Parking space limiter detection method and device, electronic equipment and storage medium

Abstract

The application discloses a kind of detection methods, device, electronic equipment and storage medium to parking space limiter;The method comprises: obtaining the image to be used collected by camera equipment deployed at multiple positions of target vehicle at current time, and each image to be used is spliced into overhead image;Based on the semantic segmentation model trained in advance, the overhead image is processed to obtain a semantic segmentation image;Determine at least one target limiter corresponding to the target parking space in the semantic segmentation image Target connected domain;Based on target connected domain, the proximal line of parking space corresponding to target parking space and two parking space segmentation lines, determine two limit points corresponding to target parking space.Solve the problem of low detection accuracy in prior art by feature extraction based on traditional image processing technology or deep learning target detection method to detect limiter, improve the accuracy of detection of parking space limiter, improve the safety of automatic parking.

Description

Technical Field

[0001] This invention relates to the field of computer processing technology, and in particular to a method, apparatus, electronic device, and storage medium for detecting parking space limiters. Background Technology

[0002] With the development of intelligent driving technology, more and more vehicles are equipped with automatic parking systems, which can replace the driver to automatically complete the functions of parking space detection and parking. However, most current automatic parking systems still maintain a relatively high speed when the wheels approach the limiter, causing a violent impact between the wheels and the limiter, resulting in a poor experience for drivers and passengers. Therefore, limiter detection is of great significance for improving the driving experience and vehicle protection.

[0003] Current technologies for detecting limiters typically employ traditional image processing algorithms and deep learning-based object detection methods. Traditional image processing algorithms primarily detect limiters based on features such as image color, grayscale, and connected components. However, these methods are prone to false positives and false negatives in complex scenarios with significant lighting variations or a limited variety of limiters. Deep learning-based object detection methods detect only the maximum bounding rectangle of the limiter, which has a large positional error and includes significant background information, resulting in large detection errors and low accuracy. Summary of the Invention

[0004] This invention provides a method, device, electronic equipment, and storage medium for detecting parking space limiters, so as to improve the accuracy of detecting parking space limiters and achieve the technical effect of improving the safety of automatic parking.

[0005] According to one aspect of the present invention, a method for detecting a parking space limiter is provided, the method comprising:

[0006] The system acquires images to be used from cameras deployed at multiple locations on the target vehicle at the current moment, and stitches these images together to form a top-down view image; wherein the locations include the front end, rear end, left end, and right end of the vehicle.

[0007] The top-view image is processed based on a pre-trained semantic segmentation model to obtain a semantic segmentation image;

[0008] Determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image;

[0009] Based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, determine the two limit points corresponding to the target parking space.

[0010] According to another aspect of the present invention, a detection device for a parking space limiter is provided, the device comprising:

[0011] The top-view image determination module is used to acquire images to be used collected by camera devices deployed at multiple locations on the target vehicle at the current time, and to stitch the images to be used into a top-view image; wherein, the locations include the front end of the vehicle, the rear end of the vehicle, the left end of the vehicle, and the right end of the vehicle;

[0012] The semantic segmentation image determination module is used to process the top view image based on a pre-trained semantic segmentation model to obtain a semantic segmentation image;

[0013] The target connected component determination module is used to determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image;

[0014] The limit point determination module is used to determine two limit points corresponding to the target parking space based on the target connected region, the parking space near-end line corresponding to the target parking space, and two parking space dividing lines.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] At least one processor; and

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the detection method for parking space limiters according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the detection method for a parking space limiter as described in any embodiment of the present invention.

[0020] The technical solution of this invention involves acquiring images to be used from cameras deployed at multiple locations on a target vehicle at the current moment, and stitching these images together into a top-view image. The top-view image is then processed based on a pre-trained semantic segmentation model to obtain a semantically segmented image. The target connected component of at least one target limiter corresponding to the target parking space is determined within the semantically segmented image. Based on the target connected component, the parking space proximal line corresponding to the target parking space, and two parking space dividing lines, two limiter locations corresponding to the target parking space are determined. This solves the problem of low detection accuracy in existing technologies that detect limiters using feature extraction based on traditional image processing techniques or deep learning target detection. The solution achieves this by using a pre-trained semantic segmentation model to obtain images of the target vehicle at multiple locations on the target vehicle. The images captured by the four cameras of the target vehicle are stitched together to form a top-down view image, increasing the detection range and improving detection accuracy. Then, the top-down view image is processed by a semantic segmentation model to obtain a semantic segmentation image, which contains the classification category of each pixel. This allows for the determination of the target connected region of at least one target limiter corresponding to the target parking space in the semantic segmentation image, enabling the identification of the target limiter region. Based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the two limiter positions corresponding to the target parking space are accurately determined, improving the accuracy of parking limiter detection and achieving the technical effects of improving automatic parking safety and enhancing the user riding experience.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a method for detecting a parking space limiter according to Embodiment 1 of the present invention;

[0024] Figure 2 This is a schematic diagram of a target parking space provided according to Embodiment 1 of the present invention;

[0025] Figure 3 This is a schematic diagram of the target limiter provided according to Embodiment 1 of the present invention;

[0026] Figure 4This is a schematic diagram of a scenario for detecting a parking space limiter according to Embodiment 2 of the present invention;

[0027] Figure 5 This is a schematic diagram of a method for detecting a parking space limiter according to Embodiment 2 of the present invention;

[0028] Figure 6 This is a schematic diagram of a method for detecting a parking space limiter according to Embodiment 2 of the present invention;

[0029] Figure 7 This is a schematic diagram of a method for detecting a parking space limiter according to Embodiment 2 of the present invention;

[0030] Figure 8 This is a schematic diagram of a method for detecting a parking space limiter according to Embodiment 2 of the present invention;

[0031] Figure 9 This is a schematic diagram of the structure of a detection device for a parking space limiter according to Embodiment 3 of the present invention;

[0032] Figure 10 This is a schematic diagram of the structure of an electronic device that implements the detection method for parking space limiters according to an embodiment of the present invention. Detailed Implementation

[0033] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0035] Before introducing this technical solution, the application scenario can be explained first. For example, in the process of automatic vehicle parking, in order to accurately and effectively park the vehicle in the parking space, the system will automatically detect the parking spaces in the surrounding scene. Based on the detection of the target parking space, the detection of the limit device in the target parking space can be implemented using the technical solution provided in the embodiments of this invention.

[0036] Example 1

[0037] Figure 1 This is a flowchart of a method for detecting parking space limiters according to Embodiment 1 of the present invention. This embodiment is applicable to situations where parking space limiters are being detected. The method can be executed by a detection device for the parking space limiters, which can be implemented in hardware and / or software and can be configured in a computing device. Figure 1 As shown, the method includes:

[0038] S110. Acquire the images to be used collected by the camera devices deployed at multiple locations on the target vehicle at the current time, and stitch the images to be used into a top-view image.

[0039] The "current moment" refers to any point in the vehicle's parking process. The implementation method of the parking space limiter is the same at each moment, and any one of these moments will be used as the "current moment" for explanation. There can be multiple locations, such as the front, rear, left, and right ends of the vehicle. The camera device can be a fisheye camera. It should be noted that the camera device can be placed anywhere on the vehicle, but the images captured by multiple cameras must be able to be stitched together to form a 360-degree top-down view. For example, the camera device can be placed at the front, rear, left, and right of the target vehicle.

[0040] In this embodiment, during the automatic parking process of the target vehicle, four cameras deployed at the front, rear, left, and right of the target vehicle can capture images of the vehicle in all directions at the current moment, all of which are used as images to be used. To improve the success rate of limit switch detection, the images to be used can be stitched together to obtain a top-down view with a wider field of view.

[0041] To improve the synchronization, speed, and accuracy of image acquisition, when acquiring images to be used from the cameras on the target vehicle, acquisition commands can be sent to each camera simultaneously to obtain multiple images to be used at the current moment.

[0042] Optionally, acquiring images to be used from cameras deployed at multiple locations of the target vehicle at the current time includes: when the target parking space is detected to be within the field of view of the camera, sending a shooting command to each camera to enable each camera to execute the shooting command and obtain each image to be used.

[0043] The system includes multiple camera devices (e.g., four or more), each with different internal and external camera parameters. These parameters include the shooting angle, camera installation location, and wide-angle parameters. The camera installation location refers to the specific location where the camera device is installed. The wide-angle parameter refers to the field of view of the camera device.

[0044] Specifically, when it is determined that the target parking space is within the field of view of the camera equipment, a shooting command can be simultaneously sent to multiple camera equipment deployed on the target vehicle. After receiving the shooting command, the multiple camera equipment can simultaneously capture images, resulting in multiple images to be used.

[0045] For example, a fisheye camera (i.e., a camera device) is set up at the front, rear, left, and right of the target vehicle. During the parking process, the original image frames captured by the four fisheye cameras are used as images to be used. According to the camera calibration parameters, the four images to be used can be stitched together to form a top view as a top view image. The parking space limiter can be identified based on the top view image. Compared with parking space images collected by only a single camera, the images obtained by multiple cameras are richer in content, have a wider detection range, and improve the detection rate of parking space limiters.

[0046] S120. The top-view image is processed based on a pre-trained semantic segmentation model to obtain a semantic segmentation image.

[0047] Semantic segmentation models can be used to classify each pixel in an image. For example, the result of semantic segmentation is to cover different objects with different colors to represent different semantic categories.

[0048] In this embodiment, the top-view image can be used as the input of the semantic segmentation model. After processing by the semantic segmentation model, a semantic segmentation image is obtained. At this time, each pixel in the semantic segmentation image is assigned a corresponding classification result, such as a limit switch, arrow, lane line, parking space line, speed bump, etc.

[0049] It's important to note that before training the semantic segmentation model, to ensure high accuracy, it's crucial to acquire as many and varied training samples as possible. In practical applications, a large number of original images containing parking spaces and limit switches can be obtained. The limit switch pixels in these images can be labeled as location labels. These original images and location labels then form the training samples for the semantic segmentation model. Specifically, the original images can be used as input samples, and the corresponding location labels as output labels. The input sample and its corresponding output label can be used as a training sample. This results in multiple training samples, which are then used to train the semantic segmentation model.

[0050] In this embodiment, the semantic segmentation model can be trained as follows: For each training sample, the original image of the current training sample is input into the semantic segmentation model to obtain the actual segmented image; based on the limiter pixel values ​​in the actual segmented image and the point labels of the current training sample, a loss value is determined, and the model parameters in the semantic segmentation model are corrected based on the loss value; the convergence of the loss function in the semantic segmentation model is taken as the training objective to obtain the semantic segmentation model. The training objective refers to the model training aiming to achieve the convergence of the preset loss function.

[0051] In this embodiment, the original image of the current training sample can be input into the semantic segmentation model to obtain the actual segmented image corresponding to the current training sample. The pixel values ​​of the limiters in the currently output actual segmented image can be compared with the corresponding point labels to calculate the similarity error value, i.e., the loss value. The model parameters can then be adjusted based on the loss value. The training error of the loss function, i.e., the loss parameter, can be used as a condition to detect whether the loss function has reached convergence. For example, whether the training error is less than a preset error, whether the error trend is stable, or whether the current iteration number is equal to a preset number. If the convergence condition is met, such as the training error of the loss function being less than the preset error or the error trend being stable, it indicates that the semantic segmentation model to be trained has completed training, and iterative training can be stopped. If the convergence condition is not met, training samples can be further obtained to train the semantic segmentation model to be trained until the training error of the loss function is within a preset range. When the training error of the loss function converges, the semantic segmentation model to be trained can be used as the final trained semantic segmentation model.

[0052] S130. Determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image.

[0053] The target parking space refers to the parking space where the vehicle is to be parked. The target parking space can be a rectangle enclosed by four lines. These four lines can be a parking space dividing line to separate it from other parking spaces, a near-end line close to the parking space entrance, and a far-end line away from the parking space entrance. For example, a schematic diagram of the target parking space can be found here. Figure 2 Vehicles can park in the target parking space by passing the near-end line. There can be one or two wheel stops in the target parking space; the specific number depends on the target parking space. Wheel stops, also known as wheel chocks, anti-skid devices, etc., guide vehicles to park accurately and neatly, and prevent collisions caused by improper operation when reversing into a parking space. They are safety devices deployed within the parking space to protect the vehicle and its occupants.

[0054] In this embodiment, each pixel in the semantic segmentation image has a certain classification result. Each classification result may be the same or different. If they are the same, it means that they have the same semantic category. The classification results of pixels in the semantic segmentation image that are consistent with the category of the limiter and are adjacent to each other can be combined to identify at least one connected region. These connected regions can be used as target connected components.

[0055] It should be noted that when determining the target connected component of at least one target limiter corresponding to the target parking space based on the semantic segmentation image, there may be other parking spaces in the semantic segmentation image that are different from the target parking space. In this case, the connected components of the limiters in other parking spaces may be identified and used as the target connected component. In order to improve the accuracy of the target connected component determination, all identified connected components can be filtered to accurately determine the target connected component belonging to the target parking space.

[0056] In this embodiment, the method for determining the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image can be as follows: based on the classification result of each pixel in the semantic segmentation image, determine the limit pixel whose classification result is a preset result; based on the limit pixel, determine the candidate connected component of at least one candidate limiter; based on the first number of candidate connected components, the positional relationship between the candidate connected components and the target parking space, and the category and size of the candidate limiter, determine at least one target limiter and the corresponding target connected component.

[0057] The preset result is a parking space limiter. The limiter can be categorized as large, small, etc. A large limiter may have one limiter per parking space, while a small limiter may have two. The specific limiter category depends on both the size of the parking space and the size of the limiter. The limiter's dimensions can include length, width, and height, for example, 560*150*100mm. Positional relationship can refer to the positional relationship in a two-dimensional or three-dimensional coordinate system.

[0058] In this embodiment, by traversing the classification results of each pixel in the semantic segmentation image, the pixel corresponding to the classification result that matches the preset result of a limiter can be found and used as the limiter pixel. Further, the limiter pixels can be divided based on their adjacent distances. Multiple adjacent limiter pixels can form a connected region as a candidate connected domain. Each candidate connected domain corresponds to a candidate limiter. The candidate limiter may be located within the target parking space or in other parking spaces. The number of candidate connected domains can be counted as a first quantity. The category of each candidate limiter can also be identified based on the number of pixels contained in the candidate connected domain. For example, if there are two candidate connected domains A and B within a parking space, and the number of pixels contained in each of A and B is less than a threshold, then the limiter category for A and B is small. The positional relationship between the candidate connected domain and the target parking space can also be obtained based on the position of the center pixel of the candidate connected domain and the positions of the four boundary lines of the target parking space. Furthermore, the connected components located within the target parking space can be selected from the candidate limiters as the target limiter and the corresponding target connected components. A schematic diagram of the target limiter can be found here. Figure 3 .

[0059] For example, the semantic segmentation image can be traversed to filter out pixels containing the semantic elements of the limiter as limit pixels, and the limiter connected components (i.e., the connected components to be selected) can be extracted based on the limit pixels. The target connected components can be filtered out according to the number of connected components to be selected, the positional relationship of the connected components in the target parking space, the limiter type, the limiter shape, and other conditions, so as to determine the limit position based on the target connected components.

[0060] It should be noted that after determining the semantic segmentation image, the historical detection positions of the two limiting points detected at the current time step and the previous time step can be retrieved. Based on the historical detection positions, the region of interest (ROI) to be detected can be delineated from the semantic segmentation image. For example, the size of the ROI can be set as a rectangular area within which the positions of the two historically detected limiting points are moved forward 0.5 meters from the near end line of the parking space and backward 0.5 meters from the far end line of the parking space. Furthermore, target connected components can be detected within the ROI. If the two historical limiting points of the target parking space were not detected at the previous time step, the target parking space can be set as the ROI to be detected, and target connected components can be detected within the ROI, improving the efficiency of target connected component detection.

[0061] S140. Based on the target connected component, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, determine the two limit points corresponding to the target parking space.

[0062] In this embodiment, after determining the target connected region, the position of the limit switch within the target parking space can be determined. Then, the points on the limit switches are positioned on the parking space dividing line of the target parking space, such that the line connecting the limit switch points positioned on the two parking space dividing lines is perpendicular to the parking space dividing line. Simultaneously, the line connecting the limit switch points positioned on the two parking space dividing lines is moved a preset distance towards the near end line of the parking space. Accordingly, the limit switch points located on the parking space dividing line are obtained. This allows the vehicle to plan its parking path based on the limit switch points.

[0063] It should be noted that there are at least two ways to determine the two limit points corresponding to the target parking space based on the target connected component, the parking space near end line corresponding to the target parking space, and the two parking space dividing lines. The two methods will be explained below.

[0064] One implementation method is to determine two limit points corresponding to the target parking space based on the target connected domain, the parking space near-end line corresponding to the target parking space, and two parking space dividing lines. This includes determining two limit points corresponding to the target parking space based on the second number of center points of the target connected domain, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line.

[0065] In this embodiment, the center point of the target connected region can be determined based on the position information of the limiting pixels associated with the target connected region. Each target connected region has a corresponding center point, and the number of center points can be counted as a second number. Every two center points are connected to obtain the connecting line between them. Combined with the two parking space dividing lines and the near end line of the parking space, the two limiting points to be screened corresponding to the target parking space are located. For example, if there are three center points, there are three connecting lines. The perpendicularity (representing the degree of perpendicularity) between each connecting line and the parking space dividing line can be calculated, and the connecting line with the highest degree of perpendicularity can be screened out. The intersection points of this connecting line with the parking space dividing line can be used as limiting points. The position of the limiting points also needs to be corrected by moving the limiting points a preset distance closer to the near end line of the parking space so that the vehicle does not collide with the limiter when parking.

[0066] In this embodiment, based on the second number of center points of the target connected region, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line, two limit point positions corresponding to the target limiter are determined, including: determining the target center point from the center points based on the second number and the first angle between the connecting line and the same parking space dividing line; determining the straight line to be used that is perpendicular to the parking space dividing line and passes through the target center point, and determining the intersection point to be corrected between the straight line to be used and the two parking space dividing lines; moving the position information of the intersection point to be corrected towards the parking space near-end line by a preset distance threshold to obtain two limit point positions.

[0067] It should be noted that when determining the first angle between each connecting line and the parking space dividing line, to ensure that the angles are in the same dimension, each connecting line only needs to intersect with the same parking space dividing line to determine the first angle. The preset distance threshold can be determined by technicians based on actual working conditions, and is not limited here.

[0068] Specifically, the line connecting any two center points intersects the same parking space dividing line, and the included angle is calculated as the first included angle. The two center points corresponding to the first included angle closest to a vertical angle (90 degrees) can be extracted, and the center point closer to the near end line of the parking space can be designated as the target center point. A straight line perpendicular to the two parking space dividing lines and passing through the target center point can be drawn and recorded as the line to be used. Further, the intersection of the line to be used with the two parking space dividing lines can be designated as the intersection point to be corrected. The position information of the intersection point to be corrected can be moved a preset distance threshold towards the near end line of the parking space, and the moved point can be designated as the limit point. At this point, the two limit points are located on the two parking space dividing lines respectively.

[0069] For example, based on the number of center points of the target connected region and the angle between the lines containing each pair of center points and the parking space dividing line, the final center point, i.e., the target center point, is calculated. A straight line passing through the target center point and perpendicular to the parking space dividing line is calculated; this is the straight line to be used. The two intersection points of this straight line and the parking space dividing line are calculated; these are the intersection points to be corrected. Moving these two intersection points along their respective parking space dividing lines towards the near end of the parking space by a preset distance threshold yields points E and F, which serve as the detection results of the limit switch within the target parking space at the current moment, i.e., the limit point location.

[0070] Another approach is to: determine the target distance between each pixel to be screened in the target connected region and the near end line of the parking space; determine the target pixel from each pixel to be screened based on each target distance and a preset ratio; and determine the two limit points corresponding to the target parking space based on the target pixel and the two parking space dividing lines.

[0071] The preset proportion can be expressed as a decimal, fraction, or percentage, for example, 10%.

[0072] In this embodiment, based on the position information of each pixel to be screened located in the target connected region, the vertical distance from each pixel to be screened to the near end line of the parking space can be calculated as the target distance. The larger the target distance value, the farther the pixel to be screened is from the near end line of the parking space; the smaller the target distance value, the closer the pixel to be screened is to the near end line of the parking space. The number of pixels to be screened can be calculated and multiplied by a preset ratio to obtain a product value. Based on the target distance, a number of pixels with smaller distances that match the product value can be selected from all the pixels to be screened as target pixels. For example, assuming the preset ratio is 10%, all target distances can be sorted from smallest to largest, and the pixels with smaller target distances at the top can be selected as target pixels. The number of target pixels accounts for 10% of the total number of pixels to be screened. Furthermore, based on the target pixels and the two parking space dividing lines, two limit points corresponding to the target parking space can be determined.

[0073] It should be noted that there are at least two ways to determine the two limit points corresponding to the target parking space based on the target pixel and the two parking space dividing lines. The two methods will be explained below.

[0074] One implementation method is to perform linear fitting on all target pixels to obtain a first fitted line, and determine the positions of two limit points based on the first fitted line and the two parking space dividing lines.

[0075] Specifically, an algorithm can be used to fit a straight line to all target pixels based on the position information of each target pixel, forming a first fitted line. The intersection of this first fitted line with the dividing lines of the two parking spaces within the target parking space can then be calculated as the limiting point location.

[0076] For example, after selecting a set of pixels in the target connected region that are close to the near end line of the parking space by a preset proportion, a straight line can be fitted to the selected set of pixels, and points E and F can be calculated as limit points.

[0077] Another approach is to perform linear fitting on the target pixels in each target connected region to obtain a second fitted line corresponding to each target connected region, and determine the two limit points based on each second fitted line and the two parking space dividing lines.

[0078] In this embodiment, after determining all target pixels, each target pixel has a corresponding target connected region. All target pixels within each target connected region can be fitted in parallel or sequentially to obtain a fitted straight line corresponding to each target connected region, which serves as the second fitted straight line. The fitted straight line most perpendicular to the parking space dividing line can be selected from the second fitted straight lines. The intersection point between the selected fitted straight line and the two parking space dividing lines within the target parking space can be used as the limit point location. Specifically, the method for determining the two limit point locations based on the second fitted straight lines and the two parking space dividing lines can be as follows: based on the third number of second fitted straight lines and the second included angle between each second fitted straight line and the same parking space dividing line, a third fitted straight line is determined from the second fitted straight lines; the two limit point locations are determined based on the intersection point between the two parking space dividing lines and the third fitted straight line.

[0079] In this embodiment, the number of second fitted lines can be counted as a third number. Furthermore, the angle between each second fitted line and the dividing line of the same parking space can be calculated and used as a second included angle. The second fitted line corresponding to the second included angle closest to the vertical angle (90 degrees) can be extracted as a third fitted line. The intersection point between the third fitted line and the dividing lines of the two parking spaces within the target parking space can be used as the limiting point location.

[0080] For example, after filtering out a set of pixels in the target connected region that are close to the near end line of the parking space by a preset proportion, a straight line can be fitted to the set of pixels after filtering each target connected region. Based on the number of straight lines and the angle between the straight lines and the parking space dividing line, straight lines are selected, and points E and F are calculated as limit points.

[0081] It should be noted that during the detection of parking space limiters, factors such as camera obstruction may cause the limit points to be unidentified, or detection malfunctions may occur. To further improve detection accuracy and parking safety, the two limit points of the target parking space identified in the previous moment can be retrieved as the limit points for the current moment, ensuring continuous and effective acquisition of the limit points. Alternatively, the limit points identified in the current moment and the limit points identified in the previous moment can be fused to obtain the final limit points.

[0082] In this embodiment, after determining the two limit points corresponding to the target parking space based on the target connected component, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the method further includes: updating the historical limit points corresponding to the historical time based on the two limit points, so that parking is performed based on the historical limit points when no limit point corresponding to the target parking space is detected; determining a reference limit area based on the historical limit points, so that after obtaining the semantic segmentation image, a detection area map matching the reference limit area is determined from the semantic segmentation image, and the detection area map is used as a new semantic segmentation image to determine the target connected component of at least one target limiter corresponding to the target parking space from the new semantic segmentation image.

[0083] Among them, a historical moment refers to a moment before the current moment. The time value corresponding to each moment can be related to the image captured by the camera device per unit time. Each frame captured by the camera device can be considered as a moment.

[0084] In practical applications, the detection of parking space limiters during vehicle parking is real-time and continuous. After determining the current limiter location, the detection results from historical time points can be combined to determine the final limiter location. After determining the final limiter location, the historical limiter locations corresponding to previous time points can be updated based on this final location. This allows for the retrieval of historical limiter locations for parking route planning when no corresponding limiter location is detected later. Furthermore, after determining the historical limiter locations, a reference limiter region can be defined. After obtaining the semantic segmentation image, a detection region map matching the reference limiter region can be determined from the semantic segmentation image. This detection region map is then used as a new semantic segmentation image to determine the target connected component of at least one target limiter corresponding to the target parking space. For example, after determining the semantic segmentation image, the historical limit position of the target limiter detected at the previous time (i.e., the historical time) can be retrieved. Based on the historical limit position, the region of interest (ROI) to be detected can be delineated from the semantic segmentation image. For example, the size of the ROI can be larger than and surround the historical limit position. Furthermore, target connected components can be detected within the ROI. If the historical limit position of the target limiter was not detected at the previous time, the target parking space can be set as the ROI to be detected, and target connected components can be detected within the ROI, improving the efficiency of target connected component detection.

[0085] The technical solution of this embodiment acquires images to be used from cameras deployed at multiple locations on the target vehicle at the current moment, and stitches these images together into a top-view image. The top-view image is then processed based on a pre-trained semantic segmentation model to obtain a semantically segmented image. The target connected component of at least one target limiter corresponding to the target parking space is determined in the semantically segmented image. Based on the target connected component, the parking space proximal line corresponding to the target parking space, and two parking space dividing lines, the locations of two limiter points corresponding to the target parking space are determined. This solves the problem of low detection accuracy in existing technologies that detect limiters using feature extraction based on traditional image processing techniques or deep learning target detection. It achieves this by using a model based on... The images captured by the four cameras of the target vehicle are stitched together to form a top-down view image, increasing the detection range and improving detection accuracy. Then, the top-down view image is processed by a semantic segmentation model to obtain a semantic segmentation image, which contains the classification category of each pixel. This allows for the determination of the target connected region of at least one target limiter corresponding to the target parking space in the semantic segmentation image, enabling the identification of the target limiter region. Based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the two limiter positions corresponding to the target parking space are accurately determined, improving the accuracy of parking limiter detection and achieving the technical effects of improving automatic parking safety and enhancing the user riding experience.

[0086] Example 2

[0087] As an optional embodiment of the above embodiments, specific application scenario examples are provided to enable those skilled in the art to further understand the technical solutions of the embodiments of the present invention. Specifically, please refer to the following detailed content.

[0088] For example, see Figure 4 , Figure 4 This diagram illustrates a scenario for detecting parking space limiters. When the automatic parking function of the target vehicle is activated, the vehicle travels from left to right. The system automatically detects parking spaces in the surrounding environment. If the system has detected a target parking space, it begins detecting the target limiter within that space. Each parking space consists of a near-end line (AB line), a far-end line (DC line), and two dividing lines (AD line and BC line). Assume... Figure 4 The left parking space is the target parking space, and the coordinates of the four corner points A, B, C, and D of the target parking space are known. The purpose is to detect the coordinates of two points E and F (i.e., the limit points). The limit points can be used as reference points for the tire position of the target vehicle after parking is completed.

[0089] For example, see Figure 5 , Figure 5This can be represented as a schematic diagram of a method for detecting parking space limiters. The implementation of parking space limiter detection can be achieved by: collecting and labeling training sample data, and training a semantic segmentation model based on deep learning. During automatic parking, raw image frames (i.e., images to be used) from four fisheye cameras (front, rear, left, and right) of the target vehicle can be acquired. Based on camera calibration parameters, the four raw image frames are stitched together to form a top-view image. This top-view image is then input into a pre-trained deep learning-based semantic segmentation model to obtain a semantic segmentation image. If a limiter has been detected in a historical frame, a region of interest (ROI) is defined in the semantic segmentation image based on the historical detection results. If no limiter has been detected in a historical frame, the target parking space is set as the ROI. The ROI in the semantic segmentation image is traversed, and pixels containing the semantic elements of the limiter (i.e., limiter pixels) are selected. Connected components of the limiter (i.e., connectable components to be selected) are extracted based on these pixels. Target connected components are filtered based on the number of target connected components, their positional relationship within the target parking space, the type of limiter, and the shape of the limiter. Finally, the positions of points E and F (i.e., limiter locations) of the limiter are calculated based on the filtered target connected components. The methods for determining the limiter locations can be categorized into three types:

[0090] The first method is: see [link / reference] Figure 6 Based on the number of center points of the target connected component and the angle between the lines containing each pair of center points and the parking space dividing line, the final center point, i.e., the target center point, is calculated. A straight line passing through the target center point and perpendicular to the parking space dividing line is calculated; this is the line to be used. The two intersection points of this line with the parking space dividing line are calculated; these are the intersection points to be corrected. These two intersection points are moved along their respective parking space dividing lines towards the near end of the parking space by a preset distance threshold, yielding points E and F. These points serve as the detection results of the limit switch within the target parking space at the current moment, i.e., the limit position. Then, combining the limit position detected at historical moments with the limit position detected at the current moment, the final detection result is returned, and the historical frame detection results are updated.

[0091] The second method is: see [link / reference] Figure 7 The system filters out a set of pixels in the target connected region that are close to the near end line of the parking space by a preset proportion. It can fit a straight line to the filtered set of pixels, calculate points E and F as limit points, and then combine the limit points detected in the past time with the limit points detected in the current time to return the final detection result and update the detection results of the historical frames.

[0092] The third method is: see [link / reference] Figure 8The system filters out a set of pixels in the target connected region that are close to the near end line of the parking space by a preset proportion. It can fit a straight line to the set of pixels after filtering each target connected region. Based on the number of straight lines and the angle between the straight line and the parking space dividing line, it filters out the straight line, calculates points E and F as limit points, and then combines the limit points detected in the historical time with the limit points detected in the current time to return the final detection result and update the historical frame detection results.

[0093] The technical solution provided in this embodiment detects parking limiters by training a deep learning-based semantic segmentation model. Compared with traditional image processing techniques, it can detect limiters in complex scenes (such as low light levels or when limiters are not commonly found), exhibiting stronger robustness. Compared to deep learning-based object detection network models, the semantic segmentation model achieves pixel-level detection, resulting in higher detection accuracy and improved accuracy in limiter detection. Using a top-down view stitched together from multiple cameras as input to the semantic segmentation model allows for simultaneous detection of image content from multiple cameras, expanding the detection range and further improving the accuracy of limiter detection. Furthermore, by incorporating historical detection results, it stably outputs limiter detection results, enhancing parking safety.

[0094] The technical solution of this embodiment acquires images to be used from cameras deployed at multiple locations on the target vehicle at the current moment, and stitches these images together into a top-view image. The top-view image is then processed based on a pre-trained semantic segmentation model to obtain a semantically segmented image. The target connected component of at least one target limiter corresponding to the target parking space is determined in the semantically segmented image. Based on the target connected component, the parking space proximal line corresponding to the target parking space, and two parking space dividing lines, the locations of two limiter points corresponding to the target parking space are determined. This solves the problem of low detection accuracy in existing technologies that detect limiters using feature extraction based on traditional image processing techniques or deep learning target detection. It achieves this by using a model based on... The images captured by the four cameras of the target vehicle are stitched together to form a top-down view image, increasing the detection range and improving detection accuracy. Then, the top-down view image is processed by a semantic segmentation model to obtain a semantic segmentation image, which contains the classification category of each pixel. This allows for the determination of the target connected region of at least one target limiter corresponding to the target parking space in the semantic segmentation image, enabling the identification of the target limiter region. Based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the two limiter positions corresponding to the target parking space are accurately determined, improving the accuracy of parking limiter detection and achieving the technical effects of improving automatic parking safety and enhancing the user riding experience.

[0095] Example 3

[0096] Figure 9 This is a schematic diagram of a detection device for a parking space limiter according to Embodiment 3 of the present invention. Figure 9 As shown, the device includes: a top view image determination module 310, a semantic segmentation image determination module 320, a target connected component determination module 330, and a limit point determination module 340.

[0097] The top-view image determination module 310 is used to acquire images to be used collected by camera devices deployed at multiple locations of the target vehicle at the current time, and stitch the images to be used into a top-view image; wherein the locations include the front end, rear end, left end, and right end of the vehicle; the semantic segmentation image determination module 320 is used to process the top-view image based on a pre-trained semantic segmentation model to obtain a semantic segmentation image; the target connected component determination module 330 is used to determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image; the limiter location determination module 340 is used to determine two limiter locations corresponding to the target parking space based on the target connected component, the parking space proximal line corresponding to the target parking space, and two parking space dividing lines.

[0098] The technical solution of this embodiment acquires images to be used from cameras deployed at multiple locations on the target vehicle at the current moment, and stitches these images together into a top-view image. The top-view image is then processed based on a pre-trained semantic segmentation model to obtain a semantically segmented image. The target connected component of at least one target limiter corresponding to the target parking space is determined in the semantically segmented image. Based on the target connected component, the parking space proximal line corresponding to the target parking space, and two parking space dividing lines, the locations of two limiter points corresponding to the target parking space are determined. This solves the problem of low detection accuracy in existing technologies that detect limiters using feature extraction based on traditional image processing techniques or deep learning target detection. It achieves this by using a model based on... The images captured by the four cameras of the target vehicle are stitched together to form a top-down view image, increasing the detection range and improving detection accuracy. Then, the top-down view image is processed by a semantic segmentation model to obtain a semantic segmentation image, which contains the classification category of each pixel. This allows for the determination of the target connected region of at least one target limiter corresponding to the target parking space in the semantic segmentation image, enabling the identification of the target limiter region. Based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the two limiter positions corresponding to the target parking space are accurately determined, improving the accuracy of parking limiter detection and achieving the technical effects of improving automatic parking safety and enhancing the user riding experience.

[0099] Optionally, based on the above-described device, the top-view image determination module 310 may include an image determination unit to be used.

[0100] The image to be used determination unit is used to send a shooting command to each of the cameras when the camera's field of view is detected to contain a target parking space, so that each of the cameras executes the shooting command and obtains each image to be used;

[0101] The internal and external parameters of each of the camera devices are different, including the shooting angle, camera installation position, and wide-angle parameters.

[0102] Based on the above-mentioned device, optionally, the target connected component determination module 330 includes: a limit pixel point determination unit, a selected connected component determination unit, and a target connected component determination unit.

[0103] A limiting pixel determination unit is used to determine limiting pixels whose classification result is a preset result based on the classification result of each pixel in the semantic segmentation image; wherein, the preset result is a limiter;

[0104] The unit for determining the connected component to be selected is used to determine the connected component to be selected for at least one selectable limiter based on the limit pixel points.

[0105] The target connected component determination unit is used to determine the at least one target limiter and the corresponding target connected component based on the first number of the selected connected components, the positional relationship between the selected connected components and the target parking space, and the type and size of the selected limiter.

[0106] Optionally, based on the above-described device, the limiting point location determination module 340 includes: a limiting point location determination first unit or a limiting point location determination second unit. The limiting point location determination second unit includes a target distance determination subunit, a target pixel point determination subunit, and a limiting point location determination second subunit.

[0107] The first unit for determining limit points is used to determine two limit points corresponding to the target parking space based on the second number of center points of the target connected domain, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line.

[0108] The target distance determination subunit is used to determine the target distance between each pixel to be screened in the target connected region and the near end line of the parking space;

[0109] The target pixel determination subunit is used to determine the target pixel from the pixels to be screened based on the target distance and the preset ratio.

[0110] The second sub-unit for determining limit points is used to determine two limit points corresponding to the target parking space based on the target pixel and the two parking space dividing lines.

[0111] Based on the above-mentioned device, optionally, the first unit for determining the location of the limiting point includes a target center point determination subunit, a subunit for determining the intersection point to be corrected, and a first subunit for determining the location of the limiting point.

[0112] The target center point determination subunit is used to determine the target center point from the center points based on the second quantity and the first included angle between the connecting line and the dividing line of the same parking space;

[0113] The sub-unit for determining the intersection point to be corrected is used to determine the straight line to be used that is perpendicular to the parking space dividing line and passes through the target center point, and to determine the intersection point to be corrected between the straight line to be used and the two parking space dividing lines.

[0114] The first sub-unit for determining the limiting point position is used to move the position information of the intersection point to be corrected towards the near end line of the parking space by a preset distance threshold to obtain the two limiting point positions.

[0115] Based on the above-mentioned device, optionally, the second sub-unit for determining the limiting point position is specifically used to perform straight line fitting on all target pixels to obtain a first fitted straight line, and determine the two limiting point positions based on the first fitted straight line and the two parking space dividing lines; or, to perform straight line fitting on the target pixels in each of the target connected regions to obtain a second fitted straight line corresponding to each of the target connected regions, and determine the two limiting point positions based on each of the second fitted straight lines and the two parking space dividing lines.

[0116] Based on the above-mentioned device, optionally, the second sub-unit for determining the location of the limiting point includes a third fitting line determination sub-unit and a limiting point location determination sub-unit.

[0117] The third fitted line determination unit is used to determine the third fitted line from the second fitted lines based on the third number of the second fitted lines and the second included angle between each of the second fitted lines and the same parking space dividing line.

[0118] The limiting point location determination unit is used to determine the two limiting point locations based on the intersection of the two parking space dividing lines and the third fitted straight line.

[0119] Optionally, based on the above-mentioned device, the device may further include a point update module, which includes a point update unit and a reference limit area determination unit.

[0120] The point update unit is used to update the historical limit point corresponding to the historical time based on the two limit point points, so that parking can be performed based on the historical limit point when no limit point corresponding to the target limit device is detected.

[0121] The reference limiting region determination unit is used to determine a reference limiting region based on the historical limiting point location, so that after obtaining the semantic segmentation image, a detection region map matching the reference limiting region is determined from the semantic segmentation image, and the detection region map is used as a new semantic segmentation image to determine the target connected region of at least one target limiter corresponding to the target parking space from the new semantic segmentation image.

[0122] The detection device for parking space limiters provided in this embodiment of the invention can execute the detection method for parking space limiters provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0123] Example 4

[0124] Figure 10 This is a schematic diagram of the structure of an electronic device implementing the detection method for a parking space limiter according to an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0125] like Figure 10 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0126] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as a camera; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as a disk, optical disk, etc.; and communication unit 19, such as a network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0127] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the detection method for a parking space limiter.

[0128] In some embodiments, the method for detecting parking space limiters may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for detecting parking space limiters described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for detecting parking space limiters by any other suitable means (e.g., by means of firmware).

[0129] Various embodiments of the systems and techniques described above herein 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), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0130] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0131] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0132] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device 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 camera through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0133] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0134] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0135] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0136] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. 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 invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting parking space limiters, characterized in that, include: The system acquires images to be used from cameras deployed at multiple locations on the target vehicle at the current moment, and stitches these images together to form a top-down view image; wherein the locations include the front end, rear end, left end, and right end of the vehicle. The top-view image is processed based on a pre-trained semantic segmentation model to obtain a semantic segmentation image; Determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image; Based on the target connected component, the parking space near-end line corresponding to the target parking space, and two parking space dividing lines, determine two limiting points corresponding to the target parking space, including: Based on the second number of center points of the target connected region, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line, two limit point positions corresponding to the target parking space are determined. The center points of the target connected region are determined based on the position information of the limit pixel points associated with the target connected region; or, Determine the target distance between each pixel to be screened in the target connected component and the near end line of the parking space; Based on the target distance and preset ratio, target pixels are determined from the pixels to be screened; Based on the target pixel and the two parking space dividing lines, two limit points corresponding to the target parking space are determined.

2. The method according to claim 1, characterized in that, The acquisition of images to be used, collected by camera devices deployed at multiple locations on the target vehicle at the current moment, includes: When the camera detects that the target parking space is within the field of view of the camera device, a shooting command is sent to each of the camera devices so that each of the camera devices executes the shooting command and obtains the image to be used. The internal and external parameters of each of the camera devices are different, including the shooting angle, camera installation position, and wide-angle parameters.

3. The method according to claim 1, characterized in that, Determining the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image includes: Based on the classification result of each pixel in the semantic segmentation image, the classification result is determined as a limiting pixel of the preset result; wherein, the preset result is a limiting device; Based on the limiting pixels, at least one selectable connected component of the selectable limiter is determined; Based on the first number of the selected connected regions, the positional relationship between the selected connected regions and the target parking space, and the type and size of the selected limiters, the at least one target limiter and the corresponding target connected region are determined.

4. The method according to claim 1, characterized in that, The determination of two limiting points corresponding to the target parking space based on the second number of center points of the target connected region, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line includes: Based on the second quantity and the first angle between the connecting line and the dividing line of the same parking space, the target center point is determined from the center point; Determine the straight line to be used that is perpendicular to the parking space dividing line and passes through the target center point, and determine the intersection point to be corrected between the straight line to be used and the two parking space dividing lines; The position information of the intersection point to be corrected is moved by a preset distance threshold towards the near end line of the parking space to obtain the two limit point positions.

5. The method according to claim 1, characterized in that, The step of determining two limiting points corresponding to the target parking space based on the target pixel and the two parking space dividing lines includes: Perform a straight line fitting on all target pixels to obtain a first fitted straight line, and determine the positions of the two limiting points based on the first fitted straight line and the two parking space dividing lines; or, Line fitting is performed on the target pixels in each of the target connected regions to obtain a second fitted line corresponding to each of the target connected regions. Based on each of the second fitted lines and the two parking space dividing lines, the positions of the two limiting points are determined.

6. The method according to claim 5, characterized in that, The determination of the two limit point positions based on each of the second fitted straight lines and the two parking space dividing lines includes: Based on the third number of the second fitted lines and the second included angle between each of the second fitted lines and the same parking space dividing line, a third fitted line is determined from the second fitted lines; The locations of the two limiting points are determined based on the intersection of the two parking space dividing lines and the third fitted straight line.

7. The method according to claim 1, characterized in that, After determining the two limiting points corresponding to the target parking space based on the target connected region, the parking space near-end line corresponding to the target parking space, and the two parking space dividing lines, the method further includes: The historical limit points corresponding to the two limit points are updated based on the historical limit points, so that parking can be performed based on the historical limit points when no limit point corresponding to the target parking space is detected. Based on the historical limit point locations, a reference limit region is determined so that after obtaining the semantic segmentation image, a detection region map matching the reference limit region is determined from the semantic segmentation image, and the detection region map is used as a new semantic segmentation image to determine the target connected region of at least one target limiter corresponding to the target parking space from the new semantic segmentation image.

8. A detection device for parking space limiters, characterized in that, include: The top-view image determination module is used to acquire images to be used collected by camera devices deployed at multiple locations on the target vehicle at the current time, and to stitch the images to be used into a top-view image; wherein, the locations include the front end of the vehicle, the rear end of the vehicle, the left end of the vehicle, and the right end of the vehicle; The semantic segmentation image determination module is used to process the top view image based on a pre-trained semantic segmentation model to obtain a semantic segmentation image; The target connected component determination module is used to determine the target connected component of at least one target limiter corresponding to the target parking space in the semantic segmentation image; The limit point determination module is used to determine two limit points corresponding to the target parking space based on the target connected region, the parking space near end line corresponding to the target parking space, and two parking space dividing lines; The limiting point location determination module includes: The first unit for determining the limiting points is used to determine two limiting points corresponding to the target parking space based on the second number of center points of the target connected region, the line connecting every two center points, the two parking space dividing lines, and the parking space near-end line. The center point of the target connected region is determined based on the position information of the limiting pixel points associated with the target connected region. Alternatively, the limiting point location determines the second unit, including: The target distance determination subunit is used to determine the target distance between each pixel to be screened in the target connected region and the near end line of the parking space; The target pixel determination subunit is used to determine the target pixel from the pixels to be screened based on the target distance and the preset ratio. The second sub-unit for determining limit points is used to determine two limit points corresponding to the target parking space based on the target pixel and the two parking space dividing lines.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the detection method for the parking space limiter according to any one of claims 1-7.

Images