A method and system for detecting illegal roadside parking of freight vehicles based on image recognition

Abstract

The present invention relates to the technical field of vehicle illegal parking detection, and discloses a method and system for detecting roadside illegal parking of freight vehicles based on image recognition, including: obtaining the GPS position information of the detected vehicle, mapping it to the road network map, and calibrating the vehicle label information; constructing a detection model for illegal roadside parking of freight vehicles to obtain the detection result of illegal roadside parking of freight vehicles; if the detection result of illegal roadside parking of the freight vehicle is in the state of roadside illegal parking, obtaining the characteristic information of the freight vehicle, calculating the occurrence frequency of illegal parking of trucks on the road, and generating a high-risk area for illegal roadside parking of freight vehicles. The present invention can efficiently identify the roadside illegal parking behavior of freight vehicles in a complex traffic environment, and improve the management efficiency and operability of detecting illegal roadside parking of freight vehicles through automatic detection of roadside illegal parking behavior and accurate risk assessment of roadside illegal parking on the road.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle illegal parking detection, and particularly to a method and system for detecting roadside illegal parking of freight vehicles based on image recognition. Background Art

[0002] With the rapid development of urbanization and freight transportation, the illegal parking behavior of freight vehicles has become one of the persistent problems in urban management. Especially in the complex environments of districts, remote areas, and areas with missing road data, the traditional detection technologies that rely on complete background data are difficult to play a role. This not only leads to the existence of traffic management blind spots but also lays a hidden danger for the occurrence of serious traffic accidents.

[0003] Due to their large volume and complex parking requirements, freight vehicles often occupy more road resources when parking. Once a roadside illegal parking behavior occurs, it will seriously hinder the normal passage of other vehicles and even trigger serious traffic accidents such as multi-vehicle chain collisions. Most of the existing illegal parking detection technologies based on road video surveillance rely on complete electronic maps and road boundary information. However, when the regional information is insufficient or the data is incomplete, the model accuracy and reliability often drop significantly, and there are bottlenecks such as a high false positive rate and many missed detection problems. Moreover, most of the existing methods are for identifying illegal parking phenomena at specific points and cannot give the high-incidence areas of roadside illegal parking of freight vehicles macroscopically, so as to optimize the police force deployment and conduct targeted governance for high-risk areas of roadside illegal parking.

[0004] The existing technology for detecting illegal parking on the sidewalk based on object detection and semantic segmentation proposes an algorithm based on deep learning and semantic segmentation for detecting illegal parking on the sidewalk. However, the limitation of this method is that it can only identify whether a vehicle is parked on the sidewalk, and its applicable range is relatively limited. In addition, during the semantic segmentation process of the image, the image is only divided into three categories: road, sidewalk, and car, and the classification method is relatively simple, resulting in the need to further improve the segmentation accuracy and processing speed in complex urban road environments. In addition, there is no license plate recognition function, and subsequent penalties cannot be imposed on illegally parked vehicles.

[0005] The existing technology for detecting illegal parking of motor vehicles from a high perspective in a complex environment proposes a method for detecting illegal parking of motor vehicles from a high perspective, which proposes a method for detecting illegal parking of motor vehicles based on an improved YOLOv3-TINY backbone network and an attention mechanism. However, the limitations of this method are as follows: its application scenario is limited to high-perspective monitoring, ignoring the potential needs of low-perspective or dynamic monitoring; it has a strong dependence on fixed-mounted high-position monitoring devices, and the implementation conditions are harsh, which is not suitable for areas with limited resources or insufficient infrastructure; in addition, although the model structure is optimized, the detection accuracy in complex scenarios (such as dense traffic, severe occlusion, etc.) still needs to be verified. Summary of the Invention

[0006] In view of the existing problems mentioned above, the present invention is proposed. Therefore, the present invention provides a method for detecting roadside illegal parking of freight vehicles based on image recognition to solve the problems of how to efficiently and accurately identify the roadside illegal parking behavior of freight vehicles in a complex traffic environment, insufficient means for supervising roadside illegal parking of freight vehicles, low efficiency and low accuracy.

[0007] To solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a method for detecting roadside illegal parking of freight vehicles based on image recognition, including: obtaining the GPS point information of the detected vehicle, mapping it to the road network map, and calibrating the vehicle label information;

[0009] Constructing a detection model for illegal roadside parking of freight vehicles, extracting key frame images of roadside parking from the road network map containing parking points, and inputting them into the detection model for illegal roadside parking of freight vehicles for processing to obtain the detection result of illegal roadside parking of freight vehicles;

[0010] If the detection result of the illegal roadside parking of the freight vehicle is the roadside illegal parking state, according to the vehicle label in the key frame image of the roadside parking, obtaining the characteristic information of the freight vehicle, calculating the occurrence frequency of illegal parking of trucks on the road, sorting the occurrence frequency of road illegal parking from high to low, and generating a high-risk area for illegal roadside parking of freight vehicles.

[0011] As a preferred solution of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to the present invention, where: obtaining the GPS point information of the detected vehicle and mapping it to the road network map and calibrating the vehicle label information includes:

[0012] Obtaining the GPS point information of the detected vehicle through the vehicle's own GPS positioning module, where the GPS point information includes the longitude, latitude, timestamp, and license plate number of the vehicle's parking position;

[0013] Mapping the real-time position of the vehicle to the road network information in the traffic management system to determine the specific position of the vehicle.

[0014] As a preferred solution of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to the present invention, where: constructing a detection model for illegal roadside parking of freight vehicles includes:

[0015] Based on an improved MTP multi-task pre-training semantic segmentation algorithm, constructing a detection model for illegal roadside parking of freight vehicles, and identifying and segmenting the road layer and parking points through training the model;

[0016] By calculating the distance between the target parking point and the road center line and comparing it with a preset tolerance value, determining whether the vehicle parks within the roadside range;

[0017] If the distance between the parking spot and the center line of the road is less than the tolerance value and is located in the roadside prohibited parking area, it will be determined as illegal roadside parking.

[0018] As a preferred embodiment of the method for detecting illegal roadside parking of freight vehicles based on image recognition according to the present invention, the step of calculating the distance between the target parking point and the road centerline and comparing the distance with the preset tolerance value specifically includes:

[0019] Calculate the relative position of the vehicle point and the road centerline, assuming the vehicle's GPS point is , find the equation of the tangent line to the road centerline , the distance between the vehicle and the centerline of the road Expressed as:

[0020] ;

[0021] in, , The coefficients representing the equation of the road centerline, represents the intercept;

[0022] The deviation tolerance value is evaluated. If the calculated distance between the vehicle and the road centerline is Greater than the preset tolerance value , then it is determined that the parking spot does not belong to the roadside illegal parking range; if the distance between the vehicle and the center line of the road Less than or equal to the preset tolerance value , it is determined to be high-risk roadside illegal parking.

[0023] As a preferred embodiment of the method for detecting illegal roadside parking of freight vehicles based on image recognition according to the present invention, obtaining the detection result of illegal roadside parking of freight vehicles includes:

[0024] The freight vehicle roadside illegal parking detection result is divided into a complete roadside illegal parking state, a high-risk roadside illegal parking state, and a non-roadside illegal parking state, wherein the complete roadside illegal parking state and the high-risk roadside illegal parking state are determined as the roadside illegal parking state;

[0025] Among them, the comparison is performed from the roadside parking key frame image. If the freight vehicle stops at a distance from the road center line If the vehicle is parked within the road width, it will be considered as illegal roadside parking.

[0026] If the freight vehicle stops at a distance from the center line of the road If the road width falls within the tolerance value area, it will be judged as a high-risk roadside illegal parking state;

[0027] If the parking position is greater than the tolerance range, or the parking behavior does not violate the roadside regulations, it is determined as a non-roadside illegal parking state.

[0028] As a preferred solution of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to the present invention, wherein: generating a high-risk area for roadside illegal parking of freight vehicles includes:

[0029] According to the complete roadside illegal parking state, high-risk roadside illegal parking state and non-roadside illegal parking state, count the occurrence times of illegal parking of freight vehicles on the road, and calculate the occurrence frequency of illegal parking of freight vehicles.

[0030] Based on the occurrence frequency of illegal parking of freight vehicles, comprehensively evaluate the road, sort the roads from high to low according to the illegal parking risk, generate a list of high-risk areas for roadside illegal parking of freight vehicles based on the risk ranking result, and at the same time mark the risk rankings of each road in the three types of complete roadside illegal parking state, high-risk roadside illegal parking state and non-roadside illegal parking state, and generate a high-risk area distribution map with the road as the core.

[0031] As a preferred solution of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to the present invention, wherein: extract the illegal parking aggregation area in the image of illegal parking of freight vehicles on the roadside, input it into the road area recognition model, and recognize the road name and the affiliated area in the case of missing road background data.

[0032] Obtain the license plate number and the aggregation area of the illegally parked vehicle, upload the license plate number and area information to the cloud to determine the driver identity information of the illegally parked vehicle, and perform key risk control processing on the high-risk area of illegal parking.

[0033] In a second aspect, the present invention provides a system for detecting roadside illegal parking of freight vehicles based on image recognition, including:

[0034] An acquisition module, configured to obtain the GPS point information of the detected vehicle, map it to the road network map, and calibrate the vehicle label information.

[0035] A detection module, configured to construct a detection model for illegal parking of freight vehicles on the roadside, extract the key frame images of roadside parking from the road network map including the parking points, and input them into the detection model for illegal parking of freight vehicles on the roadside for processing to obtain the detection result of illegal parking of freight vehicles on the roadside.

[0036] An output module, configured to, when the detection result of illegal parking of freight vehicles on the roadside is the roadside illegal parking state, obtain the feature information of the freight vehicle according to the vehicle label in the key frame image of roadside parking, calculate the occurrence frequency of illegal parking of freight vehicles on the road, sort the occurrence frequency of road illegal parking from high to low, and generate a high-risk area for roadside illegal parking of freight vehicles.

[0037] In a third aspect, the present invention provides an electronic device, comprising:

[0038] a memory and a processor;

[0039] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for detecting roadside illegal parking of freight vehicles based on image recognition are implemented.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for detecting roadside illegal parking of freight vehicles based on image recognition.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention can efficiently identify the roadside illegal parking behavior of freight vehicles in a complex traffic environment. Through the automated detection of roadside illegal parking behavior and the accurate risk assessment of roadside illegal parking on roads, it optimizes the illegal parking management and police force deployment, improves the control efficiency and operability of detecting illegal roadside parking of freight vehicles, and has a high level of intelligence and application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 It is a schematic diagram of the overall process of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to an embodiment of the present invention;

[0044] Figure 2 It is a schematic diagram of the main working process of the method for detecting roadside illegal parking of freight vehicles based on image recognition according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention shall fall within the protection scope of the present invention.

[0046] Refer to Figure 1 - Figure 2, which is an embodiment of the present invention, provides a method for detecting illegal roadside parking of freight vehicles based on image recognition, including:

[0047] S100, obtaining the GPS point information of the detected vehicle, mapping it to the road network map, and calibrating the vehicle label information;

[0048] S200, constructing a detection model for illegal roadside parking of freight vehicles, extracting key frame images of roadside parking from the road network map containing parking points, and inputting them into the detection model for illegal roadside parking of freight vehicles for processing to obtain the detection result of illegal roadside parking of freight vehicles;

[0049] S300, if the detection result of illegal roadside parking of freight vehicles is the roadside illegal parking state, obtaining the characteristic information of the freight vehicle according to the vehicle label in the key frame image of roadside parking, calculating the occurrence frequency of illegal parking of freight vehicles on the road, sorting the occurrence frequency of road illegal parking from high to low, and generating a high-risk area for illegal roadside parking of freight vehicles.

[0050] It should be noted that the present invention proposes a comprehensive solution combining GPS point position, image recognition technology and deep learning algorithm, which can evaluate whether a parking point belongs to the roadside in the case of missing road background data or information, count and sort the occurrence frequency of illegal parking of freight vehicles on the road according to the evaluation result of roadside illegal parking, generate a high-risk area for roadside illegal parking, and conduct targeted control and management on the high-risk area, so as to realize the accurate detection of illegal roadside parking of freight vehicles and the dynamic intelligent management of high-risk areas, and can achieve automatic detection and accurate risk assessment, optimize illegal parking management and police deployment, and complete roadside illegal parking detection and monitoring.

[0051] Preferably, obtaining the GPS point information of the detected vehicle, mapping it to the road network map, and calibrating the vehicle label information includes:

[0052] Obtaining the GPS point information of the detected vehicle through the vehicle's own GPS positioning module, where the GPS point information includes the longitude, latitude, timestamp, and license plate number of the vehicle's parking position;

[0053] Mapping the real-time position of the vehicle to the road network information in the traffic management system to determine the specific position of the vehicle.

[0054] Specifically, in this embodiment, the roadside is defined as the space extending inward from the curb to the driving lane. This step can collect the vehicle's precise positioning data and time information in real time, providing the basic spatial and temporal dimension basis for subsequent illegal parking judgments, and ensuring the timeliness and data integrity of the detection request; mapping the vehicle GPS point information to the road network map, calibrating vehicle labels and other information, mainly rely on map matching technology and geographic information systems. Map matching is combined with the geographic information system through geometric matching, topological matching or probabilistic matching algorithms based on hidden Markov models to accurately align discrete GPS points to the road network, providing a basis for subsequent analysis and decision-making. The discrete GPS data is aligned with the road network topology through the geographic information system to achieve accurate matching and visual annotation of the vehicle position, providing geographic spatial correlation support for subsequent image capture and road boundary analysis, and capturing key frame images of the map containing parking points and uploading them to the cloud to optimize data transmission efficiency. The redundant image processing burden is reduced through key frame extraction, and at the same time, cloud storage and computing resources are used to achieve rapid centralized processing and analysis of large-scale image data.

[0055] Preferably, building a freight vehicle roadside illegal parking detection model includes:

[0056] Based on the improved MTP multi-task pre-trained semantic segmentation algorithm, a model for detecting illegal roadside parking of freight vehicles was constructed. The trained model was used to identify and segment the road layer and parking spots.

[0057] By calculating the distance between the target parking point and the center line and comparing it with the preset tolerance value, it is determined whether the vehicle is parked within the roadside range;

[0058] If the distance between the parking spot and the center line is less than the tolerance value and is located in the roadside prohibited parking area, it will be determined as illegal roadside parking.

[0059] Specifically, the freight vehicle roadside illegal parking detection model is based on the semantic segmentation of the MTP potential illegal parking point map, including:

[0060] Multi-task annotation preparation: Annotate the map image with the label set required for semantic segmentation, including: road areas (segment the road layer), parking points (marking points), road centerlines (if there is no centerline, generate centerline annotations through geometric transformation). Generate semantic segmentation labels using the SAMRS dataset or similar annotation tools.

[0061] Feature extraction and pyramid construction: Use a pre-trained model to extract multi-scale features of the image and construct a feature pyramid. Features at different resolution levels are input to the semantic segmentation decoder.

[0062] Semantic Segmentation Task Head: In the semantic segmentation decoder, each pixel is classified into different categories (roads, parking spots, background, etc.) based on the input features, and the task loss function is used to optimize the segmentation result.

[0063] Training and Optimization: Conduct multi-task pre-training on the training dataset, and jointly optimize by combining the loss functions of other tasks to enhance the model's segmentation ability.

[0064] Model Migration and Application: After pre-training, the model is migrated to a specific task, and the model is fine-tuned to adapt to specific road segmentation and parking spot detection tasks. In the output result, the road area is marked by the segmentation mask, the parking spots are identified through classification or region detection, and the center line can be further generated through post-processing (such as extracting the midline of the road segmentation boundary).

[0065] Preferably, calculating the distance between the target parking spot and the center line and comparing it with a preset tolerance value specifically includes:

[0066] Calculating the relative position of the vehicle position to the road center line. Let the GPS position of the vehicle be , find the tangent line equation of the road center line , the distance between the vehicle and the road center line is expressed as:

[0067] ;

[0068] where , represent the coefficients of the center line straight line equation, represents the intercept to quantify the deviation degree of the vehicle relative to the center line;

[0069] Conduct an evaluation of the deviation tolerance value. If the calculated distance between the vehicle and the road center line is greater than the preset tolerance value , it is determined that this parking position does not belong to the roadside illegal parking range; if the distance between the vehicle and the road center line is less than or equal to the preset tolerance value , it is determined as a high-risk roadside illegal parking.

[0070] In an alternative embodiment, the tolerance value is a preset dynamically adjustable parameter tolerance value used to evaluate the possibility of whether the vehicle is located within the roadside parking area. The calculation formula is as follows:

[0071] ;

[0072] where is the resolution of the map image, is the GPS positioning accuracy, They are the influencing factors of real-time traffic flow or historical illegal parking records.

[0073] It should be noted that in this embodiment, the roadside illegal parking detection model for freight vehicles is built based on the improved MTP multi-task pre-trained semantic segmentation technology. By using a large-scale remote sensing dataset in the pre-training stage and combining the cosine annealing scheduling of the learning rate and hierarchical decay to optimize the performance of the large model, it shows superior performance in specific semantic segmentation tasks. The improved MTP multi-task pre-trained semantic segmentation technology solves the differences between upstream pre-training and downstream fine-tuning tasks by introducing a staged multi-task pre-training method; by pre-training the basic model, it integrates semantic segmentation and object detection tasks in a framework. Its deep learning-based image recognition model automatically analyzes the illegal parking scenario, combines with the road boundary semantic segmentation technology, improves the accuracy and robustness of illegal parking detection in complex environments, and reduces the need for manual intervention.

[0074] Preferably, obtaining the roadside illegal parking detection result of freight vehicles includes:

[0075] The roadside illegal parking detection result of freight vehicles is divided into a complete roadside illegal parking state, a high-risk roadside illegal parking state, and a non-roadside illegal parking state. Among them, the complete roadside illegal parking state and the high-risk roadside illegal parking state are determined as the roadside illegal parking state;

[0076] Among them, by comparing in the key frame image of roadside parking, if the freight vehicle stops in the road area within the road width from the center line, it is determined as the complete roadside illegal parking state;

[0077] If the freight vehicle stops in the tolerance area from the road width to the center line, it is determined as the high-risk roadside illegal parking state;

[0078] If the parking position is greater than the tolerance range, or the parking behavior does not violate the roadside regulations, it is determined as the non-roadside illegal parking state.

[0079] It should be noted that according to the vehicle label in the key frame image of parking, characteristic information such as the license plate of the freight vehicle is determined, and the license plate recognition technology (such as OCR) is integrated to accurately extract the identity identification of the illegal vehicle, providing reliable data support for subsequent law enforcement evidence collection and illegal tracing, generating the high-risk area of roadside illegal parking for freight vehicles, constructing an illegal parking heat map and a risk distribution model based on spatio-temporal data analysis, identifying the high-incidence sections, and assisting the traffic management department to optimize the deployment of police forces and the configuration of monitoring equipment.

[0080] Preferably, generating the high-risk area of roadside illegal parking for freight vehicles includes:

[0081] Based on the complete roadside illegal parking status, high-risk roadside illegal parking status, and non-roadside illegal parking status, the number of illegal parking of freight vehicles on the road is counted, and the frequency of illegal parking of trucks is calculated;

[0082] A comprehensive assessment of roads is conducted based on the frequency of illegal parking of trucks. Roads are ranked from high to low according to the risk of illegal parking. The risk ranking results are used to generate a list of high-risk areas for roadside illegal parking of freight vehicles. At the same time, the risk ranking of each road in three types of conditions, namely, complete roadside illegal parking, high-risk roadside illegal parking, and non-roadside illegal parking, is marked to generate a high-risk area distribution map with roads as the core.

[0083] Preferably, the illegally parked areas are extracted from the images of freight vehicles illegally parked on the roadside and input into the road area recognition model to identify the road names and areas in the absence of road background data. Image semantic segmentation and text recognition (OCR) technology are used to restore road section attributes (such as road names and landmarks) when road information is missing, thereby enhancing the adaptability and reliability of the system in scenarios with incomplete data.

[0084] Obtain the license plate numbers and gathering areas of illegally parked vehicles, upload the license plate numbers and area information to the cloud to determine the identity information of the drivers of illegally parked vehicles, and conduct key risk management and control on the road section based on the high-risk area for illegal parking, realizing "data-analysis-disposal" closed-loop management, triggering targeted law enforcement through identity association and area association (such as automatically notifying law enforcement personnel or initiating parking restrictions), and dynamically adjusting the control strategy of high-risk areas to improve traffic management efficiency.

[0085] Optionally, when the detected roadside illegally parked vehicles are located in a high-incidence area, the area will be subject to key risk management and control measures. Specific management and control measures include but are not limited to:

[0086] Automatically notify traffic enforcement officers to conduct on-site inspections;

[0087] Carry out key traffic control in areas with high incidence of illegal parking, such as restricting parking during certain hours or increasing the deployment of monitoring equipment.

[0088] Optionally, in situations where road background information is missing in some areas and high-risk roadside illegal parking situations cannot be matched to specific road sections, this embodiment proposes a road area recognition model. By extracting geographic feature information from a map image, the specific road sections where illegal parking is occurring are determined. The road area recognition model is constructed to extract area information from the map image, including the following steps:

[0089] Image scanning and area positioning: Use image processing technology to first scan the map image, and then use edge detection algorithm to locate the road area and related label information.

[0090] Text area recognition and label extraction: The text area in the image is extracted through image segmentation methods, and OCR technology is used to identify text labels in the image, such as road names, landmarks, shops, etc.

[0091] Matching and location calculation: The text content extracted by OCR is matched with the coordinate data of the target area to generate structured data containing road names and area labels for further analysis.

[0092] Furthermore, the detailed principle process of image text recognition includes:

[0093] Image preprocessing: grayscale the map image and use Gaussian blur to remove noise, making the image smoother and improving the accuracy of text recognition;

[0094] OCR technology: performs text recognition and identifies information such as road names and geographic tags in images.

[0095] It should be noted that the present invention can efficiently identify illegal roadside parking behaviors of freight vehicles in complex traffic environments. Through automated detection of roadside illegal parking behaviors and accurate risk assessment of roadside illegal parking, it optimizes illegal parking management and police force deployment, and improves the control efficiency and operability of roadside illegal parking detection of freight vehicles. It has a high level of intelligence and application prospects, and is particularly suitable for roadside illegal parking detection and monitoring in intelligent traffic management systems.

[0096] The above is a schematic diagram of a method for detecting illegal roadside parking of freight vehicles based on image recognition in this embodiment. It should be noted that the technical solution of this system for detecting illegal roadside parking of freight vehicles based on image recognition is based on the same concept as the technical solution of the method for detecting illegal roadside parking of freight vehicles based on image recognition. For details not described in detail in the technical solution of the system for detecting illegal roadside parking of freight vehicles based on image recognition in this embodiment, please refer to the description of the technical solution of the method for detecting illegal roadside parking of freight vehicles based on image recognition.

[0097] In this embodiment, a freight vehicle roadside illegal parking detection system based on image recognition includes:

[0098] The acquisition module is used to obtain the GPS location information of the detected vehicle, map it to the road network map, and calibrate the vehicle label information;

[0099] A detection module is used to build a freight vehicle roadside illegal parking detection model. The module extracts roadside parking keyframe images from a road network map containing parking spots and inputs them into the freight vehicle roadside illegal parking detection model for processing to obtain freight vehicle roadside illegal parking detection results.

[0100] The output module is used to obtain the characteristic information of the freight vehicle based on the vehicle label in the roadside parking key frame image if the detection result of the freight vehicle's roadside illegal parking is a roadside illegal parking state, and calculate the frequency of illegal parking of trucks on the road, sort the frequency of illegal parking on the road from high to low, and generate a high-risk area for illegal parking of freight vehicles on the roadside.

[0101] This embodiment further provides an electronic device suitable for detecting illegally parked freight vehicles on the roadside based on image recognition, including:

[0102] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the method for detecting illegal roadside parking of freight vehicles based on image recognition as proposed in the above embodiment.

[0103] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for detecting illegal roadside parking of freight vehicles based on image recognition as proposed in the above embodiment.

[0104] The storage medium proposed in this embodiment and the method for detecting illegal parking of freight vehicles on the roadside based on image recognition proposed in the above embodiment belong to the same inventive concept. For technical details not described in detail in this embodiment, please refer to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0105] Through the above description of the embodiments, those skilled in the art will clearly understand that the present invention can be implemented using software and necessary general-purpose hardware. Of course, it can also be implemented using hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk, or optical disk, and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0106] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting illegal roadside parking of freight vehicles based on image recognition, characterized in that, Including: Obtain the GPS point information of the detected vehicle, map it to the road network map, and calibrate the vehicle label information; Construct a roadside illegal parking detection model for freight vehicles. Extract the key frame images of roadside parking from the road network map containing the parking points, and input them into the roadside illegal parking detection model for freight vehicles for processing to obtain the roadside illegal parking detection results of freight vehicles; If the roadside illegal parking detection result of the freight vehicle is in the roadside illegal parking state, obtain the characteristic information of the freight vehicle according to the vehicle label in the key frame image of the roadside parking, calculate the occurrence frequency of freight vehicle illegal parking on the road, sort the occurrence frequencies of road illegal parking from high to low, and generate a high-risk area for freight vehicle roadside illegal parking; Constructing a roadside illegal parking detection model for freight vehicles includes: Based on the improved MTP multi-task pre-training semantic segmentation algorithm, construct a roadside illegal parking detection model for freight vehicles, and identify and segment the road layer and parking points through training the model; By calculating the distance between the target parking point and the road center line and comparing it with a preset tolerance value, determine whether the vehicle parks within the roadside range; If the distance between the parking point and the road center line is less than the tolerance value and it is within the roadside no-parking area, it is determined to be in the roadside illegal parking state.

2. The method for detecting illegal roadside parking of freight vehicles based on image recognition according to claim 1, characterized in that Obtain the GPS point information of the detected vehicle, and map it to the road network map, and calibrate the vehicle label information including: Obtain the GPS point information of the detected vehicle through the vehicle's own GPS positioning module. The GPS point information includes the longitude, latitude, timestamp, and license plate number of the vehicle parking position; Map the real-time position of the vehicle to the road network information in the traffic management system to determine the specific position of the vehicle.

3. The method for detecting roadside illegal parking of freight vehicles based on image recognition according to claim 2, characterized in that, The calculation of the distance between the target parking point and the road center line and the comparison with the preset tolerance value specifically includes: Calculate the relative position of the vehicle point and the road centerline, assuming the vehicle's GPS point is , find the equation of the tangent line to the road centerline , the distance between the vehicle and the centerline of the road Expressed as: ； Among them, , represent the coefficients of the tangent line equation of the road center line, represents the intercept; Evaluate the deviation tolerance value. If the calculated distance between the vehicle and the center line of the road is greater than the preset tolerance value , it is determined that this parking position does not belong to the roadside illegal parking range; if the distance between the vehicle and the center line of the road is less than or equal to the preset tolerance value , it is determined to be in the state of roadside illegal parking.

4. The method for detecting roadside illegal parking of freight vehicles based on image recognition according to claim 3, characterized in that, Obtaining the roadside illegal parking detection results of freight vehicles includes: Divide the roadside illegal parking detection results of the freight vehicle into a complete roadside illegal parking state, a high-risk roadside illegal parking state, and a non-roadside illegal parking state. Among them, the complete roadside illegal parking state and the high-risk roadside illegal parking state are determined to be in the roadside illegal parking state; Among them, by comparing from the key-frame images of roadside parking, if a freight vehicle stops within the road area within the road width from the road center line, it is determined to be in a complete roadside illegal parking state; If a freight vehicle stops within the area from the road center line to the tolerance value of the road width, it is determined to be in a high-risk roadside illegal parking state; ​ If the parking position is greater than the tolerance range, or the parking behavior does not violate the roadside regulations, it is determined to be in the non-roadside illegal parking state.

5. The method for detecting illegal roadside parking of freight vehicles based on image recognition according to claim 4, characterized in that, Generating a high-risk area for freight vehicle roadside illegal parking includes: According to the complete roadside illegal parking state, the high-risk roadside illegal parking state, and the non-roadside illegal parking state, count the occurrence times of freight vehicle illegal parking on the road, and calculate the occurrence frequency of freight vehicle illegal parking; Based on the occurrence frequency of freight vehicle illegal parking, conduct a comprehensive evaluation of the road, sort the roads from high to low according to the illegal parking risk, generate a list of high-risk areas for freight vehicle roadside illegal parking based on the risk ranking results, and at the same time mark the risk rankings of each road in the three types of complete roadside illegal parking state, high-risk roadside illegal parking state, and non-roadside illegal parking state, and generate a high-risk area distribution map centered on the road.

6. The method for detecting roadside illegal parking of freight vehicles based on image recognition according to claim 5, characterized in that Extract illegal parking clusters from images of freight vehicles parked illegally on the roadside and input them into a road area recognition model to identify road names and areas in the absence of road background data. Obtain the license plate numbers and gathering areas of illegally parked vehicles, upload the license plate numbers and area information to the cloud to determine the identity information of the drivers of illegally parked vehicles, and conduct key risk management and control in high-risk areas for illegal parking.

7. A roadside illegal parking detection system for freight vehicles based on image recognition, which applies a roadside illegal parking detection method for freight vehicles based on image recognition according to any one of claims 1 to 6, characterized in that, include, The acquisition module is used to obtain the GPS location information of the detected vehicle, map it to the road network map, and calibrate the vehicle label information; a detection module for constructing a freight vehicle roadside illegal parking detection model, extracting roadside parking key frame images from the road network map containing parking spots, and inputting the images into the freight vehicle roadside illegal parking detection model for processing to obtain freight vehicle roadside illegal parking detection results; an output module configured to obtain characteristic information of the freight vehicle based on the vehicle label in the roadside parking keyframe image if the freight vehicle roadside illegal parking detection result indicates a roadside illegal parking state, calculate the frequency of illegal parking of freight vehicles on the road, sort the frequencies of illegal parking on the road from high to low, and generate a high-risk area for illegal parking of freight vehicles on the roadside; Building a freight vehicle roadside illegal parking detection model includes: Based on the improved MTP multi-task pre-trained semantic segmentation algorithm, a model for detecting illegal roadside parking of freight vehicles was constructed. The trained model was used to identify and segment the road layer and parking spots. By calculating the distance between the target parking point and the road centerline and comparing it with the preset tolerance value, it is determined whether the vehicle is parked within the roadside range; If the distance between the parking spot and the center line of the road is less than the tolerance value and is located in the roadside prohibited parking area, it will be determined as illegal roadside parking.

8. An electronic device, characterized in that, include: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of a method for detecting illegal roadside parking of freight vehicles based on image recognition as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that, It stores computer-executable instructions, which, when executed by a processor, implement the steps of a method for detecting illegal roadside parking of freight vehicles based on image recognition as described in any one of claims 1 to 6.

Images