Prediction Method, Device, Electronic Device and Medium for Traffic Signal Switching Moment

Abstract

The present disclosure provides a method, apparatus, electronic device, and medium for predicting the switching moment of a traffic signal, which relates to the field of image processing technology, and particularly to the field of intelligent transportation technology. The specific implementation solution is as follows: for each target signal light to be scheduled, a collection task is sent to the collection client in the vehicles that can pass through the target signal light within a specified time period; based on the collection results fed back by the collection client, the switching information of the target signal light is determined, and the switching information includes the switching moment and the light state before switching; based on the switching information and the switching rule of the target signal light, the switching information of each switching of the target signal light in a future specified period is predicted, and the predicted switching information is sent to the navigation client in the vehicles passing through the target signal light within the specified period, so that the navigation client displays the countdown of the target signal light based on the received switching information. In this way, users can intuitively know when the target signal light will switch its light state.

Description

Technical Field

[0001] The present disclosure relates to the field of image processing technology, and particularly to the field of intelligent transportation technology. Background Art

[0002] When a driver is driving using in-vehicle navigation, at a traffic light intersection, the driver can judge the passing opportunity according to the display state of the traffic light. For example, for a traffic light that can display a countdown, when the driver knows that the green light is about to switch to a red light, the driver can decelerate and brake in time, which can avoid sudden braking or even rear-ending, and when the driver knows that the red light is about to switch to a green light, the driver can prepare for starting the vehicle in advance. Summary of the Invention

[0003] The present disclosure provides a method, apparatus, electronic device and medium for predicting the switching moment of a traffic light.

[0004] In a first aspect, the present disclosure provides a method for predicting the switching moment of a traffic light, including:

[0005] For each target traffic light to be scheduled, send a collection task to the collection client in the vehicle that can pass by the target traffic light within a specified time period, where the collection task is used to instruct the collection client to collect an image of the target traffic light within the specified time period;

[0006] Based on the collection result fed back by the collection client, determine the switching information of the target traffic light, where the switching information includes the switching moment and the light state before switching;

[0007] Based on the switching information and the switching rule of the target traffic light, predict the switching information for each switching of the target traffic light within a specified period in the future, and send the predicted switching information to the navigation client in the vehicle that passes by the target traffic light within the specified period, so that the navigation client displays the countdown of the target traffic light based on the received switching information.

[0008] In a second aspect, the present disclosure provides a device for predicting the switching moment of a traffic light, including:

[0009] A sending module, configured to send a collection task to the collection client in the vehicle that can pass by the target traffic light within a specified time period for each target traffic light to be scheduled, where the collection task is used to instruct the collection client to collect an image of the target traffic light within the specified time period;

[0010] A determining module, configured to determine the switching information of the target traffic light based on the collection result fed back by the collection client, where the switching information includes the switching moment and the light state before switching;

[0011] A prediction module, configured to predict the switching information of each switch of the target traffic signal in a specified period in the future based on the switching information and the switching rule of the target traffic signal;

[0012] The sending module is further configured to send the predicted switching information to the navigation clients in the vehicles passing through the target traffic signal within the specified period, so that the navigation clients display the countdown of the target traffic signal based on the received switching information.

[0013] In a third aspect, the present disclosure provides an electronic device, including:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in the first aspect.

[0017] In a fourth aspect, the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method described in the first aspect.

[0018] In a fifth aspect, the present disclosure provides a computer program product, including a computer program, where the computer program, when executed by a processor, implements the method described in the first aspect.

[0019] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. Description of the Drawings

[0020] The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. Among them:

[0021] Figure 1 is the first flowchart of the method for predicting the switching moment of a traffic signal provided by an embodiment of the present disclosure;

[0022] Figure 2 is the second flowchart of the method for predicting the switching moment of a traffic signal provided by an embodiment of the present disclosure;

[0023] Figure 3 is the third flowchart of the method for predicting the switching moment of a traffic signal provided by an embodiment of the present disclosure;

[0024] Figure 4It is the fourth schematic flow chart of the traffic signal switching time prediction method provided by the embodiments of the present disclosure;

[0025] Figure 5 It is the fifth schematic flow chart of the traffic signal switching time prediction method provided by the embodiments of the present disclosure;

[0026] Figure 6 It is an exemplary schematic diagram of the traffic signal switching time prediction method provided by the embodiments of the present disclosure;

[0027] Figure 7 It is a structural schematic diagram of the traffic signal switching time prediction device provided by the embodiments of the present disclosure;

[0028] Figure 8 It is a block diagram of an electronic device for implementing the traffic signal switching time prediction method of the embodiments of the present disclosure. Detailed Embodiments

[0029] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding and should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and structures are omitted below for clarity and conciseness.

[0030] When a user uses a vehicle-mounted navigation client for navigation, it is highly likely that the user cannot see the traffic light countdown at a traffic light intersection. Moreover, when approaching a traffic light intersection, the user cannot know the remaining time of the current red or green light, which will cause psychological uncertainty for the user and bring inconvenience to the user's driving decision-making. For example, when waiting for the red light to end at a traffic light intersection and not knowing the remaining time of the red light, it may cause distraction and result in the vehicle not starting in time after the green light switches. For another example, when approaching a traffic light intersection and not knowing that the green light is about to end, there may be sudden braking or even rear-end collisions.

[0031] In order to enable the user to know the remaining time of the current red or green light, in the embodiments of the present disclosure, the future traffic light switching time can be predicted, so as to display the traffic light countdown for the user on the navigation client, so that the user can accurately know the traffic light switching time and bring a better experience to the user.

[0032] The following provides a detailed introduction to the traffic signal switching time prediction method provided by the embodiments of the present disclosure.

[0033] As Figure 1 shown, the embodiments of the present disclosure provide a traffic signal switching time prediction method, which can be applied to a server. The method includes:

[0034] S101. For each target traffic signal to be scheduled, send a collection task to the collection clients in the vehicles that can pass by the target traffic signal within a specified time period. The collection task is used to instruct the collection clients to collect images of the target traffic signal within the specified time period.

[0035] Among them, there are multiple target traffic signals to be scheduled, and the embodiments of the present disclosure can predict the switching moments of each target traffic signal to be scheduled.

[0036] The server can determine the vehicles that can pass by the intersection where the target traffic signal is located within a specified time period based on the real-time collected driving trajectory information, and then send a collection task to the collection clients installed in these vehicles.

[0037] The collection client can be an in-vehicle terminal. The collection client can control the in-vehicle camera to collect images of the target traffic signal according to the collection task and obtain the images collected by the in-vehicle camera; or, the collection client can extract the images of the target traffic signal from the video captured by the in-vehicle camera in real time according to the collection task.

[0038] S102. Based on the collection results fed back by the collection clients, determine the switching information of the target traffic signal. The switching information includes the switching moment and the light state before switching.

[0039] Among them, the switching moment can be in the form of a timestamp, and the light state refers to the color of the traffic signal, such as red, green, or yellow. If the green light was on before switching, the light state before switching is green.

[0040] S103. Based on the switching information and the switching rule of the target traffic signal, predict the switching information of each switching of the target traffic signal within a specified future period, and send the predicted switching information to the navigation clients in the vehicles passing by the target traffic signal within the specified period, so that the navigation clients can display the countdown of the target traffic signal based on the received switching information.

[0041] Among them, the server can obtain information such as the driving trajectory and navigation route of the vehicle where the navigation client is located in real time. Based on the real-time obtained driving trajectory and navigation route information, it can know the vehicles that can pass by the target traffic signal within a specified period, and then send the predicted switching information to the navigation clients in these vehicles. In this way, when the navigation client passes by or approaches the intersection where the target traffic signal is located within the specified period, the navigation client can display the countdown of the target traffic signal on the navigation interface.

[0042] In the embodiments of the present disclosure, the specified period can be determined based on the switching rule of the target traffic signal and the above-mentioned specified time period. For example, if the switching rule of the target traffic signal is the same from 7:00 to 10:00 every morning, the specified time period can be from 7:00 to 7:10 of a day, and the specified period can be from 7:00 to 10:00 of each subsequent day.

[0043] To improve the accuracy of prediction, the method flow shown can also be re-executed for the target traffic signal every once in a while. Figure 1 For example, the specified period is from 7:00 to 10:00 every day within a week, and then again, from 7:00 to 7:10 in the morning of the first day of the second week is used as the specified time period to re-predict the switching information, and then from 7:00 to 10:00 in the morning of each day of the second week, the navigation client passing by the target traffic signal uses the re-predicted switching information to display the countdown of the target traffic signal.

[0044] The above-mentioned specified time period and specified period can be set based on the actual situation. For example, if it is found through testing that the prediction of the target traffic signal is not accurate enough, the frequency of re-prediction is increased; if it is found through testing that the prediction accuracy of the target traffic signal is high, the frequency of re-prediction can be reduced.

[0045] By adopting this method, a collection task can be sent to the collection client that can pass by the target traffic signal within the specified time period, so that the collection client collects the image of the target traffic signal within the specified time period, and then obtains the collection result fed back by the collection client. Since the collection result is obtained based on the images of the target traffic signal that have been collected within the specified time period, the switching information included in the collection result is the switching information of the actually occurred light state switching. Therefore, using this switching information and the switching rule of the target traffic signal, the switching information of each switching of the target traffic signal within the future specified period can be accurately predicted. Further, by sending the switching information to the navigation client, the navigation client can display the countdown when passing by the target traffic signal within the future specified period, so that the user can intuitively know when the target traffic signal will perform a light state switching, which is convenient for driving decisions and can improve the user experience of using the navigation client.

[0046] In some embodiments, before executing the Figure 1 method flow, the server can also obtain the real-time trajectory information returned by multiple navigation clients, and based on the real-time trajectory information and the switching rules of multiple pre-collected traffic signals, determine multiple target traffic signals to be scheduled, and the specified time period corresponding to each target traffic signal.

[0047] In the embodiments of the present disclosure, the switching rules of traffic signals can be mined using a large amount of historical trajectory information and traffic light images, or cooperation can also be carried out with relevant departments to obtain the switching rules of traffic signals provided by the relevant departments.

[0048] Optionally, the real-time trajectory information transmitted back by the navigation client can be used to determine the traffic lights with relatively high current popularity, and the traffic lights with relatively high popularity can be used as the target traffic lights. Then, based on the switching rules of the target traffic lights, the scheduling time interval corresponding to the target traffic lights, that is, the specified time period, can be determined. For example, if the switching rules of the target traffic lights include the switching rules during the morning rush hour, the switching rules during the evening rush hour, and the switching rules during the remaining time periods, then the scheduling time interval corresponding to the target signal can be determined to include a period of time during the morning rush hour, a period of time during the evening rush hour, and a period of time during the remaining time periods.

[0049] For another example, if the switching rule of the target traffic light is to switch once every 1 minute throughout the day, then any period of time throughout the day can be selected as the specified time period corresponding to the target traffic light.

[0050] It can be understood that in order for the acquisition client to know the location of the target traffic light, the geographical location area where the target traffic light is located can also be specified in the acquisition task.

[0051] By adopting this method, the server can determine multiple target traffic lights that need to be scheduled and the specified time period corresponding to each target traffic light based on the real-time trajectory information and the switching rules of multiple traffic lights collected in advance, so that the acquisition task can be issued more accurately, and the acquisition results fed back by the acquisition client based on the acquisition task can be predicted more accurately.

[0052] In the above embodiment, the target traffic light is a countdown traffic light or a non-countdown traffic light.

[0053] A countdown traffic light refers to a traffic light that can display a countdown, and a non-countdown traffic light refers to a traffic light that does not display a countdown.

[0054] The acquisition method indicated by the acquisition task of the non-countdown traffic light is: continuously collect images including the non-countdown traffic light multiple times at the specified acquisition frequency. It can be understood that for traffic lights without a countdown, multiple consecutive images including the traffic light are required to determine the switching moment of the traffic light. For example, the specified acquisition frequency can be once per second, or once every 500 milliseconds.

[0055] The acquisition method indicated by the acquisition task of the countdown traffic light is: at the specified moment, collect an image including the countdown traffic light.

[0056] By adopting this method, the characteristics of countdown signal lights and non-countdown signal lights can be utilized to select an appropriate acquisition method. For non-countdown signal lights, the client can continuously acquire images including non-countdown signal lights multiple times, so that the moment of lamp state switching can be determined using these consecutive images subsequently. For countdown signal lights, the moment of lamp state switching can be determined through the countdown numbers, so there is no need to acquire too many images. In this way, on the basis of acquiring as few images as possible, the accuracy of the determined switching information can be ensured.

[0057] In an implementation manner of the present disclosure, the first type of acquisition result fed back by the acquisition client for non-countdown signal lights includes: the lamp state before switching, the lamp state after switching, and the switching moment determined by the acquisition client based on the acquired images. For example, if among multiple continuously acquired images, the lamp states of the signal lights included in two adjacent images are different, the lamp states of the signal lights included in these two adjacent images can be respectively used as the lamp state before switching and the lamp state after switching, and the smaller timestamp among the acquisition timestamps of these two adjacent images is used as the switching moment.

[0058] The second type of acquisition result fed back by the acquisition client for countdown signal lights includes: the current lamp state, the countdown number, and the acquisition timestamp determined by the acquisition client based on the acquired images.

[0059] In the embodiments of the present disclosure, both the first type of acquisition result and the second type of acquisition result can be in the form of text.

[0060] On this basis, as Figure 2 shown, the above S102, determining the switching information of the target signal light based on the acquisition result fed back by the acquisition client, can be implemented as S1021 or S1022.

[0061] S1021, when receiving the first type of acquisition result, use the first type of acquisition result as the switching information of the target signal light.

[0062] It can be understood that in this case, the switching moment included in the switching information is the switching moment when the lamp state has actually switched. The server can use this switching moment and combine the switching rule of the target signal light to predict the future switching moment of the target signal light.

[0063] As an example, assume that the switching moment included in the switching information is 07:01:00, the lamp state before switching is red, and the switching rule is to switch once per minute. Then the server can predict that the target signal light will switch from green to red at 07:02:00 and from red to green at 07:03:00, and so on.

[0064] S1022. When the second type of acquisition result is received, use the sum of the acquisition timestamp and the countdown number as the switching moment of the target signal lamp, and use the current lamp state as the lamp state before switching to obtain the switching information of the target signal lamp.

[0065] It can be understood that in this case, the switching moment included in the switching information is the switching moment calculated based on the acquisition timestamp and the countdown number. Then, it can be determined that the target signal lamp will switch from the current lamp state to the next lamp state at this switching moment. For example, assume that the acquisition timestamp of the image is 07:01:50, the countdown number is 10, and the current lamp state is green. Then, it can be determined that the target signal lamp will switch from green to red at 07:02:00.

[0066] By adopting this method, the server can receive the first type of acquisition result and the second type of acquisition result fed back by the acquisition client based on the acquired images, that is, the acquisition client has completed the image recognition of the acquired images. In this way, the acquisition client only needs to feed back the acquisition result obtained after the image recognition to the server, without transmitting the acquired images to the server, which can reduce the amount of data transmitted, reduce the information transmission overhead, improve the transmission efficiency, and avoid excessive computing pressure on the server. The server can use different types of acquisition results to determine the accurate switching information of the target signal lamp, and then can more accurately predict the future switching information of the target signal lamp.

[0067] In another implementation manner of the embodiments of the present disclosure, the third type of acquisition result fed back by the acquisition client for non-countdown signal lamps includes multiple first images, and the first images are images including non-countdown signal lamps. Among them, the multiple first images are images acquired by the acquisition client according to the received acquisition task.

[0068] The fourth type of acquisition result fed back by the acquisition client for countdown signal lamps includes a second image, and the second image is an image including a countdown signal lamp. Among them, the second image is an image acquired by the acquisition client according to the received acquisition task.

[0069] On this basis, as Figure 3 shown, the above S102 determines the switching information of the target signal lamp based on the acquisition result fed back by the acquisition client, and can be specifically implemented as S1023 or S1024.

[0070] S1023. When the third type of acquisition result is received, perform image recognition on the multiple first images, and use the lamp state before switching, the lamp state after switching, and the switching moment obtained by the image recognition as the switching information of the target signal lamp.

[0071] S1024. When receiving the fourth type of acquisition result, perform image recognition on the second image to obtain the current light state, the countdown number, and the acquisition timestamp. Use the sum of the acquisition timestamp and the countdown number as the switching moment of the target signal lamp, and use the current light state and the calculated switching moment as the switching information of the target signal lamp.

[0072] Using this method, the server can receive the acquired images fed back by the acquisition client based on the acquisition task. Furthermore, the server can obtain the switching information of the target signal lamp through image recognition. Performing image recognition by the server can reduce the requirements for the acquisition client and make the deployment simpler. Moreover, the server can use different methods to recognize images with countdown signal lamps and images without countdown signal lamps, so that the determined switching information can be more accurate.

[0073] In some embodiments of the present disclosure, as Figure 4 shown, performing image recognition on multiple first images may specifically include the following steps:

[0074] S401. Obtain the acquisition timestamp of each first image.

[0075] Among them, the server can obtain the acquisition timestamp of the first image from the detailed information of the first image.

[0076] The detailed information of the image includes information such as the timestamp when the image was taken and the shooting position. The server can obtain the timestamp when the first image was taken from the detailed information of the first image and use this timestamp as the acquisition timestamp.

[0077] S402. For each first image, use the target detection model to recognize the first image to obtain the target position of the non-countdown signal lamp included in the first image in the first image.

[0078] Among them, the server can input each first image into the target detection model respectively and obtain the target position output by the target detection model for each first image.

[0079] The target detection model is a pre-trained neural network model capable of detecting traffic signal lamps in images.

[0080] S403. Use the light state recognition model to recognize the non-countdown signal lamp at the target position in the first image to obtain the light state of the non-countdown signal lamp.

[0081] Optionally, the server can intercept the image of the non-countdown signal lamp from the target position in the first image, and then input the image of the non-countdown signal lamp into the light state recognition model and obtain the light state output by the light state recognition model.

[0082] Alternatively, the server may input the first image and the target position into the traffic light state recognition model. Then, the traffic light state recognition model may recognize the non-countdown traffic lights included in the first image based on the target position and output the traffic light state, and the server may obtain the traffic light state output by the traffic light state recognition model.

[0083] S404. Determine two adjacent first images in which a traffic light state transition occurs based on the traffic light states of the non-countdown traffic lights included in multiple first images.

[0084] For example, if the server obtains 20 consecutively acquired first images, and the traffic light states of the traffic lights included in the first 10 first images are all red lights, and the traffic light state of the traffic lights included in the 11th first image is a green light, it can be determined that a traffic light state transition occurs between the 10th and 11th first images.

[0085] It should be noted that the embodiments of the present disclosure do not limit the execution order between S401 and S402 - S404. S401 only needs to be executed before S405. Figure 4 Here, taking the example of first executing S401 and then executing S402 - S404.

[0086] S405. Use the smaller acquisition timestamp among the acquisition timestamps of two adjacent first images as the switching moment, and use the traffic light state of the traffic lights included in the first image with the smaller acquisition timestamp among the two adjacent first images as the traffic light state before the transition.

[0087] Continuing with the example in S404, the two adjacent first images are the 10th and 11th first images, and the acquisition timestamp of the 10th first image is smaller. Then, the acquisition timestamp of the first image can be used as the switching moment, and the traffic light state of the traffic lights included in the first image can be used as the traffic light state before the transition, that is, the traffic light state before the transition is a red light.

[0088] By using this method, the server can use the object detection model to recognize the target position of the non-countdown traffic lights included in the first image in the first image, and further use the traffic light state recognition model to recognize the traffic light state of the non-countdown traffic lights included in the first image. Then, based on the recognition results, the server can accurately determine the switching moment and the traffic light state before the transition when the traffic light state transition occurs. Compared with the related art methods of cooperating with relevant departments in charge of traffic lights to obtain the switching information provided by the relevant departments, or using the moment information such as acceleration, deceleration, stop, and start in the driving trajectory to determine the switching information, the embodiments of the present disclosure can determine the switching information of the target traffic lights more accurately at a lower cost.

[0089] In some embodiments of the present disclosure, as Figure 5 shown, the image recognition of the second image may specifically include the following steps:

[0090] S501. Obtain the acquisition timestamp of the second image.

[0091] The method for obtaining the acquisition timestamp of the second image is the same as that for obtaining the acquisition timestamp of the first image. For relevant descriptions, refer to S401, which will not be elaborated here.

[0092] S502. Use the object detection model to identify the second image to obtain the target position of the non-countdown signal light in the second image.

[0093] The server can input the second image into the object detection model and obtain the target position output by the object detection model.

[0094] S503. Use the light state recognition model to identify the countdown signal light at the target position in the second image to obtain the current light state.

[0095] Optionally, the server can intercept the image of the countdown signal light from the target position in the second image, and then input the image of the countdown signal light into the light state recognition model and obtain the light state output by the light state recognition model.

[0096] Alternatively, the server can input the second image and the target position into the light state recognition model. Then, the light state recognition model can identify the countdown signal light included in the second image based on the target position and output the light state. The server can obtain the light state output by the light state recognition model.

[0097] S504. Use the optical character recognition (OCR) model to identify the countdown signal light at the target position in the second image to obtain the countdown number.

[0098] Optionally, the server can intercept the image of the countdown signal light from the target position in the second image, and then input the image of the countdown signal light into the OCR model and obtain the countdown number output by the OCR model.

[0099] Alternatively, the server can input the second image and the target position into the OCR model. Then, the OCR model can identify the countdown signal light included in the second image based on the target position and output the countdown number. The server can obtain the countdown number output by the OCR model.

[0100] It should be noted that the embodiments of the present disclosure do not limit the execution order between S503 and S504, which can be executed in parallel or in sequence. Figure 5 Here, it is taken as an example that S503 is executed first.

[0101] Moreover, the embodiments of the present disclosure do not limit the execution order between S501 and S502 - S504 either.Figure 5 Taking the case where S501 is executed first and then S502 - S504 in Zhongyi as an example.

[0102] Using this method, the server can first detect the target position of the countdown signal light included in the second image by using the target detection model, and then respectively identify the light state of the countdown signal light by using the light state recognition model, and identify the countdown number in the countdown signal light by using the OCR recognition model, so that the switching information of the countdown signal light can be accurately obtained. Compared with the related art of cooperating with the relevant departments in charge of traffic signal lights to obtain the switching information provided by the relevant departments, or using the information such as acceleration, deceleration, and start - stop moments in the driving trajectory to determine the switching information, the embodiments of the present disclosure can determine the switching information of the target signal light more accurately at a lower cost.

[0103] It should be noted that, as an optional implementation manner, the server can send the same acquisition task to multiple acquisition clients. Correspondingly, the server can receive the acquisition results feedback by multiple acquisition clients. In this case, the server can perform timestamp calibration and outlier filtering processing on the acquisition results feedback by multiple acquisition clients.

[0104] Among them, the clocks of the cameras on different vehicles may not be synchronized, and there may be clock errors. Then, the timestamps of the images in the images feedback by the acquisition clients may also be incorrect. In the embodiments of the present disclosure, the images feedback by multiple acquisition clients based on the same acquisition task can be subjected to timestamp calibration, and the images with incorrect timestamps are regarded as abnormal images with too low confidence and deleted.

[0105] For example, if the specified time period specified by the acquisition task is from 7:00 to 7:10, and the server receives the acquisition results feedback by 3 acquisition clients for the same acquisition task, the timestamps of the images feedback by acquisition client 1 and acquisition client 2 are both between 7:00 and 7:10, and the timestamp of the image feedback by acquisition client 3 is after 7:10, then the image feedback by acquisition client 3 can be deleted.

[0106] For the case where the acquisition client feedbacks multiple first images, if the server receives the first images feedback by multiple acquisition clients for the same acquisition task, after completing the outlier filtering, the first images feedback by multiple acquisition clients can be sorted in ascending order according to the acquisition timestamps, and then the sorted multiple first images can be subjected to image recognition according to the method described in the above embodiments. This can further improve the accuracy of the determined switching information.

[0107] It should be noted that, in the scenario of image recognition by the acquisition client described in the above embodiments, the acquisition client can also follow Figure 4 andFigure 5 Perform image recognition by the method of

[0108] Figure 4 and Figure 5 The object detection model, light state recognition model, and OCR model used in the corresponding embodiments are all models trained based on a traffic signal light image sample set. Optionally, the object detection model can be implemented using YOLO (full English name: You Only Look Once) or other object detection algorithms. YOLO is an object detection algorithm.

[0109] The traffic signal light image sample set includes images of high - heat traffic signal lights, traffic signal lights with countdowns, and traffic signal lights with specified switching rules. A high - heat traffic signal light is a traffic signal light at an intersection where the number of passing vehicles is greater than a preset number threshold within a preset time period.

[0110] Among them, in the embodiments of the present disclosure, the server can, in the offline calculation stage, use a large amount of user trajectories and offline - scheduled traffic signal light images to excavate a set of valuable traffic signal light images, and use the set of valuable traffic signal light images as the images in the traffic signal light image sample set.

[0111] Among them, the high - heat traffic signal lights can be determined based on user trajectories. For example, if the number of passing vehicles at a certain intersection is greater than the preset number threshold within a preset time period, the traffic signal light image including this intersection can be determined as a high - heat traffic signal light image.

[0112] The preset time period can be one week, one day, or several hours. The embodiments of the present disclosure do not make specific limitations on this.

[0113] Among them, traffic signal lights with specified switching rules can be determined through user trajectory information or by cooperating with relevant departments responsible for managing traffic signal lights to obtain some traffic signal lights with specified switching rules.

[0114] Optionally, it is also possible to screen valuable traffic signal light images based on the acquisition timestamp of the traffic signal light image and the proportion of the acquisition timestamp during the day and at night.

[0115] For example, traffic signal light images that can cover all time periods of the day can be screened, and the proportion of traffic signal lights collected during the day and the proportion of traffic signal lights collected at night can be preset, and the preset proportion is consistent with the true proportion of vehicles passing through the traffic signal lights during the day and at night.

[0116] In the embodiments of the present disclosure, the excavated traffic signal light images can be labeled by manual annotation. It is also possible to cooperate with relevant departments to obtain the true data of traffic signal lights.

[0117] For example, to train an object detection model, the positions of traffic lights in each traffic light image used as a sample can be manually labeled.

[0118] To train a light state recognition model, the light states in traffic light images can be manually labeled, or images of traffic lights and their corresponding light states provided by relevant departments can be directly obtained.

[0119] To train an OCR model, the countdown numbers in countdown traffic light images can be manually labeled, or images of countdown traffic lights and their corresponding countdown numbers provided by relevant departments can be directly obtained.

[0120] After obtaining the traffic light image sample set, the object detection model, light state recognition model, and OCR model can be supervised trained using the traffic light image sample set, so as to obtain object detection models, light state recognition models, and OCR models with higher recognition accuracy.

[0121] Optionally, during the process of training the above models, the regularization coefficient or focal-loss loss function in related technologies can also be used to reduce the phenomenon of overfitting, and by setting a reasonable sample distribution, the false recall of non-traffic light images can be reduced, and the generalization ability of the trained model for images collected at night and images collected at a long distance can be improved by means of data augmentation.

[0122] Using this method, the traffic light image sample set includes images of high-heat traffic lights, traffic lights with countdowns, and traffic lights with specified switching rules. Among them, through the images of high-heat traffic lights, this method can cover as many users as possible with fewer traffic light images. Through the images of traffic lights with countdowns, the utilization rate of the obtained images can be improved. Through the images of traffic lights with specified switching rules, the stability and accuracy of the subsequent predicted switching moments can be higher. That is, the traffic light image sample set covers many situations that may occur in actual scenarios, so that the trained model can accurately identify traffic light images, thereby improving the accuracy of the subsequent predicted switching information.

[0123] As Figure 6 shown, Figure 6 is an exemplary schematic diagram of a method for predicting the switching moment of a traffic light provided by an embodiment of the present disclosure.

[0124] The server can, based on user trajectory information, mine high-heat lights, lights with countdown information, and lights with specific rules from the full set of lights, and combine the images of these lights into a traffic light image sample set.

[0125] Then, the scheduling control module in the server can generate a real-time acquisition task based on the traffic signal light image sample set and the real-time trajectory information returned by the navigation client. The real-time acquisition task includes the specified time, specified scheduling duration, specified location area, specified acquisition frequency, and specified acquisition method for collecting the target traffic signal light.

[0126] The scheduling control module can distribute the acquisition subtasks to multiple vehicles based on the real-time trajectory returned by the navigation client and based on the real-time trajectory.

[0127] For example, if the scheduling control module determines that vehicle A and vehicle B will pass by the target traffic signal light 1 during the specified time period, it can respectively send the acquisition subtasks corresponding to the target traffic signal light 1 to the acquisition clients in vehicle A and vehicle B. If it determines that vehicle C will pass by the target traffic signal light 2 during the specified time period, it can send the acquisition subtask corresponding to the target traffic signal light 2 to vehicle C.

[0128] Then, the image recovery module in the server can recover the traffic signal light images collected by the acquisition client based on the acquisition subtasks.

[0129] Then, the image recognition module in the server can recognize the traffic signal light images. For the images of non-countdown traffic signal lights, it can recognize the positions and states of the red and green lights. For the images of countdown traffic signal lights, it can recognize the positions and states of the red and green lights and the countdown numbers. The specific recognition method can refer to the relevant description in the above embodiments and will not be elaborated here.

[0130] The server can then use the recognition results to calculate the traffic light switching points. That is, for the images with countdown numbers, the acquisition timestamp can be added to the countdown numbers to obtain the switching moment. For the images without countdown, the recognition results of multiple consecutive images can be used to determine the moment of the state switch.

[0131] Moreover, the server can also perform timestamp correction based on the recognition results of multiple images to delete the recognition results of the images with incorrect timestamps.

[0132] The traffic light switching point prediction module of the server can predict the traffic light switching information in the future specified period based on the switching information calculated by the traffic light switching point calculation module, and then send the switching information to the navigation client, so that the user can see the countdown product form, that is, the navigation client can display the countdown of the traffic lights. And the switching point calculation module and the switching point prediction module of the server can process the recognition results of a batch of images to predict the switching information of multiple traffic signal lights.

[0133] The specific prediction method can refer to the relevant description in the above embodiments and will not be elaborated here.

[0134] Using the embodiments of the present disclosure, since more than half of the map navigation users will encounter traffic lights during navigation, and on average each user can pass through more than 20 traffic lights per day, the map navigation users across the country can pass through more than 500 million traffic lights per day. Therefore, the function of displaying the countdown through the navigation client in the embodiments of the present disclosure can reach millions of users on a single day and tens of millions of page views (PV), which can bring a more comfortable user experience to users, reduce the occurrence of traffic accidents, and improve the overall reputation of users for map navigation.

[0135] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information and other processes all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0136] Corresponding to the above method embodiments, the embodiments of the present disclosure further provide a prediction device for the switching moment of traffic lights, as Figure 7 shown. The device includes:

[0137] A sending module 701, configured to send a collection task to the collection client in the vehicles that can pass through the target traffic light within a specified time period for each target traffic light to be scheduled. The collection task is used to instruct the collection client to collect images of the target traffic light within the specified time period;

[0138] A determining module 702, configured to determine the switching information of the target traffic light based on the collection result fed back by the collection client. The switching information includes the switching moment and the light state before switching;

[0139] A prediction module 703, configured to predict the switching information of each switching of the target traffic light within a specified period in the future based on the switching information and the switching rule of the target traffic light;

[0140] The sending module 701 is further configured to send the predicted switching information to the navigation client in the vehicles that pass through the target traffic light within the specified period, so that the navigation client displays the countdown of the target traffic light based on the received switching information.

[0141] Optionally, the target traffic light is a countdown traffic light or a non-countdown traffic light;

[0142] The collection method indicated by the collection task of the non-countdown traffic light is: continuously collecting images including the non-countdown traffic light multiple times at a specified collection frequency;

[0143] The collection method indicated by the collection task of the countdown traffic light is: collecting images including the countdown traffic light at a specified moment.

[0144] Optionally, the first type of acquisition result fed back by the acquisition client for non-countdown signal lights includes: the pre-switching light state, the post-switching light state, and the switching moment determined by the acquisition client based on the acquired image; the second type of acquisition result fed back by the acquisition client for countdown signal lights includes: the current light state, the countdown number, and the acquisition timestamp determined by the acquisition client based on the acquired image;

[0145] The determination module 702 is specifically configured to:

[0146] In the case of receiving the first type of acquisition result, use the first type of acquisition result as the switching information of the target signal light; or,

[0147] In the case of receiving the second type of acquisition result, use the sum of the acquisition timestamp and the countdown number as the switching moment of the target signal light, and use the current light state as the pre-switching light state to obtain the switching information of the target signal light.

[0148] Optionally, the third type of acquisition result fed back by the acquisition client for non-countdown signal lights includes multiple first images, and the first image is an image including a non-countdown signal light; the fourth type of acquisition result fed back by the acquisition client for countdown signal lights includes a second image, and the second image is an image including a countdown signal light;

[0149] The determination module 702 is specifically configured to:

[0150] In the case of receiving the third type of acquisition result, perform image recognition on the multiple first images, and use the pre-switching light state, the post-switching light state, and the switching moment obtained by the image recognition as the switching information of the target signal light; or,

[0151] In the case of receiving the fourth type of acquisition result, perform image recognition on the second image to obtain the current light state, the countdown number, and the acquisition timestamp, and use the sum of the acquisition timestamp and the countdown number as the switching moment of the target signal light, and use the current light state and the calculated switching moment as the switching information of the target signal light.

[0152] Optionally, the determination module 702 is specifically configured to:

[0153] Obtain the acquisition timestamp of each first image;

[0154] For each first image, use the target detection model to recognize the first image to obtain the target position of the non-countdown signal light included in the first image in the first image;

[0155] Use the light state recognition model to recognize the non-countdown signal light at the target position in the first image to obtain the light state of the non-countdown signal light;

[0156] Based on the light states of non-countdown signal lights included in multiple first images, determine two adjacent first images in which a light state change occurs;

[0157] Use the smaller acquisition timestamp among the acquisition timestamps of the two adjacent first images as the switching moment, and use the light state of the signal light included in the first image with the smaller acquisition timestamp among the two adjacent first images as the light state before switching;

[0158] The determination module 702 is specifically further configured to:

[0159] Obtain the acquisition timestamp of the second image;

[0160] Use the target detection model to identify the second image to obtain the target position of the non-countdown signal light in the second image;

[0161] Use the light state recognition model to identify the countdown signal light at the target position in the second image to obtain the current light state;

[0162] Use the optical character recognition OCR model to identify the countdown signal light at the target position in the second image to obtain the countdown number;

[0163] Among them, the target detection model, the light state recognition model, and the OCR model are all models trained based on the signal light image sample set.

[0164] Optionally, the signal light image sample set includes images of high-heat signal lights, images of signal lights with countdowns, and images of signal lights with specified switching rules. A high-heat signal light is a signal light at an intersection where the number of passing vehicles is greater than a preset number threshold within a preset time period.

[0165] Optionally, the device further includes:

[0166] An acquisition module, configured to acquire real-time trajectory information transmitted back by multiple navigation clients;

[0167] The determination module 702 is further configured to determine multiple target signal lights to be scheduled and the specified time period corresponding to each target signal light based on the real-time trajectory information and the switching rules of multiple traffic signal lights collected in advance.

[0168] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0169] Figure 8FIG. 0 shows a schematic block diagram of an exemplary electronic device 800 that may be used to implement embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely exemplary and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0170] As Figure 8 shown, the device 800 includes a computing unit 801 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0171] A plurality of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a magnetic disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0172] The computing unit 801 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 executes the various methods and processes described above, such as the method for predicting the traffic light switching moment. For example, in some embodiments, the method for predicting the traffic light switching moment can be implemented as a computer software program that is tangibly incorporated in a machine-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method for predicting the traffic light switching moment described above can be executed. Alternatively, in other embodiments, the computing unit 801 can be configured to execute the method for predicting the traffic light switching moment by any other suitable means (e.g., by means of firmware).

[0173] Various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0174] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to the processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as an independent software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0175] In the context of this disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0176] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0177] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0178] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The client-server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, can also be a server of a distributed system, or a server incorporating a blockchain.

[0179] It should be understood that the various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitations are imposed herein.

[0180] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub - combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A method for predicting the switching time of a traffic signal lamp, comprising: For each target signal lamp to be scheduled, send a collection task to the collection client in the vehicles that can pass by the target signal lamp within a specified time period, where the collection task is used to instruct the collection client to collect images of the target signal lamp within the specified time period; Based on the collection results fed back by the collection client, determine the switching information of the target signal lamp, where the switching information includes the switching time and the lamp state before switching; Based on the switching information and the switching rule of the target signal lamp, predict the switching information for each switching of the target signal lamp within a specified future period, and send the predicted switching information to the navigation client in the vehicles that pass by the target signal lamp within the specified period, so that the navigation client displays the countdown of the target signal lamp based on the received switching information; When the target signal lamp is a countdown signal lamp, the fourth type of collection result fed back by the collection client includes a second image, and the second image is an image including the countdown signal lamp; The determining the switching information of the target signal lamp based on the collection results fed back by the collection client includes: When receiving the fourth type of collection result, perform image recognition on the second image to obtain the current lamp state, the countdown number, and the collection timestamp, and use the sum of the collection timestamp and the countdown number as the switching time of the target signal lamp, and use the current lamp state and the calculated switching time as the switching information of the target signal lamp.

2. The method according to claim 1, wherein The target signal lamp is a countdown signal lamp or a non-countdown signal lamp; The collection method indicated by the collection task for the non-countdown signal lamp is: continuously collect images including the non-countdown signal lamp multiple times at a specified collection frequency; The collection method indicated by the collection task for the countdown signal lamp is: collect an image including the countdown signal lamp at a specified moment.

3. The method according to claim 2, wherein The first type of collection result fed back by the collection client for the non-countdown signal lamp includes: the lamp state before switching, the lamp state after switching, and the switching time determined by the collection client based on the collected images; the second type of collection result fed back by the collection client for the countdown signal lamp includes: the current lamp state, the countdown number, and the collection timestamp determined by the collection client based on the collected images; The determining the switching information of the target signal lamp based on the collection results fed back by the collection client includes: When receiving the first type of collection result, use the first type of collection result as the switching information of the target signal lamp; or, When receiving the second type of collection result, use the sum of the collection timestamp and the countdown number as the switching time of the target signal lamp, and use the current lamp state as the lamp state before switching to obtain the switching information of the target signal lamp.

4. The method according to claim 2, wherein, The third type of collection result fed back by the collection client for the non-countdown signal lamp includes multiple first images, and the first image is an image including the non-countdown signal lamp; Determining the switching information of the target traffic signal based on the acquisition results fed back by the acquisition client includes: In the case of receiving the third type of acquisition result, performing image recognition on the multiple first images, and taking the light state before switching, the light state after switching, and the switching moment obtained by the image recognition as the switching information of the target traffic signal.

5. The method according to claim 4, wherein The performing image recognition on the multiple first images includes: Obtaining the acquisition timestamp of each first image; For each first image, using a target detection model to recognize the first image to obtain the target position of the non-countdown traffic signal included in the first image in the first image; Using a light state recognition model to recognize the non-countdown traffic signal at the target position in the first image to obtain the light state of the non-countdown traffic signal; Based on the light states of the non-countdown traffic signals included in the multiple first images, determining two adjacent first images with a light state switch; Taking the smaller acquisition timestamp among the acquisition timestamps of the two adjacent first images as the switching moment, and taking the light state of the traffic signal included in the first image with the smaller acquisition timestamp among the two adjacent first images as the light state before switching; The performing image recognition on the second image includes: Obtaining the acquisition timestamp of the second image; Using a target detection model to recognize the second image to obtain the target position of the non-countdown traffic signal in the second image; Using a light state recognition model to recognize the countdown traffic signal at the target position in the second image to obtain the current light state; Using an optical character recognition OCR model to recognize the countdown traffic signal at the target position in the second image to obtain the countdown number; Wherein, the target detection model, the light state recognition model, and the OCR model are all models trained based on a traffic signal image sample set.

6. The method according to claim 5, wherein, The traffic signal image sample set includes images of high-heat traffic signals, images of traffic signals with countdowns, and images of traffic signals with specified switching rules. The high-heat traffic signal is a traffic signal at an intersection where the number of passing vehicles is greater than a preset number threshold within a preset time period.

7. According to the method described in claim 1, before sending an acquisition task to the acquisition clients in the vehicles that can pass the target traffic signal within a specified time period for each target traffic signal to be scheduled, the method further includes: Obtaining real-time trajectory information fed back by multiple navigation clients; Based on the real-time trajectory information and the switching rules of multiple pre-acquired traffic signals, determining multiple target traffic signals to be scheduled and the corresponding specified time period for each target traffic signal.

8. A prediction device for the switching moment of a traffic signal, comprising: A sending module, configured to send an acquisition task to the acquisition clients in the vehicles that can pass the target traffic signal within a specified time period for each target traffic signal to be scheduled, where the acquisition task is used to instruct the acquisition clients to acquire images of the target traffic signal within the specified time period. A determination module, configured to determine switching information of the target signal lamp based on the acquisition result fed back by the acquisition client, where the switching information includes a switching moment and a lamp state before switching; A prediction module, configured to predict switching information of each switching of the target signal lamp within a specified period in the future based on the switching information and the switching rule of the target signal lamp; The sending module is further configured to send the predicted switching information to a navigation client in a vehicle passing by the target signal lamp within the specified period, so that the navigation client displays a countdown of the target signal lamp based on the received switching information; When the target signal lamp is a countdown signal lamp, a fourth type of acquisition result fed back by the acquisition client for the countdown signal lamp includes a second image, where the second image is an image including the countdown signal lamp; The determination module is specifically configured to: When receiving the fourth type of acquisition result, perform image recognition on the second image to obtain the current lamp state, the countdown number, and the acquisition timestamp, use the sum value of the acquisition timestamp and the countdown number as the switching moment of the target signal lamp, and use the current lamp state and the calculated switching moment as the switching information of the target signal lamp.

9. The device according to claim 8, wherein, The target signal lamp is a countdown signal lamp or a non-countdown signal lamp; The acquisition method indicated by the acquisition task of the non-countdown signal lamp is: continuously acquiring images including the non-countdown signal lamp multiple times at a specified acquisition frequency; The acquisition method indicated by the acquisition task of the countdown signal lamp is: acquiring an image including the countdown signal lamp at a specified moment.

10. The device according to claim 9, wherein, A first type of acquisition result fed back by the acquisition client for the non-countdown signal lamp includes: the lamp state before switching, the lamp state after switching, and the switching moment determined by the acquisition client based on the acquired image; a second type of acquisition result fed back by the acquisition client for the countdown signal lamp includes: the current lamp state, the countdown number, and the acquisition timestamp determined by the acquisition client based on the acquired image; The determination module is specifically configured to: When receiving the first type of acquisition result, use the first type of acquisition result as the switching information of the target signal lamp; or, When receiving the second type of acquisition result, use the sum value of the acquisition timestamp and the countdown number as the switching moment of the target signal lamp, and use the current lamp state as the lamp state before switching to obtain the switching information of the target signal lamp.

11. The device according to claim 9, wherein, A third type of acquisition result fed back by the acquisition client for the non-countdown signal lamp includes multiple first images, where the first image is an image including the non-countdown signal lamp; The determination module is specifically configured to: When receiving the third type of acquisition result, perform image recognition on the multiple first images, and use the lamp state before switching, the lamp state after switching, and the switching moment obtained by the image recognition as the switching information of the target signal lamp.

12. The apparatus according to claim 11, wherein, The determination module is specifically configured to: Obtain the acquisition timestamp of each first image; For each first image, use an object detection model to recognize the first image to obtain the target positions of the non-countdown signal lights included in the first image in the first image; Use a light state recognition model to recognize the non-countdown signal lights at the target positions in the first image to obtain the light states of the non-countdown signal lights; Based on the light states of the non-countdown signal lights included in the multiple first images, determine two adjacent first images in which a light state switch occurs; Use the smaller acquisition timestamp among the acquisition timestamps of the two adjacent first images as the switching moment, and use the light state of the signal lights included in the first image with the smaller acquisition timestamp among the two adjacent first images as the light state before the switch; The determining module is specifically further configured to: Obtain the acquisition timestamp of the second image; Use an object detection model to recognize the second image to obtain the target positions of the non-countdown signal lights in the second image; Use a light state recognition model to recognize the countdown signal lights at the target positions in the second image to obtain the current light state; Use an optical character recognition OCR model to recognize the countdown signal lights at the target positions in the second image to obtain the countdown numbers; Wherein, the object detection model, the light state recognition model, and the OCR model are all models trained based on a signal light image sample set.

13. The device according to claim 12, wherein, The signal light image sample set includes images of high-heat signal lights, images of signal lights with countdowns, and images of signal lights with specified switching rules. The high-heat signal lights are signal lights at intersections where the number of passing vehicles is greater than a preset number threshold within a preset time period.

14. The device according to claim 8, wherein the device further comprises: An acquisition module, configured to acquire real-time trajectory information transmitted back by multiple navigation clients; The determining module is further configured to determine multiple target signal lights to be scheduled and a specified time period corresponding to each target signal light based on the real-time trajectory information and the switching rules of multiple pre-acquired traffic signal lights.

15. An electronic device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute the method according to any one of claims 1-7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.

17. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, it implements the method according to any one of claims 1-7.

Images