Vehicle steering scene recognition method and related equipment thereof

Abstract

The invention discloses a vehicle steering scene recognition method and related equipment thereof, and relates to the technical field of vehicles, and the vehicle steering scene recognition method comprises the steps: obtaining historical scene fragments; on the basis of each frame of data in the historical scene segment, a scene recognition result is determined, so that the situation that data of a non-key road section is mistakenly considered as a steering scene (such as slight steering when a vehicle goes straight) is avoided, and the accuracy of the scene recognition result is improved; the scene recognition result can be that a vehicle left-turn scene exists in the historical scene segment, and / or a vehicle right-turn scene exists in the historical scene segment, and / or a vehicle steering scene does not exist in the historical scene segment; according to the method, only the historical scene fragment needs to be acquired, and the scene recognition result about vehicle steering in the historical scene fragment is accurately obtained in the mode, so that the data about the vehicle steering scene is simply acquired.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a vehicle turning scene recognition method and related equipment. Background Art

[0002] With the development of intelligent driving technology, the demand for data from real traffic scenarios is increasing. These data from real traffic scenarios are not only used to train and verify autonomous driving algorithms, but can also be used to improve vehicle perception, decision-making and control algorithms.

[0003] High-quality data from real-world traffic scenarios can significantly improve the accuracy of autonomous driving algorithms, vehicle perception, decision-making, and control algorithms. However, due to technical limitations involved in the data collection process and the relatively rare and complex nature of steering scenarios, obtaining data on vehicle steering scenarios is difficult. Summary of the Invention

[0004] The main purpose of this application is to provide a vehicle turning scene recognition method and related equipment, aiming to solve the technical problem of high difficulty in obtaining data on vehicle turning scenes.

[0005] To achieve the above objectives, the present application proposes a vehicle turning scene recognition method, which includes: Get historical scene clips; When there is a road intersection scene in the historical scene segment, determining a scene recognition result based on each frame data in the historical scene segment; The scene recognition result includes whether there is a vehicle left turn scene, a vehicle right turn scene, and / or no vehicle turning scene in the historical scene segment.

[0006] In one embodiment, the step of determining the scene recognition result based on each frame data in the historical scene segment includes: Determine whether there is a possibility of a vehicle turning in each frame between a target frame and a preset frame, wherein the target frame includes a start frame and the preset frame is separated from the target frame by a preset frame length; If so, determining a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively; A scene recognition result is determined based on the change in the vehicle's driving direction.

[0007] In one embodiment, the step of determining a change in the vehicle driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively includes: Calculating the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame respectively; Based on the angle difference, a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame is determined.

[0008] In one embodiment, after the step of determining the change in the vehicle driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the angle difference, the method further includes: The target frame is updated to be a frame that slides backward by a preset number of frames, and the process returns to the step of determining whether there is a possibility of turning of the vehicle corresponding to each frame between the target frame and the preset frame until the preset frame is the end frame, thereby obtaining the change in the vehicle driving direction in the multiple scene sub-segments corresponding to the start frame and the end frame.

[0009] In one embodiment, the step of determining a scene recognition result based on a change in the vehicle's driving direction includes: determining scene recognition results for the multiple scene sub-segments based on changes in the vehicle driving directions in the multiple scene sub-segments corresponding to the start frame and the end frame; After the step of determining the scene recognition result based on the change in the vehicle's driving direction, the method further includes: The vehicle turning moment is determined based on the target scene sub-segment where the vehicle turning scene exists.

[0010] In one embodiment, the step of determining a change in the vehicle's driving direction in a scene sub-segment corresponding to a target frame and a preset frame based on the angle difference includes: performing normalization processing on the angle difference; If the normalized angle difference is within the first angle interval, determining that the change in the vehicle's driving direction is a left turn; If the normalized angle difference is within the second angle interval, determining that the change in the vehicle's driving direction is a right turn; If the normalized angle difference is within the third angle range, determining that the change in the vehicle's driving direction is no turn; There is no intersection among the first angle interval, the second angle interval and the third angle interval.

[0011] In one embodiment, the step of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame includes: Determining whether the vehicle speed corresponding to the target frame is less than a preset speed threshold; If the vehicle speed is greater than or equal to a preset speed threshold, then calculating the vehicle travel distance from the target frame to the preset frame based on the vehicle position information corresponding to each frame; Determining whether the vehicle travel distance is less than a first distance threshold; If the vehicle travel distance is greater than or equal to a first distance threshold, it is determined that there is a possibility that the vehicle corresponding to each frame between the target frame and the preset frame has turned.

[0012] In one embodiment, after the step of obtaining the historical scene fragments, the method further includes: Determine whether there is intersection identification information in each frame of data; If so, the target distance between the intersection and the vehicle is calculated based on the intersection identification information; When the target distance is less than a second distance threshold, it is determined that an intersection scene exists in the historical scene segment.

[0013] In one embodiment, the intersection identification information includes traffic light information, and the traffic light information includes first position information of at least one traffic light. The step of calculating the target distance between the intersection and the vehicle based on the intersection identification information includes: Calculating the Euclidean distance between the at least one traffic light and the vehicle, and the first relative direction between the at least one traffic light and the vehicle based on the first position information and the vehicle position information corresponding to each frame of data; determining whether there is a target traffic light in front of the vehicle based on the angle between the first relative direction and the target direction of the vehicle speed corresponding to each frame of data; If so, the minimum distance among the Euclidean distances corresponding to the target traffic light is used as the target distance between the intersection and the vehicle.

[0014] In one embodiment, the intersection identification information further includes zebra crossing information, the zebra crossing information including second position information corresponding to any corner point of a rectangular box corresponding to at least one zebra crossing, and the step of calculating the target distance between the intersection and the vehicle based on the intersection identification information includes: Calculating a projected distance between the vehicle and the at least one zebra crossing, and a second relative direction between the at least one zebra crossing and the vehicle based on the second position information and the vehicle position information corresponding to each frame of data; determining whether there is a target zebra crossing in front of the vehicle based on the angle between the second relative direction and the target direction of the vehicle speed corresponding to each frame of data; If it exists, the minimum distance among the projection distances corresponding to the target zebra crossing is used as the target distance between the intersection and the vehicle.

[0015] In addition, to achieve the above-mentioned purpose, the present application also proposes a vehicle turning scene recognition device, which includes: Acquisition module, used to obtain historical scene fragments; The recognition module is used to determine a scene recognition result based on each frame data in the historical scene segment, wherein the scene recognition result includes the presence of a vehicle left turn scene, a vehicle right turn scene, and / or the absence of a vehicle turning scene in the historical scene segment.

[0016] In addition, to achieve the above-mentioned purpose, the present application also proposes a vehicle turning scene recognition device, which includes: a memory, a processor, and a computer program stored on the memory and runnable on the processor, and the computer program is configured to implement the steps of the vehicle turning scene recognition method described above.

[0017] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the vehicle turning scene recognition method as described above are implemented.

[0018] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of the vehicle turning scene recognition method as described above.

[0019] One or more technical solutions proposed in this application have at least the following technical effects: The present application obtains historical scene fragments; when there is an intersection scene in the historical scene fragment, the scene recognition result is determined based on the data of each frame in the historical scene fragment, so as to avoid mistaking the data of non-critical sections for turning scenes (for example, a slight turn of the vehicle when going straight), thereby improving the accuracy of the scene recognition result; the scene recognition result can be that there is a vehicle left turn scene in the historical scene fragment, and / or there is a vehicle right turn scene in the historical scene fragment, and / or there is no vehicle turning scene in the historical scene fragment; the present application only needs to obtain historical scene fragments, and accurately obtain the scene recognition result about the vehicle turning in the historical scene fragment through the above method, thereby simply obtaining data about the vehicle turning scene. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0022] Figure 1 A flow chart illustrating a first embodiment of the vehicle turning scene recognition method of the present application; Figure 2 A schematic diagram of a simplified process of the vehicle turning scene recognition method provided in Example 1 of the present application; Figure 3 A flow chart illustrating a second embodiment of the vehicle turning scene recognition method of the present application; Figure 4 A flowchart of the third embodiment of the vehicle turning scene recognition method of the present application is provided; Figure 5 This is a schematic diagram of the module structure of the vehicle turning scene recognition device according to an embodiment of the present application; Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the vehicle turning scene recognition method in the embodiment of the present application.

[0023] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0025] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0026] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of implementing the above functions, a vehicle turning scene recognition device, etc. The following uses the vehicle turning scene recognition device as an example to illustrate this embodiment and the following embodiments.

[0027] Based on this, the embodiment of the present application provides a vehicle turning scene recognition method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the vehicle turning scene recognition method of the present application.

[0028] In this embodiment, the vehicle turning scene recognition method includes steps S10 to S20: Step S10, obtaining historical scene fragments; It should be noted that with the development of intelligent driving technology, the demand for data from real-world traffic scenarios is growing. This data is not only used to train and validate autonomous driving algorithms, but can also be used to improve vehicle perception, decision-making, and control algorithms. High-quality data from real-world traffic scenarios can significantly improve the accuracy of autonomous driving algorithms, vehicle perception, decision-making, and control algorithms. However, due to technical limitations in the data collection process and the relatively rare and complex nature of turning scenarios, obtaining data on vehicle turning scenarios is difficult.

[0029] In order to solve the above problems, this embodiment obtains historical scene fragments, aiming to offline identify whether there are vehicle turning scenes in the historical scene fragments, so as to use the historical scene fragments with scene recognition results as effective sample data for training and verifying autonomous driving algorithms, vehicle perception algorithms, driving decision algorithms and vehicle control algorithms.

[0030] Specifically, the historical scene fragment can be a historically collected vehicle driving record, or it can be vehicle driving information generated by simulation; the historical scene fragment includes multiple frames of data, and the scene information corresponding to each frame of data may include dynamic perception information collected by sensors such as cameras and radars, or static map information obtained by matching the historically collected vehicle latitude and longitude with a high-precision map; wherein, the dynamic perception information includes historically collected road images, timestamps, vehicle location information (for example, the vehicle's coordinates in the global coordinate system), vehicle speed and acceleration, vehicle direction and steering angle, vehicle's direction of travel, vehicle steering wheel angle, driver operations (for example, braking, throttle force, gear shifting, etc.), etc., and the static map information includes traffic sign information on the road, the location and behavior of other vehicles, pedestrians, obstacles, and other information.

[0031] Step S20, when there is an intersection scene in the historical scene segment, determine the scene recognition result based on each frame data in the historical scene segment, wherein the scene recognition result includes the presence of a vehicle left turn scene, a vehicle right turn scene, and / or no vehicle turning scene in the historical scene segment.

[0032] Since vehicle turning operations usually occur in intersection scenes, in order to improve the recognition efficiency of vehicle turning scenes for historical scene fragments, this embodiment will only continue to recognize the vehicle turning scene for the historical scene fragment when it is determined that there is an intersection scene in the historical scene fragment; if there is no intersection scene in the historical scene fragment, the vehicle turning scene will not be recognized for the historical scene fragment, and the historical scene fragment will be abandoned as training data.

[0033] Specifically, the specific implementation method of this embodiment for determining the scene recognition result of the historical scene fragment based on each frame data in the historical scene fragment may be: determining the scene recognition result of the historical scene fragment based on the dynamic perception information and static map information corresponding to each frame data in the historical scene fragment.

[0034] Among them, the scene recognition results of the historical scene fragments include whether there is a vehicle left turn scene, or a vehicle right turn scene, or no vehicle turning scene in the historical scene fragments. Therefore, the historical scene fragments with scene recognition results can be used as training data.

[0035] Since the historical scene fragment includes multiple frames of data, the scene recognition results of the historical scene fragment can also be that there is a vehicle left turn scene, a vehicle right turn scene, and / or no vehicle turning scene in each frame data; in order to increase the amount of data, each frame data with scene recognition results can be used as training data.

[0036] For details, please refer to Figure 2 , the step S20 includes steps A1 to A3: Step A1, determining whether there is a possibility of a vehicle turning in each frame between a target frame and a preset frame, wherein the target frame includes a start frame, and the preset frame and the target frame are separated by a preset frame length; Since the historical scene fragments include multiple frames of data, in order to improve data processing efficiency, we can first determine whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame. If there is no possibility of turning, we can skip further identification of the scene sub-fragments corresponding to the target frame to the preset frame, thereby saving computing resources and time.

[0037] Here, it is assumed that the total number of frames of multi-frame data in the historical scene segment is recorded as N, and the preset frame length (frame_size) is determined based on the time required for the vehicle to complete a steering operation. Specifically, the preset frame length can be 50 frames or 100 frames, etc. The target frame (frame_i frame) includes the starting frame and other frames, and the preset frame is separated from the target frame by a preset frame length; it can be understood that when the target frame is the 0th frame, the preset frame can be the 100th frame, and when the target frame is the 20th frame, the preset frame can be the 120th frame.

[0038] That is, it is determined whether the vehicle corresponding to each frame between the target frame and the preset frame has the possibility of turning.

[0039] Specifically, a feasible implementation method of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame may be: Determine whether the vehicle speed corresponding to the target frame is less than a preset speed threshold; if the vehicle speed is greater than or equal to the preset speed threshold, calculate the vehicle travel distance from the target frame to the preset frame based on the vehicle position information corresponding to each frame; determine whether the vehicle travel distance is less than a first distance threshold; if the vehicle travel distance is greater than or equal to the first distance threshold, determine the possibility that the vehicle corresponding to each frame between the target frame and the preset frame has turned.

[0040] It should be noted that since the vehicle can only perform a steering operation when the vehicle speed meets certain conditions, this embodiment determines whether the vehicle speed corresponding to the target frame is less than a preset speed threshold (vel_static), thereby excluding the situation where the vehicle is stationary, avoiding further steering judgment, and reducing the waste of computing resources.

[0041] The preset speed threshold (distance_static) may be 1.0 m / s or 1.5 m / s. If the vehicle speed corresponding to the 0th frame is less than 1.0 m / s, the vehicle corresponding to the 0th frame is considered to be stationary, and no further scene recognition is performed on the frame.

[0042] Furthermore, if the vehicle speed corresponding to the target frame is greater than or equal to a preset speed threshold, the vehicle travel distance from the target frame to the preset frame is calculated based on the vehicle position information corresponding to each frame, and it is determined whether the vehicle travel distance is less than a first distance threshold, so as to exclude the situation where the vehicle is traveling at a low speed, avoid further turning judgment, and reduce the waste of computing resources.

[0043] The first distance threshold (distance_static) can be 5.0m or 6m. If the vehicle travel distance from frame 0 to frame 100 is less than 5.0m, it is considered that the vehicle is in a low-speed driving condition from frame 0 to frame 100, and no further scene recognition is performed on this frame.

[0044] Specifically, the vehicle position information includes vehicle coordinate information, which can be the vehicle coordinate coordinate_start (start_x, start_y) under the target frame (frame_i frame), and the vehicle coordinate coordinate_end (end_x, end_y) under the preset frame (frame_i+100 frame); based on the vehicle position information corresponding to each frame, the specific implementation method of calculating the vehicle driving distance from the target frame to the preset frame can be: calculating the straight-line distance dist between coordinate_start and coordinate_end. If dist is less than distance_static, it is considered that the vehicle is in a low-speed driving condition from the target frame to the preset frame.

[0045] Furthermore, if the vehicle travel distance is greater than or equal to a first distance threshold, the possibility of the vehicle turning in each frame between the target frame and the preset frame is determined, and then whether there is a vehicle turning scene in each frame between the target frame and the preset frame is identified.

[0046] It can be understood that, by setting the preset speed threshold and the first distance threshold, this embodiment can effectively filter out false turning signals caused by sensor noise or slight vibration, ensuring that only true turning behaviors are recognized.

[0047] Step A2: If yes, determining the change of the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively; It should be noted that the vehicle heading angle can be obtained through the vehicle box or the vehicle speed. Specifically, the vehicle bounding box is usually detected by a visual or distance sensor such as a camera or radar, so the vehicle heading angle can be calculated based on the vehicle bounding box. The vehicle heading angle can also be derived by analyzing the vehicle's velocity vector (including speed magnitude and direction).

[0048] Since the vehicle heading angle can reflect the change in the vehicle's driving direction, if it is determined that the vehicle corresponding to each frame between the target frame and the preset frame has the possibility of turning, the change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame to the preset frame is determined based on the vehicle heading angles corresponding to the target frame and the preset frame respectively.

[0049] Step A3: determining a scene recognition result based on the change in the vehicle's driving direction.

[0050] Furthermore, by analyzing changes in the vehicle's driving direction, it is possible to determine whether the vehicle has a left turn scenario, a right turn scenario, and / or no vehicle turning scenario, that is, to obtain a scene recognition result.

[0051] In this embodiment, historical scene fragments are obtained; based on the data of each frame in the historical scene fragments, the scene recognition results are determined to avoid mistaking data of non-critical sections for turning scenes (for example, a slight turn of the vehicle when traveling straight). By setting a preset speed threshold and a first distance threshold, false turning signals caused by sensor noise or slight shaking can be effectively filtered out to ensure that only true turning behaviors are identified. This embodiment only needs to obtain historical scene fragments and accurately obtain the scene recognition results about vehicle turning in the historical scene fragments through the above method, thereby simply obtaining data about vehicle turning scenes.

[0052] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 3 , the step A2 includes steps B1 to B2: Step B1, calculating the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame respectively; It should be noted that when it is determined that there is a possibility that the vehicle corresponding to each frame between the target frame and the preset frame has turned, it is still impossible to determine whether the vehicle has actually turned or not. This embodiment accurately captures the changes in the vehicle's driving direction by calculating the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame.

[0053] For example, calculate the angular difference (theta_end - theta_start) between the vehicle heading angle (theta_start) corresponding to the target frame (frame_i) and the vehicle heading angle (theta_end) corresponding to the preset frame (frame_i+100).

[0054] Step B2: determining a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the angle difference; Since the angle difference can reflect the change in the vehicle's driving direction from the target frame to the preset frame, this embodiment can determine the change in the vehicle's driving direction in the corresponding scene sub-segment from the target frame to the preset frame based on the angle difference.

[0055] For example, when the angle difference is within a certain range, it can be considered that the vehicle has turned, and if it is not within the range, it is considered that the vehicle has not turned.

[0056] In addition, a specific embodiment of determining the change in the vehicle driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the angle difference may also be: The angle difference is normalized; if the normalized angle difference is within a first angle interval, it is determined that the change in the vehicle's driving direction is a left turn; if the normalized angle difference is within a second angle interval, it is determined that the change in the vehicle's driving direction is a right turn; if the normalized angle difference is within a third angle interval, it is determined that the change in the vehicle's driving direction is no turn; wherein, there is no intersection between the first angle interval, the second angle interval, and the third angle interval.

[0057] It should be noted that, in the vehicle left turn scene and the vehicle right turn scene, the angle difference between the vehicle heading angle corresponding to the target frame and the preset frame may be - , 2 , 3 , 6 In order to facilitate the subsequent judgment of the steering type corresponding to the angle difference, this embodiment normalizes the angle difference, so as to more intuitively, simply and accurately judge the change in the vehicle's driving direction.

[0058] The normalization of the angle difference may be normalizing the angle difference to the same quantization standard, for example, normalizing the angle difference to [0, 2 )or(- , ) etc.; Specifically, the angle difference is normalized to [0, 2 ) can be calculated as follows: Normalized angle difference (theta_diff) = (theta_end - theta_start + 2 )% (2 ).

[0059] Correspondingly, the first angle interval, the second angle interval and the third angle interval can be set to represent the intervals where the corresponding angle differences of the corresponding scenes are located, and there is no intersection between the first angle interval, the second angle interval and the third angle interval; it can be understood that the first angle interval can be , the second angle interval can be , the third angle interval can be and , that is, there is no intersection between the first angle interval, the second angle interval and the third angle interval.

[0060] It can be understood that since the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame is normalized to the same quantitative standard, the change in the vehicle's driving direction can be determined simply by comparing the angles without causing any errors. , 6 Improve data processing efficiency when the same standard cannot be used for comparison of equal angles.

[0061] Specifically, if the normalized angle difference is within the first angle interval, it is determined that the change in the vehicle's driving direction is a left turn; if the normalized angle difference is within the second angle interval, it is determined that the change in the vehicle's driving direction is a right turn; if the normalized angle difference is within the third angle interval, it is determined that the change in the vehicle's driving direction is no turn.

[0062] Furthermore, after the step of determining the change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the angle difference, the following steps can be performed: updating the target frame to a frame that slides backward by a preset number of frames, returning to the step of determining whether there is a possibility of turning of the vehicle corresponding to each frame between the target frame and the preset frame, until the preset frame is the end frame, to obtain the change in the vehicle's driving direction in multiple scene sub-segments corresponding to the start frame and the end frame.

[0063] Since the above only determines whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame, and the change of the vehicle's driving direction in the scene sub-segments corresponding to the target frame and the preset frame, in order to determine the change of the vehicle's driving direction in multiple scene sub-segments corresponding to the start frame and the end frame, the target frame can be updated to a frame that slides backward by a preset number of frames, and return to the step of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame until the preset frame is the end frame.

[0064] Among them, the preset number of frames can be less than the number of frames corresponding to the preset frame length, thereby increasing the number of identified scene sub-segments without missing the identification of each frame data in the historical scene segment; the preset number of frames can also be equal to the number of frames corresponding to the preset frame length, reducing repeated scenes and improving scene recognition efficiency; the preset number of frames can also be greater than the number of frames corresponding to the preset frame length and less than (N-100 frames), thereby reducing the number of scene sub-segments that need to be identified and improving data processing efficiency.

[0065] In order to increase the number of identified scene sub-segments without missing the identification of each frame of data in the historical scene segment, this embodiment can return to the step of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame every 20 frames or every 50 frames until the preset frame is the end frame (that is, until all frames are traversed).

[0066] Furthermore, the specific implementation method of determining the scene recognition result based on the change of the vehicle's driving direction may be: determining the scene recognition results of the multiple scene sub-segments based on the change of the vehicle's driving direction in the multiple scene sub-segments corresponding to the start frame to the end frame.

[0067] It can be understood that if the change in the vehicle's driving direction in the scene sub-segment from the 0th frame to the 100th frame is a left turn, the change in the vehicle's driving direction in the scene sub-segment from the 20th frame to the 120th frame is no turn, ..., the change in the vehicle's driving direction in the scene sub-segment from the (N-100)th frame to the Nth frame is a right turn; then the corresponding scene recognition results from the 0th frame to the 100th frame are that there is a vehicle left turn scene, the scene recognition results from the 20th frame to the 120th frame are that there is no vehicle turning scene, ..., the scene recognition results from the (N-100)th frame to the Nth frame are that there is a vehicle right turn scene.

[0068] Furthermore, after the step of determining the scene recognition result based on the change in the vehicle's driving direction, the method may further include: determining the vehicle turning moment based on a target scene sub-segment containing a vehicle turning scene.

[0069] Since the change of the vehicle driving direction in the multiple scene sub-segments corresponding to the start frame and the end frame is obtained in the above manner, the turning moment corresponding to the target scene sub-segment with the vehicle turning scene can be determined.

[0070] For example, the target scene sub-segment with a vehicle turning scene includes the scene sub-segment corresponding to the 0th frame to the 100th frame, the scene sub-segment corresponding to the 80th frame to the 180th frame, ..., the scene sub-segment corresponding to the (N-100)th frame to the Nth frame; in order to obtain higher-quality training data, the vehicle turning moment (including the vehicle left turn moment and the vehicle right turn moment) can be determined based on the above target scene sub-segment with a vehicle turning scene, for example, the vehicle left turn moment is from the 0th frame to the 100th frame, and the vehicle right turn moment is from the (N-100)th frame to the Nth frame.

[0071] This embodiment can quickly locate the vehicle left turn scene and the vehicle right turn scene in the historical scene segment through the above method, and can accurately locate the vehicle left turn moment and the vehicle right turn moment; thereby obtaining more detailed data about the vehicle turning scene.

[0072] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above introduction and will not be described in detail later. Figure 4 After step S10, steps C1 to C3 are further included: Step C1, determining whether there is intersection identification information in each frame of data; In order to improve the efficiency and accuracy of identifying intersection scenes, this embodiment first determines whether there is intersection identification information in each frame data; if so, it is more likely that there is an intersection in the scene corresponding to the frame; if not, it is less likely that there is an intersection in the scene corresponding to the frame.

[0073] The specific implementation method for determining whether there is intersection identification information in each frame of data can be: based on the road image in the dynamic perception information, identify whether there is intersection identification information in each frame of data through a preset target recognition model; it can also be determined whether there is intersection identification information in the traffic sign information in the static map information corresponding to each frame of data.

[0074] Among them, intersection sign information can be speed bumps, intersection warning signs, traffic lights (traffic lights), zebra crossings, etc.

[0075] Step C2: If yes, calculate the target distance between the intersection and the vehicle based on the intersection identification information; It is understandable that since a vehicle usually performs a turning operation when it is about to reach an intersection, in order to ensure effective filtering of frames where turning is impossible, this embodiment calculates the target distance between the intersection and the vehicle based on the intersection identification information.

[0076] Specifically, for each frame of data, there may be traffic light information and / or zebra crossing information. If traffic light information traffic_lights_infos does not exist in a frame, the distance from the traffic light to the vehicle is assigned a large positive value (such as 10000) to indicate that there is no intersection ahead.

[0077] If a frame contains traffic light information, the specific implementation of calculating the target distance between the intersection and the vehicle based on the intersection identification information may be: Based on the first position information and the vehicle position information corresponding to each frame data, the Euclidean distance between the at least one traffic light and the vehicle, and the first relative direction between the at least one traffic light and the vehicle are calculated; based on the angle between the first relative direction and the target direction of the vehicle's driving speed corresponding to each frame data, it is determined whether there is a target traffic light in front of the vehicle; if so, the minimum distance among the Euclidean distances corresponding to the target traffic light is used as the target distance between the intersection and the vehicle.

[0078] It can be understood that for each frame of data, when traffic light information exists in the road perception information in this embodiment, since multiple sets of traffic light information may exist at the same time, this embodiment can use the minimum distance from the vehicle to each traffic light as the target distance.

[0079] It should be noted that the traffic light information includes first position information of at least one traffic light, and the first position information may be the first coordinate of at least one traffic light in the global coordinate system.

[0080] For at least one traffic light in each frame of data, the Euclidean distance between the at least one traffic light and the vehicle can be calculated based on the first position information and the vehicle position information corresponding to each frame of data; at the same time, the first relative direction between the at least one traffic light and the vehicle can be calculated based on the first position information and the vehicle position information corresponding to each frame of data.

[0081] Specifically, the Euclidean distance between the first coordinate corresponding to at least one traffic light and the vehicle coordinates can be calculated, that is, the direction from at least one traffic light to the vehicle can be used as the first relative direction (Traffic_light_direction_vector, i.e., represented by a vector consisting of the first coordinate and the vehicle coordinates).

[0082] Based on the angle (vector angle angle) between the first relative direction and the target direction (V, i.e., represented by a vector with a speed direction) of the vehicle speed corresponding to each frame data, it is determined whether there is a target traffic light in front of the vehicle.

[0083] It can be understood that if the absolute value of the angle angle corresponding to the traffic light is less than 90 degrees, the traffic light is considered to be in front of the vehicle; if the absolute value of the angle angle corresponding to the traffic light is greater than or equal to 90 degrees, the traffic light is considered to be behind the vehicle.

[0084] In order to improve the determination efficiency, this embodiment uses the minimum distance (ego_to_traffic_dist) in the Euclidean distance corresponding to the target traffic light in front of the vehicle as the target distance between the intersection and the vehicle.

[0085] If there is no zebra crossing information (crosswalks_infos) in a certain frame, the distance from the traffic light to the vehicle (ego_to_crosswalks_dist) is assigned a large positive value (such as 10000) to indicate that there is no intersection ahead.

[0086] If a frame contains zebra crossing information, the specific implementation method of calculating the target distance between the intersection and the vehicle based on the intersection identification information may be: Based on the second position information and the vehicle position information corresponding to each frame data, the projected distance between the vehicle and the at least one zebra crossing and the second relative direction between the at least one zebra crossing and the vehicle are calculated; based on the angle between the second relative direction and the target direction of the vehicle's driving speed corresponding to each frame data, it is determined whether there is a target zebra crossing in front of the vehicle; if so, the minimum distance among the projected distances corresponding to the target zebra crossing is used as the target distance between the intersection and the vehicle.

[0087] It can be understood that for each frame of data, when zebra crossing information exists in the road perception information in this embodiment, since multiple sets of zebra crossing information may exist simultaneously, this embodiment can use the minimum distance from the vehicle to each zebra crossing as the target distance.

[0088] It should be noted that the zebra crossing includes at least one second position information corresponding to any corner point of a rectangular frame corresponding to the zebra crossing. The second position information may be the second coordinate of any corner point in the global coordinate system.

[0089] For any corner point of the rectangular box corresponding to at least one zebra crossing in each frame of data, the projected distance between the vehicle and the at least one zebra crossing can be calculated based on the second position information and the vehicle position information corresponding to each frame of data; at the same time, the second relative direction (crosswalks_direction_vector, i.e., represented by a vector consisting of the second coordinate and the vehicle coordinate) of at least one zebra crossing and the vehicle can be calculated based on the second position information and the vehicle position information corresponding to each frame of data.

[0090] Specifically, based on the angle (vector angle crosswalks_angle) between the second relative direction and the target direction of the vehicle speed corresponding to each frame data (V, i.e., represented by a vector with speed direction), it is determined whether there is a target zebra crossing in front of the vehicle.

[0091] It can be understood that if the absolute value of the zebra crossing corresponding vector angle crosswalks_angle is less than 90 degrees, the zebra crossing is considered to be in front of the vehicle; if the absolute value of the zebra crossing corresponding vector angle angle is greater than or equal to 90 degrees, the zebra crossing is considered to be behind the vehicle.

[0092] In order to improve the judgment efficiency, this embodiment uses the minimum distance (ego_to_crosswalks_dist) among the projection distances corresponding to the target zebra crossing in front of the vehicle as the target distance between the intersection and the vehicle.

[0093] If a frame contains both traffic light information and zebra crossing information, the minimum distance in the projected distance corresponding to the target zebra crossing can be used as the target distance between the intersection and the vehicle, or the minimum distance in the Euclidean distance corresponding to the target traffic light can be used as the target distance between the intersection and the vehicle, or the smaller distance between the minimum distance in the projected distance and the minimum distance in the Euclidean distance can be used as the target distance.

[0094] Step C3: When the target distance is less than a second distance threshold, it is determined that an intersection scene exists in the historical scene segment.

[0095] It can be understood that if the minimum distance in the projection distance corresponding to the target zebra crossing or the minimum distance in the Euclidean distance corresponding to the target traffic light is less than the second distance threshold (valid_dist_threshold, for example, 1 meter or 2 meters, etc.), the vehicle is considered to be close enough to the intersection and the intersection is a valid intersection.

[0096] That is, if ego_to_traffic_dist is less than valid_dist_threshold, or ego_to_crosswalks_dist is less than valid_dist_threshold, then it is considered that there is a valid intersection in the frame, otherwise there is no valid intersection in the frame.

[0097] If no valid intersection exists in all frames of the historical scene segment, no further vehicle turning scene recognition is performed on the historical scene segment.

[0098] This embodiment uses the above method to accurately identify whether there is a valid intersection in the historical scene segment, thereby improving the accuracy of obtaining data about vehicle turning scenes.

[0099] This application also provides a vehicle turning scene recognition device, please refer to Figure 5 , the vehicle turning scene recognition device includes: An acquisition module 10 is used to acquire historical scene fragments; The recognition module 20 is used to determine a scene recognition result based on each frame data in the historical scene segment, wherein the scene recognition result includes the presence of a vehicle left turn scene, a vehicle right turn scene, and / or the absence of a vehicle turning scene in the historical scene segment.

[0100] Optionally, the identification module 20 includes: A first judgment submodule is configured to judge whether there is a possibility of a vehicle turning in each frame between a target frame and a preset frame, wherein the target frame includes a start frame and the preset frame is separated from the target frame by a preset frame length; a determination submodule for determining, if any, a change in the vehicle's driving direction from the target frame to the scene sub-segment corresponding to the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively; The recognition submodule is used to determine a scene recognition result based on changes in the vehicle's driving direction.

[0101] Optionally, the determining submodule includes: A first calculation unit is used to calculate the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame respectively; A determining unit is configured to determine, based on the angle difference, a change in the vehicle's driving direction in a scene sub-segment corresponding to a target frame and a preset frame.

[0102] Optionally, after the step of determining a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the angle difference, the determining submodule further includes: The return unit is used to update the target frame to a frame that slides backward a preset number of frames, return to the step of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame, until the preset frame is the end frame, and obtain the change in the vehicle's driving direction in multiple scene sub-segments corresponding to the start frame to the end frame.

[0103] Optionally, the identification submodule includes: an identification unit, configured to determine scene recognition results of the plurality of scene sub-segments based on changes in the vehicle driving directions in the plurality of scene sub-segments corresponding to the plurality of scene sub-segments from the start frame to the end frame; Wherein, after the step of determining the scene recognition result based on the change in the vehicle's driving direction, the method further includes: The first determining module is configured to determine a vehicle turning moment based on a target scene sub-segment in which a vehicle turning scene exists.

[0104] Optionally, the determining unit includes: a normalization subunit, configured to perform normalization processing on the angle difference; a first determining subunit, configured to determine that the change in the vehicle's driving direction is a left turn if the normalized angle difference is within a first angle interval; a second determining subunit, configured to determine that the change in the vehicle's driving direction is a right turn if the normalized angle difference is within a second angle interval; a third determining subunit, configured to determine that the change in the vehicle's driving direction is no turn if the normalized angle difference is within a third angle range; There is no intersection among the first angle interval, the second angle interval and the third angle interval.

[0105] Optionally, the first judgment submodule includes: A first judging unit is configured to judge whether the vehicle speed corresponding to the target frame is less than a preset speed threshold; a second calculation unit, configured to calculate a vehicle travel distance from a target frame to a preset frame based on vehicle position information corresponding to each frame if the vehicle speed is greater than or equal to a preset speed threshold; a second judging unit, configured to judge whether the vehicle travel distance is less than a first distance threshold; The third calculation unit is configured to determine the possibility that the vehicle corresponding to each frame between the target frame and the preset frame has turned if the vehicle travel distance is greater than or equal to a first distance threshold.

[0106] Optionally, after the step of obtaining the historical scene fragments, the method further includes: A judgment module, used to judge whether there is intersection identification information in each frame of data; a calculation module, configured to calculate a target distance between the intersection and the vehicle based on the intersection identification information, if any; The second determining module is configured to determine that an intersection scene exists in the historical scene segment when the target distance is less than a second distance threshold.

[0107] Optionally, the intersection identification information includes traffic light information, the traffic light information includes first position information of at least one traffic light, and the calculation module includes: a first calculation submodule, configured to calculate the Euclidean distance between the at least one traffic light and the vehicle, and the first relative direction between the at least one traffic light and the vehicle, based on the first position information and the vehicle position information corresponding to each frame of data; a second judgment submodule, configured to judge whether there is a target traffic light in front of the vehicle based on the angle between the first relative direction and the target direction of the vehicle speed corresponding to each frame of data; The first setting submodule is configured to use, if any, the minimum distance among the Euclidean distances corresponding to the target traffic light as the target distance between the intersection and the vehicle.

[0108] Optionally, the intersection identification information further includes zebra crossing information, and the zebra crossing information includes second position information corresponding to any corner point of a rectangular box corresponding to at least one zebra crossing, and the calculation module includes: a second calculation submodule, configured to calculate a projected distance between the vehicle and the at least one zebra crossing, and a second relative direction between the at least one zebra crossing and the vehicle based on the second position information and the vehicle position information corresponding to each frame of data; a third judgment submodule, configured to judge whether there is a target zebra crossing in front of the vehicle based on the angle between the second relative direction and the target direction of the vehicle speed corresponding to each frame of data; The second setting submodule is used to use the minimum distance among the projection distances corresponding to the target zebra crossing as the target distance between the intersection and the vehicle, if it exists.

[0109] The vehicle turning scene recognition device provided in this application, utilizing the vehicle turning scene recognition method described in the aforementioned embodiments, can address the technical issue of the difficulty in acquiring data related to vehicle turning scenes. Compared to the prior art, the beneficial effects of the vehicle turning scene recognition device provided in this application are the same as those of the vehicle turning scene recognition method described in the aforementioned embodiments. Other technical features of the vehicle turning scene recognition device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0110] The present application provides a vehicle turning scene recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed 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 vehicle turning scene recognition method in the above-mentioned embodiment one.

[0111] Reference below Figure 6 , which shows a schematic diagram of the structure of a vehicle turning scene recognition device suitable for implementing embodiments of the present application. The vehicle turning scene recognition device in embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, tablet computers, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The vehicle turning scene recognition device shown is merely an example and should not impose any limitations on the functions and usage range of the embodiments of the present application.

[0112] like Figure 6As shown, the vehicle turning scene recognition device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory (ROM) 1002 or programs loaded from a storage device 1003 into a random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the vehicle turning scene recognition device. Processing device 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007, such as a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008, such as a liquid crystal display (LCD), speaker, vibrator, etc.; storage device 1003, such as a magnetic tape or hard disk; and communication device 1009. The communication device 1009 can allow the vehicle turning scene recognition device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a vehicle turning scene recognition device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0113] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0114] The vehicle turning scene recognition device provided in this application, utilizing the vehicle turning scene recognition method described in the aforementioned embodiment, can resolve the technical issue of the difficulty in acquiring data related to vehicle turning scenes. Compared to the prior art, the beneficial effects of the vehicle turning scene recognition device provided in this application are the same as those of the vehicle turning scene recognition method described in the aforementioned embodiment. Other technical features of the vehicle turning scene recognition device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0115] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0116] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any variation or substitution that a person skilled in the art can easily conceive within the technical scope disclosed in the present application should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0117] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the vehicle turning scene recognition method in the above-mentioned embodiment.

[0118] The computer-readable storage medium provided herein may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including, but not limited to, wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0119] The computer-readable storage medium may be included in the vehicle turning scene recognition device; or may exist independently without being assembled into the vehicle turning scene recognition device.

[0120] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the vehicle turning scene recognition device, the vehicle turning scene recognition device executes the vehicle turning scene recognition method.

[0121] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0122] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to the various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the prescribed logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the prescribed function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0123] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0124] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned vehicle turning scene recognition method. This computer-readable storage medium can address the technical issue of the difficulty in acquiring data related to vehicle turning scenes. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are similar to those of the vehicle turning scene recognition method provided in the aforementioned embodiments, and are not further elaborated here.

[0125] The present application also provides a computer program product, comprising a computer program, which implements the steps of the above-mentioned vehicle turning scene recognition method when executed by a processor.

[0126] The computer program product provided in this application can solve the technical problem of the high difficulty in obtaining data related to vehicle turning scenarios. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the vehicle turning scenario recognition method provided in the above-mentioned embodiment, and will not be elaborated here.

[0127] The above description is only part of the embodiments of the present application and does not limit the patent range of the present application. All equivalent structural transformations made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection range of the present application.

Claims

1. A vehicle turning scene recognition method, characterized in that: The vehicle turning scene recognition method comprises: Get historical scene clips; When there is a road intersection scene in the historical scene segment, determining a scene recognition result based on each frame data in the historical scene segment; The scene recognition result includes whether there is a vehicle left turn scene, a vehicle right turn scene, and / or no vehicle turning scene in the historical scene segment.

2. The vehicle turning scene recognition method according to claim 1, characterized in that: The step of determining the scene recognition result based on each frame data in the historical scene fragment includes: Determine whether there is a possibility of a vehicle turning in each frame between a target frame and a preset frame, wherein the target frame includes a start frame and the preset frame is separated from the target frame by a preset frame length; If so, determining a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively; A scene recognition result is determined based on the change in the vehicle's driving direction.

3. The vehicle turning scene recognition method according to claim 2, characterized in that: The step of determining a change in the vehicle driving direction in the scene sub-segment corresponding to the target frame and the preset frame based on the vehicle heading angles corresponding to the target frame and the preset frame respectively includes: Calculating the angle difference between the vehicle heading angles corresponding to the target frame and the preset frame respectively; Based on the angle difference, a change in the vehicle's driving direction in the scene sub-segment corresponding to the target frame and the preset frame is determined.

4. The vehicle turning scene recognition method according to claim 3, characterized in that: After the step of determining a change in the vehicle's driving direction in a scene sub-segment corresponding to a target frame to a preset frame based on the angle difference, the method further includes: The target frame is updated to be a frame that slides backward by a preset number of frames, and the process returns to the step of determining whether there is a possibility of turning of the vehicle corresponding to each frame between the target frame and the preset frame until the preset frame is the end frame, thereby obtaining the change in the vehicle driving direction in the multiple scene sub-segments corresponding to the start frame and the end frame.

5. The vehicle turning scene recognition method according to claim 4, characterized in that: The step of determining the scene recognition result based on the change in the vehicle's driving direction includes: determining scene recognition results for the multiple scene sub-segments based on changes in the vehicle driving directions in the multiple scene sub-segments corresponding to the start frame and the end frame; After the step of determining the scene recognition result based on the change in the vehicle's driving direction, the method further includes: The vehicle turning moment is determined based on the target scene sub-segment where the vehicle turning scene exists.

6. The vehicle turning scene recognition method according to claim 3, characterized in that: The step of determining a change in the vehicle driving direction in a scene sub-segment corresponding to a target frame and a preset frame based on the angle difference includes: performing normalization processing on the angle difference; If the normalized angle difference is within the first angle interval, determining that the change in the vehicle's driving direction is a left turn; If the normalized angle difference is within the second angle interval, determining that the change in the vehicle's driving direction is a right turn; If the normalized angle difference is within the third angle range, determining that the change in the vehicle's driving direction is no turn; There is no intersection among the first angle interval, the second angle interval and the third angle interval.

7. The vehicle turning scene recognition method according to claim 2, characterized in that: The step of determining whether there is a possibility of the vehicle turning in each frame between the target frame and the preset frame includes: Determining whether the vehicle speed corresponding to the target frame is less than a preset speed threshold; If the vehicle speed is greater than or equal to a preset speed threshold, then calculating the vehicle travel distance from the target frame to the preset frame based on the vehicle position information corresponding to each frame; Determining whether the vehicle travel distance is less than a first distance threshold; If the vehicle travel distance is greater than or equal to a first distance threshold, it is determined that there is a possibility that the vehicle corresponding to each frame between the target frame and the preset frame has turned.

8. The vehicle turning scene recognition method according to claim 1, characterized in that: After the step of obtaining the historical scene fragments, the method further includes: Determine whether there is intersection identification information in each frame of data; If so, the target distance between the intersection and the vehicle is calculated based on the intersection identification information; When the target distance is less than a second distance threshold, it is determined that an intersection scene exists in the historical scene segment.

9. The vehicle turning scene recognition method according to claim 8, characterized in that: The intersection identification information includes traffic light information, and the traffic light information includes first position information of at least one traffic light. The step of calculating the target distance between the intersection and the vehicle based on the intersection identification information includes: Calculating the Euclidean distance between the at least one traffic light and the vehicle, and the first relative direction between the at least one traffic light and the vehicle based on the first position information and the vehicle position information corresponding to each frame of data; determining whether there is a target traffic light in front of the vehicle based on the angle between the first relative direction and the target direction of the vehicle speed corresponding to each frame of data; If so, the minimum distance among the Euclidean distances corresponding to the target traffic light is used as the target distance between the intersection and the vehicle.

10. The vehicle turning scene recognition method according to claim 8, characterized in that: The intersection identification information further includes zebra crossing information, the zebra crossing information including second position information corresponding to any corner point of a rectangular box corresponding to at least one zebra crossing, and the step of calculating the target distance between the intersection and the vehicle based on the intersection identification information includes: Calculating a projected distance between the vehicle and the at least one zebra crossing, and a second relative direction between the at least one zebra crossing and the vehicle based on the second position information and the vehicle position information corresponding to each frame of data; determining whether there is a target zebra crossing in front of the vehicle based on the angle between the second relative direction and the target direction of the vehicle speed corresponding to each frame of data; If it exists, the minimum distance among the projection distances corresponding to the target zebra crossing is used as the target distance between the intersection and the vehicle.

11. A vehicle turning scene recognition device, characterized in that: The vehicle turning scene recognition device comprises: Acquisition module, used to obtain historical scene fragments; An identification module is used to determine a scene recognition result based on each frame data in the historical scene segment when there is an intersection scene in the historical scene segment, wherein the scene recognition result includes the presence of a vehicle left turn scene, a vehicle right turn scene, and / or the absence of a vehicle turning scene in the historical scene segment.

12. A vehicle turning scene recognition device, characterized in that: The vehicle turning scene recognition device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the vehicle turning scene recognition method according to any one of claims 1 to 10.

13. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the vehicle turning scene recognition method according to any one of claims 1 to 10 are implemented.

14. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the vehicle turning scene recognition method according to any one of claims 1 to 10 are implemented.

Images