Traffic target detection method, device, electronic device, storage medium and program product

Abstract

The present disclosure provides a method, device, electronic device, storage medium and program product for traffic target detection, including: determining traffic video information, classifying the video information based on different weather conditions to obtain scene classification information, and extracting and restoring the traffic video information to obtain feature extraction image restoration information, and then inputting a pre-trained blind super-resolution degradation model, obtaining high-resolution image information through generator network processing, and then performing optimization processing to obtain high-resolution image information results, and then performing target image detection to obtain target image detection results. The present disclosure can perform specific processing and optimization on traffic video frames according to different weather conditions, and enhance low-resolution images to generate high-quality high-resolution images, reduce the impact of weather on images, and improve the accuracy of traffic target detection in low-resolution road videos in bad weather scenes such as rain and snow.

Description

Technical Field

[0001] The present disclosure relates to the field of traffic safety and management technology, and in particular to a traffic target detection method, device, electronic device, storage medium and program product. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims. No description herein is admitted to be prior art by inclusion in this section.

[0003] Traffic target detection is the detection and identification of multiple targets in complex traffic scenes, such as vehicles, pedestrians, traffic signs, etc. The detection method can be achieved by first building a model for the three-dimensional scene, and then taking the scenes of the two dimensions of time and weather as prior knowledge, and selecting the corresponding model for the three-dimensional scene according to the current time and weather categories. Finally, different times and weathers can be distinguished based on the model to take specific measures.

[0004] However, existing related technologies are unable to quickly identify different weather conditions and perform specific processing based on weather changes in the real world. Different weather conditions have different impacts on images. For example, rain, snow and other weather conditions will make video images more difficult to identify, and in real scenarios, the resolution of video images is often low, which in turn affects the accuracy of traffic target detection in low-resolution road videos in severe weather scenarios such as rain and snow. Summary of the invention

[0005] In view of this, the purpose of the present disclosure is to propose a traffic target detection method, device, electronic device, storage medium and program product, which at least to a certain extent solve one of the technical problems in the related art.

[0006] Based on the above purpose, the first aspect of the exemplary embodiment of the present disclosure provides a traffic target detection method, the method comprising:

[0007] Determine traffic video information, and classify the video information according to different weather conditions to obtain scene classification information;

[0008] According to the scene classification information, feature extraction and restoration are performed on the traffic video information to obtain feature extraction image restoration information;

[0009] Inputting the feature-extracted image restoration information into a pre-trained blind super-resolution degradation model, obtaining high-resolution image information through generator network processing, and then optimizing the high-resolution image information to obtain a high-resolution image information result;

[0010] Perform target image detection on the high-resolution image information result to obtain a target image detection result.

[0011] Based on the same inventive concept, the second aspect of the exemplary embodiment of the present disclosure provides a traffic target detection device, including:

[0012] A scene classification information determination module is configured to determine traffic video information, classify the video information according to different weather conditions, and obtain scene classification information;

[0013] A feature extraction image restoration information determination module is configured to perform feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information;

[0014] A high-resolution image information result determination module is configured to input the feature extraction image restoration information into a pre-trained blind super-resolution degradation model, obtain high-resolution image information through generator network processing, and then optimize the high-resolution image information to obtain a high-resolution image information result;

[0015] The target image information determination module is configured to perform target image detection on the high-resolution image information result to obtain a target image detection result.

[0016] Based on the same inventive concept, a third aspect of the exemplary embodiment of the present disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0017] Based on the same inventive concept, a fourth aspect of the exemplary embodiments of the present disclosure provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method described in the first aspect.

[0018] Based on the same inventive concept, a fifth aspect of the exemplary embodiments of the present disclosure provides a computer program product, including computer program instructions. When the computer program instructions are executed on a computer, the computer executes the method described in the first aspect.

[0019] From the above, it can be seen that the embodiments of the present disclosure provide for determining traffic video information, classifying the video information based on different weather conditions to obtain scene classification information, extracting and restoring features of the traffic video information based on the scene classification information to obtain feature extraction image recovery information, inputting the feature extraction image recovery information into a pre-trained blind super-resolution degradation model, processing the generator network to obtain high-resolution image information, optimizing the high-resolution image information to obtain a high-resolution image information result, and performing target image detection on the high-resolution image information result to obtain a target image detection result. The present disclosure proposes a scene segmentation method for target detection in different weather conditions, which divides images into different scene categories, performs specific processing and optimization for different weather conditions, and uses a super-resolution model to pre-process data, enhance low-resolution images, and generate higher-quality high-resolution images, further reducing the impact of weather on images, and improving the accuracy of traffic target detection in low-resolution road videos in severe weather scenes such as rain and snow. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A schematic diagram of an application scenario of the traffic target detection method provided by an embodiment of the present disclosure;

[0022] Figure 2 A flow chart of a traffic target detection method provided by an embodiment of the present disclosure;

[0023] Figure 3 A schematic diagram of a structure of a traffic target detection device provided in an embodiment of the present disclosure;

[0024] Figure 4 The figure is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] It is understandable that before using the technical solutions disclosed in the embodiments of this application, the type, scope of use, usage scenarios, etc. of the personal information involved in this application should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0026] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present application according to the prompt message.

[0027] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0028] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation method of the present application. Other methods that meet the relevant laws and regulations may also be applied to the implementation method of the present application.

[0029] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0030] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, the principles and spirit of the present disclosure will be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0031] It should be understood herein that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction rather than having any limiting meaning.

[0032] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connecting" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. The article "one" or "a" before an element does not exclude the existence of multiple such elements.

[0033] The principle and spirit of the present disclosure are explained in detail below with reference to several representative embodiments of the present disclosure.

[0034] As described in the background technology, the existing related technologies cannot quickly identify different weather conditions and perform specific processing according to weather changes in the real world. Different weather conditions have different effects on images. For example, rain, snow and other weather conditions will make video images more difficult to identify, and video images in real scenes often have low resolution. Specifically:

[0035] The inventors of the present disclosure have discovered that the target detection method based on deep learning generates candidate regions for images using a specific method, and finally performs non-maximum suppression in the post-processing stage to eliminate redundant candidate regions, and finally generates the result after target detection. However, the target detection effect of these methods is based on good image data, ignoring the influence of low video pixels and weather factors in the real world, and resulting in low accuracy of traffic target detection in low-resolution road videos in severe weather scenes such as rain and snow.

[0036] In order to solve the above problems, the present disclosure provides a traffic target detection method, device, electronic device, storage medium and program product solution, which specifically include:

[0037] Determine the traffic video information, classify the video information based on different weather conditions to obtain scene classification information, extract and restore the traffic video information based on the scene classification information to obtain feature extraction image restoration information, input the feature extraction image restoration information into the pre-trained blind super-resolution degradation model, obtain high-resolution image information after generator network processing, optimize the high-resolution image information to obtain a high-resolution image information result, perform target image detection on the high-resolution image information result to obtain a target image detection result. The advantage of this solution is that the scene segmentation method in different weather conditions divides the image into different scene categories, performs specific processing and optimization for different weather conditions, and uses a super-resolution model to pre-process the data, enhance the low-resolution image, and generate a higher-quality high-resolution image, further reducing the impact of weather on the image, and improving the accuracy of traffic target detection in low-resolution road videos in bad weather scenes such as rain and snow.

[0038] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0039] refer to Figure 1 , which is an application scenario of the traffic target detection method provided by the exemplary embodiment of the present disclosure. In this application scenario, a terminal device 101, a server 102, a data storage system 103 and a client 104 are included. Among them, the terminal device 101, the server 102, the data storage system 103 and the client 104 can be connected through a wired or wireless communication network to achieve data interaction.

[0040] The terminal device 101 may be an electronic device with data transmission and multimedia input / output functions close to the user side, including but not limited to a desktop computer, a mobile phone, a mobile computer, a tablet computer, a media player, a smart wearable device, a personal digital assistant (PDA) or other electronic devices capable of realizing the above functions. The electronic device may include a processor and a display screen with a touch input function, the display screen is used to present a graphical user interface, the graphical user interface can display an application interface, and the processor is used to process application data, generate a graphical user interface, and control the display of the graphical user interface on the display screen.

[0041] Server 102 and data storage system 103 can both be independent physical servers, or a server cluster or distributed system composed of multiple physical servers. They can also be cloud servers that provide basic cloud computing services such as 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), as well as big data and artificial intelligence platforms.

[0042] In some exemplary embodiments, the traffic target detection method may be executed on the terminal device 101 or the server 102 .

[0043] When the traffic target detection method is run on the server 102 , the server 102 is used to provide traffic target detection services to users of the client 104 .

[0044] The terminal device 101 receives the traffic video information and sends it to the server 102;

[0045] The server 102 receives the traffic video information sent by the terminal device 101, and classifies the video information according to different weather conditions to obtain scene classification information;

[0046] The server 102 performs feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information;

[0047] The server 102 inputs the feature extraction image restoration information into the blind super-resolution degradation model pre-trained by the data storage system 103, and the server 102 obtains high-resolution image information through generator network processing, and then optimizes the high-resolution image information to obtain a high-resolution image information result;

[0048] The server 102 performs target image detection on the high-resolution image information result to obtain a target image detection result, and sends the target image detection result to the client 104;

[0049] The user can obtain traffic target image detection results in different weather scenarios through the client 104 .

[0050] A large amount of training data is stored in the data storage system 103, each of which includes a low-resolution and high-resolution image, synthetic data, degradation model parameters, specific types of artifact modeling, multi-frame image information fusion, and a probabilistic degradation model. The blind super-resolution degradation model can be trained based on a large amount of training data, so that the blind super-resolution degradation model can optimize the low-resolution image information. The source of the training data includes but is not limited to an existing database, data crawled from the Internet, or data uploaded by the user when using the client. When the sample information output by the blind super-resolution degradation model meets certain requirements and the label difference, the server 102 can provide users with a service of generating higher-quality high-resolution images based on the blind super-resolution degradation model.

[0051] It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this respect. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario.

[0052] refer to Figure 2 , a traffic target detection method, the method comprising the following steps:

[0053] Step S210: determine traffic video information, and classify the video information into scenes based on different weather conditions to obtain scene classification information.

[0054] Next, we will introduce how to obtain scene classification information:

[0055] In this exemplary embodiment, traffic video information is determined, and scene classification is performed on the video information based on different weather conditions to obtain scene classification information, including:

[0056] The received traffic video information is divided to obtain traffic video image frames, the traffic video image frames are scaled to a uniform size, and normalized to obtain normalized traffic video image frames, the normalized traffic video image frames are processed based on a convolutional neural network deep learning model to obtain a feature output model, and the traffic video image frames are input into the feature output model to obtain the scene classification information.

[0057] In specific implementation, the division method includes but is not limited to at least one of the following:

[0058] Python programming methods, professional frame acquisition software, and online frame acquisition tools.

[0059] In specific implementation, the normalization processing method includes but is not limited to at least one of the following:

[0060] Maximum and minimum normalization methods, normalization function, L1 normalization, and L2 normalization.

[0061] In this exemplary embodiment, the convolutional neural network deep learning model includes: a convolutional layer, a pooling layer and a fully connected layer;

[0062] The normalized traffic video image frame is processed based on the convolutional neural network deep learning model to obtain a feature output model, including:

[0063] Based on the convolution layer, scene feature extraction is performed on the normalized traffic video image frame based on weather scene feature information to obtain traffic video scene feature image information; based on the pooling layer, sampling is performed on the traffic video scene feature image information, and the information is input into the fully connected layer for mapping to obtain a feature model set; the feature model set is evaluated to obtain the feature output model.

[0064] In specific implementation, scene feature extraction processing includes:

[0065] First, it passes through a convolution layer. Assume that the image is I and its size is W*H*C, where W represents the width of the image, H represents the height of the image, and C represents the number of channels of the image. First, it passes through a convolution layer. Assume that its convolution kernel size is K and the step length is s. After convolution, the feature map F is generated, which is:

[0066]

[0067] Where i, j represent the width and height index of the feature map respectively, and C represents the number of channels of the input image. Represents the input image at position The pixel value at channel k is Indicates that the convolution kernel is at position , the parameter on channel k, b is the bias term, and σ represents the activation function.

[0068] In specific implementation, sampling processing includes:

[0069] After the pooling layer, Downsample the output of the convolutional layer to reduce the size of the feature map.

[0070] During the specific implementation, the validation set is used for evaluation, and finally the model with the highest evaluation result is selected.

[0071] Based on the above embodiment, the video information may be classified into scenes according to different weather conditions to obtain scene classification information.

[0072] Step S220: performing feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information.

[0073] Next, the scene classification information will be introduced;

[0074] The scene classification information includes at least one of the following: night image information, rainy day image information, foggy day image information and snowy day image information.

[0075] In the above exemplary embodiment, scene classification information is introduced. Now, a method for obtaining brightened night image information is introduced:

[0076] In this exemplary embodiment, obtaining the brightened night image information includes:

[0077] The night image information is subjected to denoising and downsampling processing, and image feature extraction is performed to obtain global image information and local image information of the night image information, feature fusion is performed on the global image information and the local image information, and convolution and deconvolution processing are performed to obtain brightened night image information.

[0078] In specific implementation, the denoising processing method includes but is not limited to at least one of the following:

[0079] Adaptive median filtering, Gaussian filtering, bilateral filtering and guided filtering.

[0080] In specific implementation, the downsampling processing method includes but is not limited to at least one of the following:

[0081] Pooling, convolution, and Laplacian pyramid.

[0082] In specific implementation, feature extraction includes:

[0083] Image feature extraction is performed using convolution and self-attention mechanisms to extract feature representations from the input low-resolution images, which capture both global and local information in the image.

[0084] In specific implementation, the feature fusion method includes:

[0085] The cross-layer attention mechanism is used to fuse feature representations at different levels, thereby improving the expressiveness and diversity of feature representations.

[0086] In the above exemplary embodiment, a method for obtaining the image information of the night after brightening is introduced. Now, a method for obtaining the image information of the rainy day after removing the rain is introduced:

[0087] In this exemplary embodiment, obtaining rainy day image information after rain removal includes:

[0088] Classify the rainy day image information based on a binary classifier to obtain raindrop pixel information, remove rain from the raindrop pixel information based on a rain removal module to obtain pixel information after rain removal, and perform feature fusion on the pixel information after rain removal and the rainy day image information to obtain rainy day image information after rain removal;

[0089] When implemented, the binary classifier includes:

[0090] A binary classifier is a common supervised learning model in machine learning that is used to classify data samples into two different categories. Its main purpose is to classify each data sample as one of the two different categories. The working principle of a binary classifier is usually based on a training data set, and it predicts the category of unknown data by learning from this data.

[0091] In a specific implementation, the raindrop pixel information is subjected to rain removal processing based on a rain removal module, including:

[0092] Rain Removal Module: Rain removal module models refer to a key component in deep learning frameworks or algorithms used for image rain removal. These models analyze and process elements such as raindrops and rain lines in images to restore clear background images;

[0093] After the rain removal module, each pixel is checked. If the pixel is marked as rainy, interpolation is used to remove rain from the pixel.

[0094] In specific implementation, feature fusion includes:

[0095] The attention mechanism is used to fuse the features of the original image and the derained and restored image to obtain richer and more accurate feature representation.

[0096] In the above exemplary embodiment, a method for obtaining rainy day image information after rain removal is introduced. Next, a method for obtaining foggy day image information after fog removal is introduced:

[0097] In this exemplary embodiment, obtaining foggy image information after defogging includes:

[0098] Based on multiple convolutional layers, feature extraction is performed on the foggy image information to obtain low-level feature information of the foggy image and high-level feature information of the foggy image information, feature fusion is performed on the low-level feature information of the foggy image and the high-level feature information of the foggy image, and based on guided filter processing, defogged foggy image information is obtained.

[0099] In a specific implementation, feature extraction is performed on the foggy image information based on multiple convolution layers, including:

[0100] By using multiple convolutional layers to extract foggy image features at different scales, including low-level feature extraction and high-level feature extraction, it can be expressed as:

[0101]

[0102]

[0103] in represents the input image, and Represents low-level features and high-level features, is a convolution operation.

[0104] In a specific implementation, the low-level feature information of the foggy image and the high-level feature information of the foggy image are subjected to feature fusion, including:

[0105] These features are integrated and passed through the feature perception module, so that the network can better understand the semantic information of the image and improve the defogging ability, which can be expressed as:

[0106]

[0107]

[0108] in, It is feature fusion, which combines low-level features with high-level features; It is a feature perception module that further extracts features from the fused features to improve the performance of the model.

[0109] In specific implementation, based on guided filter processing, it includes:

[0110] The weighted average of each pixel is performed according to the local structure and color information of the image, thereby reducing the impact of haze and improving the clarity and contrast of the image, and finally outputting the dehazed image. The guided filter can be expressed as:

[0111]

[0112] in Calculate the covariance of the input image, is to calculate the variance of the input image, ϵ is a very small constant to avoid division by 0; Represents the mean of the input image.

[0113] In the above exemplary embodiment, a method for obtaining foggy image information after defogging is introduced. Now, a method for obtaining snowy image information after snow removal is introduced:

[0114] In this exemplary embodiment, obtaining snowy day image information after snow removal includes:

[0115] Based on a multi-scale feature extractor, feature extraction is performed on the snow image information to obtain multi-scale feature information of the snow image, feature fusion is performed on the multi-scale feature information, and classification is performed based on an atmospheric light estimation method to obtain snowflake pixel information, and interpolation processing is performed on the snowflake pixel information to obtain snow image information after snow removal.

[0116] In specific implementation, the multi-scale feature extractor includes:

[0117] Multi-scale feature extractor is a widely used technique in computer vision tasks, especially in the fields of object detection, image classification and semantic segmentation. Its core purpose is to improve the understanding of image and video content by capturing feature information at different scales and levels, thereby improving the performance and accuracy of image processing tasks.

[0118] In specific implementation, feature fusion includes:

[0119] The attention mechanism is used to perform feature fusion on the multi-scale feature information.

[0120] In specific implementation, classification is performed based on the atmospheric light estimation method, including:

[0121] Atmospheric light estimation method: mainly based on image processing technology, by estimating the transmittance and atmospheric light components of the scene, the pixel value is adjusted to achieve the snow removal effect;

[0122] Snow removal is performed by atmospheric light estimation. In snow-covered areas, there are usually darker areas due to the reflection and occlusion effects of snow. Therefore, the brightest pixel in the image can be found as the estimated atmospheric light. After using atmospheric light estimation, the snow area in the image is marked as a darker area. In the end, each pixel will be marked as having snow or not.

[0123] In specific implementation, the interpolation process includes:

[0124] Interpolation is a method of inferring new data points between known discrete data points. It is widely used in image processing, numerical analysis, data preprocessing and other fields. Interpolation algorithms "guess" the values ​​of unknown locations by constructing new data points, thereby filling in missing parts of the data or smoothing the data.

[0125] In the above exemplary embodiments, the method of obtaining the night image information after brightening, the rainy day image information after rain removal, the foggy day image information after fog removal, and the snowy day image information after snow removal is introduced. The following is a method of obtaining the feature extraction and restoration information:

[0126] In this exemplary embodiment, feature extraction and recovery information is obtained, including: based on the night image information after brightening, the rainy day image information after rain removal, the foggy day image information after fog removal, and the snowy day image information after snow removal, the feature extraction and recovery information is obtained.

[0127] Step S230, inputting the feature extraction image restoration information into a pre-trained blind super-resolution degradation model, obtaining high-resolution image information through generator network processing, and then optimizing the high-resolution image information to obtain a high-resolution image information result;

[0128] Next, the blind super-resolution degradation model training method will be introduced;

[0129] In this exemplary embodiment, the blind super-resolution degradation model is trained by the following method, including:

[0130] Determine low-resolution image sample information and high-resolution image label information corresponding to the low-resolution image information, input the high-resolution image sample information into the blind super-resolution degradation model, obtain high-resolution image prediction information output by the model, determine the difference between the high-resolution image label information and the high-resolution image prediction information, and based on the difference, update the model parameters through a loss function until the difference between the high-resolution image label information and the high-resolution image prediction information is minimized, thereby obtaining the pre-trained blind super-resolution degradation model.

[0131] In a specific implementation, determining low-resolution image sample information and high-resolution image label information corresponding to the low-resolution image information includes:

[0132] Get a dataset containing low-resolution and high-resolution images and divide it into training, test, and validation sets in a ratio of 7:2:1.

[0133] In a specific implementation, the high-resolution image sample information is input into the blind super-resolution degradation model to obtain high-resolution image prediction information output by the model, including:

[0134] Define the generator (G) and the discriminator (D). The function of the generator is to convert the low-resolution input image into a high-resolution image. The function of the discriminator is to distinguish the generated high-resolution image from the real high-resolution image. The generator can be expressed as:

[0135]

[0136] in and Respectively represent the first convolution layer and the second convolution layer in the generator, σ represents the activation function, and the activation function used in the present invention is the ReLU activation function. Among them, the discriminator can be expressed as:

[0137]

[0138] in and They represent the first convolution layer and the second convolution layer in the discriminator respectively, σ represents the activation function, and the activation function used in the present invention is the ReLU activation function.

[0139] In a specific implementation, determining the difference between the high-resolution image label information and the high-resolution image prediction information includes:

[0140] The loss function is defined as the generator loss function and the discriminator loss function. Furthermore, the generator loss function includes adversarial loss and content loss. Among them, adversarial loss can be expressed as:

[0141]

[0142] in represents the distribution of real high-resolution images, Represents the distribution of low-resolution image inputs. The adversarial loss is used to force the generator to generate realistic images so that the discriminator cannot distinguish between generated images and real images. adv indicates that the loss is an adversarial loss. The content loss can be expressed as:

[0143]

[0144] in, represents the input low-resolution image, Represents the corresponding real high-resolution image, N represents the number of images, and content represents the loss is content loss. Content loss is used to maintain the content consistency between the generated image and the input image, and pixel-level loss is usually used to measure the difference between the two images. Furthermore, the discriminator loss also consists of two parts: real image loss and generated image loss. The real image loss indicates that the discriminator is expected to classify the real image as real. It can be expressed as:

[0145]

[0146] The real image loss indicates that we want the discriminator to classify the generated image as fake, which can be expressed as:

[0147]

[0148] Finally, the loss function of the combined generator and discriminator is called the overall loss function, which is the weighted sum of their respective losses, combined with weights:

[0149]

[0150]

[0151] in, , , and is the weight hyperparameter of the loss function.

[0152] Then train the generator G. First, input the low-resolution image into the generator G to obtain the generated high-resolution image. Then calculate the content loss between the generated high-resolution image and the real high-resolution image to optimize the parameters of the generator. Finally, calculate the loss of the generated high-resolution image after the discriminator D to train the generator adversarially. Specifically, first input the low-resolution image into the generator network, which can be expressed as:

[0153]

[0154] in Represents the generator network, continues to calculate the adversarial loss and content loss, and then obtains the total loss calculation through weighted addition.

[0155] In a specific implementation, based on the difference, the model parameters are updated through a loss function until the difference between the high-resolution image label information and the high-resolution image prediction information is minimized, thereby obtaining the pre-trained blind super-resolution degradation model, including:

[0156] The parameters of the generator are back-propagated according to the total loss and updated to reduce the total loss.

[0157] After that, the discriminator D is trained. First, the corresponding high-resolution image is input into the discriminator network to obtain the output of the discriminator for the real image, which can be expressed as:

[0158]

[0159] Then the generated high-resolution image is input into the discriminator network to obtain the output of the discriminator for the generated image, which can be expressed as:

[0160]

[0161] Then the output of the discriminator for the real image and the generated image is calculated, and the adversarial loss of the discriminator is calculated. The adversarial loss is weighted to obtain the total loss of the discriminator. Finally, the parameters of the discriminator are back-propagated according to the total loss, and the parameters are updated to reduce the total loss.

[0162] Finally, the generator and discriminator are trained alternately. Similarly, the validation set is used for evaluation, and the model with the highest evaluation result is finally selected as the final model of the classification network.

[0163] In the above exemplary embodiment, the blind super-resolution degradation model training method is introduced. The following is an introduction to the method of obtaining high-resolution image information results:

[0164] In this exemplary embodiment, the high-resolution image information result is obtained, including:

[0165] The feature-extracted image restoration information is input into a pre-trained blind super-resolution degradation model, processed by a generator network to obtain high-resolution image information, and then the high-resolution image information is optimized to obtain a high-resolution image information result.

[0166] In a specific implementation, the feature extraction image restoration information is input into a pre-trained blind super-resolution degradation model, and high-resolution image information is obtained through generator network processing, and then the high-resolution image information is optimized to obtain a high-resolution image information result, including:

[0167] First, the image output by the all-day adaptive module is obtained and input into the trained blind super-resolution degradation model (BSRGAN model) for inference. The image passes through the generator network to generate a high-resolution image. Then, the generated high-resolution image is denoised, sharpened, and color corrected to improve the image quality. Finally, the high-resolution image is output as the output result of the model.

[0168] Step S240: performing target image detection on the high-resolution image information result to obtain a target image detection result.

[0169] Next, we introduce how to get the target image detection results:

[0170] In this exemplary embodiment, the target image detection result is obtained, including:

[0171] The target is located on the high-resolution image information result based on an anchor-free method to obtain target image detection information, and the target image detection information is extracted based on a non-maximum suppression method to obtain the target image detection result.

[0172] In specific implementation, there is no anchor point method, including:

[0173] The anchor-free method is a method that does not require pre-set anchor boxes in object detection.

[0174] In specific implementation, the non-maximum suppression method includes:

[0175] Non-maximum suppression is an algorithm widely used in computer vision, especially in the field of target detection. Its main purpose is to filter out the best detection results from candidate detection frames and eliminate overlapping or redundant detection frames, thereby improving the accuracy and efficiency of target detection.

[0176] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.

[0177] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0178] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a traffic target detection device.

[0179] refer to Figure 3 , the traffic target detection device comprises:

[0180] The scene classification information determination module 310 is configured to determine the traffic video information, classify the video information according to different weather conditions, and obtain the scene classification information;

[0181] The feature extraction image restoration information determination module 320 is configured to perform feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information;

[0182] The high-resolution image information result determination module 330 is configured to input the feature extraction image restoration information into a pre-trained blind super-resolution degradation model, obtain high-resolution image information through generator network processing, and then optimize the high-resolution image information to obtain a high-resolution image information result;

[0183] The target image information determination module 340 is configured to perform target image detection on the high-resolution image information result to obtain a target image detection result.

[0184] In this exemplary embodiment, the scene classification information determination module 310 is specifically configured as follows:

[0185] The received traffic video information is divided to obtain traffic video image frames, the traffic video image frames are scaled to a uniform size, and normalized to obtain normalized traffic video image frames, scene feature extraction is performed on the normalized traffic video image frames based on the convolution layer based on weather scene feature information to obtain traffic video scene feature image information, the traffic video scene feature image information is sampled based on the pooling layer, and input into the fully connected layer for mapping to obtain a feature model set, the feature model set is evaluated to obtain the feature output model, and the traffic video image frames are input into the feature output model to obtain the scene classification information.

[0186] In this exemplary embodiment, the feature extraction image restoration information determination module 320 is specifically configured as follows:

[0187] The scene classification information includes at least one of the following: night image information, rainy day image information, foggy day image information and snowy day image information, and the traffic video information is subjected to feature extraction and recovery to obtain feature-extracted image recovery information, including at least one of the following: denoising and downsampling the night image information, and performing image feature extraction to obtain global image information and local image information of the night image information, and performing feature fusion on the global image information and the local image information, and performing convolution and deconvolution processing to obtain brightened night image information, and classifying the rainy day image information based on a binary classifier to obtain raindrop pixel information, and performing rain removal processing on the raindrop pixel information based on a rain removal module to obtain pixel information after rain removal, and performing feature fusion on the pixel information after rain removal and the rainy day image information to obtain a de-noised pixel information. The rainy day image information after the rain is performed, feature extraction is performed on the foggy day image information based on multiple convolution layers to obtain the foggy day image low-level feature information and foggy day image high-level feature information of the foggy day image information, feature fusion is performed on the foggy day image low-level feature information and the foggy day image high-level feature information, and based on the guided filter processing, the foggy day image information after defogging is obtained, feature extraction is performed on the snowy day image information based on the multi-scale feature extractor to obtain multi-scale feature information of the snowy day image, feature fusion is performed on the multi-scale feature information, and classification is performed based on the atmospheric light estimation method to obtain snowflake pixel information, interpolation processing is performed on the snowflake pixel information to obtain the snowy day image information after desnowing, and the feature extraction and recovery information is obtained based on the brightened night image information, the rainy day image information after deraining, the foggy day image information after defogging and the snowy day image information after desnowing.

[0188] In this exemplary embodiment, the high-resolution image information result determination module 330 is specifically configured as follows:

[0189] The feature extraction image restoration information is input into a pre-trained blind super-resolution degradation model, wherein the blind super-resolution degradation model is trained by the following method: determining low-resolution image sample information and high-resolution image label information corresponding to the low-resolution image information, inputting the resolution image sample information into the blind super-resolution degradation model, obtaining high-resolution image prediction information output by the model, determining the difference between the high-resolution image label information and the high-resolution image prediction information, and based on the difference, updating the model parameters through a loss function until the difference between the high-resolution image label information and the high-resolution image prediction information is minimized, thereby obtaining the pre-trained blind super-resolution degradation model. The high-resolution image information is obtained through generator network processing, and then the high-resolution image information is optimized to obtain a high-resolution image information result.

[0190] In this exemplary embodiment, the target image information determination module 340 is specifically configured as follows:

[0191] The target is located on the high-resolution image information result based on an anchor-free method to obtain target image detection information, and the target image detection information is extracted based on a non-maximum suppression method to obtain the target image detection result.

[0192] For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0193] The device of the above embodiment is used to implement the corresponding traffic target detection method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0194] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the traffic target detection method described in any of the above embodiments is implemented.

[0195] Figure 4 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0196] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0197] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0198] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0199] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0200] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0201] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0202] The electronic device of the above embodiment is used to implement the corresponding traffic target detection method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0203] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the traffic target detection method described in any of the above embodiments.

[0204] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0205] The above-mentioned non-transitory computer-readable storage medium can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)), etc.

[0206] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the traffic target detection method described in any embodiment in the above exemplary method part, and have the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0207] Based on the same inventive concept, corresponding to the traffic target detection method described in any of the above embodiments, the present disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer so that the computer and / or the processor execute the traffic target detection method. Corresponding to the execution subject corresponding to each step in each embodiment of the traffic target detection method, the processor that executes the corresponding step can belong to the corresponding execution subject.

[0208] The computer program product of the above embodiment is used to enable the computer and / or the processor to execute the traffic target detection method described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0209] Those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented as a system, method, or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." In addition, in some embodiments, the present disclosure may also be implemented in the form of a computer program product in one or more computer-readable media, which contains computer-readable program code.

[0210] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive examples) of computer-readable storage media may include, for example: 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 or 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 thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device, or device.

[0211] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media 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.

[0212] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0213] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0214] It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine, and these computer program instructions are executed by a computer or other programmable data processing device to produce a device that implements the functions / operations specified in the boxes in the flowchart and / or block diagram.

[0215] These computer program instructions may also be stored in a computer-readable medium that enables a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable medium produce a product that includes an instruction device that implements the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0216] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby enabling the instructions executed on the computer or other programmable device to provide a process for implementing the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0217] In addition, although the operations of the disclosed method are described in a particular order in the accompanying drawings, this does not require or imply that the operations must be performed in this particular order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted in the flow chart can be performed in a different order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be decomposed into multiple steps.

[0218] 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.

[0219] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0220] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0221] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power / ground connections to the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device may be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application are to be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0222] Although the present application has been described in conjunction with specific embodiments of the present application, many alternatives, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the discussed embodiments.

[0223] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.

[0224] Although the spirit and principle of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims. The scope of the attached claims conforms to the broadest interpretation, thereby including all such modifications and equivalent structures and functions.

Claims

1. A traffic target detection method, characterized in that: include: Determine traffic video information, divide the received traffic video information, and obtain traffic video image frames; Scaling the traffic video image frame to a uniform size and performing normalization processing to obtain a normalized traffic video image frame; Processing the normalized traffic video image frame based on a convolutional neural network deep learning model to obtain a feature output model; The convolutional neural network deep learning model includes: a convolutional layer, a pooling layer and a fully connected layer; Then the normalized traffic video image frame is processed based on the convolutional neural network deep learning model to obtain a feature output model, which specifically includes: Based on the convolution layer, the normalized traffic video image frame is subjected to scene feature extraction processing based on weather scene feature information to obtain traffic video scene feature image information; Based on the pooling layer, the characteristic image information of the traffic video scene is sampled and processed, and input into the fully connected layer for mapping processing to obtain a characteristic model set; Evaluating the feature model set to obtain the feature output model; Inputting the traffic video image frame into the feature output model to obtain scene classification information; According to the scene classification information, feature extraction and restoration are performed on the traffic video information to obtain feature extraction image restoration information; Inputting the feature-extracted image restoration information into a pre-trained blind super-resolution degradation model, obtaining high-resolution image information through generator network processing, and then optimizing the high-resolution image information to obtain a high-resolution image information result; Perform target image detection on the high-resolution image information result to obtain a target image detection result.

2. The method according to claim 1, characterized in that The scene classification information includes at least one of the following: night image information, rainy day image information, foggy day image information and snowy day image information; The step of performing feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information includes at least one of the following: Denoising and downsampling the night image information, and performing image feature extraction to obtain global image information and local image information of the night image information, performing feature fusion on the global image information and the local image information, and performing convolution and deconvolution processing to obtain brightened night image information; Classify the rainy day image information based on a binary classifier to obtain raindrop pixel information, remove rain from the raindrop pixel information based on a rain removal module to obtain pixel information after rain removal, and perform feature fusion on the pixel information after rain removal and the rainy day image information to obtain rainy day image information after rain removal; Extracting features of the foggy image information based on multiple convolution layers to obtain low-level feature information of the foggy image information and high-level feature information of the foggy image information, fusing the low-level feature information of the foggy image and the high-level feature information of the foggy image, and processing based on a guided filter to obtain defogged foggy image information; Extracting features of the snow image information based on a multi-scale feature extractor to obtain multi-scale feature information of the snow image, performing feature fusion on the multi-scale feature information, and classifying the multi-scale feature information based on an atmospheric light estimation method to obtain snow pixel information, and interpolating the snow pixel information to obtain snow image information after snow removal; The feature-extracted image restoration information is obtained based on the brightened night image information, the rainy day image information after rain removal, the foggy day image information after fog removal, and the snowy day image information after snow removal.

3. The method according to claim 1, characterized in that: The feature extraction image restoration information is input into a pre-trained blind super-resolution degradation model, and high-resolution image information is obtained through generator network processing, and then the high-resolution image information is optimized to obtain a high-resolution image information result, which also includes: training the blind super-resolution degradation model by the following method: Determine low-resolution image sample information and high-resolution image label information corresponding to the low-resolution image sample information; Inputting the resolution image sample information into the blind super-resolution degradation model to obtain high-resolution image prediction information output by the model; determining a difference between the high-resolution image label information and the high-resolution image prediction information; Based on the difference, the model parameters are updated through a loss function until the difference between the high-resolution image label information and the high-resolution image prediction information is minimized, thereby obtaining the pre-trained blind super-resolution degradation model.

4. The method according to claim 1, characterized in that: The target image detection is performed on the high-resolution image information result to obtain the target image detection result: Performing target positioning on the high-resolution image information result based on an anchor-free method to obtain target image detection information; The target image detection information is extracted based on a non-maximum suppression method to obtain the target image detection result.

5. A traffic target detection device, characterized in that: include: A scene classification information determination module is configured to determine traffic video information, divide the received traffic video information, and obtain traffic video image frames; Scaling the traffic video image frame to a uniform size and performing normalization processing to obtain a normalized traffic video image frame; Processing the normalized traffic video image frame based on a convolutional neural network deep learning model to obtain a feature output model; The convolutional neural network deep learning model includes: a convolutional layer, a pooling layer and a fully connected layer; Then the normalized traffic video image frame is processed based on the convolutional neural network deep learning model to obtain a feature output model, which specifically includes: Based on the convolution layer, the normalized traffic video image frame is subjected to scene feature extraction processing based on weather scene feature information to obtain traffic video scene feature image information; Based on the pooling layer, the characteristic image information of the traffic video scene is sampled and processed, and input into the fully connected layer for mapping processing to obtain a characteristic model set; Evaluating the feature model set to obtain the feature output model; Inputting the traffic video image frame into the feature output model to obtain scene classification information; A feature extraction image restoration information determination module is configured to perform feature extraction and restoration on the traffic video information according to the scene classification information to obtain feature extraction image restoration information; A high-resolution image information result determination module is configured to input the feature extraction image restoration information into a pre-trained blind super-resolution degradation model, obtain high-resolution image information through generator network processing, and then optimize the high-resolution image information to obtain a high-resolution image information result; The target image information determination module is configured to perform target image detection on the high-resolution image information result to obtain a target image detection result.

6. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 4 is implemented.

7. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 4.

8. A computer program product, characterized in that The method comprises computer program instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 4.

Images