A method, device, storage medium and electronic device for detecting hazardous chemicals vehicle

Abstract

The embodiments of the present invention provide a method, device, storage medium and electronic device for detecting hazardous chemical vehicles. The method for detecting hazardous chemical vehicles includes: obtaining a side view picture of the target vehicle, detecting the side view picture of the target vehicle, obtaining a target feature point group, wherein the target feature point group includes identification feature points and axle feature points, obtaining content attributes of the target feature point group based on the target feature point group, and confirming the category attributes of the target vehicle according to the content attributes of the target feature point group. In the scheme of the embodiments of the present invention, key point detection and target detection are combined, and in the identification of hazardous chemical vehicles, the functions of intelligent traffic monitoring and vehicle type recognition are taken into account at the same time, which is conducive to assisting staff in related work of highway toll collection.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer technology, and in particular, to a method, device, storage medium and electronic device for detecting hazardous chemicals vehicles. Background Art

[0002] Hazardous chemicals are chemicals that are flammable, explosive, toxic, harmful, and radioactive, and are prone to cause casualties and property damage during transportation, loading and unloading, and storage and custody, and therefore require special protection. Hazardous chemicals play an important role in industrial production, scientific and technological development, and people's daily lives. Due to people's growing material and cultural needs, the corresponding demand for hazardous chemicals is also increasing.

[0003] Due to the danger of hazardous chemicals, the country has formulated very strict standards for the production, transportation and use of hazardous chemicals, which has greatly limited the occurrence of accidents. However, accidents caused by hazardous chemicals are difficult to avoid, and many of them occur during transportation.

[0004] Therefore, how to effectively identify hazardous chemical vehicles during transportation, and further monitor the safe transportation of hazardous chemical vehicles, is an important issue that needs to be solved in smart transportation. Summary of the invention

[0005] In view of this, embodiments of the present invention provide a hazardous chemicals vehicle detection method, device, storage medium and electronic device to at least partially solve the above problems.

[0006] According to a first aspect of an embodiment of the present invention, a method for detecting a hazardous chemicals vehicle is provided, comprising:

[0007] Get the side image of the target vehicle;

[0008] Detecting the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point;

[0009] Based on the target feature point group, obtaining content attributes of the target feature point group;

[0010] According to the content attributes of the target feature point group, the category attribute of the target vehicle is confirmed.

[0011] In some other examples, the side image of the target vehicle is detected to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point, including:

[0012] The target vehicle side image is detected by using a pre-trained HRNet model to obtain a target feature point group.

[0013] In some other examples, obtaining content attributes of the target feature point group based on the target feature point group further includes:

[0014] Based on the coordinates of the identified feature points, determining a target frame of the identified feature points;

[0015] According to the target frame of the identified feature point, the content attribute of the identified feature point is identified, and the number of the axle feature points is used as the content attribute of the axle feature point.

[0016] In some other examples, determining a target frame of the identified feature point based on the coordinates of the identified feature point includes:

[0017] Taking the coordinates of the identified feature point as the center point;

[0018] Target offset processing is performed based on the center point to obtain four coordinate points of the target frame to determine the target frame of the identified feature point.

[0019] In some other examples, the side image of the target vehicle is detected to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point, and further includes:

[0020] After the identification feature points are identified, the hazardous chemicals vehicle warning information is reported.

[0021] According to a second aspect of an embodiment of the present invention, a hazardous chemicals vehicle detection device is provided, comprising:

[0022] An acquisition module is used to acquire a side image of a target vehicle;

[0023] A detection module, used to detect the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point;

[0024] A content attribute confirmation module, used for obtaining the content attribute of the target feature point group based on the target feature point group;

[0025] The category attribute confirmation module is used to confirm the category attribute of the target vehicle according to the content attribute of the target feature point group.

[0026] In some other examples, the detection module further includes:

[0027] Used to detect the side image of the target vehicle through a pre-trained HRNet model to obtain a target feature point group.

[0028] In some other examples, the content attribute confirmation module further includes:

[0029] A target frame determining unit, configured to determine a target frame of the identified feature point based on the coordinates of the identified feature point;

[0030] The content attribute confirmation unit is used to identify the content attribute of the identification feature point according to the target frame of the identification feature point, and use the number of the axle feature points as the content attribute of the axle feature point.

[0031] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein a computer program is stored in the computer-readable storage medium, wherein the computer program is configured to execute the steps of any of the above method embodiments when run.

[0032] According to a fourth aspect of an embodiment of the present invention, there is further provided an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0033] In the scheme of the embodiment of the present invention, a side image of the target vehicle is obtained, and the side image of the target vehicle is detected to obtain a target feature point group, wherein the target feature point group includes identification feature points and axle feature points. Based on the target feature point group, content attributes of the target feature point group are obtained, and according to the content attributes of the target feature point group, the category attributes of the target vehicle are confirmed. In the scheme of the embodiment of the present invention, key point detection and target detection are combined. In the identification of hazardous chemicals vehicles, the functions of intelligent traffic monitoring and vehicle type identification are taken into account at the same time, which is conducive to assisting staff in related work of highway toll collection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0035] Figure 1 It is a hardware structure block diagram of a mobile terminal of a hazardous chemicals vehicle detection method according to an embodiment of the present invention;

[0036] Figure 2 is a flow chart of a method for detecting a hazardous chemicals vehicle according to an embodiment of the present invention;

[0037] Figure 3 is a heat map of feature points marked according to an embodiment of the present invention;

[0038] Figure 4It is a structural block diagram of a hazardous chemicals vehicle detection device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in the field based on the embodiments in the embodiments of the present invention should fall within the scope of protection of the embodiments of the present invention.

[0040] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and in combination with the embodiments.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0042] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for detecting hazardous chemicals vehicles according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0043] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a method for detecting hazardous chemicals vehicles in an embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0044] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0045] In order to better solve the problems raised in the above background technology, the present invention discloses a hazardous chemicals vehicle detection method, device, storage medium and electronic device, which will be described in detail one by one in the following embodiments.

[0046] See also Figure 2 , Figure 2 A flowchart of a hazardous chemicals vehicle detection method provided according to an embodiment of this specification is shown, which specifically includes the following steps:

[0047] S202, obtaining a side view image of the target vehicle;

[0048] According to an embodiment of the present invention, the target vehicle side image is a frame image obtained from a video stream shot by a traffic image recording device (such as a video camera or a camera) or a picture obtained by taking a photo.

[0049] In this embodiment, the target vehicle may be any type of vehicle, such as a hazardous chemicals vehicle, a truck, a passenger car, etc.

[0050] S204, detecting the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point;

[0051] According to an embodiment of the present invention, step S204 further includes: detecting the side image of the target vehicle by using a pre-trained target detection model, and determining the identification feature points and the axle feature points respectively.

[0052] In one embodiment, the target vehicle side image is detected by a pre-trained target detection model, and the identification feature points and the axle feature points are determined respectively, so that a heat map identifying the identification feature points and the axle feature points can be obtained, such as Figure 3 As shown, Figure 3 The left half is a side view of the target vehicle, and the right half is a heat map that identifies the identification feature points and the axle feature points. According to the heat map, the coordinates of the identification feature points and the axle feature points can be obtained.

[0053] In one embodiment, the side image of the target vehicle is detected by a pre-trained target detection model, and the coordinates of the identification feature points and the axle feature points can be determined respectively.

[0054] According to an embodiment of the present invention, preferably, the target detection model includes a pre-trained HRNet neural network model.

[0055] HRNet neural network, i.e. high-resolution network, is able to maintain high-resolution representation throughout the process. Starting from the high-resolution subnet as the first stage, the high-resolution to low-resolution subnets are gradually added to form more stages, and the multi-resolution subnets are connected in parallel. Throughout the process, multi-scale repeated fusion is performed by repeatedly exchanging information on parallel multi-resolution subnets, so that key points are estimated through the high-resolution representation output by the network. HRNet neural network is widely used for two-dimensional human posture estimation, but less used in the transportation field, especially for the identification of hazardous chemical vehicles.

[0056] It can be known that hazardous chemical vehicles have special hazardous chemical signs on the top of the front of the vehicle and on the side of the vehicle. In the embodiment provided by the present invention, this special sign of the hazardous chemical vehicle is used as a key point for identification. Preferably, the special sign of the hazardous chemical vehicle can be detected by training the HRNet neural network.

[0057] In addition, as a type of truck, hazardous chemical trucks need to be charged by vehicle type during highway toll collection. By determining the axle information, the fees that need to be collected for the vehicle can be obtained, and it can assist traffic control personnel in determining whether the vehicle has evaded highway tolls by using methods such as large vehicles with small labels. It is worth mentioning that the existing vehicle type detection methods cannot correctly identify the specific type of vehicle due to insufficient image brightness or other interference. Therefore, accurately determining the number of axles is critical for truck model identification. In the embodiments provided by the present invention, the axles of hazardous chemical trucks are used as key points for identification. Preferably, by training the HRNet neural network, the axle information of hazardous chemical trucks including the number of axles can be detected.

[0058] According to some embodiments of the present invention, during the pre-training of the target detection model, firstly, sample images of the side of a hazardous chemical vehicle are collected, and the collected multiple sample images are marked, that is, the hazardous chemical identification label and axle label corresponding to the target vehicle in each sample image are marked, wherein the hazardous chemical identification label includes 10 categories of hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items, and the axle label is the specific number of axles. In some embodiments, multiple sample images that have been marked with attribute labels can also be directly obtained; then the labeled sample data set is divided into training set data, validation set data, and test set data according to a certain ratio; then the training set data is input into the HRNet network to obtain a target detection model of hazardous chemical identification features and axle features; the validation set data is used to verify the accuracy of the target detection model; the test set data is input into the trained model to correct the parameters of the target detection model. In the pre-training process, MSE is used as the loss function for training the target detection model.

[0059] In the pre-training process, the training set data is input into the HRNet network to obtain the target detection model of hazardous chemicals identification features and axle features, including:

[0060] First, the training set images are convolved through two convolution kernels with kernel 3, stride 2, and padding 1, reducing the image resolution to 1 / 4 of the original resolution.

[0061] Then, using the parallel convolution stream method, starting from the first stage, the streams from high resolution to low resolution are added one by one. The resolution of the parallel stream in the latter stage is composed of the resolution of the previous stage and the lower resolution. There are 4 stages in total, and each resolution in each stage undergoes 4 residual convolutions. In one stage, a 3x3 convolution operation is used to go from high resolution to low resolution. The smaller the resolution, the more channels there are, and it grows exponentially by 2.

[0062] Repeated fusion of multi-resolution modules to exchange information across resolutions as features for extraction;

[0063] The resulting four-resolution stream is used as the feature head output identifier, and the high resolution is output for key point detection;

[0064] MSE is used as the loss function for training and the final model is output.

[0065] It is worth mentioning that for image processing, the whole image is needed to use convolution for general input data. For general convolution layer to feature, serial stream operation is adopted, and the computing cost is acceptable. However, since the network needs to adopt parallel stream calculation to obtain high resolution, if the original image is used directly, the computing resources will be huge. Therefore, the original image is first downsampled to reduce the resolution, that is, the image resolution is reduced to 1 / 4 of the original, which can reduce the computing loss and improve the computing speed.

[0066] It is worth mentioning that most existing methods recover high-resolution representations from low-resolution representations produced by high-resolution to low-resolution networks. The classification network composed of high-to-low convolutions in series is not a good choice for regional and pixel-level classification, because it only leads to rich low-resolution representations or low-resolution representations obtained through upsampling processes. HRNet maintains high-resolution representations by connecting high-resolution to low-resolution convolutions in parallel, and enhances high-resolution representations by repeatedly performing multi-scale fusion across parallel convolutions. The HRNet network can maintain high-resolution representations throughout the process.

[0067] HRNet starts with a high-resolution subnetwork as the first stage, gradually adds high-resolution to low-resolution subnetworks, forms more stages, and connects multi-resolution subnetworks in parallel. Multiple multi-scale fusions are performed in the middle, so that each high-resolution to low-resolution representation repeatedly receives information from other parallel representations, thereby obtaining rich high-resolution representations. Therefore, the predicted key point heat map will also be more accurate and more precise in space. Based on this, we use the hazardous chemicals identification as the key point for detection, and add the axle part as the key point. The types of hazardous chemicals are divided into 10 categories: hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items. Among them, the axle part obtains the specific number of axles, which is used to assist in vehicle type classification and help highway ETC reduce the phenomenon of under-collection. Under the highway charging policy based on vehicle type, many vehicles have inconsistent vehicle types and actual vehicle types when applying for ETC, resulting in large vehicles with small labels, causing road companies to charge according to small vehicle types and incurring toll losses. Therefore, the hazardous chemicals vehicle detection method provided by the present invention can not only effectively detect the types of hazardous chemicals vehicles, but also realize the identification of vehicle types, which is beneficial for traffic control personnel to monitor highway toll evasion.

[0068] S206, obtaining content attributes of the target feature point group based on the target feature point group;

[0069] According to an embodiment of the present invention, step S206 includes:

[0070] S2062, determining a target frame of the identified feature point based on the coordinates of the identified feature point;

[0071] S2064: Identify content attributes of the identified feature points according to the target frame of the identified feature points, and use the number of axle feature points as the content attributes of the axle feature points.

[0072] According to an embodiment of the present invention, step S2062 includes:

[0073] S20622, taking the coordinates of the identified feature point as the center point;

[0074] S20624, performing target offset processing based on the center point to obtain four coordinate points of the target frame to determine the target frame of the identified feature point.

[0075] Optionally, the coordinates of the identification feature points and the axle feature points may be obtained according to a heat map identifying the identification feature points and the axle feature points.

[0076] Optionally, target offset value processing is performed based on the center point to output four coordinate information relative to the center point, namely four vertex coordinates, and the four vertex coordinates are connected to form a target frame for identifying the feature point.

[0077] According to an embodiment of the present invention, the content attribute of the identification feature point is a specific content data encoding of the identification feature graph, and the content attribute of the axle feature point is the number of axles.

[0078] It is worth mentioning that the traditional HRNet model is only used for key point detection for human posture estimation, while in the embodiment provided by the present invention, the HRNet model can not only be used to identify key points, but also to perform target frame detection based on the key points to identify the specific content of the key points. Among them, according to the identification feature points, the target frame can be further obtained for target detection, and the axle feature points only need to obtain the number of feature points, that is, the number of axles, to assist in vehicle model identification. The hazardous chemicals vehicle detection method provided by the present invention effectively combines HRNet key point detection and target detection. In the identification of hazardous chemicals vehicles, it takes into account the functions of intelligent traffic monitoring and vehicle model recognition, which is conducive to assisting staff in related work of highway toll collection.

[0079] S208: confirming the category attribute of the target vehicle according to the content attribute of the target feature point group.

[0080] According to an embodiment of the present invention, the category attributes of the target vehicle include: a specific type and model of a hazardous chemicals vehicle, such as a flammable goods vehicle-type 5.

[0081] Among them, in this embodiment, the specific types of hazardous chemical vehicles are divided into 10 categories, including hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items. In other embodiments, specific settings can also be made according to specific needs.

[0082] Among them, in this embodiment, the specific models of hazardous chemicals vehicles are divided into Type 1, Type 2, Type 3, Type 4, Type 5, and Type 6 (in this embodiment, Type 6 and above are uniformly treated as Type 6). In other embodiments, specific settings can also be made according to specific needs.

[0083] According to another embodiment of the present invention, step S204 detects the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes identification feature points and axle feature points, and further includes a step of reporting hazardous chemicals vehicle warning information after the identification feature points are identified.

[0084] It is worth mentioning that there are strict regulations on the transportation of hazardous chemical vehicles. Some roads prohibit the passage of hazardous chemical vehicles, but there are still phenomena of hazardous chemical vehicles passing through without authorization. The embodiments provided by the present invention can effectively monitor the operation of hazardous chemical vehicles by identifying the key identification feature points of hazardous chemical vehicles. When hazardous chemical vehicles pass through prohibited transportation sections and violate transportation regulations, the warning information of hazardous chemical vehicles is reported in time so that traffic management personnel can obtain relevant information in time to take management measures.

[0085] The method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, or by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0086] In this embodiment, a hazardous chemical vehicle detection device is also provided, which is used to implement the corresponding hazardous chemical vehicle detection methods in the aforementioned multiple method embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here. In addition, the functional implementation of each module of the data processing device of this embodiment can refer to the description of the corresponding parts in the aforementioned method embodiments, which will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0087] According to another embodiment of the present invention, Figure 4 , providing a hazardous chemicals vehicle detection device, comprising:

[0088] An acquisition module 30 is used to acquire a side image of a target vehicle;

[0089] A detection module 40 is used to detect the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point;

[0090] According to an embodiment of the present invention, the detection module 40 further includes: a module for detecting the side image of the target vehicle by using a pre-trained target detection model, and determining the identification feature points and the axle feature points respectively.

[0091] In one embodiment, the target vehicle side image is detected by a pre-trained target detection model, and the identification feature points and the axle feature points are determined respectively, so that a heat map identifying the identification feature points and the axle feature points can be obtained, such as Figure 3 As shown, Figure 3 The left half is a side view of the target vehicle, and the right half is a heat map that identifies the identification feature points and the axle feature points. According to the heat map, the coordinates of the identification feature points and the axle feature points can be obtained.

[0092] In one embodiment, the side image of the target vehicle is detected by a pre-trained target detection model, and the coordinates of the identification feature points and the axle feature points can be determined respectively.

[0093] According to an embodiment of the present invention, preferably, the target detection model includes a pre-trained HRNet neural network model.

[0094] According to some embodiments of the present invention, during the pre-training of the target detection model, firstly, sample images of the side of a hazardous chemical vehicle are collected, and the collected multiple sample images are marked, that is, the hazardous chemical identification label and axle label corresponding to the target vehicle in each sample image are marked, wherein the hazardous chemical identification label includes 10 categories of hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items, and the axle label is the specific number of axles. In some embodiments, multiple sample images that have been marked with attribute labels can also be directly obtained; then the labeled sample data set is divided into training set data, validation set data, and test set data according to a certain ratio; then the training set data is input into the HRNet network to obtain a target detection model of hazardous chemical identification features and axle features; the validation set data is used to verify the accuracy of the target detection model; the test set data is input into the trained model to correct the parameters of the target detection model. In the pre-training process, MSE is used as the loss function for training the target detection model.

[0095] In the pre-training process, the training set data is input into the HRNet network to obtain the target detection model of hazardous chemicals identification features and axle features, including:

[0096] First, the training set images are convolved through two convolution kernels with kernel 3, stride 2, and padding 1, reducing the image resolution to 1 / 4 of the original resolution.

[0097] Then, using the parallel convolution stream method, starting from the first stage, the streams from high resolution to low resolution are added one by one. The resolution of the parallel stream in the latter stage is composed of the resolution of the previous stage and the lower resolution. There are 4 stages in total, and each resolution in each stage undergoes 4 residual convolutions. In one stage, a 3x3 convolution operation is used to go from high resolution to low resolution. The smaller the resolution, the more channels there are, and it grows exponentially by 2.

[0098] Repeated fusion of multi-resolution modules to exchange information across resolutions as features for extraction;

[0099] The resulting four-resolution stream is used as the feature head output identifier, and the high resolution is output for key point detection;

[0100] MSE is used as the loss function for training and the final model is output.

[0101] It is worth mentioning that for image processing, the whole image is needed to use convolution for general input data. For general convolution layer to feature, serial stream operation is adopted, and the computing cost is acceptable. However, since the network needs to adopt parallel stream calculation to obtain high resolution, if the original image is used directly, the computing resources will be huge. Therefore, the original image is first downsampled to reduce the resolution, that is, the image resolution is reduced to 1 / 4 of the original, which can reduce the computing loss and improve the computing speed.

[0102] It is worth mentioning that most existing methods recover high-resolution representations from low-resolution representations produced by high-resolution to low-resolution networks. The classification network composed of high-to-low convolutions in series is not a good choice for regional and pixel-level classification, because it only leads to rich low-resolution representations or low-resolution representations obtained through upsampling processes. HRNet maintains high-resolution representations by connecting high-resolution to low-resolution convolutions in parallel, and enhances high-resolution representations by repeatedly performing multi-scale fusion across parallel convolutions. The HRNet network can maintain high-resolution representations throughout the process.

[0103] HRNet starts with a high-resolution subnetwork as the first stage, gradually adds high-resolution to low-resolution subnetworks, forms more stages, and connects multi-resolution subnetworks in parallel. Multiple multi-scale fusions are performed in the middle, so that each high-resolution to low-resolution representation repeatedly receives information from other parallel representations, thereby obtaining rich high-resolution representations. Therefore, the predicted key point heat map will also be more accurate and more precise in space. Based on this, we use the hazardous chemicals identification as the key point for detection, and add the axle part as the key point. The types of hazardous chemicals are divided into 10 categories: hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items. Among them, the axle part obtains the specific number of axles, which is used to assist in vehicle type classification and help highway ETC reduce the phenomenon of under-collection. Under the highway charging policy based on vehicle type, many vehicles have inconsistent vehicle types and actual vehicle types when applying for ETC, resulting in large vehicles with small labels, causing road companies to charge according to small vehicle types and incurring toll losses. Therefore, the hazardous chemicals vehicle detection device provided by the present invention can not only effectively detect the types of hazardous chemicals vehicles, but also realize the identification of vehicle types, which is beneficial for traffic control personnel to monitor highway toll evasion.

[0104] A content attribute confirmation module 50, configured to obtain content attributes of the target feature point group based on the target feature point group;

[0105] According to an embodiment of the present invention, the content attribute confirmation module 50 further includes:

[0106] A target frame determining unit 51 is used to determine a target frame of the identified feature point based on the coordinates of the identified feature point;

[0107] The content attribute confirmation unit 52 is used to identify the content attribute of the identification feature point according to the target frame of the identification feature point, and use the number of the axle feature points as the content attribute of the axle feature point.

[0108] According to an embodiment of the present invention, the target frame determination unit 51 further includes:

[0109] A center point confirmation unit 511 is used to take the coordinates of the identification feature point as the center point;

[0110] The target frame construction unit 512 performs target offset processing based on the center point to obtain four coordinate points of the target frame to determine the target frame of the identified feature point.

[0111] Optionally, the coordinates of the identification feature points and the axle feature points may be obtained according to a heat map identifying the identification feature points and the axle feature points.

[0112] Optionally, target offset value processing is performed based on the center point to output four coordinate information relative to the center point, namely four vertex coordinates, and the four vertex coordinates are connected to form a target frame for identifying the feature point.

[0113] According to an embodiment of the present invention, the content attribute of the identification feature point is a specific content data encoding of the identification feature graph, and the content attribute of the axle feature point is the number of axles.

[0114] It is worth mentioning that the traditional HRNet model is only used for key point detection for human posture estimation, while in the embodiment provided by the present invention, the HRNet model can not only be used to identify key points, but also to perform target frame detection based on the key points to identify the specific content of the key points. Among them, according to the identification feature points, the target frame can be further obtained for target detection, and the axle feature points only need to obtain the number of feature points, that is, the number of axles, to assist in vehicle type identification. The hazardous chemicals vehicle detection device provided by the present invention effectively combines HRNet key point detection and target detection. In the identification of hazardous chemicals vehicles, it takes into account the functions of intelligent traffic monitoring and vehicle type recognition, which is conducive to assisting staff in related work of highway toll collection.

[0115] The category attribute confirmation module 60 is used to confirm the category attribute of the target vehicle according to the content attribute of the target feature point group.

[0116] According to an embodiment of the present invention, the category attributes of the target vehicle include: a specific type and model of a hazardous chemicals vehicle, such as a flammable goods vehicle-type 5.

[0117] Among them, in this embodiment, the specific types of hazardous chemical vehicles are divided into 10 categories, including hazard lights, flammable items, explosive items, inert gases, oxides, toxic objects, infectious substances, radioactive substances, corrosive substances, and miscellaneous items. In other embodiments, specific settings can also be made according to specific needs.

[0118] Among them, in this embodiment, the specific models of hazardous chemicals vehicles are divided into Type 1, Type 2, Type 3, Type 4, Type 5, and Type 6 (in this embodiment, Type 6 and above are uniformly treated as Type 6). In other embodiments, specific settings can also be made according to specific needs.

[0119] According to another embodiment of the present invention, the hazardous chemicals vehicle detection device further includes an early warning module 70 for reporting early warning information of the hazardous chemicals vehicle after the identification feature point is identified.

[0120] It is worth mentioning that there are strict regulations on the transportation of hazardous chemical vehicles. Some roads prohibit the passage of hazardous chemical vehicles, but there are still phenomena of hazardous chemical vehicles passing through without authorization. The embodiments provided by the present invention can effectively monitor the operation of hazardous chemical vehicles by identifying the key identification feature points of hazardous chemical vehicles. When hazardous chemical vehicles pass through prohibited transportation sections and violate transportation regulations, the warning information of hazardous chemical vehicles is reported in time so that traffic management personnel can obtain relevant information in time to take management measures.

[0121] An embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0122] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0123] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0124] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0125] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0126] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0127] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting hazardous chemicals vehicles, characterized in that: include: Get the side image of the target vehicle; Detecting the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point; Based on the target feature point group, obtaining content attributes of the target feature point group; Determining the category attribute of the target vehicle according to the content attribute of the target feature point group; The method also includes: detecting the side image of the target vehicle through a pre-trained HRNet model, respectively determining the identification feature points and the axle feature points, obtaining a heat map identifying the identification feature points and the axle feature points, and determining the coordinates of the identification feature points and the axle feature points according to the heat map; determining a target frame of the identification feature points based on the coordinates of the identification feature points; identifying content attributes of the identification feature points according to the target frame of the identification feature points, and using the number of axle feature points as the content attributes of the axle feature points, wherein the pre-trained HRNet model is used to determine the target frame according to the coordinates of the identification feature points.

2. The method according to claim 1, characterized in that Detect the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point, including: The target vehicle side image is detected by the pre-trained HRNet model to obtain a target feature point group.

3. The method according to claim 1, characterized in that Determining a target frame of the identified feature point based on the coordinates of the identified feature point includes: Taking the coordinates of the identified feature point as the center point; Target offset processing is performed based on the center point to obtain four coordinate points of the target frame to determine the target frame of the identified feature point.

4. The method according to claim 1, characterized in that Detect the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point, and further includes: After the identification feature points are identified, the hazardous chemicals vehicle warning information is reported.

5. A hazardous chemicals vehicle detection device, characterized in that: include: An acquisition module is used to acquire a side image of a target vehicle; A detection module, used to detect the side image of the target vehicle to obtain a target feature point group, wherein the target feature point group includes a logo feature point and an axle feature point; A content attribute confirmation module, used for obtaining the content attribute of the target feature point group based on the target feature point group; A category attribute confirmation module, used to confirm the category attribute of the target vehicle according to the content attribute of the target feature point group; The device is also used to: detect the side image of the target vehicle through a pre-trained HRNet model, respectively determine the identification feature points and the axle feature points, obtain a heat map identifying the identification feature points and the axle feature points, determine the coordinates of the identification feature points and the axle feature points according to the heat map; and determine the target frame of the identification feature points based on the coordinates of the identification feature points; According to the target frame of the identification feature point, the content attribute of the identification feature point is identified, and the number of axle feature points is used as the content attribute of the axle feature point, wherein the pre-trained HRNet model is used to determine the target frame according to the coordinates of the identification feature point.

6. The device according to claim 5, characterized in that The detection module also includes: Used to detect the side image of the target vehicle through the pre-trained HRNet model to obtain a target feature point group.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 4 when executed.

8. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 4.

Images