A vehicle overload detection and early warning method and system

Abstract

The application discloses a kind of vehicle overload detection early warning method and system, it is related to overload detection technical field, the method includes: the vehicle image of target vehicle to be carried out vehicle overload detection is collected, image semantic segmentation is carried out, obtain vehicle area image, then vehicle height convolution feature analysis is carried out, obtain the vehicle height of target vehicle, obtain height limit detection result, judge whether it satisfies the height limit threshold value of target vehicle, when satisfying, the cargo image in target vehicle is collected, cargo image is carried out convolution feature analysis, obtain volume detection result, obtain the kind information of goods loaded in target vehicle, and input goods database, obtain fuzzy density information;According to fuzzy density information and volume detection result, calculate and obtain load detection result;Judge whether load detection result satisfies the load threshold value of target vehicle, when not satisfying, early warning is carried out.The application solves the technical problem that vehicle overload detection efficiency is low, cost is higher.

Description

Technical Field

[0001] This invention relates to the field of overload detection technology, specifically to a method and system for vehicle overload detection and early warning. Background Technology

[0002] Vehicle overloading seriously affects road traffic safety. Current overload detection technologies generally rely on weighing, requiring the deployment of weighing devices at fixed locations and guiding vehicles for enforcement. Height restrictions on trucks necessitate the installation of height-limiting facilities, consuming significant manpower and resources and resulting in low efficiency. Therefore, existing technologies suffer from the technical problems of high cost and low efficiency in vehicle overload detection. Summary of the Invention

[0003] This application provides a method and system for detecting and warning of vehicle overload, which solves the technical problems of high cost and low efficiency in vehicle overload detection in the prior art.

[0004] The first aspect of this application provides a method for detecting and warning of vehicle overload, the method comprising: acquiring a vehicle image of a target vehicle to be detected for vehicle overload;

[0005] An image segmentation channel is constructed, and the vehicle image is subjected to image semantic segmentation to obtain a vehicle region image;

[0006] Perform vehicle height convolutional feature analysis on the vehicle region image to obtain the vehicle height of the target vehicle and obtain the height restriction detection result;

[0007] Determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it does, acquire an image of the cargo inside the target vehicle, perform convolutional feature analysis on the cargo image to obtain the volume detection result, and issue an early warning if the height limit is not met.

[0008] The system obtains information on the types of goods loaded in the target vehicle and inputs it into a cargo database to obtain fuzzy density information. Based on the fuzzy density information and the volume detection result, it calculates the load detection result. It then determines whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, an early warning is issued; otherwise, no early warning is issued.

[0009] A second aspect of this application provides a vehicle overload detection and early warning system, the system comprising: a vehicle image acquisition module for acquiring vehicle images of a target vehicle to be detected for overload;

[0010] The image segmentation module is used to construct an image segmentation channel and perform image semantic segmentation on the vehicle image to obtain a vehicle region image;

[0011] The height feature analysis module is used to perform vehicle height convolution feature analysis on the vehicle area image to obtain the vehicle height of the target vehicle and obtain the height limit detection result.

[0012] The volume feature analysis module is used to determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it does, it acquires an image of the cargo inside the target vehicle, performs convolutional feature analysis on the cargo image to obtain the volume detection result, and issues an early warning if it does not meet the threshold.

[0013] The density acquisition module is used to acquire information on the types of goods loaded in the target vehicle, input the information into the goods database, and obtain fuzzy density information.

[0014] The load calculation module is used to calculate the load detection result based on the fuzzy density information and the volume detection result;

[0015] The load determination module is used to determine whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, an early warning is issued; if it does meet the threshold, no early warning is issued.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] This application provides a vehicle overload detection and early warning method. It involves acquiring images of the vehicle to be detected for overload, performing image semantic segmentation to obtain the vehicle region image, then performing convolutional feature analysis on the vehicle height to obtain the vehicle height, determining whether it meets the height restriction threshold, and issuing an early warning if it does not. If it does meet the threshold, it acquires images of the goods on the vehicle, performs convolutional feature analysis to obtain the volume detection result of the goods, and then retrieves fuzzy density information based on the type of goods to calculate the load detection result, determining whether it meets the load threshold, and issuing an early warning. This embodiment of the application improves the efficiency and accuracy of vehicle height recognition by acquiring vehicle image information, performing semantic segmentation, and analyzing the vehicle height. It eliminates the need for height restriction facilities to detect and limit vehicle height. Furthermore, when the vehicle height meets the requirements, it detects and acquires images of the goods on the vehicle and the type of goods, analyzing the vehicle load. This eliminates the need for weighing facilities, efficiently obtaining vehicle load while ensuring accuracy in load analysis, thus improving the efficiency of vehicle load detection and overload early warning, and reducing detection costs. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic flowchart of a vehicle overload detection and early warning method provided in an embodiment of this application;

[0020] Figure 2 A flowchart illustrating the process of obtaining a vehicle area image in a vehicle overload detection and early warning method provided in this application embodiment;

[0021] Figure 3 A schematic diagram illustrating the process of obtaining height restriction detection results in a vehicle overload detection and early warning method provided in this application embodiment;

[0022] Figure 4 This is a schematic diagram of a vehicle overload detection and early warning system provided in an embodiment of this application.

[0023] Explanation of reference numerals in the attached diagram: Vehicle image acquisition module 11, Image segmentation module 12, Height feature analysis module 13, Volume feature analysis module 14, Density acquisition module 15, Load calculation module 16, Load judgment module 17. Detailed Implementation

[0024] This application provides a method for detecting and warning of vehicle overload, which solves the technical problems of high cost and low efficiency in the existing technology of vehicle overload detection.

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0027] Example 1

[0028] like Figure 1 As shown, this application provides a method for detecting and warning of vehicle overload, the method comprising:

[0029] S100: Acquire vehicle images of the target vehicle to be detected for overload.

[0030] In this embodiment, the target vehicle is the vehicle to be inspected for overload, preferably a truck, and more preferably a vehicle with an open cargo box for loading and transporting goods.

[0031] The method provided in this application for vehicle overload detection includes load overload detection with height restriction detection.

[0032] When detecting vehicle overload, images of the target vehicle are captured. These images are acquired using an image acquisition device, which can be a camera installed at a road location specifically for vehicle overload detection, or a camera pre-positioned on the road.

[0033] By acquiring images of the target vehicle to be detected for overload, the data for subsequent overload detection is used as the basis. Preferably, the image acquisition device is positioned on one side of the road to acquire images of the side of the target vehicle, which are then used as the vehicle image.

[0034] S200: Construct an image segmentation channel to perform image semantic segmentation on the vehicle image and obtain a vehicle region image;

[0035] In this embodiment of the application, after acquiring the vehicle image of the target vehicle, the vehicle image includes the image of the target vehicle as well as background images such as roads, sky, and ground. In order to improve the accuracy of vehicle image analysis, the vehicle image is subjected to image semantic segmentation to divide and obtain a vehicle region image that only includes the target vehicle, thereby eliminating the influence of the background image.

[0036] In this embodiment of the application, vehicle images are segmented through image semantic segmentation.

[0037] like Figure 2 As shown, step S200 in the method provided in this application embodiment includes:

[0038] S210: Based on the detection data of vehicles that have undergone overload detection over a historical period, obtain a set of sample vehicle images;

[0039] S220: Perform vehicle region image segmentation and labeling on multiple sample vehicle images within the sample vehicle image set to obtain a sample vehicle region image set;

[0040] S230: Construct the image segmentation channel using the sample vehicle image set and the sample vehicle region image set;

[0041] S240: Input the vehicle image into the image segmentation channel to obtain the vehicle region image.

[0042] In this embodiment, detection data of vehicles that underwent overload detection within a historical period is extracted. Specifically, images of vehicles that underwent overload detection within a historical period are extracted to obtain a sample vehicle image set. For example, this can be obtained by collecting vehicle overload detection data logs within a historical period.

[0043] The vehicle region images in multiple sample vehicle images within the sample vehicle image set are segmented and labeled. Specifically, the image portions of the vehicles and the background images within the sample vehicle images are segmented and labeled to obtain the sample vehicle region image set.

[0044] The image segmentation channel is constructed using a set of sample vehicle images and a set of sample vehicle region images as the construction data.

[0045] Step S230 in the method provided in this application embodiment includes:

[0046] S231: Construct the encoder and decoder within the image segmentation channel based on a fully convolutional neural network within semantic segmentation;

[0047] S232: Using the sample vehicle image set and the sample vehicle region image set, supervised training is performed on the encoder and decoder until the convergence condition is met;

[0048] S233: Verify the encoder and decoder. If the accuracy meets the accuracy requirements, the constructed image segmentation channel is obtained.

[0049] In this embodiment of the application, the input data of the image segmentation channel is a vehicle image, and the output data is a segmented vehicle region image. Based on a fully convolutional neural network for semantic segmentation in machine learning, an encoder and decoder are constructed in the image segmentation channel. The encoder includes multiple convolutional layers and pooling layers, and the decoder includes multiple deconvolutional layers and multiple unpooling layers.

[0050] A set of sample vehicle images and a set of sample vehicle region images are used as the construction data. Supervised training is performed on the encoder and decoder. During supervised training, the network parameters are updated and adjusted based on the error of the output vehicle region images until a convergence condition is met. This convergence condition can be an accuracy of 85%, at which point training is complete. Specifically, the encoder performs convolutional feature analysis on pixels within the vehicle images to obtain the pixel characteristics of the vehicle objects. Then, the decoder, based on these characteristics, groups pixels belonging to vehicles in the vehicle images into a single category, obtaining the vehicle region images, which are then output through a fully connected layer.

[0051] After training, the encoder and decoder are validated to avoid overfitting and insufficient generalization. If the validation accuracy meets the requirements, the constructed image segmentation channel is obtained. If not, the network parameters of the image segmentation channel are adjusted, or the image segmentation channel is reconstructed.

[0052] Based on the constructed image segmentation channel, the vehicle image of the current target vehicle is input into the image segmentation channel to obtain the vehicle region image.

[0053] This application embodiment performs semantic segmentation processing on the acquired target vehicle image to obtain the vehicle region image, which can eliminate the influence of the background image on the vehicle height analysis and improve the accuracy of vehicle height limit detection and analysis.

[0054] S300: Perform vehicle height convolution feature analysis on the vehicle area image to obtain the vehicle height of the target vehicle and obtain the height limit detection result;

[0055] In this embodiment, based on the vehicle region image obtained by segmentation, convolutional feature analysis of vehicle height is performed, and the vehicle height of the target vehicle is obtained through image recognition technology as the height restriction detection result.

[0056] like Figure 3 As shown, step S300 in the method provided in this application embodiment includes:

[0057] S310: Detect and label the vehicle heights in multiple sample vehicle region images within the sample vehicle region image set to obtain a sample vehicle height set;

[0058] S320: Based on a convolutional neural network, a vehicle height recognition channel is constructed using the sample vehicle region image set and the sample vehicle height set as construction data;

[0059] S330: Input the vehicle area image into the vehicle height recognition channel to obtain the vehicle height of the target vehicle, which is used as the height restriction detection result.

[0060] In this embodiment of the application, based on the set of sample vehicle area images obtained in step S200, the height of vehicles in multiple sample vehicle area images is detected and labeled. The height of vehicles can be obtained based on height detection data within a historical time period, and labeled for supervised training of deep learning to obtain a set of sample vehicle heights.

[0061] Based on convolutional neural networks in deep learning, a vehicle height recognition channel is constructed. This channel includes multiple convolutional layers, pooling layers, and fully connected layers. The input vehicle region image is subjected to multi-layer convolutional extraction and pooling processing. Finally, based on the multi-layer features extracted by convolution, the corresponding vehicle height is obtained by classification through activation functions.

[0062] Using the sample vehicle region image set and sample vehicle height set as construction data, the vehicle height recognition channel is subjected to supervised training, validation, and testing until the accuracy meets the requirements, for example, 85%. After successful validation, the vehicle height recognition channel is obtained.

[0063] Based on the constructed vehicle height recognition channel, the vehicle area image of the current target vehicle is input into the vehicle height recognition channel for image convolution feature extraction and data processing to obtain the height restriction detection result.

[0064] This application constructs an image segmentation channel through deep learning, analyzes the vehicle height based on the vehicle region image, and obtains the height limit detection result as the basis for judging whether the vehicle exceeds the height limit. It has high recognition efficiency, does not require the deployment of height limit facilities, and saves some vehicle height recognition costs.

[0065] S400: Determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it does, acquire an image of the cargo inside the target vehicle, perform convolutional feature analysis on the cargo image to obtain a volume detection result, and issue an early warning if the height limit is not met.

[0066] In this embodiment of the application, it is determined whether the height limit detection result obtained by analysis meets the height limit threshold of the target vehicle. The height limit threshold is that the height of the target vehicle is not greater than the height threshold. The height threshold is set according to the model of the target vehicle and relevant regulations, and can be set by those skilled in the art, for example, 2.8m.

[0067] If the height restriction detection result does not meet the height restriction threshold, it means that the height of the target vehicle meets the requirements, and subsequent load overload detection will be carried out. If the height restriction detection result does not meet the height restriction threshold, it means that the height of the target vehicle exceeds the height restriction, which will affect road driving safety. An early warning will be issued to the target vehicle for rectification.

[0068] In this embodiment of the application, when the height limit detection result meets the height limit threshold, the cargo inside the target vehicle is image-captured and its quality is analyzed to identify whether the load is overloaded.

[0069] Step S400 in the method provided in this application embodiment includes:

[0070] S410: Based on the overload detection data of vehicles of the same type as the target vehicle, obtain a set of sample cargo images and a set of sample volume detection results;

[0071] S420: Using the sample cargo image set and sample volume detection result set, construct a cargo volume recognition channel based on a convolutional neural network;

[0072] S430: Input the cargo image into the cargo volume recognition channel, perform convolutional feature analysis, and obtain the volume detection result.

[0073] Step S420 in the method provided in this application embodiment includes:

[0074] S421: Based on a convolutional neural network, construct the network structure of the cargo volume recognition channel, wherein the cargo volume recognition channel includes multiple convolutional layers, pooling layers, and fully connected layers;

[0075] S422 uses the sample cargo image set and sample volume detection result set to perform supervised training, verification and testing on the cargo volume recognition channel. When the accuracy meets the requirements, the constructed cargo volume recognition channel is obtained.

[0076] In this embodiment of the application, images of loaded goods are acquired based on overload detection data of vehicles of the same type as the target vehicle over a historical period to obtain a sample goods image set.

[0077] The volume of goods within the sample goods image set is detected and extracted to obtain a sample volume detection result set.

[0078] Based on convolutional neural networks in deep learning, a network structure for the cargo volume recognition channel is constructed, which includes multiple convolutional layers, pooling layers, and fully connected layers. Using a set of sample cargo images and a set of sample volume detection results as construction data, the cargo volume recognition channel is subjected to supervised training, validation, and testing. If the accuracy meets the requirements, the constructed cargo volume recognition channel is obtained.

[0079] Based on the constructed cargo volume recognition channel, the acquired cargo image of the target vehicle is input into the cargo volume recognition channel for convolutional feature analysis to obtain the volume detection result.

[0080] This application embodiment acquires cargo images from inside the target vehicle and performs cargo volume analysis, which serves as the data basis for analyzing the target vehicle's load capacity, thereby improving the efficiency of vehicle load capacity detection and analysis.

[0081] S500: Obtain information on the type of goods loaded in the target vehicle and input it into the goods database to obtain fuzzy density information;

[0082] Furthermore, in this embodiment of the application, information on the types of goods loaded in the target vehicle is obtained, for example, by sampling or inquiry, as the data basis for analyzing and calculating the load of the target vehicle.

[0083] The obtained category information is input into the constructed cargo database to obtain the fuzzy density information of the cargo corresponding to that category information, and then the vehicle load is analyzed and calculated.

[0084] Step S500 in the method provided in this application embodiment includes:

[0085] S510: Obtain information on various types of goods based on cargo inspection data from a historical period;

[0086] S520: Perform density detection on the goods corresponding to the various types of goods information to obtain multiple density information sets;

[0087] S530: Calculate the mean of the multiple density information sets to obtain multiple fuzzy density information;

[0088] S540: Construct a mapping relationship between the multiple cargo type information and the multiple fuzzy density information to obtain the cargo database;

[0089] S550: Input the type information of the cargo loaded in the target vehicle into the cargo database for mapping and association to obtain the fuzzy density information.

[0090] In this embodiment of the application, based on cargo detection data of vehicles carrying goods over a historical period, information on multiple types of cargo loaded is obtained, and density detection is performed on the cargo corresponding to the multiple types of cargo information to obtain multiple density information sets of multiple types of cargo information.

[0091] The mean of this density information set is calculated to obtain multiple fuzzy density information values. Fuzzy density information cannot accurately reflect the actual density of different batches of goods, but its accuracy can be controlled within a certain range, and it can be used for vehicle load detection.

[0092] By constructing a mapping relationship between the various cargo type information and multiple fuzzy density information, the cargo database is obtained. By inputting the cargo type information into the cargo database and traversing and retrieving it, the corresponding fuzzy density information can be obtained.

[0093] The type information of the goods in the target vehicle is input into the goods database for mapping and association to obtain the fuzzy density information of the goods in the target vehicle.

[0094] This application embodiment constructs a cargo database based on historical data of vehicle cargo density detection, enabling the analysis and acquisition of fuzzy density information for different types of cargo, serving as the data basis for analyzing whether a vehicle is overloaded.

[0095] S600: Calculate the load detection result based on the fuzzy density information and the volume detection result;

[0096] S700: Determine whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, issue a warning; if it does meet the threshold, do not issue a warning.

[0097] In this embodiment of the application, the load detection result can be obtained by calculating the product of the fuzzy density information and the volume detection result.

[0098] Based on the calculated load test results and the target vehicle's own weight, it is determined whether the target vehicle's load threshold is met. This load threshold includes a weight not exceeding the load threshold. The target vehicle's load threshold can be set by those skilled in the art based on the target vehicle's model and relevant regulations.

[0099] If the requirements are met, the target vehicle's load and height are both within acceptable limits, and no warning is issued. If the requirements are not met, the target vehicle's load is not within acceptable limits, a warning is issued, and the target vehicle is marked for further processing.

[0100] In summary, the embodiments of this application have at least the following technical effects:

[0101] This application provides a vehicle overload detection and early warning method. It involves acquiring images of the vehicle to be detected for overload, performing image semantic segmentation to obtain the vehicle region image, then performing convolutional feature analysis on the vehicle height to obtain the vehicle height, determining whether it meets the height restriction threshold, and issuing an early warning if it does not. If it does meet the threshold, it acquires images of the goods on the vehicle, performs convolutional feature analysis to obtain the volume detection result of the goods, and then retrieves fuzzy density information based on the type of goods to calculate the load detection result, determining whether it meets the load threshold, and issuing an early warning. This embodiment of the application improves the efficiency and accuracy of vehicle height recognition by acquiring vehicle image information, performing semantic segmentation, and analyzing the vehicle height. It eliminates the need for height restriction facilities to detect and limit vehicle height. Furthermore, when the vehicle height meets the requirements, it detects and acquires images of the goods on the vehicle and the type of goods, analyzing the vehicle load. This eliminates the need for weighing facilities, efficiently obtaining vehicle load while ensuring accuracy in load analysis, thus improving the efficiency of vehicle load detection and overload early warning, and reducing detection costs.

[0102] Example 2

[0103] Based on the same inventive concept as the vehicle overload detection and early warning method in the foregoing embodiments, such as Figure 4 As shown, this application provides a vehicle overload detection and early warning system. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0104] The vehicle image acquisition module 11 is used to acquire vehicle images of the target vehicle to be detected for overload.

[0105] Image segmentation module 12 is used to construct image segmentation channels and perform image semantic segmentation on the vehicle image to obtain a vehicle region image;

[0106] The height feature analysis module 13 is used to perform vehicle height convolution feature analysis on the vehicle area image to obtain the vehicle height of the target vehicle and obtain the height limit detection result.

[0107] The volume feature analysis module 14 is used to determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it meets the threshold, it acquires an image of the cargo inside the target vehicle, performs convolution feature analysis on the cargo image, and obtains the volume detection result. If it does not meet the threshold, it issues an early warning.

[0108] The density acquisition module 15 is used to acquire information on the type of goods loaded in the target vehicle and input it into the goods database to obtain fuzzy density information.

[0109] The load calculation module 16 is used to calculate the load detection result based on the fuzzy density information and the volume detection result;

[0110] The load determination module 17 is used to determine whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, an early warning is issued; if it does meet the threshold, no early warning is issued.

[0111] Furthermore, the image segmentation module 12 is also used to perform the following steps:

[0112] Based on the detection data of vehicles that underwent overload detection over a historical period, a set of sample vehicle images was obtained;

[0113] The vehicle region images within the sample vehicle image set are segmented and labeled to obtain the sample vehicle region image set.

[0114] The image segmentation channel is constructed using the sample vehicle image set and the sample vehicle region image set;

[0115] The vehicle image is input into the image segmentation channel to obtain the vehicle region image.

[0116] The image segmentation channel is constructed using the sample vehicle image set and the sample vehicle region image set, including:

[0117] Based on a fully convolutional neural network within semantic segmentation, an encoder and decoder are constructed within the image segmentation channel;

[0118] The encoder and decoder are trained under supervised supervision using the sample vehicle image set and the sample vehicle region image set until the convergence condition is met.

[0119] The encoder and decoder are verified. If the accuracy meets the accuracy requirements, the constructed image segmentation channel is obtained.

[0120] Furthermore, the height feature analysis module 13 is also used to perform the following steps:

[0121] The vehicle heights within multiple sample vehicle region images in the sample vehicle region image set are detected and labeled to obtain a sample vehicle height set.

[0122] Based on a convolutional neural network, a vehicle height recognition channel is constructed using the sample vehicle region image set and the sample vehicle height set as construction data.

[0123] The vehicle area image is input into the vehicle height recognition channel to obtain the vehicle height of the target vehicle, which is used as the height restriction detection result.

[0124] Furthermore, the volume feature analysis module 14 is also used to perform the following steps:

[0125] Based on the overload detection data of vehicles of the same type as the target vehicle, obtain a set of sample cargo images and a set of sample volume detection results;

[0126] Using the sample cargo image set and sample volume detection result set, a cargo volume recognition channel is constructed based on a convolutional neural network;

[0127] The cargo image is input into the cargo volume recognition channel, and convolutional feature analysis is performed to obtain the volume detection result.

[0128] Specifically, using the sample cargo image set and sample volume detection result set, a cargo volume recognition channel is constructed based on a convolutional neural network, including:

[0129] Based on a convolutional neural network, a network structure for the cargo volume recognition channel is constructed, wherein the cargo volume recognition channel includes multiple convolutional layers, pooling layers, and fully connected layers;

[0130] Using the sample cargo image set and sample volume detection result set, the cargo volume recognition channel is trained, verified, and tested under supervision. When the accuracy meets the requirements, the constructed cargo volume recognition channel is obtained.

[0131] Furthermore, the density acquisition module 15 is also used to perform the following steps:

[0132] Based on cargo inspection data over a historical period, information on various cargo types is obtained;

[0133] Density detection is performed on the goods corresponding to the various types of goods information to obtain multiple density information sets;

[0134] Calculate the mean of the multiple density information sets to obtain multiple fuzzy density information;

[0135] Construct a mapping relationship between the various cargo type information and the multiple fuzzy density information to obtain the cargo database;

[0136] The type of cargo loaded in the target vehicle is input into the cargo database for mapping and association to obtain the fuzzy density information.

[0137] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0138] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0139] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for detecting and warning of vehicle overload, characterized in that, The method includes: Acquire vehicle images of the target vehicle to be detected for overload; An image segmentation channel is constructed, and the vehicle image is subjected to image semantic segmentation to obtain a vehicle region image; Perform vehicle height convolutional feature analysis on the vehicle region image to obtain the vehicle height of the target vehicle and obtain the height restriction detection result; Determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it does, acquire an image of the cargo inside the target vehicle, perform convolutional feature analysis on the cargo image to obtain the volume detection result, and issue an early warning if the height limit is not met. Obtain information on the types of goods loaded in the target vehicle and input it into the cargo database to obtain fuzzy density information; The load detection result is calculated based on the fuzzy density information and the volume detection result; Determine whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, issue an early warning; if it does meet the threshold, do not issue an early warning. The process of obtaining information on the types of goods loaded in the target vehicle and inputting this information into a cargo database to obtain fuzzy density information includes: Based on cargo inspection data over a historical period, information on various cargo types is obtained; Density detection is performed on the goods corresponding to the various types of goods information to obtain multiple density information sets; Calculate the mean of the multiple density information sets to obtain multiple fuzzy density information; Construct a mapping relationship between the various cargo type information and the multiple fuzzy density information to obtain the cargo database; The type of cargo loaded in the target vehicle is input into the cargo database for mapping and association to obtain the fuzzy density information.

2. The method according to claim 1, characterized in that, Constructing an image segmentation channel and performing image semantic segmentation on the vehicle image to obtain a vehicle region image includes: Based on the detection data of vehicles that underwent overload detection over a historical period, a set of sample vehicle images was obtained; The vehicle region images within the sample vehicle image set are segmented and labeled to obtain the sample vehicle region image set. The image segmentation channel is constructed using the sample vehicle image set and the sample vehicle region image set; The vehicle image is input into the image segmentation channel to obtain the vehicle region image.

3. The method according to claim 2, characterized in that, Using the sample vehicle image set and the sample vehicle region image set, the image segmentation channel is constructed, including: Based on a fully convolutional neural network within semantic segmentation, an encoder and decoder are constructed within the image segmentation channel; The encoder and decoder are trained under supervised supervision using the sample vehicle image set and the sample vehicle region image set until the convergence condition is met. The encoder and decoder are verified. If the accuracy meets the accuracy requirements, the constructed image segmentation channel is obtained.

4. The method according to claim 2, characterized in that, Perform convolutional feature analysis on the vehicle region image to obtain the vehicle height of the target vehicle and obtain the height restriction detection result, including: The vehicle heights within multiple sample vehicle region images in the sample vehicle region image set are detected and labeled to obtain a sample vehicle height set. Based on a convolutional neural network, a vehicle height recognition channel is constructed using the sample vehicle region image set and the sample vehicle height set as construction data. The vehicle area image is input into the vehicle height recognition channel to obtain the vehicle height of the target vehicle, which is used as the height restriction detection result.

5. The method according to claim 1, characterized in that, The cargo images inside the target vehicle are acquired, and convolutional feature analysis is performed on the cargo images to obtain volume detection results, including: Based on the overload detection data of vehicles of the same type as the target vehicle, obtain a set of sample cargo images and a set of sample volume detection results; Using the sample cargo image set and sample volume detection result set, a cargo volume recognition channel is constructed based on a convolutional neural network; The cargo image is input into the cargo volume recognition channel, and convolutional feature analysis is performed to obtain the volume detection result.

6. The method according to claim 5, characterized in that, Using the sample cargo image set and sample volume detection result set, a cargo volume recognition channel is constructed based on a convolutional neural network, including: Based on a convolutional neural network, a network structure for the cargo volume recognition channel is constructed, wherein the cargo volume recognition channel includes multiple convolutional layers, pooling layers, and fully connected layers; Using the sample cargo image set and sample volume detection result set, the cargo volume recognition channel is trained, verified, and tested under supervision. When the accuracy meets the requirements, the constructed cargo volume recognition channel is obtained.

7. A vehicle overload detection and early warning system, characterized in that, The system is used to perform the method according to any one of claims 1 to 6, the system comprising: The vehicle image acquisition module is used to acquire vehicle images of the target vehicle to be detected for overload. The image segmentation module is used to construct an image segmentation channel and perform image semantic segmentation on the vehicle image to obtain a vehicle region image; The height feature analysis module is used to perform vehicle height convolution feature analysis on the vehicle area image to obtain the vehicle height of the target vehicle and obtain the height limit detection result. The volume feature analysis module is used to determine whether the height limit detection result meets the height limit threshold of the target vehicle. If it does, it acquires an image of the cargo inside the target vehicle, performs convolutional feature analysis on the cargo image to obtain the volume detection result, and issues an early warning if it does not meet the threshold. The density acquisition module is used to acquire information on the types of goods loaded in the target vehicle, input the information into the goods database, and obtain fuzzy density information. The load calculation module is used to calculate the load detection result based on the fuzzy density information and the volume detection result; The load determination module is used to determine whether the load detection result meets the load threshold of the target vehicle. If it does not meet the threshold, an early warning is issued; if it does meet the threshold, no early warning is issued.

Images