Road condition risk detection method and device, electronic equipment and storage medium

Abstract

A road condition risk detection method and device, electronic equipment and storage medium are disclosed. The road condition risk detection method comprises: collecting a road condition image of a target road condition through a target sensor; performing semantic segmentation on the input road condition image through a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image, wherein the target semantic segmentation model comprises an encoding end, a decoding end and a hollow space pyramid pooling-channel attention mechanism module; and determining a risk result of the target road condition based on the semantic segmentation image. According to the embodiment of the present application, the accuracy of road condition risk detection can be improved.

Description

Technical Field

[0001] This invention relates to the field of computer application technology, and in particular to a road condition risk detection method, device, electronic device, and storage medium. Background Technology

[0002] With the advancement of technology, intelligent vehicles can make intelligent judgments on road conditions to help them understand the road and reduce the risk of accidents. However, with the development of transportation, the diversification of transportation tools, the complexity of traffic roads, and the dispersion of pedestrians, the scenarios for intelligent vehicle recognition of road conditions are becoming increasingly complex.

[0003] In existing technologies, intelligent vehicles use sensors to collect road condition image information and use trained semantic segmentation models to classify the collected road condition images at the pixel level to obtain semantic segmentation results, thus achieving the purpose of road condition risk detection. However, existing semantic segmentation models have problems such as long training time and delayed output, i.e., poor training efficiency and semantic segmentation efficiency. In addition, road condition risk detection errors often occur, i.e., the accuracy of road condition risk detection is poor. Summary of the Invention

[0004] This invention provides a road condition risk detection method, device, electronic device, and storage medium to solve the technical problem of poor efficiency and accuracy in road condition risk detection.

[0005] According to one aspect of the present invention, a road condition risk detection method is provided, wherein the method includes:

[0006] Road condition images of the target road are acquired using target sensors;

[0007] The input road condition image is semantically segmented by a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole space pyramid pooling-channel attention mechanism module.

[0008] The risk outcome of the target road condition is determined based on the semantic segmentation image.

[0009] According to another aspect of the present invention, a road condition risk detection device is provided, wherein the device comprises:

[0010] The image acquisition module is used to acquire road condition images of the target road using the target sensor;

[0011] The semantic segmentation module is used to perform semantic segmentation on the input road condition image through a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole space pyramid pooling-channel attention mechanism module.

[0012] The risk detection module is used to determine the risk outcome of the target road condition based on the semantic segmentation image.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the road condition risk detection method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the road condition risk detection method according to any embodiment of the present invention.

[0018] The technical solution of this invention involves acquiring road condition images of a target road condition using a target sensor; performing semantic segmentation on the input road condition images using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole-space pyramid pooling-channel attention mechanism module, which improves the accuracy of the target semantic segmentation model and the accuracy of the semantically segmented image; and determining the risk result of the target road condition based on the semantically segmented image. This improves the accuracy of road condition risk detection.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] 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.

[0021] Figure 1 This is a flowchart of a road condition risk detection method provided in Embodiment 1 of the present invention;

[0022] Figure 2 This is a flowchart illustrating the workflow of implementing the target semantic segmentation model in this embodiment of the invention.

[0023] Figure 3 This is a flowchart illustrating the workflow of the void space pyramid pooling-channel attention mechanism module implemented in this embodiment of the invention.

[0024] Figure 4 This is a flowchart of a road condition risk detection method provided in Embodiment 2 of the present invention;

[0025] Figure 5 This is a schematic diagram of the structure of a road condition risk detection device according to Embodiment 3 of the present invention;

[0026] Figure 6 This is a schematic diagram of the structure of an electronic device that implements the road condition risk detection method of this invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention 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 the invention 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 a non-exclusive inclusion; for example, a process, method, system, product, or apparatus 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 units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] Figure 1This is a flowchart of a road condition risk detection method provided in Embodiment 1 of the present invention. This embodiment is applicable to image segmentation. The method can be executed by a road condition risk detection device, which can be implemented in hardware and / or software and can be configured in the target vehicle. Figure 1 As shown, the method includes:

[0031] S110. Acquire road condition images of the target road using the target sensor.

[0032] The target sensor can be understood as a sensor used to acquire the road condition image. Optionally, it is a device with image acquisition function installed on the target vehicle. In this embodiment of the invention, the target sensor can be preset according to scenario requirements, and is not specifically limited here. For example, the target sensor can be an image sensor.

[0033] The target road condition can be understood as the road condition to be detected. Optionally, the target road condition may be a road condition that may pose a risk. In this embodiment of the invention, the target road condition is related to the travel distance of the target vehicle, and is not specifically limited here.

[0034] The road condition image can be understood as the image corresponding to the target road condition.

[0035] S120. The input road condition image is semantically segmented using a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole space pyramid pooling-channel attention mechanism module.

[0036] The target semantic segmentation model can be understood as a model with semantic segmentation function.

[0037] The semantic segmentation image can be understood as performing semantic segmentation on the input road condition image using a target semantic segmentation model to obtain an image corresponding to the road condition image. Optionally, the semantic segmentation image includes semantic tags corresponding to the road condition image. The semantic tags can be tags representing the risk level of the target road condition. In this embodiment of the invention, the representation of the semantic tags can be preset according to scenario requirements and is not specifically limited here. For example, the semantic tags can be 0 or 1. 0 indicates that the target road condition corresponding to the road condition image has a risk; 1 indicates that the target road condition corresponding to the road condition image does not have a risk.

[0038] The encoding end can be understood as converting the road condition image into a low-dimensional representation to capture the model structure of the key features of the road condition image. The decoding end, the decoder, then converts this low-dimensional representation back into the road condition image space to recover the model structure of the road condition image.

[0039] The Atrous Spatial Pyramid Pooling-Channel Attention Mechanism module can be understood as a module that combines the Atrous Spatial Pyramid Pooling (ASPP) module and the Squeeze-and-Excitation Networks (SENet) mechanism. Specifically, the Atrous Spatial Pyramid Pooling-Channel Attention Mechanism module is located at the bridging part between the encoder and decoder (see reference). Figure 2 ).

[0040] In this embodiment of the invention, the hollow spatial pyramid pooling-channel attention mechanism module in the target semantic segmentation model can achieve feature extraction at different scales in deep layers, focusing on useful features and suppressing irrelevant features. This improves the accuracy of feature extraction in the feature processing of the target semantic segmentation model.

[0041] Optionally, the encoding end includes a target number of cascaded encoding layers, and the decoding end includes a target number of cascaded decoding layers matching the encoding end, wherein each encoding layer in the encoding end includes a 3×3 convolutional kernel, an activation function, and a 2×2 max-pooling downsampling layer.

[0042] The target number can be understood as the number of encoding layers in the encoding end. In this embodiment of the invention, the target number can be preset according to scenario requirements and is not specifically limited here. Optionally, the target number can be a value less than 4. For example, the target number can be 3. It should be understood that the traditional encoding end includes 4 encoding / decoding layers. Based on the target semantic segmentation model including the hollow space pyramid pooling-channel attention mechanism module, this invention can use the encoding end with 3 encoding / decoding layers to improve the efficiency of semantic segmentation. This achieves both improved feature extraction accuracy and improved semantic segmentation efficiency. It also improves the efficiency and accuracy of road condition risk detection.

[0043] Optionally, the hollow spatial pyramid pooling-channel attention mechanism module includes multiple feature processing layers and 1×1 convolutional kernels; wherein,

[0044] The feature multilayer processing layer includes a feature processing layer consisting of a 1x1 convolutional layer and a channel attention mechanism, a 3×3 convolutional layer with a first dilatation rate, a 3×3 convolutional layer with a second dilatation rate, a 3×3 convolutional layer with a third dilatation rate, and a global average pooling layer.

[0045] Figure 3 This is a flowchart illustrating the workflow of the void space pyramid pooling-channel attention mechanism module implemented in this embodiment of the invention. (Reference) Figure 3 The structure of the dilated spatial pyramid pooling-channel attention mechanism module can be as follows: The first layer is a 1x1 convolution with a channel attention mechanism to adjust channel attention, thereby enhancing useful features and suppressing useless features; the second layer is a 3x3 convolution with a dilation rate of 6; the third layer is a 3x3 convolution with a dilation rate of 12; the fourth layer is a 3x3 convolution with a dilation rate of 18; and the fifth layer is global average pooling. Further, specifically, the feature maps output from the 5-layer structure can be concatenated, and the number of channels can be adjusted using 1x1 convolutions.

[0046] The channel attention mechanism can be understood as a channel attention model that enhances the attention to important channels by adjusting the weights of each channel, thereby improving the model's performance.

[0047] The global average pooling layer can be understood as a special type of average pooling that averages all elements in the entire feature map and outputs them to the next layer.

[0048] Optionally, the road condition risk detection method further includes: before performing semantic segmentation on the input road condition image using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image,

[0049] The target semantic segmentation model is obtained by training the deep learning model using a sample image set, cross-entropy loss function, and Dice loss function.

[0050] The sample image set can be understood as a sample set used to train the deep learning model to obtain the target semantic segmentation model.

[0051] The cross-entropy loss function can be understood as a loss function applied to classification problems in information theory. The Dice loss function can be understood as a loss function related to sample similarity. The deep learning model can be understood as a learning model that learns high-order representation features of the inherent patterns in data through a large number of vector calculations and uses these features to make decisions.

[0052] In this embodiment of the invention, during the training of the deep learning model based on the sample image set, the cross-entropy loss function and the Dice loss function are combined simultaneously. This can ensure the accuracy of the trained target semantic segmentation model even when the sample image set is imbalanced, thereby reducing the adverse effects of the imbalanced sample image set on model training.

[0053] In this embodiment of the invention, the deep learning model used to train the target semantic segmentation model can be a model with the same structure as the target semantic segmentation model. Using a deep learning model with the same structure as the target semantic segmentation model, i.e., a 3-layer structure, can improve the training efficiency of the model.

[0054] S130. Determine the risk result of the target road condition based on the semantic segmentation image.

[0055] The risk result can be understood as the risk detection result of the target road condition. In this embodiment, the risk result can be determined based on the semantic segmentation image corresponding to the road condition image. Optionally, the risk result can include: presence of risk and no risk.

[0056] The technical solution of this invention involves acquiring road condition images of a target road condition using a target sensor; performing semantic segmentation on the input road condition images using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole-space pyramid pooling-channel attention mechanism module, which improves the accuracy of the target semantic segmentation model and the accuracy of the semantically segmented image; and determining the risk result of the target road condition based on the semantically segmented image. This improves the accuracy of road condition risk detection.

[0057] Example 2

[0058] Figure 4 This is a flowchart of a road condition risk detection method provided in Embodiment 2 of the present invention. This embodiment focuses on refining the semantic segmentation image obtained by performing semantic segmentation on the input road condition image using a target semantic segmentation model as described in the above embodiments. Figure 4 As shown, the method includes:

[0059] S210. Acquire road condition images of the target road using the target sensor.

[0060] In this embodiment of the invention, the process of extracting features from the input road condition image using a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image based on the extracted features can be referred to. Figure 2 and Figure 3 .like Figure 2 and Figure 3 As shown, the specific process of performing semantic segmentation on the input road condition image using a target semantic segmentation model to obtain the semantic segmentation image corresponding to the road condition image can be referred to in the following steps.

[0061] S220. The road condition image is extracted layer by layer through each encoding layer in the encoding end to obtain the first feature corresponding to each encoding layer and the second feature output by the last encoding layer.

[0062] The first feature can be understood as the feature corresponding to each coding layer obtained by extracting features from the road condition image layer by layer through each coding layer in the coding end.

[0063] The second feature can be understood as the feature output through the last coding layer.

[0064] S230. Based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end, a semantic segmentation image corresponding to the road condition image is obtained.

[0065] Optionally, obtaining the semantic segmentation image based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end includes:

[0066] The second feature is input into the void space pyramid pooling-channel attention mechanism module, and feature processing is performed on the input second feature to obtain multiple preliminary features;

[0067] Multiple preliminary features are concatenated, and the number of channels is adjusted using a 1×1 convolution kernel to obtain the target features corresponding to the road condition image;

[0068] The target features are input to the decoding end to obtain the semantic segmentation image corresponding to the road condition image.

[0069] The preliminary features can be understood as features obtained by processing the input second feature through the void space pyramid pooling-channel attention mechanism. It is understood that there can be multiple preliminary features.

[0070] The target feature can be understood as concatenating multiple preliminary features and adjusting the number of channels using a 1×1 convolution kernel to obtain the feature corresponding to the road condition image. Optionally, the target feature can be the output result determined by the hollow spatial pyramid pooling-channel attention mechanism based on the input second feature.

[0071] Optionally, inputting the target features into the decoding end to obtain the semantic segmentation image corresponding to the road condition image includes:

[0072] The target features are upsampled layer by layer by each decoding layer in the decoding end to obtain the third feature corresponding to each decoding layer;

[0073] The first and third features are stacked in corresponding layers, and the number of channels is adjusted by a 3×3 convolution kernel to obtain the semantic segmentation image output by the last decoding layer.

[0074] The third feature can be understood as the feature corresponding to each decoding layer obtained by upsampling the input target feature layer by layer through each decoding layer in the decoding end.

[0075] The convolution kernel can be understood as a weighted summation of a certain local area. In this embodiment of the invention, the number of channels can be adjusted using the convolution kernel.

[0076] S240. Determine the risk result of the target road condition based on the semantic segmentation image.

[0077] The technical solution of this invention involves extracting features from the road condition image layer by layer through each encoding layer in the encoding end, obtaining a first feature corresponding to each encoding layer, and a second feature output by the last encoding layer. Based on the second feature, the hollow spatial pyramid pooling-channel attention mechanism module, and the decoding end, a semantic segmentation image corresponding to the road condition image is obtained. This method can enhance useful features and suppress irrelevant features while extracting deep multi-scale features, ensuring the accuracy of feature extraction and thus improving the accuracy of the determined semantic segmentation image.

[0078] Example 3

[0079] Figure 5 This is a schematic diagram of a road condition risk detection device provided in Embodiment 3 of the present invention. Figure 5 As shown, the device includes: an image acquisition module 310, a semantic segmentation module 320, and a risk detection module 330; wherein,

[0080] Image acquisition module 310 is used to acquire road condition images of the target road condition through a target sensor; semantic segmentation module 320 is used to perform semantic segmentation on the input road condition image through a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image, wherein the target semantic segmentation model includes an encoding end, a decoding end, and a hole space pyramid pooling-channel attention mechanism module; risk detection module 330 is used to determine the risk result of the target road condition based on the semantic segmentation image.

[0081] The technical solution of this invention involves acquiring road condition images of a target road condition using a target sensor; performing semantic segmentation on the input road condition images using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole-space pyramid pooling-channel attention mechanism module, which improves the accuracy of the target semantic segmentation model and the accuracy of the semantically segmented image; and determining the risk result of the target road condition based on the semantically segmented image. This improves the accuracy of road condition risk detection.

[0082] Optionally, the encoding end includes a target number of cascaded encoding layers, and the decoding end includes a target number of cascaded decoding layers matching the encoding end, wherein each encoding layer in the encoding end includes a 3×3 convolutional kernel, an activation function, and a 2×2 max-pooling downsampling layer.

[0083] Optionally, the hollow spatial pyramid pooling-channel attention mechanism module includes multiple feature processing layers and 1×1 convolutional kernels; wherein,

[0084] The feature multilayer processing layer includes a feature processing layer consisting of a 1x1 convolutional layer and a channel attention mechanism, a 3×3 convolutional layer with a first dilatation rate, a 3×3 convolutional layer with a second dilatation rate, a 3×3 convolutional layer with a third dilatation rate, and a global average pooling layer.

[0085] Optionally, the semantic segmentation module 320 includes: a feature processing unit and a segmentation image determination unit; wherein,

[0086] The feature processing unit is used to extract features from the road condition image layer by layer through each encoding layer in the encoding end, to obtain the first feature corresponding to each encoding layer, and the second feature output through the last encoding layer.

[0087] The segmentation image determination unit is used to obtain a semantic segmentation image corresponding to the road condition image based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end.

[0088] Optionally, the image segmentation determination unit includes: a first feature processing subunit, a second feature processing subunit, and an image segmentation determination subunit; wherein,

[0089] The first feature processing subunit is used to input the second feature into the void space pyramid pooling-channel attention mechanism module, and perform feature processing on the input second feature to obtain multiple preliminary features;

[0090] The second feature processing subunit is used to stitch together multiple preliminary features and adjust the number of channels through a 1×1 convolution kernel to obtain the target features corresponding to the road condition image;

[0091] The segmentation image determination subunit is used to input the target features to the decoding end to obtain the semantic segmentation image corresponding to the road condition image.

[0092] Optionally, the segmented image determining subunit is used for:

[0093] The target features are upsampled layer by layer by each decoding layer in the decoding end to obtain the third feature corresponding to each decoding layer;

[0094] The first and third features are stacked in corresponding layers, and the number of channels is adjusted by a 3×3 convolution kernel to obtain the semantic segmentation image output by the last decoding layer.

[0095] Optionally, the road condition risk detection device further includes a model training module, used for:

[0096] Before performing semantic segmentation on the input road condition image using the target semantic segmentation model to obtain the semantic segmentation image corresponding to the road condition image, the deep learning model is trained using a sample image set, a cross-entropy loss function, and a Dice loss function to obtain the target semantic segmentation model.

[0097] The road condition risk detection device provided in this embodiment of the invention can execute the road condition risk detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0098] Example 4

[0099] Figure 6 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0100] like Figure 6As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0101] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0102] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as road condition risk detection methods.

[0103] In some embodiments, the road condition risk detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the road condition risk detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the road condition risk detection method by any other suitable means (e.g., by means of firmware).

[0104] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0105] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0106] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0107] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0108] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0109] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0110] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0111] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A road condition risk detection method, characterized in that, include: Road condition images of the target road are acquired using target sensors; The input road condition image is semantically segmented by a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole space pyramid pooling-channel attention mechanism module. The risk outcome of the target road condition is determined based on the semantic segmentation image; The hollow spatial pyramid pooling-channel attention mechanism module includes multiple feature processing layers and 1×1 convolutional kernels; wherein... The multi-layer feature processing layer includes a feature processing layer consisting of a 1x1 convolutional layer and a channel attention mechanism, a 3×3 convolutional layer with a first dilatation rate, a 3×3 convolutional layer with a second dilatation rate, a 3×3 convolutional layer with a third dilatation rate, and a global average pooling layer. The channel attention mechanism adjusts the weights of each channel; the global average pooling layer averages all elements in the feature map and outputs the average to the next layer. The step of performing semantic segmentation on the input road condition image using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image includes: The road condition image is extracted layer by layer by each encoding layer in the encoding end to obtain the first feature corresponding to each encoding layer and the second feature output by the last encoding layer. Based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end, a semantic segmentation image corresponding to the road condition image is obtained.

2. The method according to claim 1, characterized in that, The encoding end includes a target number of cascaded encoding layers, and the decoding end includes a target number of cascaded decoding layers matching the encoding end. Each encoding layer in the encoding end includes a 3×3 convolutional kernel, an activation function, and a 2×2 max-pooling downsampling layer.

3. The method according to claim 1, characterized in that, The semantic segmentation image is obtained based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end, including: The second feature is input into the void space pyramid pooling-channel attention mechanism module, and feature processing is performed on the input second feature to obtain multiple preliminary features; Multiple preliminary features are concatenated, and the number of channels is adjusted using a 1×1 convolution kernel to obtain the target features corresponding to the road condition image; The target features are input to the decoding end to obtain the semantic segmentation image corresponding to the road condition image.

4. The method according to claim 3, characterized in that, The step of inputting the target features into the decoding end to obtain the semantic segmentation image corresponding to the road condition image includes: The target features are upsampled layer by layer by each decoding layer in the decoding end to obtain the third feature corresponding to each decoding layer; The first and third features are stacked in corresponding layers, and the number of channels is adjusted by a 3×3 convolution kernel to obtain the semantic segmentation image output by the last decoding layer.

5. The method according to claim 1, further comprising, before performing semantic segmentation on the input road condition image using a target semantic segmentation model to obtain a semantically segmented image corresponding to the road condition image: The target semantic segmentation model is obtained by training the deep learning model using a sample image set, cross-entropy loss function, and Dice loss function.

6. A road condition risk detection device, characterized in that, include: The image acquisition module is used to acquire road condition images of the target road using the target sensor; The semantic segmentation module is used to perform semantic segmentation on the input road condition image through a target semantic segmentation model to obtain a semantic segmentation image corresponding to the road condition image. The target semantic segmentation model includes an encoder, a decoder, and a hole space pyramid pooling-channel attention mechanism module. The risk detection module is used to determine the risk outcome of the target road condition based on the semantic segmentation image; The semantic segmentation module is specifically used for: The hollow spatial pyramid pooling-channel attention mechanism module includes multiple feature processing layers and 1×1 convolutional kernels; wherein... The multi-layer feature processing layer includes a feature processing layer consisting of a 1x1 convolutional layer and a channel attention mechanism, a 3×3 convolutional layer with a first dilatation rate, a 3×3 convolutional layer with a second dilatation rate, a 3×3 convolutional layer with a third dilatation rate, and a global average pooling layer. The channel attention mechanism adjusts the weights of each channel; the global average pooling layer averages all elements in the feature map and outputs the average to the next layer. The semantic segmentation module is further used for: The road condition image is extracted layer by layer by each encoding layer in the encoding end to obtain the first feature corresponding to each encoding layer and the second feature output by the last encoding layer. Based on the second feature, the void space pyramid pooling-channel attention mechanism module, and the decoding end, a semantic segmentation image corresponding to the road condition image is obtained.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the road condition risk detection method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the road condition risk detection method according to any one of claims 1-5.

Images