Day and night vehicle re-identification method and device based on monitoring scene

Abstract

The invention relates to the technical field of image processing, and discloses a day and night vehicle re-identification method and device based on a monitoring scene, and the method comprises the steps: obtaining video image data sets of the same vehicle in different domains in the daytime and at night; preprocessing the video image data set; performing two-dimensional target detection on the preprocessed video image data set by constructing a preset target vehicle detection network to obtain vehicle image data; and constructing a preset day and night cross-domain vehicle re-identification model, wherein the preset day and night cross-domain vehicle re-identification model comprises a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module which carry out vehicle re-identification on the vehicle image data. According to the method provided by the invention, in model design, the preset night glare suppression module is constructed, and the attention of the model is guided to a non-glare area by utilizing glare visual prompt, so that the influence of glare of a vehicle headlamp is reduced, and the vehicle re-identification capability of the model in difficult scenes such as day and night domain difference is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a daytime and nighttime vehicle re-identification method and device based on monitoring scenes. Background Art

[0002] In recent years, urban development has led to an increase in population, car ownership, and traffic flow, and increased traffic complexity. At the same time, with the advancement of software and hardware technology and algorithms, data can be collected by installing traffic cameras to effectively support traffic scene analysis and improve analysis accuracy. Among them, the vehicle re-identification (ReID) task has important applications. Vehicle re-identification refers to the technology of finding the specified vehicle in images taken by different cameras given an image of a vehicle. This plays an important role in vehicle tracking, vehicle behavior judgment, suspicious vehicle search, vehicle automatic parking fee management, urban security and other applications in traffic scenes.

[0003] At present, vehicle ReID methods mainly use visible light cameras as sensors to find the same target vehicle in videos or images captured by different visible light cameras. With the advancement of computer vision algorithms, the ReID methods based on visible light cameras have achieved good results in vehicle re-identification tasks in day-to-day scenes. However, in scenes with poor lighting conditions or complex scenes such as day-to-night scenes, the ReID methods based on visible light cameras still have some limitations. Therefore, some methods introduce infrared cameras to improve the vehicle re-identification accuracy in day-to-night cross-domain scenes. Infrared sensors can capture the thermal radiation of objects, but are easily affected by noise and have difficulty capturing the texture of objects. By fusing infrared images and visible light images, the two modal data can be complemented to obtain a multimodal fusion image containing both thermal objects and texture details, combining the advantages of the two spectral images to improve the accuracy of vehicle re-identification. However, this method has certain limitations in practical applications. First, thermal infrared cameras are expensive and have limited resolution, which makes their application in traffic conditions challenging. Second, thermal infrared cameras rely on the emitted near-infrared light, which is easily affected by car headlights, street lights, and building lights in night traffic scenes. Summary of the invention

[0004] In view of this, the present invention provides a day and night vehicle re-identification method and device based on monitoring scenes to overcome the problems of high price and susceptibility to noise caused by the use of thermal infrared cameras, resulting in low vehicle re-identification accuracy.

[0005] In a first aspect, the present invention provides a day and night vehicle re-identification method based on a monitoring scene, the method comprising:

[0006] Obtain video image datasets of the same vehicle in different domains during the day and at night;

[0007] Preprocess the video image dataset;

[0008] By constructing a preset target vehicle detection network, two-dimensional target detection is performed on the preprocessed video image data set to obtain vehicle image data;

[0009] A preset day and night cross-domain vehicle re-identification model is constructed to perform vehicle re-identification on vehicle image data. The preset day and night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module.

[0010] In an optional implementation, the daytime-nighttime cross-domain image of the same vehicle is expressed by the following formula:

[0011] I={I day ,I night ,y}

[0012] Among them, I represents the day-night cross-domain image of the same vehicle, I day ,I night They represent vehicle images during the day and at night respectively, and y represents the identification label of the vehicle.

[0013] In an optional implementation, the preset target vehicle detection network includes: a feature extraction network and a target detection network, wherein:

[0014] Feature extraction network, used to output feature representation of the image;

[0015] The target detection network is used to output the category and location information of the target vehicle and extract the vehicle image.

[0016] In an optional implementation, the output of the target detection network is expressed by the following formula:

[0017] O i =[c i ,x i ,y i ,w i ,h i ]

[0018] Among them, O i Represented as the i-th target vehicle, c i Indicates the vehicle category of the extraction target, x i ,y i are the center point coordinates of the target vehicle rectangular frame, wi ,h i are the length and width of the target vehicle respectively.

[0019] In an optional implementation, the preset day-night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day-night cross-domain vehicle re-identification module, wherein:

[0020] A preset nighttime glare suppression module is used to direct the attention of the preset daytime and nighttime cross-domain vehicle re-identification model to the glare-free area;

[0021] A preset day and night cross-domain vehicle re-identification module is used to extract vehicle image features from images after glare suppression, and to perform vehicle re-identification on the vehicle image data when preset conditions are met.

[0022] In an optional implementation, the preset conditions include: a classification loss function, and / or a metric loss function, wherein:

[0023] The classification loss function is used to determine whether the vehicles in the two domains are the same vehicle;

[0024] Metric loss function, used to measure the distance between the same vehicle in different domains.

[0025] In a second aspect, the present invention provides a daytime and nighttime vehicle re-identification device based on a monitoring scene, the device comprising:

[0026] A data acquisition module is used to acquire video image datasets of the same vehicle in different domains during the day and at night;

[0027] A preprocessing module, used for preprocessing the video image data set;

[0028] The target detection module is used to perform two-dimensional target detection on the preprocessed video image data set by constructing a preset target vehicle detection network to obtain vehicle image data;

[0029] The vehicle re-identification module is used to construct a preset day and night cross-domain vehicle re-identification model to perform vehicle re-identification on vehicle image data. The preset day and night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module.

[0030] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the day and night vehicle re-identification method based on the monitoring scene of the above-mentioned first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0031] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the monitoring scene-based day and night vehicle re-identification method of the first aspect or any corresponding embodiment thereof.

[0032] The present invention provides a method and device for day and night vehicle re-identification based on monitoring scenarios. By constructing matching data sets of the same vehicle during the day and at night, a preset night glare suppression module is constructed in the model design. The glare visual cues are used to guide the model's attention to the glare-free area, reduce the impact of vehicle headlight glare, and improve the model's vehicle re-identification capability in difficult scenarios such as day and night domain differences. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 is a flow chart of a method for day and night vehicle re-identification based on a monitoring scene according to an embodiment of the present invention;

[0035] Figure 2 is a structural block diagram of a day and night vehicle re-identification device based on a monitoring scene according to an embodiment of the present invention;

[0036] Figure 3 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0038] According to an embodiment of the present invention, an embodiment of a day and night vehicle re-identification method based on a monitoring scene is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] In this embodiment, a daytime and nighttime vehicle re-identification method based on a monitoring scene is provided, which can be used in the above-mentioned mobile terminals, such as mobile phones, tablet computers, etc. Figure 1 is a flow chart of a method for re-identifying a vehicle during the day and at night based on a monitoring scene according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0040] Step S101, obtaining video image datasets of the same vehicle in different domains during the day and at night.

[0041] In the embodiment of the present invention, the video image data set includes: video image data of different scenes and different monitoring angles. Different scenes include: visible light camera data of different seasons, different time periods, different lighting conditions, and different urban areas. Different monitoring angles refer to different camera positions, such as parallel installation, overhead installation, etc. This is only an example and is not limited to this. In actual applications, the corresponding data set is selected according to the actual situation.

[0042] Specifically, by using the target detection algorithm or manual annotation, the vehicle area in the monitoring scene image is cropped out to obtain the images of the same vehicle in daytime and nighttime scenes. The daytime-nighttime cross-domain image of the same vehicle is expressed by the following formula:

[0043] I={I day ,I night ,y}

[0044] Among them, I day ,I night They represent vehicle images during the day and at night respectively, and y represents the identification label of the vehicle.

[0045] Step S102: pre-processing the video image data set.

[0046] In an embodiment of the present invention, preprocessing includes performing sample balancing on the daytime-nighttime data of the same vehicle. This is only an example and is not intended to be limiting. In practical applications, a corresponding preprocessing method is selected according to actual conditions, and the means of sample balancing are not limited here, and are selected accordingly according to actual conditions. For example: In the existing daytime-nighttime cross-domain vehicle re-identification dataset, the amount of nighttime data for the same vehicle is limited, so there is a large sample imbalance problem for the daytime-nighttime data of the same vehicle. Therefore, the present invention performs sample balancing on the daytime-nighttime data of the same vehicle, so that the amount of data during the day and at night is balanced, thereby reducing the adverse effects of sample imbalance on the model.

[0047] Step S103, by constructing a preset target vehicle detection network, two-dimensional target detection is performed on the preprocessed video image data set to obtain vehicle image data.

[0048] In an embodiment of the present invention, in order to reduce manual labeling and improve the efficiency of data construction, a target detection method is used to crop the vehicle from the monitoring image, and a preset target vehicle detection network is constructed to perform two-dimensional target detection to obtain vehicle image data.

[0049] Specifically, the preset target vehicle detection network includes: a feature extraction network and a target detection network, wherein the feature extraction network is used to output the feature representation of the image; the target detection network is used to output the category and position information of the target vehicle and extract the vehicle image.

[0050] Specifically, the feature extraction part uses a backbone network, including but not limited to convolutional neural networks such as ResNet, Mobile Net, and VGG, with the input being an image and the output being a feature representation of the image. For the target detection network part, it includes but is not limited to using commonly used target detection methods such as YOLO, SSD, and CenterNet, and the output is the category and location information of the target;

[0051] The output is defined as:

[0052] O i =[c i ,x i ,y i ,w i ,h i ]

[0053] O i Represented as the i-th target vehicle, c i Indicates the category of the target vehicle to be extracted, x i ,y i are the center point coordinates of the target vehicle rectangular frame, w i ,h i are the length and width of the target vehicle respectively.

[0054] Specifically, according to the target detection result, the coordinate position of the vehicle two-dimensional target detection box is obtained, and the vehicle image is extracted from the source image as the image data of the vehicle re-identification data set.

[0055] Step S104, constructing a preset day and night cross-domain vehicle re-identification model to perform vehicle re-identification on vehicle image data. The preset day and night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module.

[0056] In an embodiment of the present invention, the preset day-night cross-domain vehicle re-identification model includes: a preset night-time glare suppression module and a preset day-night cross-domain vehicle re-identification module, wherein the preset night-time glare suppression module is used to direct the attention of the preset day-night cross-domain vehicle re-identification model to the glare-free area, thereby reducing the impact of vehicle headlight glare; the preset day-night cross-domain vehicle re-identification module is used to extract vehicle image features from the image after glare suppression, and when the preset conditions are met, perform vehicle re-identification on the vehicle image data.

[0057] Specifically, in night scenes, vehicles usually turn on their headlights for illumination, resulting in excessive brightness in a certain local area of ​​the image, thus forming glare, which affects the accuracy of day-night vehicle re-identification. Therefore, a preset night glare suppression module is added to direct the model's attention to the glare-free area, thereby reducing the impact of vehicle headlight glare.

[0058] Specifically, first, the night image is transformed into features, which can be expressed as:

[0059]

[0060] Among them, I night Represents the input night image, Conv() represents a convolution operation with a convolution kernel size of 7X7, BN() represents a normalization operation, ReLU represents an activation function layer, and MaxPooling represents a maximum pooling operation. Represents the feature vector after feature transformation;

[0061] Specifically, the night image I night Perform grayscale transformation operation, which is specifically expressed as:

[0062] I gray = rgb2gray(I night )

[0063] Specifically, the highlight pixels in the night image are distinguished by setting the brightness threshold thre. Thre can be set to 220. It is set accordingly according to the specific situation and is not limited here. After identifying the highlighted pixels, the pixels adjacent to the highlighted pixels are combined to form a highlight area, and the areas with fewer highlighted pixels are discarded. Finally, a binary mask image is obtained from the original night image, which is represented by M gray , where the area with a pixel value of 1 is a highlight (i.e., glare) area, and the area with a pixel value of 0 is a non-glare area.

[0064] Specifically, the transformed feature vector is then With glare mask figure M grayPerform feature concatenation and input a learnable projection vector for feature learning, which is specifically expressed as:

[0065]

[0066] in, M gray ∈R H0×W0×1 , the feature vector With glare mask figure M gray Concatenate in the channel dimension, V represents a learnable projection vector, concat means concatenation in the channel dimension, Sigmoid means the sigmoid function; represents the learned domain suppression mask.

[0067] Specifically, the daytime image is used to guide glare suppression, which is specifically expressed as:

[0068]

[0069] Among them, zeros() means setting all pixels to 0. Represents a daytime image without the effects of glare.

[0070] Based on domain suppression mask, The final inhibition process is expressed as:

[0071]

[0072] in, It represents the feature representation after glare suppression, α is the weight coefficient, which can be set to 0.5 to balance the original features and the suppressed features.

[0073] In an embodiment of the present invention, a preset day-night cross-domain vehicle re-identification module is constructed, and vehicle image features are extracted from images after glare suppression through a backbone network.

[0074] Specifically, the function of the backbone network is to extract image features. The input image size is H*W*C, where the input night image is the above-mentioned image after glare suppression, where H and W represent the height and width of the image respectively, and C represents the number of channels. In this case, C is 3, indicating an RGB three-channel image.

[0075] Specifically, the backbone network uses a convolutional combination method such as convolution layer-normalization layer-activation function layer to perform superposition operations of multiple convolutional combination layers, and performs downsampling once in each convolutional combination operation, including but not limited to using backbone networks such as ResNet, VGG, and MobileNet. Normalization layers include but are not limited to instance normalization layers, adaptive instance normalization layers, etc. Non-linear activation layers include but are not limited to non-linear activation functions such as ReLU and Leaky ReLU. Make corresponding selections based on actual conditions.

[0076] Specifically, in order to enhance the feature extraction of daytime and nighttime image pairs, the present invention uses ResNet50 as the backbone network. ResNet solves the gradient vanishing and performance saturation problems in deep networks by using residual connections. In ResNet50, each residual block contains multiple convolutional layers and a skip connection, so that the network can maintain good performance while deepening. The output feature map of each residual block in the backbone network is defined as:

[0077]

[0078] Among them, s∈{1,2,3,4,5} represents the different stages of residual block output, and m∈{Day,Night} represents the cross-domain images during the day and at night; among them, the night image is the image after glare suppression.

[0079] In the embodiment of the present invention, the preset conditions include: a classification loss function, and / or a measurement loss function, wherein the classification loss function is used to determine whether the vehicles in two domains are the same vehicle; the measurement loss function is used to measure the distance between the same vehicles in different domains. This is only an example, and is not limited to this. In actual applications, the corresponding conditions are selected according to the actual situation. For example: the classification loss function and the measurement loss function are both less than a certain value, and the value range is not limited here.

[0080] Specifically, the last layer of features in the above steps is first processed, and the final corresponding feature vector is obtained by using the global average pooling operation, which is specifically expressed as:

[0081]

[0082] Among them, GAP represents the global average pooling operation.

[0083] Specifically, the loss function consists of two parts. The first part is the classification loss function, which uses the vehicle ID as a training label to determine whether the vehicles in the two domains are the same vehicle. The cross entropy loss function is used for calculation, which is expressed as:

[0084]

[0085] in, Represents the vehicle ID category of the predicted output, and L represents the true value of the annotation;

[0086] The second is the metric learning loss function, which aims to make the distance between two positive samples (the same vehicle in different domains) as small as possible and the distance between negative samples as large as possible (different vehicles in different domains). The triplet loss function is used. The triplet consists of Anchor (A), Negative (N), and Positive (P). The base point is A, the positive match is P, and the negative match is N. The purpose of the triplet loss function is to make the distance between features of the same class as small as possible, and the distance between features of different classes as large as possible, which can be expressed as:

[0087]

[0088] Represents the distance between two identical vehicle features, the goal is to continuously reduce, It represents the distance between two different vehicle features. The purpose is to provide continuous reminders. α is a constant parameter, which is selected according to the actual situation. When the preset conditions are met, the model training is completed and the vehicle image data is re-identified.

[0089] The day and night vehicle re-identification method based on monitoring scenes provided in this embodiment uses only visible light camera data to construct matching data sets of the same vehicle during the day and at night. In terms of model design, a preset night glare suppression module is constructed, and glare visual cues are used to guide the model's attention to glare-free areas, thereby reducing the impact of vehicle headlight glare and improving the model's vehicle re-identification capability in difficult scenarios such as day and night domain differences.

[0090] In this embodiment, a daytime and nighttime vehicle re-identification device based on a monitoring scene is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0091] This embodiment provides a daytime and nighttime vehicle re-identification device based on monitoring scenes, such as Figure 2 As shown, including:

[0092] The data acquisition module 201 is used to acquire video image datasets of the same vehicle in different domains during the day and at night;

[0093] A preprocessing module 202, used for preprocessing the video image data set;

[0094] The target detection module 203 is used to perform two-dimensional target detection on the preprocessed video image data set by constructing a preset target vehicle detection network to obtain vehicle image data;

[0095] The vehicle re-identification module 204 is used to construct a preset day-night cross-domain vehicle re-identification model to perform vehicle re-identification on vehicle image data. The preset day-night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day-night cross-domain vehicle re-identification module.

[0096] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0097] The day and night vehicle re-identification device based on the monitoring scene in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0098] The embodiment of the present invention also provides a computer device having the above Figure 3 The day and night vehicle re-identification device based on monitoring scenes is shown.

[0099] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 3 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to an interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 3 A processor 10 is taken as an example.

[0100] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0101] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0102] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage devices. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0103] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0104] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0105] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0106] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A day and night vehicle re-identification method based on monitoring scenes, characterized in that: The method comprises: Obtain video image datasets of the same vehicle in different domains during the day and at night; Preprocess the video image dataset; By constructing a preset target vehicle detection network, two-dimensional target detection is performed on the preprocessed video image data set to obtain vehicle image data; A preset day and night cross-domain vehicle re-identification model is constructed to perform vehicle re-identification on vehicle image data. The preset day and night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module.

2. The method according to claim 1, characterized in that The day-night cross-domain image of the same vehicle is expressed by the following formula: I={I day ,I night ,y} Among them, I represents the day-night cross-domain image of the same vehicle, I day ,I night They represent vehicle images during the day and at night respectively, and y represents the identification label of the vehicle.

3. The method according to claim 1, characterized in that The preset target vehicle detection network includes: a feature extraction network and a target detection network, wherein: Feature extraction network, used to output feature representation of the image; The target detection network is used to output the category and location information of the target vehicle and extract the vehicle image.

4. The method according to claim 3, characterized in that The output of the object detection network is expressed by the following formula: O i =[c i ,x i ,y i ,w i ,h i ] Among them, O i Represented as the i-th target vehicle, c i Indicates the vehicle category of the extraction target, x i ,y i are the center point coordinates of the target vehicle rectangular frame, w i ,h i are the length and width of the target vehicle respectively.

5. The method according to claim 1, characterized in that The preset day-night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day-night cross-domain vehicle re-identification module, wherein: A preset nighttime glare suppression module is used to direct the attention of the preset daytime and nighttime cross-domain vehicle re-identification model to the glare-free area; A preset day and night cross-domain vehicle re-identification module is used to extract vehicle image features from images after glare suppression, and to perform vehicle re-identification on the vehicle image data when preset conditions are met.

6. The method according to claim 5, characterized in that The preset conditions include: classification loss function, and / or measurement loss function, where: The classification loss function is used to determine whether the vehicles in the two domains are the same vehicle; Metric loss function, used to measure the distance between the same vehicle in different domains.

7. A day and night vehicle re-identification device based on monitoring scenes, characterized in that: The device comprises: A data acquisition module is used to acquire video image datasets of the same vehicle in different domains during the day and at night; A preprocessing module, used for preprocessing the video image data set; The target detection module is used to perform two-dimensional target detection on the preprocessed video image data set by constructing a preset target vehicle detection network to obtain vehicle image data; The vehicle re-identification module is used to construct a preset day and night cross-domain vehicle re-identification model to perform vehicle re-identification on vehicle image data. The preset day and night cross-domain vehicle re-identification model includes: a preset night glare suppression module and a preset day and night cross-domain vehicle re-identification module.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the daytime and nighttime vehicle re-identification method based on monitoring scenes as described in any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the day and night vehicle re-identification method based on monitoring scenes according to any one of claims 1 to 6.

Images