send an alarm signal to the surrounding environment in an acoustic reminding mode, attract the attention of related personnel, send an alarm notification, send alarm information to appointed personnel or a monitoring center through a communication means, and timely take necessary safety measures and actions.
8. A safety belt wearing detection system in a high-altitude power operation scene, based on the safety belt wearing detection method in the high-altitude power operation scene according to any one of claims 1 to 7, characterized in that: comprising the steps of (a) a step of,
The acquisition module is used for acquiring a data set in an electric power operation scene and preprocessing the data set;
The optimizing module is used for optimizing a loss function of the safety belt wearing detection network applicable to the high-altitude power operation scene and constructing a safety belt wearing detection network structure;
the training module is used for training the target detection network by using the preprocessed data set and the optimized loss function to obtain a safety belt wearing detection model;
The detection module is used for detecting the image acquired on site in real time by using the trained safety belt wearing detection model and carrying out safety belt wearing detection under the high-altitude power operation scene.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: the steps of the method for detecting the wearing of the safety belt in the high-altitude power operation scene according to any one of claims 1 to 7 are realized when the processor executes the computer program.
10. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program when executed by a processor realizes the steps of the seatbelt wearing detection method in the high-altitude power operation scene according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311835286.0A CN118038119A (en) | 2023-12-28 | 2023-12-28 | Safety belt wearing detection method and system in high-altitude power operation scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311835286.0A CN118038119A (en) | 2023-12-28 | 2023-12-28 | Safety belt wearing detection method and system in high-altitude power operation scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118038119A true CN118038119A (en) | 2024-05-14 |
Family
ID=90983041
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311835286.0A Pending CN118038119A (en) | 2023-12-28 | 2023-12-28 | Safety belt wearing detection method and system in high-altitude power operation scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118038119A (en) |
-
2023
- 2023-12-28 CN CN202311835286.0A patent/CN118038119A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110084165B (en) | Intelligent identification and early warning method for abnormal events in open scene of power field based on edge calculation | |
| CN111563557B (en) | Method for detecting target in power cable tunnel | |
| CN109977921B (en) | Method for detecting hidden danger of power transmission line | |
| CN107742099A (en) | A kind of crowd density estimation based on full convolutional network, the method for demographics | |
| CN110516529A (en) | It is a kind of that detection method and system are fed based on deep learning image procossing | |
| CN111222478A (en) | Construction site safety protection detection method and system | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| CN115761537B (en) | Power transmission line foreign matter intrusion identification method oriented to dynamic feature supplementing mechanism | |
| CN108664875A (en) | Underground belt-conveying monitoring method based on image recognition | |
| CN116416281A (en) | Grain depot AI video supervision and analysis method and system | |
| CN115546742A (en) | Rail foreign matter identification method and system based on monocular thermal infrared camera | |
| CN118411724A (en) | Icing form segmentation method for overhead transmission line | |
| CN118447442A (en) | Intelligent safety helmet system based on image reverse operation and AI identification | |
| CN118038021A (en) | Transformer substation operation site foreign matter intrusion detection method based on improvement yolov4 | |
| CN117423157A (en) | Mine abnormal video action understanding method combining migration learning and regional invasion | |
| CN118038119A (en) | Safety belt wearing detection method and system in high-altitude power operation scene | |
| CN117636129A (en) | Identification method of power construction supervision image | |
| CN116546287A (en) | Multi-linkage wild animal online monitoring method and system | |
| CN116091976A (en) | Station room defect identification detection method, system, device and storage medium | |
| CN115909144A (en) | Method and system for detecting abnormity of surveillance video based on counterstudy | |
| CN114694090A (en) | Campus abnormal behavior detection method based on improved PBAS algorithm and YOLOv5 | |
| Thevarasa et al. | Weighted Ensemble Algorithm for Aerial Imaging Based Mosquito Breeding Sites Classification | |
| Torsvik et al. | Detection of macroplastic on beaches using drones and object-based image analysis | |
| Sládek | Application of K‐Nearest Neighbor Classification for Static Webcams Visibility Observation | |
| CN118314488B (en) | Extra-high voltage transformer station space-earth multi-scale re-decision target detection method |
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 |