Method and System for Detecting Road Construction Status Based on Dynamic Data Monitoring

Abstract

The present invention discloses a method and system for detecting road construction status based on dynamic data monitoring. The method includes: acquiring road images collected by road traffic vehicles; performing target recognition on the road images and segmenting the road area, construction area, and construction element features; tracking the operation of construction vehicles to determine the movement of the construction vehicles; determining the road construction status; when the road construction status belongs to the construction state, combining the enclosure features, construction vehicle features, construction personnel features, and traffic cone features in the target recognition result to form a multi-channel scene feature vector and performing feature fusion; calculating the feature influence factor; calculating the environmental weight factor; and determining the construction type based on the feature influence factor and the environmental weight factor in combination with historical information. By implementing the method of the present invention, the problems existing in the prior art can be solved, and the ability of urban governance and traffic warning can be improved.

Description

Technical Field

[0001] The present invention relates to a method for detecting the state of road construction, and more specifically to a method and system for detecting the state of road construction based on dynamic data monitoring. Background Art

[0002] At present, the methods for detecting road construction behaviors mainly adopt object detection models or judgment methods based on multi-element detection. However, for object detection models, the accuracy of construction scene detection depends on the comprehensiveness of the object detection model and the labeled data, which makes the detection effect vulnerable to the influence of data quality and model recognition rate. Secondly, in specific scenarios, a small number of devices can be used for efficient detection, but this method lacks generalizability and scalability and is only applicable in cases with consistent scenarios. In addition, current mainstream methods often rely on single-object detection, which is simple to operate and easy to expand, but cannot comprehensively reflect the overall state of the construction scene. The judgment method based on multi-element detection can make up for the deficiencies of single features, but it is still a simple superposition of multiple models and fails to effectively integrate the mutual influences among various elements.

[0003] Therefore, it is necessary to design a new method to achieve the fusion of multiple elements, realize the positioning and judgment of the construction scene, conduct full-process tracking and monitoring, generate auxiliary decision-making information, be applicable to different urban road conditions, quickly connect and save costs, and improve the ability of urban governance and traffic warning. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects of the prior art and provide a method and system for detecting the state of road construction based on dynamic data monitoring.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: A method for detecting the state of road construction based on dynamic data monitoring, including:

[0006] Obtain road images collected by road traffic vehicles;

[0007] Perform object recognition on the road images and segment the road area, construction area, and construction element features to obtain an object recognition result;

[0008] Adopt a multi-object tracking method to track the operation of the construction vehicle to determine the movement of the construction vehicle;

[0009] Combine the object recognition result and the movement of the construction vehicle to determine the road construction state according to a set decision tree;

[0010] Fuse the construction element features according to the road construction status, calculate the characteristic influence factor and environmental weight factor corresponding to the construction element features based on the fusion result, and make a judgment by integrating the characteristic influence factor, environmental weight factor and historical information to determine the construction type.

[0011] Its further technical solution is:

[0012] The step of fusing the construction element features according to the road construction status, calculating the characteristic influence factor and environmental weight factor corresponding to the construction element features based on the fusion result, and making a judgment by integrating the characteristic influence factor, environmental weight factor and historical information to determine the construction type includes:

[0013] When the road construction status belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector, and perform the fusion of feature weights to obtain the fusion result; wherein, the construction element features include fence features, construction vehicle features, construction personnel features, and traffic cone features.

[0014] Calculate the characteristic influence factor corresponding to the construction element features according to the fusion result;

[0015] Calculate the environmental weight factor corresponding to the construction element features;

[0016] Make a comprehensive judgment based on the characteristic influence factor and environmental weight factor, combined with historical information, to obtain the construction type.

[0017] Its further technical solution is: The decision tree includes eight different construction states and non-construction states. Among them, the construction states include: having a fence; having no fence, having a construction vehicle, the construction vehicle is stationary, having construction personnel; having no fence, having a construction vehicle, the construction vehicle is stationary, having no construction personnel, having traffic cones; having no fence, having no construction vehicle, having construction personnel, having traffic cones; the non-construction states include: having no fence, having a construction vehicle, the construction vehicle is stationary, having no construction personnel, having no traffic cones; having no fence, having a construction vehicle, the construction vehicle is moving; having no fence, having no construction vehicle, having construction personnel, having no traffic cones; having no fence, having no construction vehicle, having no construction personnel.

[0018] Its further technical solution is: The step of combining the construction element features to form a multi-channel scene feature vector and performing the fusion of feature weights to obtain the fusion result when the road construction status belongs to the construction state includes:

[0019] When the road construction status belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector; where the weight of each channel is represented by the segmentation result corresponding to the feature.

[0020] Perform the fusion of feature weights on the multi-channel scene feature vector to obtain the fusion result.

[0021] Its further technical solution is: the multi-channel scene feature vectors are subjected to fusion of feature weights to obtain a fusion result, including:

[0022] The multi-channel scene feature vectors are subjected to fusion of feature weights by using at least one of Gaussian fusion, regional mean feature fusion, regional clustering mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion to obtain a fusion result.

[0023] Its further technical solution is: calculating the feature influence factor corresponding to the construction element feature according to the fusion result, including:

[0024] Using to calculate the feature influence factor corresponding to the construction element feature, where refers to the enclosure feature, construction vehicle feature, construction personnel feature, and traffic cone feature; refers to the weight coefficient; is the fusion result; is the feature influence factor corresponding to the construction element feature.

[0025] Its further technical solution is: calculating the environmental weight factor corresponding to the construction element feature, including:

[0026] Calculating the relevant parameters of the construction element feature;

[0027] Using prior parameters to perform pre-normalization processing on the relevant parameters of the construction element feature to obtain a processing result;

[0028] Performing Softmax processing on the processing result to obtain the environmental weight factor.

[0029] Its further technical solution is: making a comprehensive judgment based on the feature influence factor and the environmental weight factor in combination with historical information to obtain the construction type, including:

[0030] Applying the sigmoid function to activate the feature influence factor corresponding to the construction element feature to obtain an activation result;

[0031] Determining the construction type according to the activation result, the environmental weight factor, and historical information.

[0032] Its further technical solution is: determining the construction type according to the activation result, the environmental weight factor, and historical information, including:

[0033] Using to determine the construction type, where is the construction type; refers to the importance weight of current and historical data; , , , are respectively the environmental weight factors corresponding to the construction element features; , , , are respectively the feature influence factors corresponding to the construction element features; and are 0.5, refers to the influence degree of historical data; is the activation result.

[0034] The present invention also provides a road construction state detection device based on dynamic data monitoring, including:

[0035] An image acquisition unit, configured to acquire road images collected by road traffic vehicles;

[0036] An identification and segmentation unit, configured to perform target identification and segment the road area, construction area, and construction element features according to the road images to obtain a target identification result;

[0037] A tracking unit, configured to perform running tracking on the construction vehicle by using a multi-target tracking method to determine the movement of the construction vehicle;

[0038] A state determination unit, configured to determine the road construction state according to the set decision tree by combining the target identification result and the movement of the construction vehicle;

[0039] A fusion unit, configured to, when the road construction state belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector and perform fusion of feature weights to obtain a fusion result;

[0040] A feature influence factor calculation unit, configured to calculate the feature influence factor corresponding to the construction element feature according to the fusion result;

[0041] An environmental weight factor calculation unit, configured to calculate the environmental weight factor corresponding to the construction element feature;

[0042] A comprehensive judgment unit, configured to perform comprehensive judgment by combining the historical information according to the feature influence factor and the environmental weight factor to obtain the construction type.

[0043] The beneficial effects of the present invention compared with the prior art are as follows: The present invention obtains road images through cameras on urban public transport vehicles, performs target recognition and segmentation, extracts construction-related features, then uses multi-target tracking technology to monitor the movement of construction vehicles, and combines the target recognition results to determine the construction status using a decision tree. If it is in the construction state, various features are combined to form a multi-channel scene feature vector, and feature weight fusion is performed. Then, according to the fusion result, the feature influence factor and the environmental weight factor are calculated, and combined with historical information for comprehensive judgment, and finally the construction type is obtained. This process can adapt to different urban road conditions, quickly access, save costs, and improve urban governance and traffic warning capabilities.

[0044] The following further describes the present invention in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are 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.

[0046] Figure 1 It is a flowchart of the road construction status detection method based on dynamic data monitoring provided by the embodiment of the present invention;

[0047] Figure 2 It is a schematic diagram of the decision tree provided by the embodiment of the present invention;

[0048] Figure 3 It is a schematic diagram of target segmentation provided by the embodiment of the present invention;

[0049] Figure 4 It is a schematic diagram of the network output provided by the embodiment of the present invention;

[0050] Figure 5 It is a schematic diagram of Gaussian fusion provided by the embodiment of the present invention;

[0051] Figure 6 It is a schematic diagram of regional mean feature fusion provided by the embodiment of the present invention;

[0052] Figure 7 It is a schematic diagram of regional clustering mode feature fusion provided by the embodiment of the present invention;

[0053] Figure 8 It is a schematic diagram of regional median feature fusion provided by the embodiment of the present invention;

[0054] Figure 9 It is a schematic diagram of double-layer dynamic feature activation fusion provided by the embodiment of the present invention;

[0055] Figure 10 A schematic flowchart of a road construction status detection method based on dynamic data monitoring provided by another embodiment of the present invention;

[0056] Figure 11 A schematic block diagram of a road construction status detection device based on dynamic data monitoring provided by an embodiment of the present invention;

[0057] Figure 12 A schematic block diagram of a road construction status detection device based on dynamic data monitoring provided by another embodiment of the present invention;

[0058] Figure 13 A schematic block diagram of a computer device provided by an embodiment of the present invention. Detailed implementation manners

[0059] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0061] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in this specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0062] It should be further understood that the term " / and" as used in this specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0063] Please refer to Figure 1 , Figure 1Schematic flowchart of the road construction status detection method based on dynamic data monitoring provided by an embodiment of the present invention. The road construction status detection method based on dynamic data monitoring is applied to a server, which interacts with cameras on public transportation vehicles to obtain road images using the cameras on urban public transportation vehicles, realizing target recognition and region segmentation; performing multi-target tracking on construction vehicles to master their dynamic movement conditions; determining the road construction status through a decision tree according to the target recognition results and vehicle movement conditions; if the status is construction, combining multi-channel scene feature vectors for feature weight fusion; calculating feature influence factors and environmental weight factors for comprehensive judgment; recording the target recognition results to form historical information for subsequent analysis; applying various fusion methods to improve the feature weight fusion effect and ensure accuracy; achieving rapid judgment of construction types and decision support through the combination of real-time data analysis and historical information, applicable to diverse urban road conditions.

[0064] Figure 1 is a schematic flowchart of the road construction status detection method based on dynamic data monitoring provided by an embodiment of the present invention. As Figure 1 shown, the method includes the following steps S110 to S180.

[0065] S110. Obtain road images collected by road traffic vehicles.

[0066] In this embodiment, the road images are dynamic images of the road collected by road traffic vehicles.

[0067] Using cameras installed on public transportation means (such as buses and taxis), collect and analyze the conditions of the road surface ahead. The device needs to operate in an environment with an open view to ensure clear images are collected, and adjust the focal length and GPS sensor to confirm the accuracy of the position information. At fixed time intervals, the cameras on urban public transportation vehicles will continuously capture images, capturing n images to form road images.

[0068] S120. Perform target recognition on the road images and segment the road area, construction area, and construction element features to obtain the target recognition results.

[0069] In this embodiment, the target recognition results include the image areas corresponding to the road area, construction area, and construction element features, where the construction element features include multiple types of features such as fences, construction vehicles, construction personnel, and traffic cones.

[0070] Specifically, use deep learning technology to perform target segmentation on these images to identify six types of targets: road area, construction area, fence, construction vehicle, construction personnel, and traffic cone. The segmented and recognized images are as Figure 3 shown.

[0071] Among them, road area characteristics refer to lanes and sidewalks.

[0072] Construction Area Features: Identify where construction is taking place.

[0073] Fence features: used to isolate the construction area, protect the safety of passers-by and prevent interference.

[0074] Construction Vehicles Features: Vehicles used on construction sites, such as excavators and transporters.

[0075] Construction crew characteristics: Workers and managers involved in construction work to ensure the project progresses.

[0076] Traffic cone feature: used to guide and warn pedestrians and vehicles, usually set up near construction sites to draw attention.

[0077] Specifically, we can use Calculate the corresponding features, It is the output of the neural network corresponding to the deep learning technology, reflecting the superposition of multiple features. Under the influence of the convolution activation map, the mapping relationship between the position of the target and its features is relatively stable and usually shows similarity. Therefore, the target representation area is not only suitable for target detection and segmentation, but also can provide support for subsequent fine type judgment.

[0078] By comparing images collected by the same bus at different time points (such as 9 a.m. and 6 p.m.), it is possible to infer the cycle and status of road construction, assess the size of the construction area, the presence of fences, and the number of personnel and equipment on site, thereby providing strong data support for traffic management and construction planning.

[0079] S130: Track the construction vehicle using a multi-target tracking method to determine the movement of the construction vehicle.

[0080] In this embodiment, the movement of the construction vehicle includes two situations: the construction vehicle is moving or stationary.

[0081] Specifically, a motion reference system is established to more accurately capture and analyze the motion state of the construction vehicle; the motion reference system is set relative to the motion state of the observer, which means that the system will track the position of the surrounding construction vehicles based on the motion trajectory of the bus.

[0082] Obtain the real-time location information of the bus through sensors (such as GPS and accelerometers), thereby setting a dynamic reference system so that the system can update the location in real time; use multi-object tracking algorithms (such as Kalman filter, SORT, DeepSORT, etc.) to identify and track multiple construction vehicles; from the continuous image stream captured by the camera, identify the construction vehicles and assign a unique ID to each vehicle to achieve continuous tracking.

[0083] By analyzing the trajectories of the construction vehicles, determine their status as "running" or "stationary"; the "running" status indicates that the construction vehicle is moving, possibly during construction or changing positions; the "stationary" status indicates that the vehicle is in a stopped state.

[0084] Adopting the multi-object tracking method in the moving reference system can not only accurately monitor the running status of construction vehicles, but also provide a scientific basis for traffic management and construction scheduling, improving the overall construction management efficiency and safety.

[0085] S140. Combine the target recognition result and the movement of the construction vehicle to determine the road construction status according to the set decision tree.

[0086] In this embodiment, the road construction status refers to whether the road is currently under construction or not.

[0087] Specifically, please refer to Figure 2 , the decision tree includes eight different construction and non-construction states. Among them, the construction states include: with enclosures; without enclosures, with construction vehicles, construction vehicles stationary, with construction workers; without enclosures, with construction vehicles, construction vehicles stationary, without construction workers, with traffic cones; without enclosures, without construction vehicles, with construction workers, with traffic cones; the non-construction states include: without enclosures, with construction vehicles, construction vehicles stationary, without construction workers, without traffic cones; without enclosures, with construction vehicles, construction vehicles moving; without enclosures, without construction vehicles, with construction workers, without traffic cones; without enclosures, without construction vehicles, without construction workers.

[0088] The decision tree can refine different construction states, enabling the system to accurately distinguish various construction scenarios, such as with enclosures, without enclosures, and the status of construction vehicles, etc.; using the decision tree to analyze the construction status can not only improve the intelligent level of construction management, but also provide strong support for urban traffic and safety management.

[0089] S150. According to the road construction status, fuse the construction element features, calculate the characteristic influence factor and environmental weight factor corresponding to the construction element features according to the fusion result, and make a judgment by integrating the characteristic influence factor, environmental weight factor and historical information to determine the construction type.

[0090] In one embodiment, the above step S150 may include steps S151 to S154.

[0091] S151. When the road construction state belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector, and perform the fusion of feature weights to obtain a fusion result; wherein, the construction element features include fence features, construction vehicle features, construction personnel features, and traffic cone features.

[0092] In this embodiment, the fusion result refers to the comprehensive feature representation formed by the fusion of the feature weights of the multi-channel scene feature vector.

[0093] In one embodiment, the above step S151 may include steps S1511 to S1512.

[0094] S1511. When the road construction state belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector; wherein the weight of each channel is represented by the segmentation result corresponding to the feature.

[0095] In this embodiment, the four element features of fence features, construction vehicle features, construction personnel features, and traffic cone features respectively constitute the feature vectors of four channels. This structure allows the system to process different types of information simultaneously, thereby providing a richer scene description.

[0096] The generation of the weight coefficient matrix depends on the segmentation result of each element, usually implemented through a binary mask.

[0097] Taking a construction vehicle as an example, the segmentation result generates a binary image, where the pixel value of the area where the vehicle is located is 1 (with weight), while the pixel value of the area outside the vehicle is 0 (without weight); similarly, the segmentation result corresponding to the construction vehicle feature generates a binary image, where the pixel value of the area where the construction vehicle feature is located is 1, and the others are 0; the segmentation result corresponding to the construction personnel feature generates a binary image, where the pixel value of the area where the construction personnel feature is located is 1, and the others are 0; the segmentation result corresponding to the traffic cone feature generates a binary image, where the pixel value of the area where the traffic cone feature is located is 1, and the others are 0.

[0098] S1512. Perform the fusion of the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0099] In this embodiment, at least one of Gaussian fusion, regional mean feature fusion, regional clustering mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion is used to perform the fusion of the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0100] Specifically, the Gaussian fusion calculation mainly uses a Gaussian matrix for fusion, such as Figure 5As shown, the product of the network output result and the corresponding Gaussian matrix and then summing them is the final eigenvalue. The right side is the Gaussian matrix, and the sum of all elements is 1. During the Gaussian fusion process, mainly the Gaussian weights corresponding to the features are calculated, using , where A is the weight amplitude. The weight for the fence feature is set to 7, the weight for the vehicle feature is set to 5, the weight for the construction worker feature is set to 3, and the weight for the traffic cone feature is set to 1. For other feature fusion methods, they can be calculated independently without manually specifying the weight coefficients. is the coordinate of the center, is the variance.

[0101] In the regional mean feature fusion, the mean of the eigenvalue of different regions in the result matrix output by the neural network is used as the weight of feature fusion, that is, the final eigenvalue. As Figure 6 shown, where the right side is the mean matrix.

[0102] In the regional clustering mode feature fusion, the features in the region are classified according to the extremely small intervals, and the feature interval with the largest number is used as the final feature weight. As Figure 7 shown, that is, the mode of the network output is statistically calculated as the final eigenvalue.

[0103] As Figure 8 shown, in the regional median feature fusion, the features output by the network are sorted, and the median feature is used as the final feature weight.

[0104] In the double-layer dynamic feature activation fusion, two feature matrices with the same shape are output for each target. One feature matrix represents the confidence of the target, and the other feature matrix represents the weight. Only when the confidence is in the activatable state can the rationality of the weight sum be determined. If the confidence is low, there is no need to judge the feature weight. Finally, the mean of the activation region is used as the final weight. As Figure 9 shown, for the two matrices on the right, the first matrix is the Gaussian matrix M1. The value of the matrix from the boundary to the center point is (0 - n), where n is the mean of the network output. The second Gaussian matrix M2, the sum of all elements is 1. First, the network output is compared with all elements of M1, that is, the corresponding features one by one, and the elements in the network output result greater than M1 are statistically calculated, and the remaining elements are discarded. Then the retained elements are summed with the Gaussian matrix M2, and this value is used as the final eigenvalue. That is, M1 acts as a filtering role, only filtering the results that meet the requirements of the threshold in the network output.

[0105] Through different fusion methods (such as Gaussian fusion, regional mean, mode, and median feature fusion), various feature information can be effectively integrated, improving the model's ability to understand complex scenarios.

[0106] Fusion methods such as regional mean and clustering mode feature fusion do not require manual weight specification, enhancing the system's adaptability and enabling it to automatically adapt to the feature distributions of different scenarios.

[0107] The double-layer dynamic feature activation fusion method can dynamically filter features based on confidence, effectively suppressing the influence of noise and irrelevant features and improving the rationality of feature selection.

[0108] The combination of multiple fusion methods enables the model to select the most suitable feature fusion strategy according to specific task requirements, improving the overall performance.

[0109] Adopting different fusion strategies makes the model more robust in the face of data changes or uncertainties and enables it to maintain good performance.

[0110] S152. Calculate the feature influence factor corresponding to the construction element feature according to the fusion result.

[0111] In this embodiment, the feature influence factor refers to the construction level in the current scenario expressed in relative terms.

[0112] Specifically, use to calculate the feature influence factor corresponding to the construction element feature, where refers to the fence feature, construction vehicle feature, construction worker feature, traffic cone feature; refers to the weight coefficient; is the fusion result; is the feature influence factor corresponding to the construction element feature.

[0113] Finally, perform operation on the fusion result, that is, the activation operation. Sum the graphs corresponding to the fence feature, construction vehicle feature, construction worker feature, and traffic cone feature.

[0114] Through the above steps, the influence of each feature in a specific scenario can be quantified, which helps to accurately evaluate the construction environment and improve the effectiveness of safety management and decision support.

[0115] S153. Calculate the environmental weight factor corresponding to the construction element feature.

[0116] In this embodiment, the environmental weight factor is used to evaluate the relative importance of each feature to construction safety and efficiency in a specific environment. Specifically, it refers to the degree of influence of each feature on the overall construction state in the construction environment. They can help identify which features are more important in the current environment, thus providing support for subsequent decisions.

[0117] In one embodiment, the above step S1530 may include steps S1531 to S1533.

[0118] S1531. Calculate the relevant parameters of the construction element features.

[0119] In this embodiment, when judging the construction level, in addition to paying attention to the feature dimensions, it is also necessary to consider the proportion of each element in the construction scenario, because these factors directly affect the safety and efficiency of construction. The specific steps are as follows:

[0120] Calculate the area enclosed by the enclosure : First, measure the total area of the area enclosed by the enclosure. This provides the basic data for subsequent judgment.

[0121] Divide the lanes and calculate the lane widths : In the enclosure area, identify the lanes and measure their widths. The lane width is an important factor affecting the flow and safety of construction vehicles.

[0122] Calculate the vehicle footprint area : Count the number of vehicles in the construction area and calculate their total footprint area according to the vehicle type and size. This helps to evaluate the congestion level of the construction site.

[0123] Calculate the number of construction workers : Record the number of people at the construction site. The density of people is closely related to construction safety.

[0124] Calculate the area of the polygon area enclosed by traffic cones : Traffic cones are usually used to indicate the boundaries of the construction area. Calculate the area of the polygon area enclosed by them according to the positions of the traffic cones to understand the effectiveness and safety of traffic guidance.

[0125] Through these calculations, various elements of the construction scenario can be comprehensively considered, so as to more accurately judge the construction level and potential risks.

[0126] S1532. Use the prior parameters to perform pre-normalization processing on the relevant parameters of the construction element features to obtain the processing results.

[0127] In this embodiment, the processing results include the following:

[0128] The intermediate weight of the enclosure feature ;

[0129] The intermediate weight of the construction vehicle feature ;

[0130] The intermediate weight of the construction area feature ;

[0131] The intermediate weight of the traffic cone feature ;

[0132] Among them, , , , are prior parameters.

[0133] The above steps only limit the weights of the four elements of the enclosure feature, construction vehicle feature, construction area feature, and traffic cone feature within a relatively fixed range and normalize them to between 0 and 1.

[0134] S1533. Perform Softmax processing on the processing result to obtain an environmental weight factor.

[0135] Specifically, use ; ; ; , perform sofmax processing on the above weights to obtain an environmental weight factor , , .

[0136] The above steps can comprehensively understand the construction environment by calculating the relevant parameters of features such as enclosures, construction vehicles, construction personnel, and traffic cones, and help identify potential risks; using prior parameters and normalization processing enables various features to be compared on the same scale, which helps to make more scientific decisions; by quantifying the congestion level, personnel density, etc. at the construction site, potential safety hazards can be predicted in advance, and corresponding measures can be taken; by calculating the environmental weight factor, the most important factors during the construction process can be identified, thereby optimizing the allocation of human and material resources and improving construction efficiency; this method allows the weights to be dynamically updated according to real-time data, making construction management more flexible and adaptable to changing on-site conditions.

[0137] Present the feature influence factor and the environmental weight factor in the segmented image to form a heat map, as Figure 4 shown, which can be used for the classification of construction types.

[0138] S154. Based on the feature influence factor and the environmental weight factor, make a comprehensive judgment in combination with historical information to obtain the construction type.

[0139] In this embodiment, the construction type includes construction behavior categories, which can be specifically divided into 3 types: small-scale construction, medium-scale construction, and large-scale construction.

[0140] In this embodiment, construction is a continuous process. Therefore, it not only depends on the currently collected image data but also needs to refer to historical information to obtain a more comprehensive judgment. The collected sequence of images of the current vehicle is a complete monitoring, but by comparing with historical collection data, the accuracy of recognition can be improved.

[0141] In one embodiment, the above step S154 may include steps S1541 to S1542.

[0142] S1541. Apply the sigmoid function to activate the feature influence factor corresponding to the construction element feature to obtain an activation result.

[0143] In this embodiment, the activation result refers to the value of the feature influence factor after being transformed by the sigmoid function, and these values will be restricted between 0 and 1 for subsequent processing.

[0144] Specifically, introducing a β weight (with a value of 0.5) and the Sigmoid activation function can effectively comprehensively evaluate four element features (such as the area of the enclosure, lane width, vehicle footprint area, and the number of construction workers). Through the Sigmoid function, when the four features are obvious, the activation result is close to 1, indicating that the construction behavior is ongoing; when the features are small, the activation result is small, indicating that the construction activity is weakening; if the features are extremely unobvious and the result is close to 0, it means that the construction is approaching completion.

[0145] Specifically, the network output is a matrix of h*w*4, where 4 represents the elements, including the enclosure feature, construction vehicle feature, personnel feature, and traffic cone feature. The feature matrix is normalized to between 0 and 1 through the sigmoid function, and the feature vectors corresponding to 4 mask regions are statistically calculated.

[0146] Before this, the "Gaussian weight weighting method" or "regional mean feature fusion" or "regional clustering mode feature fusion" or "regional median feature fusion" or "double-layer dynamic feature activation fusion", etc., are used to statistically calculate the features of each channel respectively to determine the feature influence factor corresponding to the construction element feature. , , , , at this time, the feature values may take positive, negative, or 0 values, not between 0 and 1. Then, the weights of the 4 elements are summed up, and through the sigmoid method or the tanh function normalization method or the linear normalization method, they are normalized to between 0 and 1 to obtain this part.

[0147] Linear normalization method , where xMin and xMax are estimated maximum values. For example, the width and height of the image size are 100*100; it can be estimated that xMin is -10000 and xMax is 10000. That is, when the feature matrix is all -1, the sum is -10000, and when the feature matrix is all 1, the sum is 10000.

[0148] Linear normalization can effectively integrate the influences of various features and enhance the scientificity and accuracy of overall construction management.

[0149] S1542. Determine the construction type according to the activation result and the environmental weight factor in combination with historical information.

[0150] Specifically, use to determine the construction type, where is the construction type; refers to the importance weight of current and historical data; 、 、 、 are the environmental weight factors corresponding to the construction element features respectively; 、 、 、 are the feature influence factors corresponding to the construction element features respectively; and are 0.5, refers to the influence degree of historical data; is the activation result.

[0151] In this embodiment, in addition to considering the simple superposition of the feature influence factors corresponding to the construction element features, environmental factors are also considered. The environmental factors are multiplied and summed with the corresponding feature influence factors respectively, corresponding to .

[0152] Actually, first name the result of as Res1, the result as Res2; α*Res1 + β*Res2 (where both α and β take the value of 0.5), which represents the detection result of the current vehicle this time. The current detection result factor and the historical result factor are: k and 1 - k respectively. k is set to 0.9, indicating that the confidence of the current result is very high. 1 - k is 0.1, indicating that the historical result is only for reference (the historical result is the result of the previous vehicle, and the result of the current vehicle can become the historical result of the next vehicle's detection result, which is a relative concept). In this way, the historical result and the current result are combined to judge this scenario. Finally, the value range of p is 0 - 1. Classification can be carried out according to the value of grade P for the result. For example, 0 - 0.3 represents a small construction site, 0.3 - 0.6 represents a medium-sized construction site, and 0.6 - 1 represents a large construction site.

[0153] By combining current and historical data and using environmental weight factors and feature influence factors, the accuracy and credibility of construction type determination are improved. It can more comprehensively reflect the on-site situation, make the classification more precise, contribute to optimizing construction management and resource allocation, and ultimately enhance safety and efficiency.

[0154] The above road construction status detection method based on dynamic data monitoring obtains road images through cameras on urban public transport vehicles, performs object recognition and segmentation, extracts construction-related features, and then uses multi-object tracking technology to monitor the movement of construction vehicles. Combining the object recognition results, a decision tree is used to determine the construction status. If it is in the construction status, various features are combined to form a multi-channel scene feature vector, and feature weight fusion is performed. Then, according to the fusion result, a feature influence factor and an environmental weight factor are calculated, and a comprehensive judgment is made in combination with historical information. Finally, the construction type is obtained. This process can adapt to different urban road conditions, quickly access, save costs, and improve urban governance and traffic warning capabilities.

[0155] Figure 10 It is a schematic flowchart of a road construction status detection method based on dynamic data monitoring provided by another embodiment of the present invention. As Figure 10 shown, the road construction status detection method based on dynamic data monitoring in this embodiment includes steps S210-S260. Among them, steps S210-S250 are similar to steps S110-S150 in the above embodiment and will not be described in detail here. The following will detail the added step S260 in this embodiment.

[0156] S260. Record the object recognition result to form historical information.

[0157] Specifically, the current classification features are continuously updated to the cloud platform to form historical information . These data can not only be used to determine whether there is a construction behavior and the scale of the construction type in the current scene, but also provide a basis for the subsequent comprehensive judgment of vehicles at the current location. Subsequent vehicles can combine historical information with real-time data for more accurate analysis and decision-making.

[0158] Figure 11 It is a schematic block diagram of a road construction status detection device 300 based on dynamic data monitoring provided by an embodiment of the present invention. As Figure 11 shown, corresponding to the above road construction status detection method based on dynamic data monitoring, the present invention also provides a road construction status detection device 300 based on dynamic data monitoring. The road construction status detection device 300 based on dynamic data monitoring includes units for executing the above road construction status detection method based on dynamic data monitoring, and this device can be configured in a server. Specifically, please refer to Figure 11 , the road construction status detection device 300 based on dynamic data monitoring includes an image acquisition unit 301, an identification and segmentation unit 302, a tracking unit 303, a status determination unit 304, and a construction type determination unit 305.

[0159] An image acquisition unit 301 for acquiring road images collected by road traffic vehicles; an identification and segmentation unit 302 for performing target identification and segmenting road regions, construction regions, and construction element features based on the road images to obtain a target identification result; a tracking unit 303 for performing operation tracking on construction vehicles using a multi-target tracking method to determine the movement of the construction vehicles; a status determination unit 304 for determining the road construction status according to a set decision tree in combination with the target identification result and the movement of the construction vehicles; a construction type determination unit 305 for fusing the construction element features according to the road construction status, calculating the feature influence factor and environmental weight factor corresponding to the construction element features according to the fusion result, and making a judgment by synthesizing the feature influence factor, environmental weight factor, and historical information to determine the construction type.

[0160] In one embodiment, the construction type determination unit 305 includes:

[0161] A fusion subunit for, when the road construction status belongs to the construction state, combining the construction element features to form a multi-channel scene feature vector and performing fusion of feature weights to obtain a fusion result; wherein the construction element features include fence features, construction vehicle features, construction personnel features, and traffic cone features.

[0162] A feature influence factor calculation subunit for calculating the feature influence factor corresponding to the construction element features according to the fusion result.

[0163] An environmental weight factor calculation subunit for calculating the environmental weight factor corresponding to the construction element features.

[0164] A comprehensive judgment subunit for making a comprehensive judgment according to the feature influence factor and environmental weight factor in combination with historical information to obtain the construction type.

[0165] In one embodiment, the fusion single subunit includes:

[0166] A combination module for, when the road construction status belongs to the construction state, combining the construction element features to form a multi-channel scene feature vector; wherein the weight of each channel is represented by the segmentation result corresponding to the feature; a weight fusion module for performing fusion of feature weights on the multi-channel scene feature vector to obtain a fusion result.

[0167] In one embodiment, the weight fusion module is used to perform fusion of feature weights on the multi-channel scene feature vector by at least one of Gaussian fusion, regional mean feature fusion, regional clustering mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion to obtain a fusion result.

[0168] In one embodiment, the feature influence factor calculation unit is configured to use to calculate the feature influence factor corresponding to the construction element feature, where refers to the enclosure feature, the construction vehicle feature, the construction personnel feature, and the traffic cone feature; refers to the weight coefficient; is the fusion result; is the feature influence factor corresponding to the construction element feature.

[0169] In one embodiment, the environmental weight factor calculation sub-unit includes:

[0170] A parameter calculation module for calculating the relevant parameters of the construction element feature; a normalization module for pre-normalizing the relevant parameters of the construction element feature using prior parameters to obtain a processing result; and a processing module for performing Softmax processing on the processing result to obtain an environmental weight factor.

[0171] In one embodiment, the comprehensive judgment sub-unit includes:

[0172] An activation module for activating the feature influence factor corresponding to the construction element feature using the sigmoid function to obtain an activation result;

[0173] A determination module for determining the construction type according to the activation result and the environmental weight factor in combination with historical information. Specifically, use to determine the construction type, where is the construction type; refers to the importance weight of the current and historical data; , , , are respectively the environmental weight factors corresponding to the construction element features; , , , are respectively the feature influence factors corresponding to the construction element features; is 0.5, refers to the influence degree of historical data; is the activation result.

[0174] Figure 12 is a schematic block diagram of a road construction status detection device 300 based on dynamic data monitoring provided in another embodiment of the present invention. As Figure 12 shown, the road construction status detection device 300 based on dynamic data monitoring in this embodiment adds a recording unit on the basis of the above embodiment.

[0175] A recording unit 306 for recording the target recognition result to form historical information.

[0176] It should be noted that those skilled in the art can clearly understand that the specific implementation processes of the above-mentioned road construction status detection device 300 based on dynamic data monitoring and each unit can refer to the corresponding descriptions in the foregoing method embodiments. For the sake of convenience and brevity of description, they will not be elaborated herein.

[0177] The above-mentioned road construction status detection device 300 based on dynamic data monitoring can be implemented in the form of a computer program, and this computer program can run on a computer device as shown in Figure 13 Figure.

[0178] Please refer to Figure 13 , Figure 13 which is a schematic block diagram of a computer device provided by an embodiment of the present application. This computer device 500 can be a server. Among them, the server can be an independent server or a server cluster composed of multiple servers.

[0179] Refer to Figure 13 , this computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501. Among them, the memory can include a non-volatile storage medium 503 and an internal memory 504.

[0180] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. This computer program 5032 includes program instructions. When the program instructions are executed, the processor 502 can be made to execute a road construction status detection method based on dynamic data monitoring.

[0181] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500.

[0182] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can be made to execute a road construction status detection method based on dynamic data monitoring.

[0183] The network interface 505 is used for network communication with other devices. Those skilled in the art can understand that Figure 13 the structure shown in

[0184] Among them, the processor 502 is used to run the computer program 5032 stored in the memory to implement the following steps:

[0185] Obtain road images collected by road traffic vehicles; perform object recognition on the road images and segment the road area, construction area, and construction element features to obtain an object recognition result; use a multi-object tracking method to track the operation of the construction vehicle to determine the movement of the construction vehicle; combine the object recognition result and the movement of the construction vehicle to determine the road construction status according to a set decision tree; according to the road construction status, fuse the construction element features, calculate the feature influence factor and environmental weight factor corresponding to the construction element features according to the fusion result, and make a judgment by synthesizing the feature influence factor, environmental weight factor, and historical information to determine the construction type.

[0186] Among them, the decision tree includes eight different construction states and non-construction states. Among them, the construction states include: having a fence; having no fence, having a construction vehicle, the construction vehicle is stationary, and having construction personnel; having no fence, having a construction vehicle, the construction vehicle is stationary, having no construction personnel, and having traffic cones; having no fence, having no construction vehicle, having construction personnel, and having traffic cones; the non-construction states include: having no fence, having a construction vehicle, the construction vehicle is stationary, having no construction personnel, and having no traffic cones; having no fence, having a construction vehicle, and the construction vehicle is moving; having no fence, having no construction vehicle, having construction personnel, and having no traffic cones; having no fence, having no construction vehicle, and having no construction personnel.

[0187] In one embodiment, when the processor 502 implements the step of fusing the construction element features according to the road construction status, calculating the feature influence factor and environmental weight factor corresponding to the construction element features according to the fusion result, and making a judgment by synthesizing the feature influence factor, environmental weight factor, and historical information to determine the construction type, the specific implementation steps are as follows:

[0188] When the road construction status belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector, and perform feature weight fusion to obtain a fusion result; among them, the construction element features include fence features, construction vehicle features, construction personnel features, and traffic cone features; calculate the feature influence factor corresponding to the construction element features according to the fusion result; calculate the environmental weight factor corresponding to the construction element features; according to the feature influence factor and environmental weight factor, make a comprehensive judgment in combination with historical information to obtain the construction type.

[0189] In one embodiment, when the processor 502 implements the step of combining the construction element features to form a multi-channel scene feature vector and performing feature weight fusion to obtain a fusion result when the road construction status belongs to the construction state, the specific implementation steps are as follows:

[0190] When the road construction state belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector; where the weight of each channel is represented by the result of the segmentation corresponding to the feature; fuse the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0191] In one embodiment, when the processor 502 implements the step of fusing the feature weights of the multi-channel scene feature vector to obtain a fusion result, the specific implementation steps are as follows:

[0192] Use at least one of Gaussian fusion, regional mean feature fusion, regional clustering mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion to fuse the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0193] In one embodiment, when the processor 502 implements the step of calculating the feature influence factor corresponding to the construction element feature according to the fusion result, the specific implementation steps are as follows:

[0194] Use Calculate the feature influence factor corresponding to the construction element feature, where refers to the fence feature, construction vehicle feature, construction personnel feature, and traffic cone feature; refers to the weight coefficient; is the fusion result; is the feature influence factor corresponding to the construction element feature.

[0195] In one embodiment, when the processor 502 implements the step of calculating the environmental weight factor corresponding to the construction element feature, the specific implementation steps are as follows:

[0196] Calculate the relevant parameters of the construction element feature; use the prior parameters to pre-normalize the relevant parameters of the construction element feature to obtain a processing result; perform Softmax processing on the processing result to obtain the environmental weight factor.

[0197] In one embodiment, when the processor 502 implements the step of making a comprehensive judgment based on the feature influence factor and the environmental weight factor in combination with historical information to obtain the construction type, the specific implementation steps are as follows:

[0198] Apply the sigmoid function to activate the feature influence factor corresponding to the construction element feature to obtain an activation result; determine the construction type according to the activation result, the environmental weight factor, and historical information.

[0199] In one embodiment, when the processor 502 implements the step of determining the construction type according to the activation result and the environmental weight factor in combination with historical information, the following steps are specifically implemented:

[0200] Adopt Determine the construction type, where is the construction type; refers to the importance weight of the current and historical data; 、 、 、 are respectively the environmental weight factors corresponding to the construction element features; 、 、 、 are respectively the feature influence factors corresponding to the construction element features; and are 0.5, refers to the influence degree of historical data; is the activation result.

[0201] It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0202] Those of ordinary skill in the art can understand that all or part of the processes in the methods of implementing the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, and the storage medium is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0203] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the following steps:

[0204] Obtain road images collected by road traffic vehicles; perform target recognition on the road images and segment the road area, construction area, and construction element features to obtain the target recognition result; use a multi-target tracking method to track the operation of the construction vehicle to determine the movement of the construction vehicle; combine the target recognition result and the movement of the construction vehicle to determine the road construction status according to a set decision tree; according to the road construction status, fuse the construction element features, calculate the feature influence factor and environmental weight factor corresponding to the construction element features based on the fusion result, and make a judgment by integrating the feature influence factor, environmental weight factor, and historical information to determine the construction type.

[0205] Among them, the decision tree includes eight different construction states and non-construction states. Among them, the construction states include: with a fence; without a fence, with a construction vehicle, the construction vehicle is stationary, with construction workers; without a fence, with a construction vehicle, the construction vehicle is stationary, without construction workers, with traffic cones; without a fence, without a construction vehicle, with construction workers, with traffic cones; the non-construction states include: without a fence, with a construction vehicle, the construction vehicle is stationary, without construction workers, without traffic cones; without a fence, with a construction vehicle, the construction vehicle is moving; without a fence, without a construction vehicle, with construction workers, without traffic cones; without a fence, without a construction vehicle, without construction workers.

[0206] In one embodiment, when the processor executes the computer program to implement the step of fusing the construction element features according to the road construction status, calculating the feature influence factor and environmental weight factor corresponding to the construction element features based on the fusion result, and making a judgment by integrating the feature influence factor, environmental weight factor, and historical information to determine the construction type, the specific implementation steps are as follows:

[0207] When the road construction status belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector and perform feature weight fusion to obtain the fusion result; among them, the construction element features include fence features, construction vehicle features, construction worker features, and traffic cone features; calculate the feature influence factor corresponding to the construction element features according to the fusion result; calculate the environmental weight factor corresponding to the construction element features; make a comprehensive judgment based on the feature influence factor and environmental weight factor in combination with historical information to obtain the construction type.

[0208] In one embodiment, when the processor executes the computer program to implement the step of combining the construction element features to form a multi-channel scene feature vector and performing feature weight fusion to obtain the fusion result when the road construction status belongs to the construction state, the specific implementation steps are as follows:

[0209] When the road construction state belongs to the construction state, combine the construction element features to form a multi-channel scene feature vector; where the weight of each channel is represented by the result of the segmentation corresponding to the feature; fuse the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0210] In one embodiment, when the processor executes the computer program to implement the step of fusing the feature weights of the multi-channel scene feature vector to obtain a fusion result, the specific implementation steps are as follows:

[0211] Use at least one of Gaussian fusion, regional mean feature fusion, regional clustering mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion to fuse the feature weights of the multi-channel scene feature vector to obtain a fusion result.

[0212] In one embodiment, when the processor executes the computer program to implement the step of calculating the feature influence factor corresponding to the construction element feature according to the fusion result, the specific implementation steps are as follows:

[0213] Use Calculate the feature influence factor corresponding to the construction element feature, where refers to the enclosure feature, construction vehicle feature, construction personnel feature, and traffic cone feature; refers to the weight coefficient; is the fusion result; is the feature influence factor corresponding to the construction element feature.

[0214] In one embodiment, when the processor executes the computer program to implement the step of calculating the environmental weight factor corresponding to the construction element feature, the specific implementation steps are as follows:

[0215] Calculate the relevant parameters of the construction element feature; use the prior parameters to pre-normalize the relevant parameters of the construction element feature to obtain a processing result; perform Softmax processing on the processing result to obtain the environmental weight factor.

[0216] In one embodiment, when the processor executes the computer program to implement the step of making a comprehensive judgment based on the feature influence factor and the environmental weight factor in combination with historical information to obtain the construction type, the specific implementation steps are as follows:

[0217] Apply the sigmoid function to activate the feature influence factor corresponding to the construction element feature to obtain an activation result; determine the construction type according to the activation result, the environmental weight factor, and historical information.

[0218] In one embodiment, when the processor executes the computer program to implement the step of determining the construction type according to the activation result and the environmental weight factor in combination with historical information, the following steps are specifically implemented:

[0219] Adopt Determine the construction type, where is the construction type; refers to the importance weight of the current and historical data; 、 、 、 are respectively the environmental weight factors corresponding to the construction element features; 、 、 、 are respectively the feature influence factors corresponding to the construction element features; and are 0.5, refers to the influence degree of historical data; is the activation result.

[0220] The storage medium can be various computer-readable storage media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disc that can store program codes.

[0221] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0222] In several embodiments provided by the present invention, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0223] The steps in the method of the embodiments of the present invention can be adjusted in sequence, combined, and deleted according to actual needs. The units in the device of the embodiments of the present invention can be combined, divided, and deleted according to actual needs. In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0224] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention.

[0225] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A road construction status detection method based on dynamic data monitoring, characterized in that: include: Acquire road images collected by road traffic vehicles; Performing target recognition and segmenting road areas, construction areas, and construction element features according to the road image to obtain target recognition results; Tracking the construction vehicle using a multi-target tracking method to determine the movement of the construction vehicle; Determine the road construction status according to a set decision tree based on the target recognition result and the movement of the construction vehicle; According to the road construction status, the construction element characteristics are integrated, and the characteristic influencing factors corresponding to the construction element characteristics are calculated according to the integration results, and the environmental weighting factors corresponding to the construction element characteristics are calculated. The characteristic influencing factors, environmental weighting factors and historical information are comprehensively judged to determine the construction type; the construction types include small-scale construction, medium-scale construction and large-scale construction; The construction type is obtained by making a comprehensive judgment based on the characteristic influencing factors and the environmental weight factors in combination with historical information, including: Applying a sigmoid function to activate the characteristic influencing factors corresponding to the construction element characteristics to obtain an activation result; use Determine the type of construction, where For construction type; It refers to the importance weight of current and historical data; , , , are the environmental weight factors corresponding to the characteristics of the construction elements; , , , are characteristic influencing factors corresponding to the characteristics of the construction elements respectively; and is 0.5, It refers to the degree of influence of historical data; Activation result.

2. The road construction status detection method based on dynamic data monitoring according to claim 1 is characterized in that: According to the road construction status, the construction element characteristics are integrated, and the characteristic influence factors corresponding to the construction element characteristics are calculated according to the integration results, and the environmental weight factors corresponding to the construction element characteristics are calculated. The characteristic influence factors, the environmental weight factors and the historical information are comprehensively judged to determine the construction type, including: When the road construction state belongs to the construction state, the construction element features are combined to form a multi-channel scene feature vector, and feature weight fusion is performed to obtain a fusion result; Calculate the characteristic influencing factor corresponding to the construction element characteristic according to the fusion result; Calculating the environmental weight factor corresponding to the construction element characteristics; A comprehensive judgment is made based on the characteristic influencing factors and environmental weight factors in combination with historical information to obtain the construction type.

3. The road construction status detection method based on dynamic data monitoring according to claim 1 is characterized in that: The decision tree includes eight different construction states and non-construction states, wherein the construction state includes: with fences; without fences, with construction vehicles, the construction vehicles are stationary, and with construction personnel; without fences, with construction vehicles, the construction vehicles are stationary, without construction personnel, and with traffic cones; without fences, without construction vehicles, with construction personnel, and with traffic cones; the non-construction state includes: without fences, with construction vehicles, the construction vehicles are stationary, without construction personnel, and without traffic cones; without fences, with construction vehicles, and with moving construction vehicles; without fences, without construction vehicles, with construction personnel, and without traffic cones; without fences, without construction vehicles, and without construction personnel.

4. The road construction status detection method based on dynamic data monitoring according to claim 1 is characterized in that: When the road construction state belongs to the construction state, the construction element features are combined to form a multi-channel scene feature vector, and the feature weights are fused to obtain a fusion result, including: When the road construction state belongs to the construction state, the construction element features are combined to form a multi-channel scene feature vector; wherein the weight of each channel is represented by the segmentation result corresponding to the feature; The multi-channel scene feature vectors are fused by feature weights to obtain a fusion result.

5. The road construction status detection method based on dynamic data monitoring according to claim 4 is characterized in that: The multi-channel scene feature vectors are fused with feature weights to obtain a fusion result, including: The feature weights of the multi-channel scene feature vectors are fused by at least one of Gaussian fusion, regional mean feature fusion, regional cluster mode feature fusion, regional median feature fusion, and double-layer dynamic feature activation fusion to obtain a fusion result.

6. The road construction status detection method based on dynamic data monitoring according to claim 1 is characterized in that: The calculating, according to the fusion result, the characteristic influencing factor corresponding to the construction element characteristic comprises: use Calculate the characteristic influencing factor corresponding to the construction element characteristic, where: It refers to the characteristics of fences, construction vehicles, construction workers, and traffic cones; is the weight coefficient; is the fusion result; is the characteristic influencing factor corresponding to the construction element characteristics.

7. The road construction status detection method based on dynamic data monitoring according to claim 1 is characterized in that: The calculating of the environmental weight factor corresponding to the construction element feature includes: Calculating relevant parameters of the construction element characteristics; Using a priori parameters to pre-normalize the relevant parameters of the construction element characteristics to obtain a processing result; The processing result is subjected to Softmax processing to obtain an environment weight factor.

8. A road construction status detection device based on dynamic data monitoring, the device using the road construction status detection method based on dynamic data monitoring according to any one of claims 1 to 7, characterized in that: include: An image acquisition unit, used to acquire a road image collected by a road traffic vehicle; An identification and segmentation unit, used to perform target identification and segment the road area, construction area, and construction element features according to the road image to obtain a target identification result; A tracking unit, used to track the construction vehicle using a multi-target tracking method to determine the movement of the construction vehicle; A state determination unit, configured to determine the road construction state according to a set decision tree in combination with the target recognition result and the movement of the construction vehicle; The construction type determination unit is used to fuse the construction element characteristics according to the road construction status, and calculate the characteristic influence factors corresponding to the construction element characteristics according to the fusion results, calculate the environmental weight factors corresponding to the construction element characteristics, and make a judgment based on the comprehensive characteristic influence factors, environmental weight factors and historical information to determine the construction type.

Images