Input adjustment method, device, equipment and medium for intelligent driving perception network

Abstract

The present invention relates to the field of intelligent driving technology, and discloses an input adjustment method, device, equipment and medium of an intelligent driving perception network. The present application first obtains the spatiotemporal parameters required for the vehicle to reach the target intersection, and can accurately and flexibly adjust the sparsification degree of the traffic light image according to the spatiotemporal relationship between the vehicle and the intersection, thereby laying the foundation for the subsequent accurate identification of the traffic light type. This process does not need to rely on high-precision maps, avoiding the problem of lagging traffic light information caused by the slow update of high-precision maps. Furthermore, the method of adjusting the sparsification degree using spatiotemporal parameters does not rely on simply increasing the GPU computing power to improve image resolution and recognition accuracy. Instead, it cleverly and accurately optimizes the image according to the actual situation, meets the signal light image recognition requirements without increasing the hardware cost, and solves the high cost problem caused by increasing the GPU computing power.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to an input adjustment method, device, equipment and medium for an intelligent driving perception network. Background Art

[0002] In the field of intelligent driving, the current intelligent driving perception neural network mainly identifies obstacles (such as vehicles, two-wheeled vehicles, pedestrians, etc.) and signal signs (such as traffic lights, taillights of the front vehicle, speed limit signs, etc.). Because the obstacles are large in size and there is millimeter-wave radar to assist in judgment, the requirements for image resolution can be relatively reduced. However, the traffic light target is small, and in complex scenes such as glare and rainy days, the image resolution needs to be improved to ensure recognition accuracy. At the same time, the higher the image resolution, the higher the computing power required.

[0003] At present, in order to ensure the accuracy of traffic light recognition and help vehicles pass smoothly in urban navigation assisted driving (Navigate on Autopilot, NOA) scenarios, conventional solutions include using signal light information from high-precision maps, increasing GPU computing power, and combining the two. However, the slow update of high-precision maps greatly limits their guiding role in vehicle traffic. The commonly used methods in the industry, such as increasing computing power or combining the two, can solve the problem to a certain extent, but it increases the computing cost. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides an input adjustment method for an intelligent driving perception network to solve the problems of slow updates or high computing costs in existing solutions using high-precision maps or increasing GPU computing power.

[0005] In a first aspect, an embodiment of the present invention provides an input adjustment method for an intelligent driving perception network, the method comprising:

[0006] Acquire the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass;

[0007] Acquire a traffic light image corresponding to the target intersection collected by the vehicle, and adjust an original sparsification degree of the traffic light image using the spatiotemporal parameters to obtain an adjusted traffic light image;

[0008] The adjusted traffic light image is recognized through an intelligent driving perception network to obtain the current traffic light type of the target intersection.

[0009] Furthermore, the adjusting the original sparsification degree of the traffic light image by using the spatiotemporal parameters to obtain the adjusted traffic light image includes:

[0010] Comparing the spatiotemporal parameter with a first preset threshold, if the spatiotemporal parameter is less than or equal to the first preset threshold, adjusting the signal light image from an original sparsification degree to a first sparsification degree to obtain an adjusted signal light image, wherein the first sparsification degree is less than the original sparsification degree;

[0011] or, comparing the spatiotemporal parameter with a second preset threshold, if the spatiotemporal parameter is greater than or equal to the second preset threshold, adjusting the signal light image from an original sparsification degree to a second sparsification degree to obtain an adjusted signal light image, wherein the second sparsification degree is greater than the original sparsification degree;

[0012] The first preset threshold is smaller than the second preset threshold.

[0013] Furthermore, the step of adjusting the signal light image from an original sparseness level to a first sparseness level to obtain an adjusted signal light image includes:

[0014] Obtaining a mapping relationship between a preset threshold and a sparsification degree;

[0015] A first thinning degree corresponding to the first preset threshold is determined based on the mapping relationship, and the thinning degree of the signal light image is adjusted to the first thinning degree to obtain an adjusted signal light image.

[0016] Furthermore, the step of adjusting the sparsification degree of the traffic light image to the first sparsification degree to obtain the adjusted traffic light image includes:

[0017] Obtaining illumination data of the road section where the vehicle is located and the lane position corresponding to the vehicle;

[0018] Determine a target area in the signal light image that needs to be thinned using the illumination data and the lane position;

[0019] The thinning degree of the target area in the traffic light image is adjusted to the first thinning degree to obtain an adjusted traffic light image.

[0020] Furthermore, the step of adjusting the signal light image from an original sparseness level to a first sparseness level to obtain an adjusted signal light image includes:

[0021] Detecting traffic condition data of the road section where the vehicle is located and vehicle driving data of the vehicle;

[0022] Analyzing the influence of the traffic condition data and the vehicle driving data on signal light recognition, and obtaining an adjustment ratio corresponding to the influence;

[0023] A calculation is performed based on the adjustment ratio and the original thinning degree to obtain a first thinning degree, and the thinning degree of the traffic light image is adjusted to the first thinning degree to obtain an adjusted traffic light image.

[0024] Furthermore, the analyzing the influence of the traffic condition data and the vehicle driving data on the signal light recognition includes:

[0025] Extracting road section environment data and road section traffic flow data from the traffic condition data;

[0026] Using the traffic flow data of the road section, predicting the probability of the traffic light being blocked;

[0027] Using the vehicle driving data and the road section environment data, predicting the accuracy of the signal light recognition;

[0028] The influence degree is calculated based on the probability and the accuracy rate.

[0029] Furthermore, the step of adjusting the signal light image from an original sparseness level to a second sparseness level to obtain an adjusted signal light image includes:

[0030] Obtaining a mapping relationship between a preset threshold and a sparsification degree;

[0031] A second thinning degree corresponding to the second preset threshold is determined based on the mapping relationship, and the thinning degree of the traffic light image is adjusted to the second thinning degree to obtain an adjusted traffic light image.

[0032] Furthermore, after obtaining the time and space parameters required for the vehicle to reach the target intersection, the method further includes:

[0033] comparing the spatiotemporal parameter with a third preset threshold;

[0034] If the spatiotemporal parameter is less than the third preset threshold and greater than the first preset threshold, the vehicle is adjusted from an original driving speed to a target driving speed, and the vehicle is controlled to execute a step of acquiring a signal light image, wherein the target driving speed is less than the original driving speed;

[0035] If the traffic light type identified based on the traffic light image is a passable type, after the vehicle passes through the target intersection, the vehicle is adjusted from the target driving speed to the original driving speed.

[0036] In a second aspect, an embodiment of the present invention provides an input adjustment device for an intelligent driving perception network, the device comprising:

[0037] An acquisition module, used to acquire the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass;

[0038] An adjustment module, used for acquiring the traffic light image corresponding to the target intersection collected by the vehicle, and adjusting the original sparsification degree of the traffic light image by using the spatiotemporal parameters to obtain an adjusted traffic light image;

[0039] The recognition module is used to recognize the adjusted traffic light image through an intelligent driving perception network to obtain the current traffic light type of the target intersection.

[0040] In a third aspect, an embodiment of 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 method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0041] In a fourth aspect, an embodiment of 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 method of the first aspect or any corresponding embodiment thereof.

[0042] The method provided in the embodiment of the present application has the following beneficial effects:

[0043] This application first obtains the spatiotemporal parameters required for the vehicle to reach the target intersection, and can flexibly adjust the sparsification degree of the traffic light image based on the spatiotemporal relationship between the vehicle and the intersection, laying the foundation for the subsequent accurate identification of the traffic light type. This process does not rely on high-precision maps, avoiding the problem of lagging traffic light information caused by slow updates of high-precision maps. Furthermore, the method of using spatiotemporal parameters to adjust the sparsification degree does not rely on simply increasing GPU computing power to improve image resolution and recognition accuracy. Instead, it cleverly and accurately optimizes the image according to the actual situation, meets the needs of traffic light image recognition without increasing hardware costs, and solves the high cost problem caused by increasing GPU computing power. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] 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.

[0045] Figure 1 is a flow chart of an input adjustment method of an intelligent driving perception network according to some embodiments of the present invention;

[0046] Figure 2 is a schematic diagram of another sparsification degree adjustment according to some embodiments of the present invention;

[0047] Figure 3 is a flowchart of yet another (another / another) intelligent driving perception network input adjustment method according to some embodiments of the present invention;

[0048] Figure 4 is a flowchart of yet another (another / another) intelligent driving perception network input adjustment method according to some embodiments of the present invention;

[0049] Figure 5 is a structural block diagram of an input adjustment device of an intelligent driving perception network according to an embodiment of the present invention;

[0050] Figure 6 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] 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.

[0052] According to an embodiment of the present invention, an embodiment of an input adjustment method for an intelligent driving perception network 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 may be executed in an order different from that shown here.

[0053] In this embodiment, a method for adjusting the input of an intelligent driving perception network is provided. Figure 1 is a flow chart of an input adjustment method of an intelligent driving perception network according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0054] Step S101, obtaining the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass.

[0055] In the embodiment of the present application, the spatiotemporal parameter refers to an indicator that comprehensively considers the spatial distance between the vehicle and the target intersection and the time required for the vehicle to reach the intersection. It plays an important guiding role in the subsequent operations such as dynamically adjusting the sparsification degree of the traffic light image and the vehicle's driving speed according to the relative position and time relationship between the vehicle and the intersection. By obtaining this parameter, the state and needs of the vehicle at different stages can be predicted in advance, so as to sense and control it. For example, when the vehicle is close to the intersection and is about to arrive (the spatiotemporal parameter is small), the traffic light image needs to be processed more finely to ensure accurate recognition, and the speed is appropriately reduced to ensure safe passage; when the vehicle is far away from the intersection (the spatiotemporal parameter is large), a relatively low image resolution processing strategy can be adopted to save computing resources, while maintaining a high speed to improve driving efficiency.

[0056] Step S102, obtaining a traffic light image corresponding to the target intersection collected by the vehicle, and adjusting the original sparsification degree of the traffic light image using the spatiotemporal parameters to obtain an adjusted traffic light image.

[0057] In the embodiment of the present application, the vehicle collects image information of the target intersection ahead in real time through the on-board camera during driving, including the image of the traffic light. These images are the basic data source for subsequent traffic light recognition and processing. The viewing angle and shooting range of the camera will be determined according to the installation position and angle of the vehicle to ensure that relevant information of the target intersection can be captured, including the status and position of the traffic light.

[0058] In addition, the reason for using spatiotemporal parameters to adjust the degree of image sparsification is that the requirements for image resolution are different in different situations. For example, when a vehicle approaches an intersection, in order to more accurately identify traffic lights, the image resolution needs to be increased; when the vehicle is far away from the intersection, in order to save computing resources and increase processing speed, the image resolution can be appropriately reduced. By using spatiotemporal parameters to adjust the original sparsification degree of the traffic light image, the system performance and resource utilization can be optimized while ensuring accurate traffic light recognition.

[0059] In the embodiment of the present application, the original sparseness degree of the traffic light image is adjusted by using the spatiotemporal parameters to obtain the adjusted traffic light image, including the following cases:

[0060] Case 1, compare the spatiotemporal parameters with the first preset threshold. If the spatiotemporal parameters are less than or equal to the first preset threshold, the traffic light image is adjusted from the original sparsification degree to the first sparsification degree to obtain an adjusted traffic light image, and the first sparsification degree is less than the original sparsification degree.

[0061] It should be noted that the first preset threshold can be set to 40m or T1 time to define the relative position and time relationship between the vehicle and the target intersection. The spatiotemporal parameter comprehensively reflects the spatial distance and time required for the vehicle to reach the target intersection. By comparing it with the first preset threshold, the stage of the vehicle approaching the intersection can be determined. If the distance in the spatiotemporal parameter is ≤40m, it means that the vehicle is close to the intersection in space; if the time is ≤T1, it means that the vehicle will arrive in the short term.

[0062] In addition, the degree of sparsification determines the image resolution. During normal driving, the vehicle processes the collected traffic light images using the original sparsification degree to balance computing resources and image recognition requirements. When the vehicle approaches an intersection (the spatiotemporal parameter ≤ the first preset threshold), in order to improve the accuracy of traffic light recognition, it is necessary to adjust to the first sparsification degree, which is less than the original sparsification degree, that is, reduce the image sparsification level and increase image details and resolution.

[0063] Specifically, the operation of adjusting the degree of image sparsification is as follows: First, starting from a certain pixel point in the image, an area is selected to expand outward from its periphery. In this selected area, a target value can be obtained by calculating the maximum value, minimum value or average value. Then, according to the target value, the original larger area is replaced with a smaller area to adjust the degree of image sparsification. For example, Figure 2 As shown in the figure, for an 8*8 bitmap, you can select 4*4 as a selection box to reduce the bitmap to a 4*4 small image, thus completing the adjustment of the image resolution. In general, assuming that the resolution of the original image is 800 pixels, its size is 3300*1800, and the minimum pixel that the system can normally use is 3 million pixels.

[0064] When the vehicle is in normal driving state, the image is usually compressed to a size of 2000*1500. In this process, for the horizontal size, every 165 pixels can be processed into 100 pixels on average.

[0065] However, when it comes to traffic light scenes, in order to more accurately identify traffic lights, it is desirable to obtain a higher resolution traffic light image. One way is to simultaneously increase the resolution of the full-frame image to the highest, in which case the thinning operation can be canceled to restore the image pixels from 3 million pixels to 8 million pixels; another way is to adjust the image pixels from 3 million pixels to a lower pixel value by adjusting the coefficient parameters. In addition, if you only want to increase the pixel resolution of a specific target area in the traffic light image (such as the upper right area), you can choose to cancel the thinning operation of the upper right area, or change the original thinning degree of this area.

[0066] This adjustment process aims to flexibly change the image sparsification and resolution according to different needs and scenarios to meet the different processing requirements of traffic light images at different driving stages. When the vehicle is far away from the intersection, a higher sparsification degree can save computing resources, while when the vehicle approaches the intersection, the sparsification degree can be reduced or the pixel size of some areas of the traffic light image can be increased to ensure better traffic light recognition effect.

[0067] In the embodiment of the present application, the signal light image is adjusted from the original sparseness degree to the first sparseness degree to obtain the adjusted signal light image, such as Figure 3 As shown, the following steps A1-A2 are included:

[0068] Step A1, obtaining a mapping relationship between a preset threshold and a sparsification degree.

[0069] Specifically, in the process of adjusting the sparsification degree of the traffic light image by the intelligent driving perception network, the preset threshold is used to define the relative position or time relationship between the vehicle and the target intersection. The vehicle stores the correspondence between different preset thresholds and sparsification degrees, which is set based on the different requirements for resolution of traffic light image recognition under different distances or time conditions. By establishing such a mapping relationship, it is possible to accurately determine to what level the sparsification degree of the traffic light image should be adjusted according to the actual spatiotemporal parameters of the vehicle. For example, when the vehicle is close to the intersection (the spatiotemporal parameters meet specific threshold conditions), a higher image resolution is required to accurately identify the traffic light, and the corresponding sparsification degree should be lower; when the vehicle is far away from the intersection, the image resolution requirement is relatively low, and the sparsification degree can be higher.

[0070] Step A2: determining a first thinning degree corresponding to a first preset threshold based on the mapping relationship, and adjusting the thinning degree of the signal light image to the first thinning degree to obtain an adjusted signal light image.

[0071] Specifically, after obtaining the mapping relationship between the preset threshold and the sparsification degree, when it is determined that the spatiotemporal parameters of the vehicle are less than or equal to the first preset threshold, the first sparsification degree corresponding to this time can be determined according to the mapping relationship. For example, if the mapping relationship indicates that when the distance between the vehicle and the intersection is less than or equal to 40m (preset threshold), the image sparsification degree needs to be adjusted to a specific lower level, namely the first sparsification degree. The first sparsification degree is determined for the special case of the vehicle approaching the intersection to meet higher signal light recognition accuracy requirements.

[0072] In the embodiment of the present application, adjusting the sparsification degree of the traffic light image to a first sparsification degree to obtain the adjusted traffic light image includes the following steps A201-A203:

[0073] Step A201, obtaining the lighting data of the road section where the vehicle is located and the lane position corresponding to the vehicle.

[0074] It should be noted that the lighting data of the road section where the vehicle is located contains the lighting information of the current environment, such as light intensity, light direction, etc. Different lighting conditions will affect the quality of the signal light image. The display effect of the signal light in the image will be different under different lighting conditions such as daytime, night, cloudy, sunny, backlight or front light. For example, in the case of backlight, the signal light may appear too bright or some areas may be too dark, and at night, it may be necessary to enhance the image recognition ability under lighting conditions.

[0075] In addition, vehicles in different lanes have different viewing angles and relative positions when viewing the signal light, which will affect the position of the signal light in the image captured by the camera. For example, the angle at which the vehicle in the leftmost lane views the signal light is different from that of the vehicle in the rightmost lane, and the position of the signal light in the image will also be offset. Determining the lane position of the vehicle can help determine the approximate position of the signal light in the image, providing a basis for subsequently determining the area where the signal light is located.

[0076] Step A202, using the illumination data and the lane position, determine the target area in the signal light image that needs to be thinned.

[0077] It should be noted that lighting conditions will affect image quality. When the image is backlit, the overall image is darker, and the area where the signal light is located may be darker, making the image details difficult to discern. The lane position of the vehicle can provide clues for determining the approximate position of the signal light. Different lanes correspond to different driving directions, and the positions of signal lights in each direction in the image are usually relatively fixed. For example, if the vehicle is in the leftmost lane and is preparing to turn left, the left-turn signal light is usually in the upper left part of the image.

[0078] Specifically, by combining the illumination data and the lane position information, it is possible to accurately determine the target area in the signal light image that needs to be thinned out. For example, because the backlight makes the overall image darker, especially the upper left area (where the left turn signal light is usually located) is difficult to identify clearly. At the same time, the vehicle is preparing to turn left in the leftmost lane, and it is necessary to focus on identifying the left turn signal light in the upper left area. Therefore, considering these two factors, the upper left part of the image is determined as the target area that needs to be thinned out. The purpose of doing this is to perform targeted processing on this area later, and by adjusting the degree of thinning, the image resolution of this area can be improved, so that the signal light can be presented more clearly for accurate identification.

[0079] Step A203: Adjust the thinning degree of the target area in the traffic light image to a first thinning degree to obtain an adjusted traffic light image.

[0080] Specifically, according to the principle of image thinning, the pixel processing method of the target area is changed. For example, previously, thinning may be achieved by performing pixel averaging and other operations on a larger area. Now, for the target area, the range of the averaging area is reduced, or a more complex algorithm is used to retain more details, so that the pixels in the target area are denser and the image resolution is improved. After adjusting the degree of thinning in some areas, the left turn signal light that was originally blurred due to backlighting and thinning processing will have clearer outlines, colors, brightness and other details. The adjusted signal light image obtained in this way provides a better image basis for the subsequent accurate identification of the signal light type through the intelligent driving perception network, which helps the vehicle make correct driving decisions.

[0081] The method provided in the embodiment of the present application obtains the lighting data of the road section where the vehicle is located and the lane position corresponding to the vehicle, and uses this information to determine the target area in the traffic light image that needs to be thinned out, and adjusts the thinning degree of the target area to a first thinning degree to obtain an adjusted traffic light image. The method takes into account the impact of environmental factors on the traffic light image, making the image thinning adjustment more targeted, which helps to improve the processing effect of the traffic light image under different lighting and lane positions, and improves the subsequent traffic light recognition accuracy.

[0082] The method provided in the embodiment of the present application obtains the mapping relationship between the preset threshold and the sparsification degree, and determines the first sparsification degree corresponding to the first preset threshold based on the mapping relationship, and adjusts the sparsification degree of the traffic light image to the first sparsification degree, thereby providing a clear adjustment mechanism, which makes the adjustment of the sparsification degree of the traffic light image more standardized and accurate, and helps to better adjust the image according to the spatiotemporal relationship between the vehicle and the intersection, thereby improving the accuracy of traffic light recognition.

[0083] In the embodiment of the present application, the thinning degree of the traffic light image is adjusted to a first thinning degree to obtain an adjusted traffic light image, such as Figure 4 As shown, the following steps B1-B3 are included:

[0084] Step B1, detecting traffic condition data of the road section where the vehicle is located and vehicle driving data of the vehicle.

[0085] It should be noted that traffic condition data includes but is not limited to road section environment data and road section traffic flow data, among which, road section environment data may include road conditions (whether there is water accumulation, road construction, potholes, etc.), weather conditions (sunny, rainy, snowy, foggy, etc.), surrounding facilities (such as roadside billboards, trees, buildings, etc. that may block the line of sight or affect the lighting); road section traffic flow data includes the vehicle flow of the road section where the vehicle is located, the distribution of vehicles (such as whether the vehicles are concentrated in certain lanes, whether there is congestion, etc.) and the distribution of vehicle driving speeds. This is because high traffic flow may affect the direct observation of the vehicle on the signal light and increase the possibility of the signal light being blocked by other vehicles. In addition, different traffic density also reflects the complexity of traffic. When the traffic flow is large, the identification of the signal light needs to be more accurate and timely to ensure the safety of the vehicle and the smoothness of traffic.

[0086] In addition, vehicle driving data involves the vehicle's own driving status information, such as the vehicle's current speed, acceleration, braking status, driving mode (such as automatic driving mode or manual driving mode), etc. The vehicle's driving speed and acceleration will affect the time requirements for signal light recognition. For example, when a vehicle approaches an intersection quickly, it needs to recognize signal lights more promptly and accurately in order to make correct driving decisions; in automatic driving mode, the reliance on signal light recognition is higher, requiring higher accuracy and reliability.

[0087] Step B2: analyzing the influence of traffic condition data and vehicle driving data on signal light recognition, and obtaining an adjustment ratio corresponding to the influence.

[0088] Specifically, the impact of various factors in traffic condition data and vehicle driving data on traffic light recognition is comprehensively considered, and different factors are assigned different weights due to their different importance to traffic light recognition. For example, heavy traffic volume, high vehicle speed, bad weather (such as rain, fog, and snow), poor lighting conditions (such as backlight and low light), being in automatic driving mode, and vehicles accelerating near intersections have a greater impact on the accurate recognition of traffic lights and will be assigned higher weights, while some factors with relatively smaller impacts will be assigned lower weights.

[0089] Subsequently, a weighted calculation is performed by comprehensively considering various factors and their weights to obtain a comprehensive value to indicate the degree of influence on signal light recognition. The higher the comprehensive value, the greater the negative impact of the current traffic and driving conditions on signal light recognition, indicating that stronger measures need to be taken to ensure accurate signal light recognition. For example, when the vehicle is in foggy weather, in autonomous driving mode and approaches an intersection quickly, the comprehensive value will be higher, and the processing of the signal light image will need to be adjusted accordingly.

[0090] Finally, a corresponding adjustment ratio is determined based on the calculated impact degree, and the adjustment ratio will be used for subsequent adjustments to the sparsification degree. The greater the impact degree, the greater the proportion of sparsification adjustment required to ensure the quality and recognition accuracy of the traffic light image. The smaller the impact degree, the smaller the proportion of sparsification adjustment required.

[0091] In another embodiment of the present application, analyzing the influence of traffic condition data and vehicle driving data on signal light recognition further includes the following steps B201-B204:

[0092] Step B201, extracting road section environment data and road section traffic flow data from traffic condition data.

[0093] Specifically, the relevant concepts of the road section environment data and the road section traffic flow data in this step are the same as those in the corresponding embodiment of the above-mentioned step B1, and will not be repeated here.

[0094] Step B202, using the road section traffic data, predict the probability of the traffic light being blocked.

[0095] Specifically, by analyzing the traffic data of a road section, such as the density of vehicles, the type and size of vehicles, the probability of a signal light being blocked can be roughly estimated. Large vehicles (such as buses and trucks) are more likely to block the view of vehicles behind them, and when the vehicles are dense, the probability of mutual blocking between vehicles will also increase. For example, if there are many vehicles in a lane and the distance between vehicles is small, the probability of the signal light being blocked when the vehicles in the lane observe the signal light will be relatively high. In addition, the dynamic changes of traffic flow, such as vehicle acceleration, deceleration and lane change behavior, will also affect the blocking situation. If the vehicle changes lanes frequently or decelerates suddenly, the signal light may be blocked in an instant, thereby affecting the continuous observation and recognition of the signal light. Through the comprehensive analysis of these factors, a model can be established to predict the probability of a signal light being blocked, providing a basis for the subsequent signal light recognition processing.

[0096] Step B203, using the vehicle driving data and the road environment data, predicting the accuracy of traffic light recognition.

[0097] Specifically, the specific process of predicting the accuracy of traffic light recognition may include:

[0098] First, the influencing factors of vehicle driving data are analyzed.

[0099] As for the speed factor, as the vehicle speed increases, the number of signal light image frames that the image acquisition system can capture per unit time will decrease, and the time each image frame stays during processing will also be shortened, which will affect the system's extraction and recognition of signal light features. Assuming the vehicle speed is (Unit: km / h), a linear function can be used to represent the effect of speed on recognition accuracy, for example:

[0100]

[0101] in, is a preset maximum speed. When the speed exceeds this speed, the impact of speed on recognition accuracy reaches its maximum. =120, vehicle speed =80, , indicating that the impact of speed on recognition accuracy is a 33.3% decrease.

[0102] As for the braking and acceleration status factors, the braking and acceleration status of the vehicle will cause changes in the vehicle posture, thereby affecting the camera's shooting angle and image stability. (Unit: m / s²) to measure its impact. Assume that when the acceleration exceeds a certain threshold (For example, =2m / s²), will have a significant impact on the recognition accuracy. Use the following function to calculate its impact on the accuracy:

[0103]

[0104] Among them, when hour, , since its value is negative, it can be set to 0, which means that when the acceleration exceeds the threshold, the impact on the accuracy is great, and the accuracy contribution of this part is directly set to 0.

[0105] As for the driving direction factor, different driving directions (left turn, right turn, straight ahead) lead to different signal light layouts and relative positions, thus affecting the difficulty of recognition. Assume that a classification function is used To represent the impact of different driving directions, for example: Indicates the accuracy impact when going straight, Indicates the accuracy impact when turning left or right.

[0106] Secondly, the influencing factors of road section environmental data are analyzed.

[0107] As for weather conditions, different weather conditions will significantly affect the image quality and visibility of the signal lights. For example, on sunny days, visibility is best. ; In rainy days, visibility decreases, set 0.7; in foggy or snowy weather, visibility is further reduced, set to 0.5, etc. It can be determined based on the weather information obtained by the sensor. The value of .

[0108] As for the lighting condition, the light intensity and direction will affect the brightness and contrast of the signal light, thus affecting the recognition. Assuming the light intensity is (Unit: lux), you can use a function To express its impact on accuracy, for example:

[0109] ,in and are the preset minimum and maximum light intensities. When the light intensity is in the middle range, its impact on accuracy is 0.5; when the light intensity is close to When , the impact on accuracy is close to 1.

[0110] For surrounding obstacles, the number of obstacles can be and size (You can use indicators such as area or height) to measure its impact. Assume that the function is used:

[0111]

[0112] Among them, when hour, ,and , hour, , indicating that the impact of surrounding obstacles on accuracy is a 20% decrease.

[0113] Finally, taking all the above factors into consideration, the final traffic light recognition accuracy is calculated using a weighted average method:

[0114]

[0115] in, is the weight of each factor, and .

[0116] Step B204, calculating the degree of influence based on the probability and accuracy.

[0117] Specifically, when the probability of a traffic light being blocked is high, it indicates that the complete traffic light information cannot be obtained at present, which will have a serious impact on recognition and will increase the value of the degree of impact accordingly. For example, if the predicted probability of a traffic light being blocked reaches 50% or higher, it means that there is a 50% chance that the complete traffic light cannot be seen. A lower traffic light recognition accuracy will also increase the degree of impact. For example, if the predicted traffic light recognition accuracy is only 60%, it means that there is a high risk of misjudging the state of the traffic light, which will also be reflected in the calculation of the degree of impact. These two factors are combined according to certain weights and calculation methods to obtain a final degree of impact value. The higher the value, the greater the negative impact of the current traffic conditions and vehicle driving status on traffic light recognition. In the future, it is necessary to make more detailed adjustments to the processing of the traffic light image (such as the adjustment of the degree of sparsification) to improve the accuracy and reliability of recognition and ensure the safety and smoothness of intelligent driving.

[0118] The method provided in the embodiment of the present application extracts road section environment data and road section traffic flow data from traffic condition data, uses the road section traffic flow data to predict the probability of traffic lights being blocked, uses vehicle driving data and road section environment data to predict the accuracy of traffic light recognition, and calculates the degree of influence based on probability and accuracy, thereby providing a more detailed basis for adjusting the degree of sparsification of traffic light images. The degree of sparsification can be reasonably adjusted according to the comprehensive situation of traffic conditions and vehicle driving, thereby improving the reliability of traffic light recognition.

[0119] Step B3, performing calculation based on the adjustment ratio and the original thinning degree to obtain a first thinning degree, and adjusting the thinning degree of the traffic light image to the first thinning degree to obtain an adjusted traffic light image.

[0120] Specifically, the adjustment ratio is calculated based on the degree of influence of traffic condition data and vehicle driving data on signal light recognition. The original sparsification degree is a state of processing the signal light image during normal vehicle driving. When the accuracy of signal light recognition needs to be improved (for example, when a vehicle approaches an intersection), the original sparsification degree is adjusted according to the adjustment ratio to obtain a first sparsification degree. The specific calculation can be performed through a mathematical formula or a mapping function.

[0121] For example, if the adjustment ratio is (the value range is between 0 and 1, 0 means no adjustment, 1 means maximum adjustment), the original sparsification degree is expressed as Represents that the first sparsification degree The formula can be Calculated. Indicates that when the ratio is adjusted The larger the value, the higher the first degree of sparsification. The smaller it is, the lower the degree of sparsification is and the higher the image resolution is.

[0122] After obtaining the first thinning degree, the thinning degree of the traffic light image is adjusted from the original state to the degree. This process involves pixel processing operations on the image. For example, if the pixels of the image are merged and simplified according to a larger area under the original thinning degree, when adjusting to the first thinning degree, the area range of pixel merging can be reduced, or a more sophisticated pixel processing algorithm can be used to make the pixels denser and the image details richer.

[0123] For traffic light images, especially when approaching intersections, such adjustments are made to display traffic lights more clearly, because higher resolution can make the outline, color, status and other information of traffic lights clearer, which is helpful for subsequent traffic light type recognition. For example, the outline of a traffic light that may have been blurred becomes clearer after adjusting the sparsification degree, making it easier for the intelligent driving perception network to accurately determine whether it is a red light, green light or yellow light.

[0124] The method provided in the embodiment of the present application detects the traffic condition data of the road section where the vehicle is located and the vehicle driving data of the vehicle, analyzes the influence of these data on traffic light recognition and obtains the corresponding adjustment ratio, calculates the first sparsification degree based on the adjustment ratio and the original sparsification degree and makes adjustments, taking into account the comprehensive influence of traffic conditions and vehicle driving data on traffic light recognition, so that the adjustment of the sparsification degree is more in line with the actual situation, which helps to improve the effect of traffic light image recognition in complex traffic environments.

[0125] The method provided in the embodiment of the present application compares the spatiotemporal parameters with a first preset threshold value. If the spatiotemporal parameters are less than or equal to the first preset threshold value, the traffic light image is adjusted from the original sparsity degree to the first sparsity degree to obtain an adjusted traffic light image. The first sparsity degree is less than the original sparsity degree. When the vehicle approaches the target intersection, the sparsity degree of the traffic light image can be reduced, the image details and resolution can be increased, and the accuracy of traffic light recognition can be improved, so as to better assist the subsequent recognition of the traffic light image through the intelligent driving perception network, and then accurately determine the type of traffic light at the target intersection.

[0126] Case 2, comparing the spatiotemporal parameters with the second preset threshold, if the spatiotemporal parameters are greater than or equal to the second preset threshold, the traffic light image is adjusted from the original sparsity level to the second sparsity level to obtain an adjusted traffic light image, and the second sparsity level is greater than the original sparsity level; wherein the first preset threshold is less than the second preset threshold.

[0127] Specifically, the second preset threshold is an important judgment indicator, and its value is greater than the first preset threshold. It can be set to, for example, an 80m distance value or a T2 time value. Its specific setting is intended to divide the vehicle's driving process into different stages, so as to adopt different processing strategies for the signal light image at different distances or time stages. When the spatiotemporal parameter is compared with the second preset threshold, and it is determined that the spatiotemporal parameter is greater than or equal to the second preset threshold, it means that the vehicle is far away from the target intersection in terms of distance or time. For example, if the second preset threshold is a distance of 80m, the condition is met when the vehicle is greater than or equal to 80m from the intersection, or from the time dimension, it is met when the vehicle is expected to arrive at the intersection at a time greater than or equal to T2. Because the vehicle is far away from the intersection at this time, the recognition accuracy of the signal light image is not high. In order to optimize the system performance, the sparsification degree of the signal light image will be adjusted to the second sparsification degree, which is greater than the original sparsification degree but less than the first sparsification degree. This means that the vehicle is currently in a transitional state. In this state, the resolution of the traffic light image is reduced to improve system performance. However, it has not yet reached the low-resolution state corresponding to the first sparsification degree. That is, under the premise of ensuring a certain signal light image information processing capability, the traffic light image is subjected to a certain degree of sparse processing to reduce the computing resources required for the system to process the image, increase the number of frames per second that the system processes, speed up the system response, and allocate more computing resources to other tasks, such as monitoring the surrounding environment and road conditions, to achieve a broader environmental perception, and provide the vehicle with a smoother driving experience and safety protection.

[0128] In the embodiment of the present application, adjusting the signal light image from the original sparseness degree to the second sparseness degree to obtain the adjusted signal light image includes the following steps C1-C2:

[0129] Step C1, obtaining a mapping relationship between a preset threshold and a sparsification degree.

[0130] Specifically, there may be multiple preset thresholds, including a first preset threshold, a second preset threshold, etc. Each threshold corresponds to a different vehicle position or time node, reflecting the relative relationship between the vehicle and the target intersection at different driving stages. The degree of sparsification involves the processing method of the traffic light image, including pixel merging, the degree of detail retention, etc. Different degrees of sparsification will affect the resolution of the image and the amount of resources required by the system to process the image.

[0131] Step C2: determining a second thinning degree corresponding to a second preset threshold based on the mapping relationship, and adjusting the thinning degree of the traffic light image to the second thinning degree to obtain an adjusted traffic light image.

[0132] Specifically, according to the mapping relationship, the sparsification degree corresponding to the second preset threshold is searched, that is, the second sparsification degree. The second preset threshold indicates that the vehicle is in a state relatively far away from the intersection but different from the state represented by the first preset threshold. When the spatiotemporal parameter of the vehicle is greater than or equal to the second preset threshold, the vehicle is in a relatively loose driving state and is far away from the intersection.

[0133] As an example, a vehicle is driving normally on the road at a certain speed. When there is still a long distance (greater than or equal to 80m) to the next intersection, it is judged by the spatiotemporal parameters that the second preset threshold is reached. According to the mapping relationship, it is determined that the second sparsification degree should be used at this time, and then the collected signal light image is processed. Assuming that the resolution of the original image is 1280x720 pixels, at the second sparsification degree, the image is processed to 384x216 pixels (according to the sparsification degree adjustment, the full image or part of the image area is compressed to 30% of the original resolution). Although the image processed in this way loses some details of the signal light, at this stage, more attention is paid to the overall traffic environment perception and performance optimization rather than the precise details of the signal light, so as to ensure that the vehicle can travel smoothly, and when approaching the intersection later, according to other preset thresholds and corresponding sparsification adjustment strategies, the signal light image is processed more accurately to ensure the accuracy of signal light recognition.

[0134] The method provided in the embodiment of the present application obtains the mapping relationship between the preset threshold and the sparsification degree, determines the second sparsification degree corresponding to the second preset threshold, and adjusts the sparsification degree of the traffic light image to the second sparsification degree. The adjustment logic is similar to that of the first preset threshold, and also makes the adjustment of the sparsification degree more standardized, ensuring that the traffic light image can be more appropriately adjusted under different time and space conditions to meet the recognition requirements of the intelligent driving perception network.

[0135] The method provided in the embodiment of the present application compares the spatiotemporal parameters with a second preset threshold value. If the spatiotemporal parameters are greater than or equal to the second preset threshold value, the traffic light image is adjusted from the original sparsity degree to the second sparsity degree to obtain an adjusted traffic light image. The second sparsity degree is greater than the original sparsity degree. The sparsity degree of the traffic light image can be flexibly adjusted according to different spatiotemporal parameter ranges to ensure that the image sparsity degree adapts to the recognition requirements of the intelligent driving perception network under different relative positions of vehicles and intersections, thereby optimizing the processing effect of the traffic light image.

[0136] Step S103, recognizing the adjusted traffic light image through the intelligent driving perception network to obtain the current traffic light type of the target intersection.

[0137] In the embodiment of the present application, the intelligent driving perception network is the core part. It processes the traffic light image after the sparsification degree is adjusted according to the vehicle's spatiotemporal parameters before processing. It uses pre-trained algorithms and models to analyze the image's pixel distribution, color, shape, brightness and other features. In order to ensure recognition accuracy, a multi-frame verification method is adopted. The type of traffic light is determined only when the recognition results of multiple frames of images are consistent. If it is identified as a red light, the braking system is triggered to make the vehicle stop and wait. If it is a green light, the vehicle is allowed to continue or accelerate through. If it is a yellow light, different decisions are made based on factors such as speed and distance from the intersection. At the same time, the adaptability of the system is reflected. When approaching an intersection, the sparsification parameters are adjusted to improve the resolution of the traffic light area and reduce the risk of misjudgment. When far away from the intersection, the resolution is reduced to save system resources, improve the system's response speed and ability to handle other tasks, and ensure the safe and efficient passage of vehicles at intersections.

[0138] In the embodiment of the present application, after obtaining the time and space parameters required for the vehicle to reach the target intersection, the method further includes steps D1-D3:

[0139] Step D1, comparing the spatiotemporal parameter with a third preset threshold.

[0140] It should be noted that the third preset threshold is a critical value pre-set by the system, and its specific value is determined according to the system performance, traffic rules and the needs of actual driving scenarios. This threshold plays an important role in distinguishing different driving states and operations of the vehicle, and its purpose is to further refine the different situations when the vehicle approaches the intersection.

[0141] Step D2, if the spatiotemporal parameter is less than the third preset threshold and greater than the first preset threshold, the vehicle is adjusted from the original driving speed to the target driving speed, and the vehicle is controlled to execute the signal light image acquisition step, wherein the target driving speed is less than the original driving speed.

[0142] Specifically, when the spatiotemporal parameter of the vehicle is between the first preset threshold and the third preset threshold, it means that the vehicle has approached the intersection, but has not yet reached the most critical position or time point. In this case, in order to better process the signal light information and ensure safety, the vehicle's driving speed is adjusted from the original driving speed to the target driving speed, wherein the target driving speed is less than the original driving speed. This speed reduction is a preventive and adaptive measure, which is intended to give the vehicle more time and a more stable state to process the signal light information. At the same time, the vehicle is controlled to perform the step of collecting the signal light image. As the vehicle's driving speed slows down, the vehicle has more time and more stable conditions to collect the signal light image. For example, at a slower speed, the camera can capture the image of the signal light more clearly and stably, reducing the image blur or information loss caused by excessive speed. Moreover, in this case, the sparseness degree of the collected signal light image is adjusted to ensure that the signal light can be effectively recognized under different circumstances (such as different traffic conditions and environmental conditions).

[0143] Step D3: If the traffic light type identified based on the traffic light image is a passable type, after the vehicle passes the target intersection, the vehicle is adjusted from the target driving speed to the original driving speed.

[0144] Specifically, when the intelligent driving perception network determines that the signal light type is a passable type (such as a green light), it means that the vehicle can safely pass through the intersection. When the vehicle successfully passes the target intersection, the vehicle speed is restored from the target driving speed to the original driving speed, allowing the vehicle to continue to drive at a normal speed, maintaining the smoothness and efficiency of traffic.

[0145] The method provided in the embodiment of the present application adjusts the vehicle's speed according to the temporal and spatial parameter relationship between the vehicle and the target intersection, and combines the recognition results of the signal light image to ensure the safety, accuracy and traffic flow of the vehicle when approaching and passing the intersection. By reasonably adjusting the speed and performing the corresponding signal light image acquisition operation, and restoring the speed according to the signal light type, the intelligent driving system can achieve flexible operation and safety assurance in different traffic scenarios.

[0146] In this embodiment, an input adjustment device for an intelligent driving perception network is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated hereafter. As used below, the term "module" may implement a combination of software and / or hardware for 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.

[0147] This embodiment provides an input adjustment device for an intelligent driving perception network, such as Figure 5As shown, including:

[0148] An acquisition module 501 is used to acquire the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass;

[0149] An adjustment module 502 is used to obtain a traffic light image corresponding to a target intersection collected by a vehicle, and adjust the original sparsification degree of the traffic light image using a spatiotemporal parameter to obtain an adjusted traffic light image;

[0150] The recognition module 503 is used to recognize the adjusted traffic light image through the intelligent driving perception network to obtain the current traffic light type of the target intersection.

[0151] Furthermore, the adjustment module 502 also includes a first comparison submodule and a second comparison submodule;

[0152] The first comparison submodule is used to compare the spatiotemporal parameters with a first preset threshold. If the spatiotemporal parameters are less than or equal to the first preset threshold, the traffic light image is adjusted from the original sparsity level to the first sparsity level to obtain an adjusted traffic light image, and the first sparsity level is less than the original sparsity level.

[0153] The first comparison submodule is used to obtain a mapping relationship between a preset threshold and a sparsification degree; determine a first sparsification degree corresponding to the first preset threshold based on the mapping relationship, and adjust the sparsification degree of the traffic light image to the first sparsification degree to obtain an adjusted traffic light image.

[0154] The first comparison submodule is used to obtain the lighting data of the road section where the vehicle is located and the lane position corresponding to the vehicle; use the lighting data and the lane position to determine the target area in the traffic light image that needs to be thinned out; adjust the thinning degree of the target area in the traffic light image to a first thinning degree to obtain an adjusted traffic light image.

[0155] The first comparison submodule is used to detect the traffic condition data of the road section where the vehicle is located and the vehicle driving data of the vehicle; analyze the influence of the traffic condition data and the vehicle driving data on the traffic light recognition, and obtain the adjustment ratio corresponding to the influence degree; calculate based on the adjustment ratio and the original sparsification degree to obtain the first sparsification degree, and adjust the sparsification degree of the traffic light image to the first sparsification degree to obtain the adjusted traffic light image.

[0156] The first comparison submodule is used to extract road section environment data and road section traffic flow data from the traffic condition data; use the road section traffic flow data to predict the probability of traffic lights being blocked; use the vehicle driving data and road section environment data to predict the accuracy of traffic light recognition; and calculate the degree of influence based on the probability and accuracy.

[0157] The second comparison submodule is used to compare the spatiotemporal parameters with a second preset threshold. If the spatiotemporal parameters are greater than or equal to the second preset threshold, the traffic light image is adjusted from the original sparsity level to the second sparsity level to obtain an adjusted traffic light image, and the second sparsity level is greater than the original sparsity level; wherein the first preset threshold is less than the second preset threshold.

[0158] The second comparison submodule is used to obtain a mapping relationship between a preset threshold and a sparsification degree; determine a second sparsification degree corresponding to a second preset threshold based on the mapping relationship, and adjust the sparsification degree of the traffic light image to the second sparsification degree to obtain an adjusted traffic light image.

[0159] Furthermore, the device also includes: an adjustment module, used to compare the spatiotemporal parameters with a third preset threshold; if the spatiotemporal parameters are less than the third preset threshold and greater than the first preset threshold, the vehicle is adjusted from an original driving speed to a target driving speed, and the vehicle is controlled to execute a signal light image acquisition step, wherein the target driving speed is less than the original driving speed; if the signal light type identified based on the signal light image is a passable type, then after the vehicle passes the target intersection, the vehicle is adjusted from the target driving speed to the original driving speed.

[0160] See also Figure 6 , Figure 6 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 6 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. 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 the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the 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).

[0161] 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.

[0162] 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.

[0163] 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 by the use of a computer device based on the presentation of a small program landing page, 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 device. 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.

[0164] 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.

[0165] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0166] 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.

[0167] 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 method for adjusting input of an intelligent driving perception network, characterized in that: The method comprises: Acquire the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass; Acquire a traffic light image corresponding to the target intersection collected by the vehicle, and adjust an original sparsification degree of the traffic light image using the spatiotemporal parameters to obtain an adjusted traffic light image; Recognize the adjusted signal light image through an intelligent driving perception network to obtain the current signal light type of the target intersection; The method of using the spatiotemporal parameters to adjust the original sparsification degree of the traffic light image to obtain an adjusted traffic light image includes: comparing the spatiotemporal parameters with a first preset threshold value, and if the spatiotemporal parameters are less than or equal to the first preset threshold value, detecting the traffic condition data of the road section where the vehicle is located and the vehicle driving data of the vehicle; analyzing the traffic condition data and the vehicle driving data to obtain the probability of the traffic light being blocked and the accuracy of traffic light recognition, and obtaining the degree of influence based on the probability and the accuracy; obtaining an adjustment ratio corresponding to the degree of influence; calculating based on the adjustment ratio and the original sparsification degree to obtain a first sparsification degree, and adjusting the sparsification degree of the traffic light image to the first sparsification degree to obtain an adjusted traffic light image.

2. The method according to claim 1, characterized in that The step of adjusting the original sparseness degree of the traffic light image by using the spatiotemporal parameters to obtain an adjusted traffic light image includes: Comparing the spatiotemporal parameter with a second preset threshold, if the spatiotemporal parameter is greater than or equal to the second preset threshold, adjusting the signal light image from an original sparsification degree to a second sparsification degree to obtain an adjusted signal light image, wherein the second sparsification degree is greater than the original sparsification degree; The first preset threshold is smaller than the second preset threshold.

3. The method according to claim 2, characterized in that The step of adjusting the signal light image from an original sparseness degree to a first sparseness degree to obtain an adjusted signal light image includes: Obtaining a mapping relationship between a preset threshold and a sparsification degree; A first thinning degree corresponding to the first preset threshold is determined based on the mapping relationship, and the thinning degree of the signal light image is adjusted to the first thinning degree to obtain an adjusted signal light image.

4. The method according to claim 3, characterized in that The step of adjusting the thinning degree of the traffic light image to the first thinning degree to obtain an adjusted traffic light image includes: Obtaining illumination data of the road section where the vehicle is located and the lane position corresponding to the vehicle; Determine a target area in the signal light image that needs to be thinned using the illumination data and the lane position; The thinning degree of the target area in the traffic light image is adjusted to the first thinning degree to obtain an adjusted traffic light image.

5. The method according to claim 1, characterized in that The analyzing the traffic condition data and the vehicle driving data to obtain the probability of a signal light being blocked and the accuracy of signal light recognition includes: Extracting road section environment data and road section traffic flow data from the traffic condition data; Using the traffic flow data of the road section, predicting the probability of the traffic light being blocked; The accuracy of the signal light recognition is predicted using the vehicle driving data and the road section environment data.

6. The method according to claim 2, characterized in that The step of adjusting the signal light image from an original sparseness degree to a second sparseness degree to obtain an adjusted signal light image includes: Obtaining a mapping relationship between a preset threshold and a sparsification degree; A second thinning degree corresponding to the second preset threshold is determined based on the mapping relationship, and the thinning degree of the traffic light image is adjusted to the second thinning degree to obtain an adjusted traffic light image.

7. The method according to claim 1, characterized in that After obtaining the time and space parameters required for the vehicle to reach the target intersection, the method further includes: comparing the spatiotemporal parameter with a third preset threshold; If the spatiotemporal parameter is less than the third preset threshold and greater than the first preset threshold, the vehicle is adjusted from an original driving speed to a target driving speed, and the vehicle is controlled to execute a step of acquiring a signal light image, wherein the target driving speed is less than the original driving speed; If the traffic light type identified based on the traffic light image is a passable type, after the vehicle passes through the target intersection, the vehicle is adjusted from the target driving speed to the original driving speed.

8. An input adjustment device for an intelligent driving perception network, characterized in that: The device comprises: An acquisition module, used to acquire the time and space parameters required for the vehicle to reach the target intersection, wherein the target intersection is the intersection that the vehicle is about to pass; An adjustment module, used for acquiring the traffic light image corresponding to the target intersection collected by the vehicle, and adjusting the original sparsification degree of the traffic light image by using the spatiotemporal parameters to obtain an adjusted traffic light image; A recognition module, used to recognize the adjusted signal light image through an intelligent driving perception network to obtain the current signal light type of the target intersection; The adjustment module is used to compare the spatiotemporal parameters with a first preset threshold value. If the spatiotemporal parameters are less than or equal to the first preset threshold value, the traffic condition data of the road section where the vehicle is located and the vehicle driving data of the vehicle are detected; the traffic condition data and the vehicle driving data are analyzed to obtain the probability of the traffic light being blocked and the accuracy of traffic light recognition, and the degree of influence is obtained based on the probability and the accuracy; an adjustment ratio corresponding to the degree of influence is obtained; a first sparsification degree is obtained by calculation based on the adjustment ratio and the original sparsification degree, and the sparsification degree of the traffic light image is adjusted to the first sparsification degree to obtain an adjusted traffic light image.

9. A vehicle, 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 method according to any one of claims 1 to 7 by executing the computer instructions.

Images