Safety vest target detection system based on power construction recognition environment
By designing a safety vest object detection system based on the power construction identification environment and using the convolutional neural network model for automated detection, the problem of time-consuming and labor-intensive human detection in the power construction environment in the prior art is solved, and efficient and accurate safety vest detection is achieved.
Patent Information
- Application Number
- CN202510048496.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-13
- Publication Date
- 2025-05-09
AI Technical Summary
The methods for identifying whether power construction personnel wear safety vests in existing power construction environments mainly rely on human testing, which is time-consuming and labor-intensive and has a limited range of testing.
Design a safety vest target detection system based on the power construction identification environment, and realizes automated safety vest detection through modules such as historical data acquisition, object detection model construction and training, real-time data acquisition and detection, judgment and early warning.
It realizes automatic inspection, saves time and effort, has fast detection speed, high detection accuracy and wide detection range, and can effectively identify whether power construction personnel wear safety vests.
Smart Images
Figure CN119963819A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of electric power construction, and in particular relates to a safety vest target detection system based on an electric power construction identification environment. Background Art
[0002] Power construction refers to a series of work involving the construction, installation, maintenance and overhaul of power equipment and facilities, which is of great significance for ensuring power supply and safe operation.
[0003] In order to ensure safety during power construction, power construction workers must wear safety vests. In some cases, some power construction workers may forget to wear safety vests due to various factors. At this time, it is particularly important to identify whether power construction workers are wearing safety vests in the power construction environment.
[0004] Currently, the method for identifying whether power construction workers are wearing safety vests in power construction environments is generally manual detection, which is time-consuming and labor-intensive, and has a limited detection range.
[0005] In view of this, a safety vest target detection system based on the power construction identification environment is designed to solve the above problems. Summary of the Invention
[0006] To solve the problems raised in the above background technology, the present invention provides a safety vest target detection system based on the power construction identification environment, which has the characteristics of automated detection, time and labor saving, fast detection speed, high detection accuracy and wide detection range.
[0007] To achieve the above objectives, the present invention provides the following technical solutions: a safety vest target detection system based on power construction identification environment, comprising:
[0008] A historical data collection module collects historical power construction environment images, which include power construction workers and safety vests;
[0009] The target detection model construction module builds the human target detection model and the safety vest target detection model;
[0010] The target detection model training module trains the human target detection model and the safety vest target detection model based on the collected historical power construction environment images until convergence;
[0011] A real-time data acquisition module collects images of the power construction environment, which include power construction workers and safety vests;
[0012] The target detection module detects and outputs human targets and safety vest targets in the collected power construction environment images based on the human target detection model and the safety vest target detection model;
[0013] The judgment module compares the number of detected human targets with the number of safety vest targets to determine whether the power construction workers have violated the regulations by not wearing safety vests;
[0014] The early warning module issues an early warning when it determines that power construction workers are operating in violation of regulations by not wearing safety vests.
[0015] Furthermore, in the target detection model construction module, the person target detection model and the safety vest target detection model are both convolutional neural network models.
[0016] Furthermore, in the target detection model training module, the specific steps of training the person target model include:
[0017] The collected historical power construction environment images are divided into two parts, and the characters in the images are annotated in one part.
[0018] The historical power construction environment images with people annotated are input into the constructed human target detection model for training;
[0019] Calculate the loss function based on the detection and annotation results, backpropagate, and optimize the parameters until the model converges.
[0020] Furthermore, in the target detection model training module, the specific steps of training the safety vest target detection model include:
[0021] The collected historical power construction environment images are divided into two parts, and the safety vests in the other part are labeled;
[0022] The historical power construction environment images marked with safety vests are input into the constructed safety vest target detection model for training;
[0023] Calculate the loss function based on the detection and annotation results, backpropagate, and optimize the parameters until the model converges.
[0024] Furthermore, in the real-time data acquisition module, the power construction environment image is collected in multiple directions, and the collection is continuous and uninterrupted in the same direction.
[0025] Furthermore, in the real-time data acquisition module, the electric power construction environment images collected from multiple directions are spliced based on coordinates to improve the electric power construction environment images.
[0026] Compared with the prior art, the present invention has the following beneficial effects:
[0027] The present invention trains and constructs a person target detection model and a safety vest target detection model based on the collected historical power construction environment images, and then detects people and safety vests in the real-time collected power construction environment images based on the converged person target detection model and the safety vest target detection model. Then, the number of people and safety vests is compared to realize the detection of whether the power construction personnel in the power construction environment are wearing safety vests. Compared with the existing technology, the detection is automated, which saves time and labor, has a fast detection speed, high detection accuracy and a wide detection range. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 It is a system framework diagram of the present invention;
[0029] In the figure: 1. Historical data acquisition module; 2. Target detection model construction module; 3. Target detection model training module; 4. Real-time data acquisition module; 5. Target detection module; 6. Judgment module; 7. Early warning module. DETAILED DESCRIPTION
[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0031] See attached Figure 1 The present invention provides the following technical solutions: a safety vest target detection system based on the power construction identification environment, comprising:
[0032] The historical data collection module 1 collects historical power construction environment images, which include power construction workers and safety vests;
[0033] Target detection model construction module 2, builds a person target detection model and a safety vest target detection model;
[0034] Target detection model training module 3: Based on the collected historical power construction environment images, the human target detection model and the safety vest target detection model are trained until convergence;
[0035] The real-time data acquisition module 4 collects images of the power construction environment, which include power construction workers and safety vests;
[0036] Target detection module 5, based on the human target detection model and the safety vest target detection model, detects the human target and the safety vest target in the collected power construction environment image and outputs them;
[0037] Determination module 6, based on the comparison of the number of detected human targets and the number of safety vest targets, determines whether the power construction workers have violated the regulations by not wearing safety vests;
[0038] The early warning module 7 issues an early warning when it is determined that the power construction personnel have violated the regulations by not wearing safety vests.
[0039] Specifically, in the target detection model construction module 2, the person target detection model and the safety vest target detection model are both convolutional neural network models.
[0040] A convolutional neural network model refers to a deep learning model specifically designed for processing grid-structured data including images. It consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives input data including images and converts it into numerical values. The convolutional layer performs local connections and convolution operations with the input data through convolution kernels to extract the features of the input data. The pooling layer downsamples the feature map output by the convolutional layer to reduce the spatial size of the input data, reduce the computational complexity of the model, and extract important features. The fully connected layer integrates the features of the aforementioned outputs, and the output layer outputs the final result.
[0041] The convolutional neural network model as a target detection model has the effects of fast detection speed and high detection accuracy.
[0042] Specifically, in the target detection model training module 3, the specific steps of training the person target model include:
[0043] The collected historical power construction environment images are divided into two parts, and the characters in the images are annotated in one part.
[0044] The historical power construction environment images with people annotated are input into the constructed human target detection model for training;
[0045] Calculate the loss function based on the detection and annotation results, backpropagate, and optimize the parameters until the model converges.
[0046] Specifically, in the target detection model training module 3, the specific steps of the safety vest target detection model training include:
[0047] The collected historical power construction environment images are divided into two parts, and the safety vests in the other part are labeled;
[0048] The historical power construction environment images marked with safety vests are input into the constructed safety vest target detection model for training;
[0049] Calculate the loss function based on the detection and annotation results, backpropagate, and optimize the parameters until the model converges.
[0050] Specifically, in the real-time data acquisition module 4, the power construction environment image is collected in multiple directions, and the collection is continuous and uninterrupted in the same direction.
[0051] Multi-directional acquisition can avoid the problem of incomplete images of people and safety vests in the collected images of power construction environments;
[0052] Continuous and uninterrupted acquisition in one direction can avoid the interference of excessive redundant image information.
[0053] Specifically, in the real-time data acquisition module 4, the electric power construction environment images collected from multiple directions are spliced based on coordinates to improve the electric power construction environment images.
[0054] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A safety vest target detection system based on power construction recognition environment, characterized in that: include: A historical data collection module (1) collects historical power construction environment images, wherein the historical power construction environment images include power construction workers and safety vests; The target detection model building module (2) builds a human target detection model and a safety vest target detection model; The target detection model training module (3) trains the human target detection model and the safety vest target detection model based on the collected historical power construction environment images until convergence; A real-time data acquisition module (4) collects an image of the power construction environment, wherein the image of the power construction environment includes power construction workers and safety vests; A target detection module (5) detects and outputs human targets and safety vest targets in the collected power construction environment image based on a human target detection model and a safety vest target detection model; A determination module (6) determines whether the power construction personnel have violated the regulations by not wearing safety vests based on the comparison between the number of detected human targets and the number of safety vest targets; The early warning module (7) issues an early warning when it is determined that the electric power construction personnel are operating in violation of regulations by not wearing a safety vest.
2. According to claim 1, a safety vest target detection system based on electric power construction identification environment is characterized in that: In the target detection model construction module (2), the person target detection model and the safety vest target detection model are both convolutional neural network models.
3. According to claim 1, a safety vest target detection system based on electric power construction recognition environment is characterized in that: In the target detection model training module (3), the specific steps of training the person target model include: The collected historical power construction environment images are divided into two parts, and one of them is used to mark the people in the image; The historical power construction environment images with people annotated are input into the constructed person target detection model for training; The loss function is calculated based on the detection results and annotation results, and backpropagation is performed to optimize the parameters until the model converges.
4. The safety vest target detection system based on the power construction recognition environment according to claim 3 is characterized by: In the target detection model training module (3), the specific steps of training the safety vest target detection model include: The collected historical power construction environment images are divided into two parts, and the other part is taken to mark the safety vests in the image; The historical power construction environment images annotated with safety vests are input into the constructed safety vest object detection model for training; The loss function is calculated based on the detection results and annotation results, and backpropagation is performed to optimize the parameters until the model converges.
5. The safety vest target detection system based on the power construction recognition environment according to claim 1 is characterized by: In the real-time data acquisition module (4), the power construction environment image is collected in multiple directions, and the collection is continuous and uninterrupted in the same direction.
6. A safety vest target detection system based on electric power construction identification environment according to claim 5, characterized in that: In the real-time data acquisition module (4), the power construction environment images acquired from multiple directions are spliced based on coordinates to improve the power construction environment images.