Urban area treatment method based on visual model

Abstract

The invention relates to the technical field of urban area treatment, in particular to an urban area treatment method based on a visual model, which comprises the following steps: S1, acquiring previous urban road data, and analyzing and processing the data: acquiring the urban road data through a network technology, and extracting, concluding and summarizing the data to obtain related data; according to the treatment method designed by the invention, feature extraction can be performed on conditions on urban roads through a sight distance detection technology, risks on the roads can be accurately detected, subsequent generation of an accurate management and control scheme is facilitated, and meanwhile, pedestrians can be remotely controlled to assist management and control through supervision equipment, so that the timeliness of management and control is ensured, and the management and control efficiency is improved. Compared with the prior art, large loss caused by untimely management and control can be avoided, a detailed risk management and control method can be obtained through network search, an appropriate management and control scheme can be automatically generated, the perfection degree of the generated scheme is ensured, road traffic risks can be perfectly processed conveniently, and the loss caused by the road traffic risks is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of urban area governance, and in particular to an urban area governance method based on a visual model. Background Art

[0002] Urban regional governance can maintain social stability, improve public services, promote infrastructure construction and environmental protection, among which the governance of urban roads is the focus of urban regional governance.

[0003] For example, a digital urban traffic management method based on public opinion knowledge graph with authorization announcement number CN115050187B and authorization announcement date 20221101. In order to overcome the fact that the existing technology only controls urban traffic in a short period of time, and lacks long-term urban traffic planning and management. The present invention includes the following steps: S1: Obtain multi-dimensional historical traffic public opinion information of different cities and construct a traffic public opinion knowledge graph; S2: Divide the city into several control areas, collect traffic public opinion data of each control area in real time, and identify the processing category; S3: When the processing category is real-time control, determine the specific location and specific event of the anomaly according to the traffic public opinion data, associate and execute the corresponding traffic control operation; S4: When the processing category is long-term planning, determine the specific event according to the traffic public opinion data, match the planning processing plan and feedback. Real-time control and long-term planning suggestions for urban traffic are carried out based on the knowledge graph, making full use of the collected public opinion information to improve the level of urban traffic management.

[0004] Although the above-mentioned urban traffic management methods and those in the prior art can meet the needs of urban traffic management, the precision of urban traffic management is poor during the management process, and urban traffic problems cannot be accurately judged, resulting in the inability to timely control urban traffic problems, which will cause major accidents and losses. Therefore, it is urgent to design an urban area management method based on visual models to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide an urban area governance method based on a visual model to solve the above-mentioned deficiencies in the prior art.

[0006] In order to achieve the above object, the present invention provides the following technical solutions: A method for urban area governance based on a visual model comprises the following steps: S1. Obtaining previous urban road data and analyzing and processing the data: Obtaining urban road data through network technology, extracting, summarizing and processing the data to obtain relevant data; S2. Establish models through programming technology and divide model functions: According to relevant data, select appropriate algorithms, establish supervision models, and divide the functions of the supervision models; S3. Based on the analysis of urban road data, divide the area and arrange the supervision equipment: formulate a road planning scheme based on urban road data, and then arrange the supervision equipment on the road according to the planning scheme; S4. Supervision equipment monitors the road in real time and collects data: After the supervision equipment is deployed, the supervision equipment will monitor the road conditions and collect road information; S5. The model processes the data and confirms the road risk problem: the supervision model receives the information collected by the supervision equipment, and the supervision module analyzes, extracts and processes the collected road information to confirm the road risk; S6. Rapidly formulate control plans based on dangerous situations and conduct control: Based on the confirmed road risks, the supervision module automatically generates a treatment plan and controls the road according to the treatment plan.

[0007] Specifically, in this embodiment, the following steps are included: S1. Obtaining previous urban road data and analyzing and processing the data: obtaining urban road data through network technology, where urban road data refers to urban road traffic data and urban road environment data, and relevant data refers to risk data in urban road traffic data and urban road environment data, and extracting, summarizing and processing the data to obtain relevant data; S2. Establish a model through programming technology and divide the model functions: According to relevant data, select appropriate algorithms to establish a supervision model. The supervision model is presented in the form of software and the supervision model is divided into image optimization module, image extraction module, problem judgment module, text link module, and engine search module; The image optimization module is established based on image enhancement technology and is used to optimize the image introduced into the supervision module; It should be noted that image enhancement technology mainly uses three types: color transformation, histogram equalization and homomorphic filtering. The color transformation type includes brightness adjustment, contrast adjustment, saturation adjustment, and hue adjustment; histogram equalization is used to improve the contrast of the image, especially when the contrast of the image is low; homomorphic filtering is used to eliminate the impact of uneven illumination and enhance image details.

[0008] The image extraction module is established based on the vehicle detection algorithm of FasterR-CNN and the human posture algorithm based on OpenPose, and is used to extract vehicle and human posture features from the image optimized by the image optimization module; It should be noted that the vehicle detection algorithm based on FasterR-CNN is a deep learning method. The core idea of ​​FasterR-CNN is to introduce the region proposal network (RPN) to generate candidate regions and combine it with FastR-CNN for target classification and bounding box regression. The specific steps are as follows: Feature extraction: The input image is passed through a convolutional neural network (such as VGG16, ResNet, etc.) to extract image features; ‌Candidate Region Generation‌: Use RPN to generate a series of candidate target regions based on image features. RPN extracts multiple anchor boxes of different scales on the feature map through a sliding window, and performs classification and regression prediction on each anchor box; ‌Target classification and localization‌: For each candidate region, RoIPooling is used to convert it into a fixed-size feature map, and the convolutional layer is used to perform target classification and position regression‌2; The human posture algorithm uses OpenPose, which is an open source library for human key point detection and full body posture estimation. It uses a deep learning model for real-time human key point detection. The core principles of the algorithm include the following steps: ‌Feature map generation‌: Input the original image, initialize and fine-tune it through the first 10 layers of ‌VGG-19, and generate a set of feature maps; ‌Keypoint Detection‌: The network generates the orientation (limbs) of pixels in the skeleton, i.e. pixel association fields (‌PAFs); Confidence calculation: At each stage, the predictions from the previous stage are concatenated with the original image features to produce refined predictions. ‌Multi-person parsing‌: Use PAFs for multi-person parsing, transform the multi-person detection problem into a bipartite graph matching problem, and use the ‌Hungarian algorithm to obtain the optimal matching of connected key points‌.

[0009] The problem judgment module is established based on relevant data and is used to judge and analyze the features extracted by the image extraction module. The engine search module is established based on a keyword search algorithm and searches for relevant information according to the results of the judgment and analysis of the problem judgment module. It should be noted that the keyword search algorithm uses the BM25 algorithm. The basic principle of the BM25 algorithm is to evaluate the relevance of the document and the query by calculating the frequency and position information of the keywords in the document. The BM25 algorithm considers the following factors: ‌Term Frequency (TF)‌: the frequency of a keyword appearing in a document; ‌Inverse Document Frequency (IDF): The frequency of occurrence of a keyword in all documents, used to adjust the weight of common words; ‌Document length‌: The total number of words in the document, which affects the weight of keywords.

[0010] The text link module is established based on the link algorithm and the report automatic generation algorithm, and is used to generate text for the relevant information searched by the engine search module, and link the generated text with the relevant data; It should be noted that the image-text linking algorithm uses the CLIP (Contrastive Language–Image Pre-training) model, which is based on contrastive learning and uses a large number of image and text pairs for training. The model learns how to link images and text by comparing positive samples (matching images and text) and negative samples (unmatched images and text). During the training process, CLIP will learn how to identify objects in images based on text descriptions; The report automatic generation algorithm uses a deep learning algorithm. The steps for automatically generating a report include: ‌Data preprocessing‌: including data cleaning, integration and data standardization to ensure data accuracy and consistency.

[0011] ‌Feature Extraction‌: Extract key information from raw data, such as values, text, images, etc.

[0012] ‌Model training and optimization‌: Use machine learning algorithms to train the extracted features, generate models for report generation, and continuously adjust model parameters to optimize model performance.

[0013] Report generation and output: Generate structured, high-quality reports based on the trained model and output them in the form of text, charts, PDF, etc.

[0014] S3. Based on the analysis of urban road data, divide the area to arrange supervision equipment: formulate a road planning scheme based on urban road data. The road planning scheme is the location and quantity of the supervision equipment. The supervision equipment is then arranged on the road according to the planning scheme. The supervision equipment uses AI cameras and broadcast speakers; S4. Supervision equipment monitors roads in real time and collects data: After the supervision equipment is deployed, it will monitor the road conditions and collect road information. The road information is all the data during the operation of urban roads; S5. The model processes the data and confirms the road risk problem: the supervision model receives the information collected by the supervision equipment, the supervision module analyzes and extracts the collected road information, confirms the road risk, and the supervision equipment transmits the data to the main control server equipped with the supervision module through optical cable, network communication, and satellite communication; S6. Rapidly formulate control plans based on dangerous situations for control: Based on the confirmed road risks, the supervision module automatically generates a treatment plan, and controls the road based on the treatment plan. The treatment plan is divided into remote control of road pedestrian auxiliary control and control of traffic police to location control.

[0015] In the above technical solution, the present invention provides a method for urban area governance based on a visual model, which has the following beneficial effects: (1) The urban area governance method designed in the present invention can extract features of the conditions on urban roads through line-of-sight detection technology, accurately detect risks on the roads, and facilitate the subsequent generation of accurate management and control plans. At the same time, pedestrian auxiliary management and control can be remotely controlled through supervision equipment, ensuring the timeliness of management and control and avoiding large losses caused by untimely management and control.

[0016] (2) The urban area governance method designed by the present invention can obtain detailed risk control methods through network search, and can also automatically generate appropriate control plans, thereby ensuring the perfection of the generated plans, facilitating the perfect handling of road traffic risks, and reducing the losses caused by road traffic risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0018] Figure 1 A schematic diagram of a method flow chart provided for an embodiment of a method for urban area governance based on a visual model of the present invention.

[0019] Figure 2 A schematic diagram of the composition of a supervision model provided for an embodiment of an urban area governance method based on a visual model of the present invention. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0021] like Figure 1-2 As shown, an urban area governance method based on a visual model provided by an embodiment of the present invention includes the following steps: S1. Obtaining previous urban road data and analyzing and processing the data: Obtaining urban road data through network technology, extracting, summarizing and processing the data to obtain relevant data; S2. Establish models through programming technology and divide model functions: According to relevant data, select appropriate algorithms, establish supervision models, and divide the functions of the supervision models; S3. Based on the analysis of urban road data, divide the area and arrange the supervision equipment: formulate a road planning scheme based on urban road data, and then arrange the supervision equipment on the road according to the planning scheme; S4. Supervision equipment monitors the road in real time and collects data: After the supervision equipment is deployed, the supervision equipment will monitor the road conditions and collect road information; S5. The model processes the data and confirms the road risk problem: the supervision model receives the information collected by the supervision equipment, and the supervision module analyzes, extracts and processes the collected road information to confirm the road risk; S6. Rapidly formulate control plans based on dangerous situations and conduct control: Based on the confirmed road risks, the supervision module automatically generates a treatment plan and controls the road according to the treatment plan.

[0022] Specifically, in this embodiment, the following steps are included: S1. Obtaining previous urban road data and analyzing and processing the data: obtaining urban road data through network technology, where urban road data refers to urban road traffic data and urban road environment data, and relevant data refers to risk data in urban road traffic data and urban road environment data, and extracting, summarizing and processing the data to obtain relevant data; S2. Establish a model through programming technology and divide the model functions: According to relevant data, select appropriate algorithms to establish a supervision model. The supervision model is presented in the form of software and the supervision model is divided into image optimization module, image extraction module, problem judgment module, text link module, and engine search module; The image optimization module is built based on image enhancement technology and is used to optimize the images introduced into the supervision module; It should be noted that image enhancement technology mainly uses three types: color transformation, histogram equalization and homomorphic filtering. The color transformation type includes brightness adjustment, contrast adjustment, saturation adjustment, and hue adjustment; histogram equalization is used to improve the contrast of the image, especially when the contrast of the image is low; homomorphic filtering is used to eliminate the impact of uneven illumination and enhance image details.

[0023] The image extraction module is built based on the vehicle detection algorithm of FasterR-CNN and the human posture algorithm based on OpenPose, and is used to extract vehicle and human posture features from the images optimized by the image optimization module; It should be noted that the vehicle detection algorithm based on FasterR-CNN is a deep learning method. The core idea of ​​FasterR-CNN is to introduce the region proposal network (RPN) to generate candidate regions and combine it with FastR-CNN for target classification and bounding box regression. The specific steps are as follows: Feature extraction: The input image is passed through a convolutional neural network (such as VGG16, ResNet, etc.) to extract image features; ‌Candidate Region Generation‌: Use RPN to generate a series of candidate target regions based on image features. RPN extracts multiple anchor boxes of different scales on the feature map through a sliding window, and performs classification and regression prediction on each anchor box; ‌Target classification and localization‌: For each candidate region, RoIPooling is used to convert it into a fixed-size feature map, and the convolutional layer is used to perform target classification and position regression‌2; The human posture algorithm uses OpenPose, which is an open source library for human key point detection and full body posture estimation. It uses a deep learning model for real-time human key point detection. The core principles of the algorithm include the following steps: ‌Feature map generation‌: Input the original image, initialize and fine-tune it through the first 10 layers of ‌VGG-19, and generate a set of feature maps; ‌Keypoint Detection‌: The network generates the orientation (limbs) of pixels in the skeleton, i.e. pixel association fields (‌PAFs); Confidence calculation: At each stage, the predictions from the previous stage are concatenated with the original image features to produce refined predictions. ‌Multi-person parsing‌: Use PAFs for multi-person parsing, transform the multi-person detection problem into a bipartite graph matching problem, and use the ‌Hungarian algorithm to obtain the optimal matching of connected key points‌.

[0024] The problem judgment module is established based on relevant data and is used to judge and analyze the features extracted by the image extraction module. The engine search module is established based on a keyword search algorithm and searches for relevant information according to the results of the judgment and analysis of the problem judgment module. It should be noted that the keyword search algorithm uses the BM25 algorithm. The basic principle of the BM25 algorithm is to evaluate the relevance of the document and the query by calculating the frequency and position information of the keywords in the document. The BM25 algorithm considers the following factors: ‌Term Frequency (TF)‌: the frequency of a keyword appearing in a document; ‌Inverse Document Frequency (IDF): The frequency of occurrence of a keyword in all documents, used to adjust the weight of common words; ‌Document length‌: The total number of words in the document, which affects the weight of keywords.

[0025] The text link module is established based on the link algorithm and the report automatic generation algorithm, and is used to generate text for the relevant information searched by the engine search module, and link the produced text with the relevant data; It should be noted that the image-text linking algorithm uses the CLIP (Contrastive Language–Image Pre-training) model, which is based on contrastive learning and uses a large number of image and text pairs for training. The model learns how to link images and text by comparing positive samples (matching images and text) and negative samples (unmatched images and text). During the training process, CLIP will learn how to identify objects in images based on text descriptions; The report automatic generation algorithm uses a deep learning algorithm. The steps for automatically generating a report include: ‌Data preprocessing‌: including data cleaning, integration and data standardization to ensure data accuracy and consistency.

[0026] ‌Feature Extraction‌: Extract key information from raw data, such as values, text, images, etc.

[0027] ‌Model training and optimization‌: Use machine learning algorithms to train the extracted features, generate models for report generation, and continuously adjust model parameters to optimize model performance.

[0028] Report generation and output: Generate structured, high-quality reports based on the trained model and output them in the form of text, charts, PDF, etc.

[0029] S3. Based on the analysis of urban road data, divide the area to arrange supervision equipment: formulate a road planning scheme based on urban road data. The road planning scheme is the location and quantity of the supervision equipment. The supervision equipment is then arranged on the road according to the planning scheme. The supervision equipment uses AI cameras and broadcast speakers; S4. Supervision equipment monitors roads in real time and collects data: After the supervision equipment is deployed, it will monitor the road conditions and collect road information. The road information is all the data during the operation of urban roads; S5. The model processes the data and confirms the road risk problem: the supervision model receives the information collected by the supervision equipment, the supervision module analyzes and extracts the collected road information, confirms the road risk, and the supervision equipment transmits the data to the main control server equipped with the supervision module through optical cable, network communication, and satellite communication; S6. Rapidly formulate control plans based on dangerous situations for control: Based on the confirmed road risks, the supervision module automatically generates a treatment plan, and controls the road based on the treatment plan. The treatment plan is divided into remote control of road pedestrian auxiliary control and control of traffic police to location control.

[0030] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for urban area governance based on visual model, characterized in that: The following steps are involved: S1. Obtaining previous urban road data and analyzing and processing the data: Obtaining urban road data through network technology, extracting, summarizing and processing the data to obtain relevant data; S2. Establish models through programming technology and divide model functions: According to relevant data, select appropriate algorithms, establish supervision models, and divide the functions of the supervision models; S3. Based on the analysis of urban road data, divide the area and arrange the supervision equipment: formulate a road planning scheme based on urban road data, and then arrange the supervision equipment on the road according to the planning scheme; S4. Supervision equipment monitors the road in real time and collects data: After the supervision equipment is deployed, the supervision equipment will monitor the road conditions and collect road information; S5. The model processes the data and confirms the road risk problem: the supervision model receives the information collected by the supervision equipment, and the supervision module analyzes, extracts and processes the collected road information to confirm the road risk; S6. Rapidly formulate control plans based on dangerous situations and conduct control: Based on the confirmed road risks, the supervision module automatically generates a treatment plan and controls the road according to the treatment plan.

2. According to the method of urban area governance based on visual model in claim 1, it is characterized in that: In the step S1, the urban road data refers to urban road traffic data and urban road environment data, and the relevant data refers to risk data in the urban road traffic data and urban road environment data.

3. The urban area governance method based on visual model according to claim 1 is characterized in that: In the step S2, the supervision model is divided into an image optimization module, an image extraction module, a problem judgment module, a text link module, and an engine search module, and the supervision model is presented in the form of software.

4. The urban area governance method based on visual model according to claim 3 is characterized in that: The image optimization module is established based on image enhancement technology, and is used to optimize the image introduced into the supervision module. The image extraction module is established based on the vehicle detection algorithm of FasterR-CNN and the human posture algorithm based on OpenPose, and is used to extract vehicle and human posture features from the image optimized by the image optimization module.

5. The urban area governance method based on visual model according to claim 3 is characterized in that: The problem judgment module is established based on relevant data and is used to judge and analyze the features extracted by the image extraction module. The engine search module is established based on a keyword search algorithm and searches for relevant information according to the results of the judgment and analysis of the problem judgment module.

6. The urban area governance method based on visual model according to claim 1 is characterized in that: The text link module is established based on the link algorithm and the report automatic generation algorithm, and is used to generate text for the relevant information searched by the engine search module, and link the generated text with the relevant data.

7. The urban area governance method based on visual model according to claim 1 is characterized in that: In step S3, the road planning plan is the location and quantity of the monitoring equipment, and the monitoring equipment uses AI cameras and broadcast speakers.

8. The urban area governance method based on visual model according to claim 1 is characterized in that: In step S4, the road information is all data during the operation of the urban road.

9. The urban area governance method based on visual model according to claim 1 is characterized in that: In step S5, the monitoring device transmits the data to the main control server equipped with the monitoring module via optical cable, network communication or satellite communication.

10. The urban area governance method based on visual model according to claim 1 is characterized in that: In step S6, the processing scheme for road control is divided into remote control of road pedestrian auxiliary control and regulation of traffic police to location control.

Images