Traffic sign recognition system, method and vehicle
By combining vehicle behavior information acquired by camera modules and sensors, and using training modules and neural networks for traffic sign recognition, the problem of low recognition accuracy in existing technologies has been solved, achieving higher recognition accuracy and robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUANGZHOU AUTOMOBILE GROUP CO LTD
- Filing Date
- 2022-02-25
- Publication Date
- 2026-05-29
AI Technical Summary
In existing technologies, traffic sign recognition relies solely on cameras and cannot be combined with vehicle behavior information, resulting in low recognition accuracy.
Traffic sign recognition results are obtained through a camera module, and the behavior information of the vehicle and nearby vehicles is obtained through sensors. Information is fused using a training module and a recurrent neural network module, including a long short-term memory artificial neural network and a Kalman filter, to improve recognition accuracy.
It improves the accuracy and robustness of traffic sign recognition by integrating vehicle behavior information and camera recognition results, thereby reducing the error caused by relying solely on camera recognition.
Smart Images
Figure CN116868246B_ABST
Abstract
Description
Technical Field
[0001] This application relates to road safety, and more particularly to a traffic sign recognition system, method, and vehicle. Background Technology
[0002] Detecting and recognizing traffic signs is fundamental to driving. Traffic sign detection and recognition are also essential for autonomous driving; therefore, autonomous driving technology requires vehicles to be able to recognize traffic signs. Currently, traffic sign recognition technology only uses cameras to detect and recognize traffic signs and cannot be combined with vehicle behavior information.
[0003] Therefore, there is room for improvement. Summary of the Invention
[0004] In view of this, it is necessary to provide a traffic sign recognition system, method, and vehicle that can combine traffic signs recognized by cameras with vehicle behavior information for traffic sign recognition.
[0005] Firstly, this application provides a traffic sign recognition system, comprising: a camera module for acquiring a first traffic sign recognition result; a sensor for acquiring behavioral information of the vehicle and nearby vehicles; a training module connected to the sensor, the training module being used to output traffic sign recognition parameters based on the behavioral information of the vehicle and nearby vehicles; and a recurrent neural network module connected to the training module and the camera module; wherein the recurrent neural network module is used to output a second traffic sign recognition result based on the traffic sign recognition parameters and the first traffic sign recognition result. Clearly, the traffic sign recognition system can be trained using the behavioral information of the vehicle and nearby vehicles and the first traffic sign recognition result acquired by the camera module, thereby improving the accuracy of traffic sign recognition.
[0006] In one possible design, the training module is a long short-term memory artificial neural network module.
[0007] In one possible design, the long short-term memory artificial neural network module includes an information input layer; the information input layer is connected to the sensor to acquire behavioral information of the vehicle and nearby vehicles.
[0008] In one possible design, the long short-term memory artificial neural network module further includes an information filtering layer; the information filtering layer is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0009] In one possible design, the traffic sign recognition parameters include: the vehicle center position, yaw angle, speed, yaw rate, vehicle acceleration, vehicle width and height, and vehicle category.
[0010] In one possible design, the traffic sign recognition system further includes a Kalman filter connected to the recurrent neural network module, wherein the Kalman filter is used to fuse the first traffic sign recognition result with the second traffic sign recognition result to generate a third traffic sign recognition result.
[0011] In one possible design, the training module includes a first convolutional neural network module and a second convolutional neural network module. The first convolutional neural network module is connected to the camera module and the recurrent neural network module, and the second convolutional neural network module is connected to the sensor and the recurrent neural network module. The camera module is used to acquire image information; the first convolutional neural network module is used to generate a first traffic sign recognition result based on the image information; and the second convolutional neural network module is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0012] Secondly, this application provides a traffic sign recognition method, comprising: obtaining a first traffic sign recognition result; obtaining behavioral information of the vehicle and nearby vehicles; outputting traffic sign recognition parameters based on the behavioral information of the vehicle and nearby vehicles; and outputting a second traffic sign recognition result based on the traffic sign recognition parameters and the first traffic sign recognition result, wherein the traffic sign recognition parameters include the traffic sign recognition parameters and the first traffic sign recognition result.
[0013] In one possible design, the method further includes: outputting the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0014] In one possible design, the traffic sign recognition parameters include: the vehicle center position, yaw angle, speed, yaw rate, vehicle acceleration, vehicle width and height, and vehicle category.
[0015] In one possible design, the traffic sign recognition method further includes: fusing the first traffic sign recognition result with the second traffic sign recognition result to generate a third traffic sign recognition result.
[0016] In one possible design, the method further includes: acquiring image information; and generating a first traffic sign recognition result based on the image information.
[0017] In one possible design, the method further includes: outputting the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0018] Thirdly, this application provides a vehicle, including: a traffic sign recognition system, the traffic sign recognition system including: a camera module for acquiring a first traffic sign recognition result; a sensor for acquiring behavioral information of the vehicle and nearby vehicles; a training module connected to the sensor, the training module being used to output traffic sign recognition parameters based on the behavioral information of the vehicle and nearby vehicles; and a recurrent neural network module connected to the training module and the camera module;
[0019] The recurrent neural network module is used to output a second traffic sign recognition result based on the traffic sign recognition parameters and the first traffic sign recognition result.
[0020] In one possible design, the training module is a long short-term memory artificial neural network module.
[0021] In one possible design, the long short-term memory artificial neural network module includes an information input layer; the information input layer is connected to the sensor to acquire behavioral information of the vehicle and nearby vehicles.
[0022] In one possible design, the long short-term memory artificial neural network module further includes an information filtering layer; the information filtering layer is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0023] In one possible design, the traffic sign recognition parameters include: the vehicle center position, yaw angle, speed, yaw rate, vehicle acceleration, vehicle width and height, and vehicle category.
[0024] In one possible design, the vehicle further includes a Kalman filter connected to the recurrent neural network module, wherein the Kalman filter is used to fuse the first traffic sign recognition result with the second traffic sign recognition result to generate a third traffic sign recognition result.
[0025] In one possible design, the training module includes a first convolutional neural network module and a second convolutional neural network module. The first convolutional neural network module is connected to the camera module and the recurrent neural network module, and the second convolutional neural network module is connected to the sensor and the recurrent neural network module. The camera module is used to acquire image information, the first convolutional neural network module is used to generate a first traffic sign recognition result based on the image information, and the second convolutional neural network module is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
[0026] The technical effects of the second and third aspects can be found in the description of the traffic sign recognition system mentioned in the first aspect above, and will not be repeated here. Attached Figure Description
[0027] The implementation of this application will now be described with reference to the accompanying drawings and embodiments.
[0028] Figure 1 This application provides a traffic sign recognition system for autonomous driving technology, as one embodiment of the present application.
[0029] Figure 2 A traffic sign recognition system provided in another embodiment of this application.
[0030] Figure 3 A traffic sign recognition system provided in another embodiment of this application.
[0031] Figure 4 This is a schematic flowchart of a traffic sign recognition method for autonomous vehicles provided in an embodiment of this application.
[0032] Figure 5 This is a schematic diagram of an autonomous vehicle provided in one embodiment of this application.
[0033] Explanation of main component symbols
[0034] Traffic sign recognition system 10; 20 camera modules 100; sensors 200
[0035] Training module 300; 300a Long Short-Term Memory Network module 310 First CNN module 320
[0036] The second CNN module has 330 RNN modules, 400 Kalman filters, and 500 vehicles.
[0037] The following detailed description, in conjunction with the accompanying drawings, will further illustrate this application. Detailed Implementation
[0038] It should be understood that, for the sake of simplicity and clarity, reference numerals are repeated in different figures where appropriate to denote corresponding or similar elements. Furthermore, numerous specific details are set forth to provide a thorough understanding of the embodiments described herein. However, those skilled in the art will understand that the embodiments described herein can be practiced without these specific details. In other instances, methods, processes, and components have not been described in detail so as not to obscure the relevant features described herein. The figures are not necessarily to scale, and some portions may be enlarged to better show details and features. This description should not be considered as limiting the scope of the embodiments described herein.
[0039] The following are definitions of several terms used throughout this application.
[0040] The term "coupling" is defined as a connection, whether direct or indirect through intermediate components, and is not necessarily limited to a physical connection. A connection can be such that objects are permanently connected or releasably connected. The term "include" means "to include, but not necessarily to include"; it specifically indicates an open inclusion or membership relationship in a combination, group, series, etc., as described herein.
[0041] Figure 1 The traffic sign recognition system 10 provided in one embodiment of this application includes a camera module 100, a sensor 200, a training module 300, and a recurrent neural network (RNN) module 400. The sensor 200 is sequentially connected to the training module 300 and the RNN module 400, and the training module 300 is connected to the RNN module 400.
[0042] In one embodiment of this application, the traffic sign recognition system 10 obtains the recognition result of traffic signs through the camera module 100. The traffic sign recognition system 10 obtains information about the behavior of its own vehicle and nearby vehicles through the sensor 200. The traffic sign recognition system 10 inputs the behavior information of its own vehicle and nearby vehicles from the sensor 200 to the training module 300. The training module 300 outputs traffic sign recognition parameters to the RNN module 400. The RNN module 400 fuses the recognition result obtained by the camera module 100 and the traffic sign recognition parameters from the training module 300, and outputs the traffic sign recognition result based on the information input from the training module 300 and the camera module 100.
[0043] In one embodiment of this application, the sensor 200 includes, but is not limited to, radar, locator, and lidar sensors. Since traffic signs have the same constraining effect on the vehicle and nearby vehicles, analyzing the behavior of the vehicle and nearby vehicles can improve the recognition rate of traffic signs. For example, after the camera module 100 detects a traffic sign, the sensor 200 collects behavioral information of the vehicle and nearby vehicles, and the training module 300 then learns and filters information based on the detected traffic signs and vehicle behavior. The training module 300 further outputs the learned information to the RNN module 400. The RNN module 400 determines the type of traffic sign based on the information obtained by the camera module 100 and the information trained by the training module 300.
[0044] In one embodiment of this application, the behavioral information of the vehicle and nearby vehicles collected by the sensor 200 includes, but is not limited to, the distance, speed, and headlight information of the vehicle and nearby vehicles. The sensor 200 integrates the acquired information to form parameters of the vehicle and nearby vehicles. The parameters of the vehicle and nearby vehicles output by the sensor 200 to the training module 300 are X = {(x,y),Φ,v,ω,a,(w,h),c}. In this formula, the parameters (x,y) represent the center point position of the vehicle. The parameter Φ represents the yaw angle. The parameter v represents the speed, and the parameter ω represents the yaw rate. The parameter a represents the acceleration of the vehicle. The parameters (w,h) represent the width and height of the vehicle. The parameter c represents the type of vehicle. In one embodiment, only information of moving vehicles is collected to train the module 300 to recognize traffic signs based on the vehicle and nearby vehicles.
[0045] See Figure 2 In one embodiment, the training module 300 is a Long Short-Term Memory (LSTM) network module 310. The traffic sign recognition system 10 further includes a Kalman filter 500. The LSTM module 310 is connected to the sensor 200 and the RNN module 400. The Kalman filter 500 is connected to the RNN module 400 and the camera module 100 to output traffic sign recognition results.
[0046] In one embodiment of this application, the Kalman filter 500 can fuse the traffic sign recognition result obtained by the camera module 100 (hereinafter referred to as the first recognition result) and the traffic sign recognition result inferred by the sensor 200 (hereinafter referred to as the second recognition result) with the behavior of the vehicle and nearby vehicles. Therefore, the Kalman filter 500 can fuse these two recognition results to obtain a new traffic sign recognition result (hereinafter referred to as the third recognition result). Since nearby vehicles observe traffic signs from different angles, taking their behavior into account reduces the possibility of errors in recognizing individual traffic signs. Therefore, compared with related technologies that only recognize traffic signs through cameras, the traffic sign recognition result has higher robustness and accuracy.
[0047] In one embodiment of this application, the Kalman filter 500 can fuse the first recognition result and the second recognition result, namely the traffic sign recognition result derived from image recognition and the behavior of the vehicle and nearby vehicles, so that the traffic sign recognition system 10 has high accuracy.
[0048] In one embodiment of this application, the LSTM module 310 typically has a two-layer structure. One layer is an information input layer, and the other is an information filtering layer. The sensor 200 continuously collects behavioral information about the vehicle and nearby vehicles, or periodically collects such information. The sensor 200 inputs the behavioral information about the vehicle and nearby vehicles into the information input layer of the LSTM module 310.
[0049] In this embodiment, when the sensor 200 inputs the behavior information of the vehicle and nearby vehicles into the information input layer of the LSTM module 310, the behavior information of the vehicle and nearby vehicles is assigned different weights. For example, the weights of the vehicle's information and the information of nearby vehicles can be set to 1:1 to increase the importance of the vehicle's behavior information during training. In other embodiments, the weights can also be set according to the number of nearby vehicles, with the vehicle and each nearby vehicle having the same weight. The weights of the vehicle and nearby vehicles can be set in the gate structure of the information filtering layer in the LSTM module 310.
[0050] In this embodiment, since there are no other moving vehicles nearby, training can be performed solely based on the vehicle's behavior information. However, training based solely on the vehicle's behavior information may lead to inaccurate training results. Therefore, if there are periods when there are no other moving vehicles nearby, but other periods when there are many other vehicles, the weight of the periods with more other vehicles is increased to improve the accuracy of the training results. That is, the overall weight of the behavior information when only the vehicle is nearby is reduced in the information filtering layer's gate structure. Conversely, the overall weight of the behavior information when there are many nearby vehicles is increased in the information filtering layer's gate structure to improve the accuracy of the training results.
[0051] In this embodiment, sensor 200 is also used to acquire driver behavior information of the vehicle itself and nearby vehicles. LSTM module 310 filters the driver behavior information to remove information that would negatively impact training. Due to the characteristics of LSTM networks, driver behavior information can be added to or removed from cell states through the gate structure of the information filtering layer. Therefore, LSTM module 310 can retain the information that needs to be retained and delete the driver behavior information that does not need to be retained, thereby achieving the filtering of driver behavior information.
[0052] For example, when analyzing driver behavior information, the results may be interfered with by drivers who violate traffic rules. For instance, such violations can lead to mislearning and affect the accuracy of the RNN module 400. Therefore, it is necessary to filter driver behavior information. For example, there may be a no-right-turn traffic sign on the road. However, among the driver behaviors of the vehicle and nearby vehicles acquired and analyzed by sensor 200, most are either going straight or turning left. Therefore, if a right-turn behavior is observed, it can be considered a violation of traffic rules. The LSTM module 310 can remove this right-turn behavior information from the cell state through the gate structure in the information filtering layer, thereby improving the accuracy of the RNN module 400 and reducing its training time.
[0053] In one embodiment of this application, if the behavior of the vehicle and nearby vehicles becomes abnormal due to other factors, the weight of vehicle behavior within a certain period of time can be reduced during the fusion of recognition results. For example, when a traffic accident occurs in the middle of the road, nearby vehicles will slow down and drive around the accident vehicle. The LSTM module 310 may identify this behavior as the corresponding deceleration and detour traffic sign, thus causing a misjudgment. Therefore, reducing the weight of the recognition results within that period of time when an abnormal state is detected can improve the recognition accuracy of the traffic sign recognition system 10.
[0054] Figure 3This application provides a traffic sign recognition system 20 according to another embodiment. It is understood that the traffic sign recognition system 20 includes a camera module 100, a sensor 200, a convolutional neural network (CNN) module 300a, and an RNN module 400.
[0055] In this embodiment, the traffic sign recognition system 20 and Figure 1 The traffic sign recognition system 10 shown is similar, except that 300a is a CNN module. This CNN module 300a includes a first CNN module 320 and a second CNN module 330. The first CNN module 320 is connected to the camera module 100 and the RNN module 400. The second CNN module 330 is connected to the sensor 200 and the RNN module 400.
[0056] In one embodiment of this application, the image information acquired by the camera module 100 can be input to the first CNN module 320 to obtain a first traffic sign recognition result. The first CNN module 320 then inputs the first traffic sign recognition features to the RNN module 400. The sensor 200 acquires the behavior information of its own vehicle and nearby vehicles and inputs it to the second CNN module 330 to obtain traffic sign recognition features. The second CNN module 330 then inputs the traffic sign recognition features to the RNN module 400, and the RNN module 400 outputs the second traffic sign recognition feature result.
[0057] In this embodiment, due to the introduction of a first CNN module 320 and a second CNN module 330, the first CNN module 320 can directly process the image information acquired by the camera module 100 and perform feature extraction based on the image information. The second CNN module 330 is trained based on the vehicle behavior information of the vehicle and nearby vehicles acquired by the sensor 200 to obtain traffic sign recognition results based on the vehicle behavior information of the vehicle and nearby vehicles. The RNN module 400 is trained based on the information input to the first CNN module 320 and the second CNN module 330 to obtain the traffic sign recognition results. The traffic sign recognition system 20 does not need to perform feature fusion based on the extracted traffic signs, which improves the applicability of the traffic sign recognition system 20 and enables traffic sign recognition training based on the original images acquired by the camera module 100.
[0058] Figure 4 This is a flowchart of a traffic sign recognition method. The embodiments are provided by way of example because there are multiple ways to implement the method. For example, one could use... Figure 1-3 The system shown is used to perform the methods described below, and reference is made to the various elements of these figures when explaining the embodiments. Figure 4Each step shown represents one or more processes, methods, or sub-processes performed in this embodiment. Furthermore, the order of the steps shown is merely illustrative, and the order of the steps can be changed. Additional steps may be added, or fewer steps may be used, without departing from this application. This embodiment may begin at step S100.
[0059] In step S100, the first traffic sign recognition result is obtained.
[0060] In one embodiment of this application, the recognition result of the first traffic sign can be obtained through the camera module 100.
[0061] In step S200, the behavior information of the vehicle and nearby vehicles is obtained.
[0062] In one embodiment of this application, the behavior information of the vehicle and nearby vehicles can be obtained through sensor 200.
[0063] In one embodiment of this application, the order of steps S100 and S200 is not limited.
[0064] In step S300, traffic sign recognition parameters are generated based on the behavioral information.
[0065] In one embodiment of this application, the training module 300 can be trained using behavioral information of the vehicle and nearby vehicles collected by the sensor 200, so as to output traffic sign recognition parameters to the RNN module 400. It is understood that the method for outputting traffic sign recognition parameters is the same as that in traffic sign recognition system 10 and traffic sign recognition system 20, and will not be described again here.
[0066] In step S400, the second traffic sign recognition result is output based on the traffic sign recognition parameters and the first traffic sign recognition result.
[0067] In one embodiment of this application, the first traffic sign recognition result from the camera module 100 and the output traffic sign recognition parameters from the training module 300 can be received by the RNN module 400, and a second traffic sign recognition result can be generated.
[0068] In one embodiment of this application, the training module 300 may be Figure 2 The LSTM module 310 shown in the figure or Figure 3 The specific functions and electrical connections of the CNN module 330 shown in the figure can be found in [reference needed]. Figure 2 and Figure 3 The description will not be repeated here.
[0069] The traffic sign recognition system 10 provided in this application embodiment can be used in L3, L4 or L5 level autonomous driving, and can recognize traffic signs based on the behavior of nearby vehicles and the camera module.
[0070] like Figure 5 As shown, in one embodiment of this application, a vehicle 30 is also provided. The vehicle 30 includes a traffic sign recognition system 10. The vehicle 30 provided in this embodiment can recognize traffic signs based on information collected by different types of sensors on the vehicle. Compared with related technologies that only use cameras to recognize traffic signs, combining other sensors for traffic sign recognition training can improve the accuracy of the training results.
[0071] Although many features and advantages of the present technology, as well as details of the structure and function of this application, have been set forth in the foregoing description, this application is merely illustrative, and changes in detail may be made within the principles of this application, particularly in terms of the shape, size, and arrangement of components, up to and including the full scope established by the broad general meaning of the terms used in the claims. Therefore, it should be understood that modifications may be made to the above exemplary embodiments within the scope of the claims.
Claims
1. A traffic sign recognition system, comprising: The camera module is used to acquire the recognition results of the first traffic sign; Sensors are used to acquire behavioral information about the vehicle and nearby vehicles; The training module is connected to the sensor. The training module is used to assign weights to the behavior information of the vehicle and nearby vehicles and output traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles. Specifically, when there is only the vehicle, the overall weight of the vehicle's behavior information is reduced; when there are many nearby vehicles, the overall weight of the behavior information of nearby vehicles is increased; and when there is an abnormal state, the weight of the behavior information of nearby vehicles is reduced. A recurrent neural network module is provided, which is connected to the training module and the camera module; wherein the recurrent neural network module is used to output a second traffic sign recognition result based on the traffic sign recognition parameters and the first traffic sign recognition result; and A Kalman filter is connected to the recurrent neural network module, wherein the Kalman filter is used to fuse the first traffic sign recognition result and the second traffic sign recognition result to generate a third traffic sign recognition result; The training module includes a first convolutional neural network module and a second convolutional neural network module. The first convolutional neural network module is connected to the camera module and the recurrent neural network module, and the second convolutional neural network module is connected to the sensor and the recurrent neural network module. The camera module is used to acquire image information. The first convolutional neural network module is used to generate a first traffic sign recognition result based on the image information. The second convolutional neural network module is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
2. The traffic sign recognition system as described in claim 1, wherein, The training module is a long short-term memory artificial neural network module.
3. The traffic sign recognition system as described in claim 2, wherein, The long short-term memory artificial neural network module includes an information input layer; The information input layer is connected to the sensor to obtain behavioral information of the vehicle and nearby vehicles.
4. The traffic sign recognition system as described in claim 3, wherein, The long short-term memory artificial neural network module also includes an information filtering layer; The information filtering layer is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
5. The traffic sign recognition system as described in claim 4, wherein, The traffic sign recognition parameters include: The vehicle's center position, yaw angle, speed, yaw rate, acceleration, width and height, and class of vehicle.
6. A traffic sign recognition method, applied to a traffic sign recognition system as described in any one of claims 1 to 5, the traffic sign recognition method comprising: Obtain the first traffic sign recognition result; Obtain behavioral information of this vehicle and nearby vehicles; Assign weights to the behavior information of the vehicle and nearby vehicles and output traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles. Specifically, when there is only the vehicle, reduce the overall weight of the vehicle's behavior information; when there are many nearby vehicles, increase the overall weight of the behavior information of nearby vehicles; and reduce the weight of the behavior information of nearby vehicles in abnormal conditions. A second traffic sign recognition result is output based on the traffic sign recognition parameters and the first traffic sign recognition result. The traffic sign recognition parameters include the traffic sign recognition parameters and the first traffic sign recognition result. The first traffic sign recognition result and the second traffic sign recognition result are combined to generate a third traffic sign recognition result.
7. The traffic sign recognition method as described in claim 6, wherein, The method further includes: The traffic sign recognition parameters are output based on the behavior information of the vehicle and nearby vehicles.
8. The traffic sign recognition method as described in claim 7, wherein, The traffic sign recognition parameters include: The vehicle's center position, yaw angle, speed, yaw rate, acceleration, width and height, and class of vehicle.
9. The traffic sign recognition method as described in claim 6, wherein, The method further includes: Acquire image information; A first traffic sign recognition result is generated based on the image information.
10. The traffic sign recognition method as described in claim 9, wherein, The method further includes: The traffic sign recognition parameters are output based on the behavior information of the vehicle and nearby vehicles.
11. A vehicle comprising: Traffic sign recognition system, the traffic sign recognition system includes: The camera module is used to acquire the recognition results of the first traffic sign; Sensors are used to acquire behavioral information about the vehicle and nearby vehicles; The training module is connected to the sensor and is used to assign weights to the behavior information of the vehicle and nearby vehicles and output traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles. Specifically, when there is only the vehicle, the overall weight of the vehicle's behavior information is reduced; when there are many nearby vehicles, the overall weight of the behavior information of nearby vehicles is increased; and when there is an abnormal state, the weight of the behavior information of nearby vehicles is reduced. A recurrent neural network module is provided, which is connected to the training module and the camera module; wherein the recurrent neural network module is used to output a second traffic sign recognition result based on the traffic sign recognition parameters and the first traffic sign recognition result; and A Kalman filter is connected to the recurrent neural network module, wherein the Kalman filter is used to fuse the first traffic sign recognition result and the second traffic sign recognition result to generate a third traffic sign recognition result; The training module includes a first convolutional neural network module and a second convolutional neural network module. The first convolutional neural network module is connected to the camera module and the recurrent neural network module, and the second convolutional neural network module is connected to the sensor and the recurrent neural network module. The camera module is used to acquire image information. The first convolutional neural network module is used to generate a first traffic sign recognition result based on the image information. The second convolutional neural network module is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
12. The vehicle as claimed in claim 11, wherein, The training module is a long short-term memory artificial neural network module.
13. The vehicle as claimed in claim 12, wherein, The long short-term memory artificial neural network module includes an information input layer; The information input layer is connected to the sensor to obtain behavioral information of the vehicle and nearby vehicles.
14. The vehicle as claimed in claim 13, wherein, The long short-term memory artificial neural network module also includes an information filtering layer; The information filtering layer is used to output the traffic sign recognition parameters based on the behavior information of the vehicle and nearby vehicles.
15. The vehicle as claimed in claim 11, wherein, The traffic sign recognition parameters include: The vehicle's center position, yaw angle, speed, yaw rate, acceleration, width and height, and class of vehicle.