Vehicle window glass fogging identification method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a vehicle window glass fogging identification method and device, electronic equipment and a storage medium. The method comprises the steps that the internal and external temperature difference of a vehicle window and humidity information corresponding to the interior of a vehicle are obtained; if the internal and external temperature difference reaches a preset temperature difference threshold value and the humidity information reaches a preset humidity threshold value, obtaining a vehicle window image corresponding to the vehicle window glass; configuration information corresponding to the vehicle is acquired, and a target convolutional neural network model is determined based on the configuration information and the vehicle window image; and extracting image features in the vehicle window image based on the convolutional neural network model, and if the image features represent that the vehicle window image is a foggy image, determining that the vehicle window glass is foggy. The embodiment of the invention can adapt to vehicles of different brands and models and vehicle window fogging conditions under different environmental conditions, the applicability and the flexibility of the vehicle window fogging control method are improved, and the perception capability of the vehicle to the environment is improved.

Description

Technical Field

[0001] The present application relates to the field of computer program technology, and in particular to a method, device, electronic device, and computer-readable storage medium for identifying fogging of vehicle window glass. Background Art

[0002] With the development of automobile technology and the improvement of people's living standards, people's requirements for automobile intelligence and user experience are getting higher and higher. Car windows, such as the front windshield, are important channels for users to perceive information from the outside world while driving a car. Their importance is self-evident. The fogging of car windows affects the driver's visual judgment and has become an important reason affecting the user's driving safety.

[0003] In the existing technical solutions, the dew point temperature of the air in the car is mainly calculated by the air conditioning and humidity in the car, and the calculated dew point temperature is compared with the temperature of the windshield, so as to control the car air conditioning to defog the windshield. However, this solution cannot accurately determine whether the windows are fogged, or it may cause incomplete defoggering, affecting the user experience. Summary of the invention

[0004] In order to solve the above technical problems, the embodiments of the present application provide a method and device for identifying fogging of vehicle window glass, an electronic device, a computer-readable storage medium, and a computer program product.

[0005] According to one aspect of an embodiment of the present application, a method for identifying fogged vehicle window glass is provided, including: obtaining the inside-outside temperature difference of the vehicle window and the corresponding humidity information inside the vehicle; if the inside-outside temperature difference reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold, obtaining a vehicle window image corresponding to the vehicle window glass; obtaining configuration information corresponding to the vehicle, and determining a target convolutional neural network model based on the configuration information and the vehicle window image; extracting image features in the vehicle window image based on the convolutional neural network model, and if the image features characterize that the vehicle window image is a foggy image, determining that the vehicle window glass is fogged.

[0006] According to one aspect of an embodiment of the present application, the configuration information includes the operating load of the on-board terminal of the vehicle, and determining the target convolutional neural network model based on the configuration information and the window image includes: determining the convolution kernel of the target convolutional neural network model based on the operating load of the on-board terminal; determining the image input parameters of the target neural network based on the image parameters corresponding to the window image; and determining a matching target convolutional neural network model based on the convolution kernel and the image input parameters.

[0007] According to one aspect of an embodiment of the present application, the convolution kernel of the target convolutional neural network model is determined based on the operating load of the vehicle terminal, including: if the operating load of the vehicle terminal is greater than a preset operating load threshold, then determining the convolution kernel of the target convolutional neural network model based on the vehicle terminal operating load; if the operating load of the vehicle terminal is not greater than the preset operating load threshold, then using the preset convolution kernel as the convolution kernel of the target convolutional neural network model.

[0008] According to one aspect of an embodiment of the present application, the image parameters include pixel size, and determining the image input parameters of the target neural network based on the image parameters corresponding to the vehicle window image includes: if the pixel size corresponding to the vehicle window image is larger than a preset pixel size, compressing the vehicle window image to obtain a vehicle window image that matches the preset pixel size; if the pixel size corresponding to the vehicle window image is smaller than the preset pixel size, pixel padding the vehicle window image to obtain a vehicle window image that matches the preset pixel size.

[0009] According to one aspect of an embodiment of the present application, the compression processing of the vehicle window image includes: determining interpolation parameters corresponding to the vehicle window image based on the preset pixel size; and performing interpolation processing on the vehicle window image based on the interpolation parameters to obtain a vehicle window image matching the preset pixel size.

[0010] According to one aspect of an embodiment of the present application, the method further includes: determining a transparency level of the vehicle window image based on the image features; if the transparency level is less than a preset transparency level threshold, determining that the vehicle window image is a foggy image.

[0011] According to one aspect of an embodiment of the present application, the method also includes: obtaining the external temperature of the vehicle, and determining a defog strategy for the vehicle window glass based on the external temperature and the transparency level; controlling the vehicle to execute the defog strategy, and obtaining an image of the vehicle window after defog; if the transparency level corresponding to the image of the vehicle window after defog reaches a preset transparency level threshold, determining that the vehicle defog is completed.

[0012] According to one aspect of an embodiment of the present application, a device for identifying fogged vehicle window glass is provided, including: an acquisition module, used to obtain the inside-outside temperature difference of the vehicle window and the corresponding humidity information inside the vehicle; an image acquisition module, used to obtain a vehicle window image corresponding to the vehicle window glass if the inside-outside temperature difference reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold; a determination module, used to obtain configuration information corresponding to the vehicle, and determine a target convolutional neural network model based on the configuration information and the vehicle window image; an extraction module, used to extract image features in the vehicle window image based on the convolutional neural network model, and if the image features characterize that the vehicle window image is a foggy image, it is determined that the vehicle window glass is fogged.

[0013] According to one aspect of an embodiment of the present application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements the method for treating fogging of vehicle window glass as described above.

[0014] According to one aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a computer, the computer executes the method for fogging a vehicle window glass as described above.

[0015] According to one aspect of an embodiment of the present application, a computer program product is also provided, including a computer program, which, when executed by a processor, implements the steps in the method for identifying fogging of vehicle window glass as described above.

[0016] In the technical solution provided in the embodiments of the present application, by using the temperature difference between the inside and outside of the car windows and the humidity information inside the vehicle, it is possible to ensure in a timely and accurate manner that the car windows can be understood and detected when they begin to fog, and by automatically selecting a suitable convolutional neural network model for analysis based on the vehicle configuration information and the window image, the driver's burden is greatly reduced and the driving experience is improved. On the other hand, by selecting a suitable convolutional neural network model based on the vehicle configuration information and the window image, the present application can adapt to vehicles of different brands and models, as well as window fogging under different environmental conditions, thereby improving its applicability and flexibility and enhancing the vehicle's ability to perceive the environment.

[0017] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0019] Figure 1 is a schematic diagram of an implementation environment for identifying fogging of vehicle window glass, shown in an exemplary embodiment of the present application;

[0020] Figure 2 is a flow chart of a method for identifying fogging of vehicle window glass shown in an exemplary embodiment of the present application;

[0021] Figure 3 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0022] Figure 4 is a schematic diagram of the structure of a convolutional neural network model shown in an exemplary embodiment;

[0023] Figure 5 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0024] Figure 6 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0025] Figure 7 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0026] Figure 8 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0027] Fig. 9 is a flow chart of a method for identifying fogging of vehicle window glass shown in another exemplary embodiment of the present application;

[0028] Fig.10 It is a brief schematic diagram of the process of identifying fogging of vehicle window glass in an exemplary application scenario;

[0029] Fig.11 is a block diagram of a vehicle window glass fogging recognition device shown in an exemplary embodiment of the present application;

[0030] Fig.12 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0031] Here, exemplary embodiments will be described in detail, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0032] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0033] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0034] The term "multiple" as used in this application refers to two or more than two. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the related objects are in an "or" relationship.

[0035] First of all, it should be noted that fogging of car windows is a common phenomenon, especially in environments with large temperature differences and high humidity. When the temperature inside the car is higher than the temperature outside and the humidity inside the car is high, a layer of fog is likely to form on the inner surface of the car window glass. This is because when the warm and humid air inside the car encounters the colder glass surface, the water vapor in the air condenses into tiny water droplets, forming a layer of foggy substance on the glass.

[0036] Convolutional Neural Network (CNN) is a special deep learning model that is mainly used to process data with a grid structure, such as images or videos. Convolutional neural networks have achieved remarkable results in image recognition, image classification, object detection, image segmentation and other fields.

[0037] The core idea of ​​the convolutional neural network model is to extract local features in the image through the convolutional layer. Each neuron in the convolutional layer is only connected to a local area of ​​the input image (called the receptive field), and a learnable convolution kernel (or filter) is used to perform a convolution operation on the area to extract the features of the area. This local connection and weight sharing method enables the convolutional neural network to effectively reduce the number of model parameters, reduce computational complexity, and extract translation invariant features in the image. In addition to the convolutional layer, the convolutional neural network model usually includes the following key components: Activation layer: After the convolutional layer, a nonlinear activation function (such as ReLU, Sigmoid or Tanh) is usually used to increase the nonlinear ability of the model, so that the model can learn more complex features; Pooling layer: The pooling layer is used to downsample the feature map output by the convolutional layer, reducing the amount of calculation by reducing the dimension of the feature map while retaining important feature information. Common pooling operations include max pooling and average pooling; Fully connected layer: After multiple convolutional layers and pooling layers, one or more fully connected layers are usually connected. Each neuron in the fully connected layer is connected to all neurons in the previous layer, which is used to combine the extracted local features into global features and perform tasks such as classification or regression; Output layer: The output layer is the last layer of the convolutional neural network model, which is used to output the prediction results of the model. For classification tasks, the output layer usually uses the softmax function to output the predicted probability of each category; for regression tasks, the output layer directly outputs the predicted value.

[0038] The convolutional neural network model forms a deep network structure by stacking multiple convolutional layers, activation layers, pooling layers and fully connected layers. It can automatically extract complex feature representations from the original image and be used for various image-related tasks.

[0039] Figure 1 FIG. 1 is a schematic diagram of an implementation environment for identifying fogging of vehicle window glass during vehicle operation, according to an exemplary embodiment of the present application. Figure 1 As shown, during the driving of the vehicle, if the intelligent terminal 110 detects that the temperature difference between the inside and outside of the window reaches a preset temperature difference threshold and the humidity information inside the vehicle reaches a preset humidity threshold, it can further obtain the window image corresponding to the window glass and obtain the configuration information corresponding to the vehicle, and then determine the corresponding target convolutional neural network model based on the vehicle configuration information and the window image, so as to recognize the window image through the target convolutional neural network model and determine whether the window glass is fogged through the recognition result.

[0040] In addition, in some feasible embodiments, the implementation environment diagram of the identification of fogged vehicle window glass can be jointly implemented by the smart terminal 110 and the server 120. During the driving of the vehicle, the temperature difference between the inside and outside of the vehicle window and the corresponding humidity information inside the vehicle are collected by the smart terminal 110 and sent to the server 120. If the server 120 determines that the temperature difference between the inside and outside of the vehicle window and the humidity information inside the vehicle reach the preset threshold, the smart terminal 110 can be triggered to collect the window image corresponding to the vehicle window glass, and the collected vehicle window glass can be sent to the server 120 through the terminal device 110, so that the server 120 can determine the corresponding target convolutional neural network model according to the configuration information of the vehicle and the window image, and then determine whether the vehicle window glass is fogged through the target convolutional neural network model.

[0041] in, Figure 1 The smart terminal 110 shown can be any terminal device that supports the installation of navigation map software, such as a smart phone, a car computer, a tablet computer, a laptop computer, or a wearable device, but is not limited thereto. Figure 1 The server 120 shown, for example, can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms, which are not limited here. The smart terminal 110 can communicate with the server 120 through wireless networks such as 3G (third generation mobile information technology), 4G (fourth generation mobile information technology), and 5G (fifth generation mobile information technology), which are not limited here.

[0042] In the existing technical solutions, the dew point temperature of the air in the car is mainly calculated by the air conditioning and humidity in the car, and the calculated dew point temperature is compared with the temperature of the windshield, so as to control the car air conditioning to defog the windshield. However, this solution cannot accurately determine whether the windows are fogged, or it may cause incomplete defoggering, affecting the user experience.

[0043] The above-mentioned problems are generally applicable in common travel scenarios. To solve these problems, the embodiments of the present application respectively propose a method for identifying fogged vehicle window glass, a device for identifying fogged vehicle window glass, an electronic device, a computer-readable storage medium, and a computer program product, which will be described in detail below.

[0044] See also Figure 2 , Figure 2is a flow chart of a method for identifying fogging of vehicle window glass shown in an exemplary embodiment of the present application. The method can be applied to Figure 1 The implementation environment shown is specifically executed by the smart terminal 110 in the implementation environment. It should be understood that the method can also be applied to other exemplary implementation environments and be specifically executed by devices in other implementation environments. This embodiment does not limit the implementation environment to which the method is applicable.

[0045] like Figure 2 As shown, in an exemplary embodiment, the method for identifying fogging of vehicle window glass includes at least steps S210 to S240, which are described in detail as follows:

[0046] Step S210, obtaining the temperature difference between the inside and outside of the vehicle window and the corresponding humidity information inside the vehicle.

[0047] Specifically, the fog in the car interior is mainly caused by the difference in humidity inside and outside the car. When the humidity inside the car is high and the humidity outside the car is low, the water vapor on the car windows will condense into small water droplets and form fog. The high humidity inside the car is usually because people release water vapor when breathing in the car, especially when the temperature inside the car is high, these water vapors will condense on the car windows and form fog. In addition, if the air humidity in the car is relatively high, it will also cause fog on the windows. Using the air conditioner in the car for a long time can also easily lead to an increase in humidity in the car, because the air conditioner will increase the water vapor content in the air, and fog is easy to form when the temperature difference occurs.

[0048] For example, the temperature outside the window and the temperature inside the window can be obtained through temperature sensors inside and outside the vehicle's window glass, and then the temperature difference between the inside and outside of the vehicle's window can be calculated through the temperature outside the window and the temperature inside the window. In addition, the humidity information inside the vehicle can also be obtained through the corresponding humidity sensor inside the vehicle.

[0049] Step S220: If the temperature difference between the inside and outside reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold, a window image corresponding to the window glass is obtained.

[0050] Specifically, the calculated temperature difference between the inside and outside of the window is compared with a preset temperature difference threshold, wherein the preset temperature difference threshold may be related to factors such as season and climate. It is also necessary to compare the acquired humidity information of the vehicle with the preset humidity threshold. Similarly, the setting of the preset humidity threshold may also be related to factors such as season and climate. If it is determined through comparison that the temperature difference between the inside and outside of the window reaches the preset temperature difference threshold, and the humidity information inside the vehicle also reaches the preset humidity threshold, then it is determined that the window glass is at a high risk of fogging, and a window image corresponding to the window glass is further obtained to further determine whether the window glass is fogged through the window image.

[0051] Exemplarily, the preset temperature difference threshold is dynamically set according to factors such as season, climate, and geographical location. For example, in winter, due to the low outside temperature, the preset temperature difference threshold may be set to a smaller value to prevent frosting of the car windows; in summer, due to the high outside temperature, a larger threshold may need to be set to prevent overheating in the car. Similarly, the preset internal humidity threshold is set according to factors such as season and climate. A high humidity environment may cause fogging of the car windows, so the threshold needs to be adjusted according to actual conditions. If the temperature difference between the inside and outside of the car window reaches the preset temperature difference threshold, and the humidity information inside the vehicle reaches the preset humidity threshold, the window image corresponding to the window glass can be collected by the corresponding image acquisition device in the car.

[0052] Step S230, obtaining configuration information corresponding to the vehicle, and determining a target convolutional neural network model based on the configuration information and the vehicle window image.

[0053] Specifically, after obtaining the window image corresponding to the window glass, it is necessary to further determine the convolutional neural network model used to process the window image. This can be done by obtaining the configuration information corresponding to the vehicle, where the configuration information includes the vehicle model, the computing power of the on-board terminal, etc., and then the target convolutional neural network model can be determined based on the vehicle configuration information and the obtained window image.

[0054] Exemplarily, vehicle configuration information may include, but is not limited to: vehicle model: the window design, size, and location of different vehicle models may vary; year: newer models may be equipped with more advanced sensors and image processing technology, or their windows may be made of different materials and designs; additional configurations: such as the configuration of advanced driver assistance systems (ADAS), which may affect the available sensor data and image processing capabilities; regions and markets: different regions and markets may have different preferences for vehicle configurations, which may affect the selection of data sets for model training. Based on the vehicle configuration information and the specific tasks of window image analysis (such as detecting fogging, identifying stains, detecting the open and closed status of windows, etc.), determine the requirements of the convolutional neural network model: input data: resolution and color depth of the window image, processing performance: real-time requirements, computing resource limitations, etc.; model accuracy: requirements for accuracy, whether a certain degree of false positives or negatives is allowed.

[0055] In addition, if the pre-trained model does not meet specific needs, or if you need to process very special car window images, you can consider training a new convolutional neural network model from scratch. This requires the use of a large number of labeled car window image datasets, and may require a lot of experiments to optimize the model structure and parameters, and select one or more suitable convolutional neural network models based on vehicle configuration information and task requirements. For example, if computing resources are limited, you can choose a lightweight model (such as MobileNet); if you pursue high accuracy, you can choose a deeper and more complex model (such as ResNet).

[0056] Step S240: extracting image features from the vehicle window image based on the target convolutional neural network model. If the image features indicate that the vehicle window image is a foggy image, it is determined that the vehicle window glass is fogged.

[0057] Specifically, image features in the car window image can be extracted based on the selected target convolutional neural network model, and then the image features can be analyzed. If the image features characterize that the car window image is a foggy image, it can be determined that the car window glass is fogged.

[0058] Exemplarily, first, the car window image to be analyzed is provided as input to the convolutional neural network model, and the image is preprocessed as necessary, such as resizing to meet the input requirements of the model, normalizing pixel values, etc. The preprocessed image is sent to the previously determined target convolutional neural network model. The model is trained to automatically extract features from the image that are useful for judging whether the car window is foggy. The convolutional neural network model converts the original image into a series of high-level feature representations through multi-layer convolution, pooling and other operations. These features usually encode the texture, edge, shape and other attributes of the image. The last few layers of the convolutional neural network model usually include a fully connected layer (FC layer) or a global average pooling layer (GAP layer) for converting the extracted features into a fixed-length feature vector, and further classifying or regressing. In addition, during the training process, the convolutional neural network model has learned the difference between foggy images and non-fog images. Therefore, when a car window image is input, the model can output the probability or score that the image belongs to the foggy category. According to the probability or score output by the model, a threshold is set to determine whether the car window glass is fogged. If the probability or score exceeds the threshold, the car window glass is considered to be fogged; otherwise, the car window glass is considered to be not fogged.

[0059] In this embodiment, by using the temperature difference between the inside and outside of the windows in the lake area and the humidity information inside the vehicle, it is possible to ensure that the windows can be understood and detected in a timely and accurate manner when they begin to fog up, and by automatically selecting a suitable convolutional neural network model for analysis based on the vehicle configuration information and the window image, the driver's burden is greatly reduced and the driving experience is improved. On the other hand, by selecting a suitable convolutional neural network model based on the vehicle configuration information and the window image, the solution can adapt to vehicles of different brands and models, as well as window fogging under different environmental conditions, thereby improving its applicability and flexibility and enhancing the vehicle's ability to perceive the environment.

[0060] Further, based on the above embodiments, please refer to Figure 3 In one of the exemplary embodiments provided in the present application, the configuration information includes the vehicle terminal operation load of the vehicle, and the specific implementation process of determining the target convolutional neural network model based on the configuration information and the window image may also include steps S310 to S330, which are described in detail as follows:

[0061] Step S310, determining the convolution kernel of the target convolutional neural network model based on the operating load of the vehicle terminal.

[0062] Specifically, when running the target convolutional neural network (CNN) model on the vehicle terminal, considering the computing resources, memory limitations and real-time requirements of the vehicle terminal, choosing the right convolution kernel is crucial to optimizing model performance and reducing operating load. Analyze the hardware resources of the vehicle terminal, for example, by analyzing the processor type and performance of the vehicle terminal, including the instruction set, number of cores, main frequency, etc. supported by the vehicle terminal, evaluate the memory size and storage speed of the vehicle terminal to determine the model size and amount of data that can be loaded into the memory, and consider the impact of long-term high-load computing on the power consumption and heat dissipation of the vehicle terminal. In addition, smaller convolution kernels can usually reduce the amount of calculation and the number of parameters, but more convolution layers may be required to achieve the same receptive field. Larger convolution kernels may capture features faster, but the computational cost is higher. Deeper networks can capture more complex features, but they will also increase the amount of calculation and memory usage. During the training process, experiments are conducted with convolution kernels of different sizes and numbers to find a model configuration that strikes a balance between accuracy and speed.

[0063] For example, the operating environment of the model can be simulated on hardware similar to an in-vehicle terminal, the performance under different convolution kernel configurations can be tested, and the processing speed of the model on the in-vehicle terminal can be evaluated to ensure that the real-time requirements can be met. According to the results of the performance test, the size, number and arrangement of the convolution kernels can be adjusted to optimize the performance and resource usage of the model.

[0064] Step S320, determining image input parameters of the target neural network based on image parameters corresponding to the vehicle window image.

[0065] Specifically, when processing the target neural network (such as a convolutional neural network) based on the car window image, determining the image input parameters is an important step to ensure that the model can efficiently receive and process image data. These image input parameters are usually related to the characteristics of the car window image itself and the input requirements of the neural network. Among them, the image size includes: Resolution: Determine the resolution (width and height) of the input image. This usually depends on the architectural design of the neural network, because different network layers may have specific requirements for the input size. At the same time, considering the processing power and real-time requirements of the vehicle terminal, it may be necessary to select a suitable resolution balance point. In addition, the color format also needs to be considered: Determine the color format of the input image (such as RGB, grayscale, etc.). Most convolutional neural networks are designed based on the RGB color space because they can process richer color information. However, in some specific tasks (such as edge detection, binarization, etc.), grayscale images may be more effective. Color normalization: In order to accelerate model training and improve model generalization ability, it is usually necessary to normalize the color of the input image, that is, scale the pixel value to a specific range (such as 0-1 or -1 to 1).

[0066] Step S330, determining a matching target convolutional neural network model based on the convolution kernel and image input parameters.

[0067] Specifically, the convolution kernel of the target convolutional neural network model can be determined according to the operating load of the vehicle's on-board terminal, and the image input parameters of the target convolutional neural network can be determined by the image parameters corresponding to the collected window image, and then the target convolutional neural network model that matches it can be determined by the convolution kernel and the image input parameters.

[0068] For example, see Figure 4 , Figure 4 This is a schematic diagram of the structure of a convolutional neural network model shown in an exemplary embodiment of the present application, for example:

[0069] Input image shape: [inChannels, height, weight] / [3,8,8];

[0070] Convolution kernel shape: [outChannels, inChannels, height, weight] / [5, 3, 3, 3];

[0071] The shape of the output image is: [outChannels, outHeight, outweight] / [5,6,6];

[0072] When the convolution kernel is a unit convolution kernel with size f=1, stride=1 and no padding, the cross-correlation calculation within the convolution layer is equivalent to matrix multiplication, and a fully connected network is constructed between the convolution layers, and the softmax feature value of the window is extracted through the convolutional neural network to determine whether it is foggy.

[0073] In this embodiment, the size and number of convolution kernels are determined according to the operating load of the vehicle terminal, which can ensure that the convolution neural network model runs more efficiently under limited computing resources, and when the vehicle terminal resources are tight, a smaller convolution kernel or a simplified network structure can be selected to reduce the computational complexity and energy consumption. The image input parameters of the convolution neural network model are determined according to the actual parameters of the window image (such as resolution, color depth, etc.), which can ensure that the model can make full use of the image information and improve the accuracy of recognition or detection.

[0074] Further, based on the above embodiments, please refer to Figure 5 In one of the exemplary embodiments provided in the present application, the specific implementation process of determining the convolution kernel of the target convolutional neural network model based on the vehicle terminal operating load may also include the following steps S510 and S520, which are described in detail as follows:

[0075] Step S510, if the vehicle terminal operating load is greater than a preset operating load threshold, determining the convolution kernel of the target convolutional neural network model based on the vehicle terminal operating load;

[0076] Step S520: If the operating load of the vehicle terminal is not greater than the preset operating load threshold, the preset convolution kernel is used as the convolution kernel of the target convolutional neural network model.

[0077] Specifically, first, the system needs to monitor the current operating load of the vehicle terminal in real time or regularly. This load may include CPU usage, memory occupancy, network bandwidth occupancy and other aspects, depending on the system design and requirements. Then, the monitored operating load is compared with the preset operating load threshold. This threshold is set in advance based on factors such as the hardware configuration of the vehicle terminal, the expected application scenario and performance requirements. If the operating load of the vehicle terminal is greater than the preset operating load threshold, it means that the current system resources are relatively tight and measures need to be taken to reduce the computing burden. At this time, the system will dynamically determine a lighter convolution kernel based on the current operating load for use in the target convolutional neural network model. This lightweight convolution kernel may have fewer parameters and a simpler structure, so that it can reduce the amount of calculation and resource consumption while ensuring a certain processing accuracy. If the operating load of the vehicle terminal is not greater than the preset operating load threshold, it means that the current system resources are relatively sufficient and can support more complex computing tasks. At this time, the system will use the preset convolution kernel (usually a convolution kernel with better performance and higher accuracy) as the convolution kernel of the target convolutional neural network model. This can make full use of system resources and improve processing efficiency and accuracy.

[0078] In this embodiment, through this mechanism of dynamically adjusting the convolution kernel, the vehicle-mounted terminal can flexibly select a suitable convolutional neural network model configuration according to its own operating conditions, thereby optimizing system performance and resource utilization while ensuring the processing effect.

[0079] Further, based on the above embodiments, please refer to Figure 6 In one of the exemplary embodiments provided in the present application, the above-mentioned image parameters include pixel size, and the specific implementation process of determining the image input parameters of the target neural network based on the image parameters corresponding to the vehicle window image may also include steps S610 and S620, which are described in detail as follows:

[0080] Step S610: If the pixel size corresponding to the vehicle window image is larger than the preset pixel size, the vehicle window image is compressed to obtain a vehicle window image matching the preset pixel size.

[0081] Specifically, when the pixel size of the window image is larger than the preset pixel size, it means that the resolution of the image is higher than the resolution required for processing or analysis. In order to save storage space and improve processing efficiency, the system will compress the image. Compression processing may involve a variety of technologies, such as reducing the color depth of the image, using a more efficient encoding method, and reducing the resolution of the image. These technologies can reduce the size of the image and the required storage space without significantly reducing the image quality. After compression processing, the pixel size of the window image will match the preset pixel size, thereby meeting the requirements of subsequent processing or analysis.

[0082] Step S620: If the pixel size corresponding to the vehicle window image is smaller than the preset pixel size, pixel padding is performed on the vehicle window image to obtain a vehicle window image matching the preset pixel size.

[0083] Specifically, when the pixel size of the car window image is smaller than the preset pixel size, it means that the resolution of the image is not sufficient to meet the processing or analysis requirements. In order to make up for this deficiency, the system will perform pixel padding on the image. Pixel padding usually adds additional pixels at the border or specified area of ​​the image to increase the size of the image. These newly added pixels can be filled in a variety of ways, such as using the border pixel values ​​of the image, performing interpolation calculations to generate new pixel values, etc. The filled car window image will have a size that matches the preset pixel size, so that it can better adapt to the subsequent processing or analysis process.

[0084] Exemplarily, in some feasible embodiments, the edge of the picture may be filled with pixels of 0, so that the window image that is smaller than the preset pixel size after pixel filling meets the requirement of the preset pixel size.

[0085] In this embodiment, by unifying the pixel size of the car window image to a preset standard, it is possible to ensure that the target convolutional neural network model has a consistent input size when processing different images, which is crucial for the training and reasoning process of the model. Unifying the input standard helps improve the stability and accuracy of the model. By ensuring the consistency of the input image and optimizing the image quality, the accuracy and reliability of the model processing results can be improved, thereby providing users with a better user experience.

[0086] Further, based on the above embodiments, please refer to Figure 7 In one of the exemplary embodiments provided in the present application, the specific implementation process of compressing the vehicle window image may further include step S710 and step S720, which are described in detail as follows:

[0087] Step S710, determining interpolation parameters corresponding to the vehicle window image based on a preset pixel size;

[0088] Step S720: performing interpolation processing on the vehicle window image based on the interpolation parameters to obtain a vehicle window image matching a preset pixel size.

[0089] Specifically, first, it is necessary to determine the scaling ratio required from the original window image to the target preset pixel size. This is usually done by comparing the width and height of the original image with the width and height of the preset pixel size. The interpolation parameters may not be specified directly, but are implicitly determined by the scaling ratio and the selected interpolation algorithm (such as bilinear interpolation, bicubic interpolation, etc.). The choice of interpolation algorithm affects the quality and processing time of the final image. Using the selected interpolation algorithm and the calculated scaling ratio, the original window image is interpolated. The interpolation process generates a new image whose pixel size matches the preset pixel size. In this process, the pixel values ​​in the original image are used to calculate the pixel values ​​at the corresponding positions in the new image. The result of the interpolation process is a resized window image that can be directly used for subsequent processing or analysis.

[0090] In this embodiment, the interpolation process can flexibly adjust the image size according to the preset pixel size, without being restricted by the original image size. This means that no matter what the size of the original window image is, the interpolation process can be used to obtain an image size that meets the system requirements, which enhances the flexibility and adaptability of the system. Compared with simple cropping or scaling, the interpolation process can reduce information loss when adjusting the image size. Cropping directly removes part of the image content, while scaling may cause the image to be blurred or distorted. Interpolation processing fills the image by calculating new pixel values, which can retain the features and details of the original image to a certain extent.

[0091] Further, based on the above embodiments, please refer to Figure 8 In one of the exemplary embodiments provided in the present application, the specific implementation process of the above-mentioned vehicle window glass fogging identification method may also include step S810 and step S820, which are described in detail as follows:

[0092] Step S810, determining the transparency level of the vehicle window image based on the image features;

[0093] Step S820: If the transparency level is less than the preset transparency level threshold, the vehicle window image is determined to be a foggy image.

[0094] Specifically, first, key features can be extracted from the car window image through the target convolutional neural network model, which are crucial for evaluating the transparency of the image. Common image features include but are not limited to: color histogram: it can reflect the distribution of colors in the image, which is very helpful for detecting gray tones in foggy weather, contrast: the reduction of contrast is an obvious sign that the image is affected by fog, dark channel prior: this theory points out that in a clear natural image, except for a few areas such as the sky, there will be at least one color channel with a very low intensity value in each local small window. In foggy images, due to the scattering effect of fog, such low-intensity areas will be greatly reduced, texture features: the texture details of clear images are richer, while foggy images are relatively blurred. Based on the above extracted features, an algorithm can be designed to calculate the transparency level of the car window image. This usually involves a comprehensive consideration of one or more features, and the level is determined by some weighting or scoring mechanism. For example, the dark channel prior algorithm can be used to estimate the concentration of fog, and then a quantitative indicator of the transparency level can be obtained.

[0095] Then, a preset transparency level threshold is set, which can be adjusted according to actual conditions to adapt to different environments and needs. Under different weather and lighting conditions, the transparency of the car window image will be different. The method of determining the transparency level based on image features can adapt to this change and accurately judge whether the image is foggy by dynamically adjusting the evaluation criteria. This adaptability enables this method to be effectively applied in a variety of practical scenarios. The calculated transparency level is compared with this threshold: if the transparency level is less than the preset transparency level threshold, the car window image is considered to be affected by fog, that is, it is judged as a "foggy image"; if the transparency level is greater than or equal to the preset transparency level threshold, the car window image is considered to be relatively clear and not significantly affected by fog.

[0096] In this embodiment, by analyzing and extracting features in the car window image (such as contrast, brightness distribution, edge blur, etc.), the transparency of the image can be more accurately evaluated. Compared with methods that rely on a single threshold or simple rules, this method can more comprehensively consider various attributes of the image, thereby improving the accuracy of identifying foggy images.

[0097] Further, based on the above embodiments, please refer to Fig. 9 In one of the exemplary embodiments provided in the present application, the specific implementation process of the above-mentioned vehicle window glass fogging identification method may also include steps S910 to S930, which are described in detail as follows:

[0098] Step S910, obtaining the external temperature of the vehicle, and determining the defog strategy of the vehicle window glass based on the external temperature and the transparency level.

[0099] Specifically, the vehicle's external ambient temperature can be obtained, and a defog strategy for the vehicle windows can be determined based on the vehicle's external ambient temperature and the transparency level of the windows. The defog strategy includes using the air-conditioning system defog button, and the cold air defog method (turn on the air-conditioning cooling switch in summer, use the defog mode to blow towards the windshield) or the warm air defog method (use warm air defog in winter, turn on the air-conditioning warm air and defog switches after the water temperature is normal, and then blow towards the windshield) can be used to reduce the temperature difference between inside and outside the vehicle to achieve the purpose of quickly removing fog.

[0100] Exemplary, low transparency level and low temperature: If the transparency level is very low and the outside temperature is also low, then thick fog is likely to have formed on the car window glass. At this time, active defog measures need to be taken, such as turning on the front windshield defog function, adjusting the air-conditioning temperature in the car, etc. Low transparency level but moderate or high temperature: Although the transparency level is low, if the outside temperature is moderate or high, the fog may be mainly caused by excessive humidity or temperature difference in the car. At this time, the fog can be reduced by adjusting the temperature and humidity in the car, and the defog function can be turned on to speed up the defog process. High transparency level: If the transparency level is high, it means that the car window glass is relatively clear and there is no serious fog. However, it is still necessary to adjust the interior environment of the car appropriately according to the external temperature to prevent the sudden formation of fog.

[0101] Exemplarily, the following defog strategies can be formulated: Turn on the defog function: For low transparency levels, the defog function of the front windshield and rearview mirrors should be turned on first, regardless of the outside temperature. This usually involves heating the window glass or blowing hot air to the window surface. Adjust the temperature and humidity in the car: According to the external temperature and humidity in the car, adjust the temperature and humidity settings of the air-conditioning system in the car. For example, in a low temperature and high humidity environment, the temperature in the car can be appropriately increased and the humidity can be lowered to reduce fog formation. Use anti-fog agent: Before the vehicle departs or when necessary, anti-fog agent can be applied to the car window glass. Anti-fog agent can form a thin film on the glass surface to prevent water vapor from condensing into fog. Ventilation: When conditions permit, open the windows for ventilation to reduce the humidity and temperature difference in the car, thereby reducing fog formation.

[0102] In addition, after the defogging strategy is implemented, the transparency of the car window glass should be monitored in real time, and the defogging measures should be adjusted as needed. For example, if the defogging effect is not good or the fog re-forms, the power of the defogging function can be increased or the in-car environment settings can be adjusted.

[0103] Step S920, controlling the vehicle to execute the defogging strategy and obtaining a defogging window image;

[0104] Step S930: If the transparency level corresponding to the defogged vehicle window image reaches a preset transparency level threshold, it is determined that the vehicle defog is completed.

[0105] Specifically, the vehicle's control system (such as ECU, Engine Control Unit) sends instructions to related actuators (such as the front windshield defogger, air conditioning system, etc.) to start the defog function. Adjust parameters: According to the defog strategy, adjust the temperature, wind speed, wind direction and other parameters of the air-conditioning system, as well as the heating power of the defogger, to ensure the best defog effect. Monitor the execution of the defog function to ensure that all instructions are executed correctly and the system is in normal working condition. Use the vehicle's built-in camera or other image acquisition equipment to obtain the image of the defogged window. Perform necessary processing on the acquired image, such as denoising, contrast enhancement, etc., to more accurately evaluate the transparency level of the window.

[0106] In this embodiment, by combining the external temperature and the transparency level of the window image, the system can automatically determine whether defog is needed and select an appropriate defog strategy. This automated and intelligent management method not only improves efficiency, but also reduces the need for human intervention, avoids unnecessary energy consumption when defog is not needed, and helps save energy and reduce emissions. At the same time, by accurately controlling the defog time and intensity, it can also reduce unnecessary energy waste and improve the energy efficiency ratio of the vehicle.

[0107] Fig.10 It is a simple flow chart for identifying fogged car window glass in an exemplary application scenario. In the application scenario shown in 10, the temperature difference between the inside and outside of the car window and the corresponding humidity information inside the vehicle are obtained; if the temperature difference between the inside and outside reaches the preset temperature difference threshold and the humidity information reaches the preset humidity threshold, the car window image corresponding to the car window glass is obtained; the configuration information corresponding to the vehicle is obtained, and the convolution kernel of the target convolution neural network model is determined based on the operating load of the vehicle terminal; the image input parameters of the target neural network are determined based on the image parameters corresponding to the car window image; and the matching target convolution neural network model is determined based on the convolution kernel and the image input parameters. If the operating load of the vehicle terminal is greater than the preset operating load threshold, the convolution kernel of the target convolution neural network model is determined based on the operating load of the vehicle terminal; if the operating load of the vehicle terminal is not greater than the preset operating load threshold, the preset convolution kernel is used as the convolution kernel of the target convolution neural network model. If the pixel size corresponding to the window image is larger than the preset pixel size, the window image is compressed to obtain a window image that matches the preset pixel size; if the pixel size corresponding to the window image is smaller than the preset pixel size, the window image is pixel-filled to obtain a window image that matches the preset pixel size. Image features in the window image are extracted based on the target convolutional neural network model, and the transparency level of the window image is determined based on the image features; if the transparency level is smaller than the preset transparency level threshold, the window image is determined to be a foggy image. For the detailed implementation process, please refer to the records in the aforementioned embodiments, which will not be repeated here.

[0108] Fig.11 is a block diagram of an exemplary embodiment of the present application showing the identification of fogging of vehicle windows. The device can be applied to Figure 1 The implementation environment shown is specifically configured in the intelligent terminal 110. The device may also be applicable to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment applicable to the device.

[0109] like Fig.11 As shown, the exemplary vehicle window glass fogging identification device includes: an acquisition module 1110, used to obtain the inside and outside temperature difference of the vehicle window and the corresponding humidity information inside the vehicle; an image acquisition module 1120, used to obtain a vehicle window image corresponding to the vehicle window glass if the inside and outside temperature difference reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold; a determination module 1130, used to obtain configuration information corresponding to the vehicle, and determine a target convolutional neural network model based on the configuration information and the vehicle window image; an extraction module 1140, used to extract image features in the vehicle window image based on the convolutional neural network model, and if the image features characterize that the vehicle window image is a foggy image, it is determined that the vehicle window glass is fogged.

[0110] According to one aspect of an embodiment of the present application, the above-mentioned determination module 1130 is also used to determine the convolution kernel of the target convolution neural network model based on the operating load of the vehicle terminal; determine the image input parameters of the target neural network based on the image parameters corresponding to the vehicle window image; and determine the matching target convolution neural network model based on the convolution kernel and the image input parameters.

[0111] According to one aspect of an embodiment of the present application, the above-mentioned determination module 1130 is also used to determine the convolution kernel of the target convolution neural network model based on the operating load of the vehicle terminal if the operating load of the vehicle terminal is greater than a preset operating load threshold; if the operating load of the vehicle terminal is not greater than the preset operating load threshold, the preset convolution kernel is used as the convolution kernel of the target convolution neural network model.

[0112] According to one aspect of an embodiment of the present application, the above-mentioned determination module 1130 is also used to, if the pixel size corresponding to the vehicle window image is larger than the preset pixel size, compress the vehicle window image to obtain a vehicle window image matching the preset pixel size; if the pixel size corresponding to the vehicle window image is smaller than the preset pixel size, fill the vehicle window image with pixels to obtain a vehicle window image matching the preset pixel size.

[0113] According to one aspect of an embodiment of the present application, the above-mentioned determination module 1130 is also used to determine the interpolation parameters corresponding to the vehicle window image based on the preset pixel size; and perform interpolation processing on the vehicle window image based on the interpolation parameters to obtain a vehicle window image matching the preset pixel size.

[0114] According to one aspect of an embodiment of the present application, the extraction module 1140 is further used to determine the transparency level of the vehicle window image based on image features; if the transparency level is less than a preset transparency level threshold, the vehicle window image is determined to be a foggy image.

[0115] According to one aspect of an embodiment of the present application, the above-mentioned extraction module 1140 is also used to obtain the external temperature of the vehicle, determine the defog strategy of the vehicle window glass based on the external temperature and the transparency level; control the vehicle to execute the defog strategy, and obtain the image of the vehicle window after defog; if the transparency level corresponding to the image of the vehicle window after defog reaches a preset transparency level threshold, it is determined that the vehicle defog is completed.

[0116] It should be noted that the vehicle window glass fogging identification device provided in the above embodiment and the vehicle window glass fogging identification method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiment, and will not be repeated here. In actual applications, the vehicle window glass fogging identification device provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0117] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by one or more processors, the electronic device implements the method for identifying fogged vehicle window glass provided in the above-mentioned embodiments.

[0118] Fig.12 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Fig.12 The computer system 1200 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0119] like Fig.12As shown, the computer system 1200 includes a central processing unit (CPU) 1201, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1202 or the program loaded from the storage part 1208 to the random access memory (RAM) 1203, such as executing the method in the above embodiment. In the RAM 1203, various programs and data required for system operation are also stored. The CPU 1201, the ROM 1202 and the RAM 1203 are connected to each other through the bus 1204. The input / output (I / O) interface 1205 is also connected to the bus 1204.

[0120] The following components are connected to the I / O interface 1205: an input section 1206 including a keyboard, a mouse, etc.; an output section 1207 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1208 including a hard disk, etc.; and a communication section 1209 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet. A drive 1210 is also connected to the I / O interface 1205 as needed. A removable medium 1211, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1210 as needed so that a computer program read therefrom is installed into the storage section 1208 as needed.

[0121] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication section 1209, and / or installed from a removable medium 1211. When the computer program is executed by a central processing unit (CPU) 1201, various functions defined in the system of the present application are executed.

[0122] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0123] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0124] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0125] Another aspect of the present application further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned vehicle window glass fogging identification method is implemented. The computer-readable storage medium may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device.

[0126] Another aspect of the present application also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the vehicle window glass fogging identification method provided in each of the above embodiments.

[0127] The above content is only a preferred exemplary embodiment of the present application and is not intended to limit the implementation scheme of the present application. A person skilled in the art can easily make corresponding changes or modifications based on the main concept and spirit of the present application. Therefore, the scope of protection of the present application shall be based on the scope of protection required by the claims.

Claims

1. A method for identifying fogging of vehicle window glass, characterized in that: include: Obtaining the temperature difference between the inside and outside of the vehicle window and the corresponding humidity information inside the vehicle; If the inside-outside temperature difference reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold, acquiring a vehicle window image corresponding to the vehicle window glass; Obtaining configuration information corresponding to the vehicle, and determining a target convolutional neural network model based on the configuration information and the vehicle window image; Image features in the vehicle window image are extracted based on the target convolutional neural network model, and if the image features indicate that the vehicle window image is a foggy image, it is determined that the vehicle window glass is fogged.

2. The method according to claim 1, characterized in that The configuration information includes the operating load of the vehicle-mounted terminal of the vehicle, and the determining of the target convolutional neural network model based on the configuration information and the vehicle window image includes: Determining the convolution kernel of the target convolutional neural network model based on the operating load of the vehicle-mounted terminal; Determining image input parameters of the target neural network based on image parameters corresponding to the vehicle window image; A matching target convolutional neural network model is determined based on the convolution kernel and the image input parameters.

3. The method according to claim 2, characterized in that The determining the convolution kernel of the target convolutional neural network model based on the operating load of the vehicle-mounted terminal includes: If the vehicle terminal operating load is greater than a preset operating load threshold, determining the convolution kernel of the target convolutional neural network model based on the vehicle terminal operating load; If the operating load of the vehicle-mounted terminal is not greater than a preset operating load threshold, the preset convolution kernel is used as the convolution kernel of the target convolutional neural network model.

4. The method according to claim 2, characterized in that The image parameters include pixel size, and determining the image input parameters of the target neural network based on the image parameters corresponding to the vehicle window image includes: If the pixel size corresponding to the vehicle window image is larger than the preset pixel size, compressing the vehicle window image to obtain a vehicle window image matching the preset pixel size; If the pixel size corresponding to the vehicle window image is smaller than the preset pixel size, the vehicle window image is pixel-filled to obtain a vehicle window image matching the preset pixel size.

5. The method according to claim 4, characterized in that The compressing process of the vehicle window image comprises: Determining an interpolation parameter corresponding to the vehicle window image based on the preset pixel size; The vehicle window image is interpolated based on the interpolation parameters to obtain a vehicle window image matching the preset pixel size.

6. The method according to claim 1, characterized in that: The method further comprises: Determining a transparency level of the vehicle window image based on the image features; If the transparency level is less than a preset transparency level threshold, the vehicle window image is determined to be a foggy image.

7. The method according to claim 6, characterized in that The method further comprises: Acquiring an external temperature of the vehicle, and determining a defog strategy for the vehicle window glass based on the external temperature and the transparency level; Controlling the vehicle to execute a defog strategy and obtaining a defogged vehicle window image; If the transparency level corresponding to the defogged vehicle window image reaches a preset transparency level threshold, it is determined that the vehicle defog is completed.

8. A device for identifying fogging of vehicle windows, characterized in that: The device comprises: An acquisition module, used to acquire the temperature difference between the inside and outside of the vehicle window and the corresponding humidity information inside the vehicle; An image acquisition module, configured to acquire a vehicle window image corresponding to the vehicle window glass if the inside-outside temperature difference reaches a preset temperature difference threshold and the humidity information reaches a preset humidity threshold; A determination module, used to obtain configuration information corresponding to the vehicle, and determine a target convolutional neural network model based on the configuration information and the vehicle window image; An extraction module is used to extract image features in the vehicle window image based on the convolutional neural network model, and if the image features indicate that the vehicle window image is a foggy image, it is determined that the vehicle window glass is fogged.

9. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the method for identifying fogging of vehicle window glass as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Computer-readable instructions are stored thereon, and when the computer-readable instructions are executed by a processor of a computer, the computer is caused to execute the method for identifying fogging of vehicle window glass according to any one of claims 1 to 7.

Images