Vehicle special scene detection method and system and vehicle

Abstract

The invention provides a vehicle special scene detection method and device, equipment and a storage medium, and the method comprises the steps: obtaining N detection results after a current detection result under the condition that the first M detection results of a scene where a vehicle is located indicate that the vehicle is located in an indoor parking lot, and the current detection result indicates that the vehicle is not located in a parking lot, m and N are positive integers; and determining whether the vehicle is located in a special scene of the indoor parking lot according to N detection results after the current detection result. The detection accuracy of the vehicle operation domain can be improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle control, and in particular to a vehicle special scene detection method, system and vehicle. Background Art

[0002] With the continuous improvement of the level of automobile intelligence, advanced driver assistance technology has developed rapidly. Home-zone Parking Assist (HPA) can help users automatically cruise and park their vehicles from the current location to the parking space or from the parking space to the designated location according to the pre-memorized route through automated means.

[0003] In order to ensure driving safety and effectiveness, the memory parking function can only be activated when the vehicle is detected to be in a safe operating domain. Currently, the global positioning system (GPS) positioning combined with map data, or the use of sensors to detect specific targets in the parking lot, is mainly used to determine whether the current environment meets the operating domain of memory parking, that is, to determine whether the vehicle is in a parking lot environment.

[0004] However, both of the above two methods suffer from misidentification, resulting in low detection accuracy in the vehicle operation domain. Summary of the invention

[0005] Based on the above technical status, the present application provides a vehicle special scene detection method, system and vehicle, which can improve the detection accuracy in the vehicle operation domain.

[0006] In order to achieve the above technical objectives, this application specifically proposes the following technical solutions:

[0007] According to a first aspect of an embodiment of the present application, a vehicle special scene detection method is provided, comprising: when the previous M detection results of the scene in which the vehicle is located all indicate that the vehicle is located in an indoor parking lot, and the current detection result indicates that the vehicle is located in a non-parking lot, obtaining N detection results after the current detection result, wherein M and N are both positive integers; and determining whether the vehicle is located in a special scene of the indoor parking lot based on the N detection results after the current detection result.

[0008] In some implementations, the N detection results include multiple detection results, and determining whether the vehicle is located in the special scene of the indoor parking lot based on the N detection results after the current detection result includes: if, among the multiple detection results after the current detection result, the number of detection results indicating that the vehicle is located in the non-parking lot is greater than or equal to a preset number, then it is determined that the vehicle is located in the special scene of the indoor parking lot.

[0009] In some implementations, after determining that the vehicle is located in the special scene of the indoor parking lot, the method further includes: when the detection result after the multiple detection results indicates that the vehicle is located in the special scene of the indoor parking lot, obtaining the position change information of the vehicle in the special scene of the indoor parking lot; if the distance change corresponding to the position change information is greater than or equal to a preset distance, determining that the vehicle is outside the special scene of the indoor parking lot; if the distance change corresponding to the position change information is less than the preset distance, determining that the vehicle is located in the special scene of the indoor parking lot.

[0010] In some implementations, after determining that the vehicle is located in the special scenario of the indoor parking lot, the method further includes: in P detection results after the multiple detection results, if the number of detection results indicating that the vehicle is located in an outdoor parking lot is greater than a preset number, then it is determined that the vehicle is located in the outdoor parking lot, and P is a positive integer.

[0011] In some implementations, the current detection result is obtained by the following steps: acquiring a current frame image captured of the vehicle's surroundings; performing feature extraction on the current frame image to obtain image features corresponding to the current frame image; and determining the current detection result based on the image features.

[0012] In some implementations, the current frame image includes: various images of the vehicle surrounding environment collected from different angles at the moment of current frame image acquisition; wherein, feature extraction is performed on the current frame image to obtain image features corresponding to the current frame image; based on the image features, the current detection result is determined, including: extracting image features from each image corresponding to the moment of current frame image acquisition, and fusing the image features corresponding to each image to obtain fused image features; based on the fused image features, the current detection result is determined.

[0013] In some implementations, the feature extraction is performed on the current frame image to obtain image features corresponding to the current frame image; and the current detection result is determined based on the image features, including: inputting the current frame image into a scene detection model, and performing feature extraction on the current frame image through the scene detection model to obtain image features corresponding to the current frame image; and determining the current detection result based on the image features; wherein the scene detection model is trained based on each frame image sample collected from the vehicle's surrounding environment and the corresponding annotated scene categories.

[0014] In some implementations, after determining that the vehicle is located in a special scenario of the indoor parking lot, the method further includes: initiating a memory parking function when determining that the vehicle is located in a special scenario of the indoor parking lot.

[0015] According to the second aspect of an embodiment of the present application, a vehicle special scene detection system is provided, comprising: a control device, the control device comprising a memory and a processor, the memory storing a control program, and the control program, when executed by the processor, is used to implement the vehicle special scene detection method described in the first aspect.

[0016] According to a third aspect of an embodiment of the present application, a vehicle is provided, wherein the vehicle is equipped with the vehicle special scene detection system as described in the second aspect.

[0017] The embodiments of the present application provide a vehicle special scene detection method, system and vehicle, which, when it is determined that the vehicle is located in an indoor parking lot based on the previous M detection results of the scene in which the vehicle is located, and the current detection result indicates that the vehicle is located in a non-parking lot, obtains N detection results after the current detection result, and determines whether the vehicle is located in a special scene of the indoor parking lot based on the N detection results after the current detection result. Since the previous M detection results indicate that the vehicle has entered the indoor parking lot, after the vehicle enters the indoor parking lot, if the next detection result indicates that the vehicle is located in a non-parking lot, it means that the vehicle is likely to have entered a special scene of the indoor parking lot at this time, and by combining the next N detection results, it can be more accurately determined whether the vehicle is located in a special scene of the indoor parking lot. In this way, the special situation of the vehicle in the indoor parking lot can be effectively identified, thereby improving the detection accuracy of the vehicle operation domain. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0019] Figure 1 A flowchart of a vehicle special scene detection method provided in an embodiment of the present application;

[0020] Figure 2 A flowchart of a specific implementation method of vehicle special scene detection provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the framework of the scene detection model provided in the embodiment of the present application;

[0022] Figure 4 A flowchart of the training process of the scene detection model provided in the embodiment of the present application;

[0023] Figure 5 A flowchart of an example of a vehicle scene detection method provided in an embodiment of the present application;

[0024] Figure 6 A schematic diagram of the structure of a vehicle special scene detection device provided in an embodiment of the present application;

[0025] Figure 7 A schematic diagram of the structure of a vehicle special scene detection system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The technical solution proposed in the embodiment of this application is applicable to the memory parking scenario, and aims to significantly improve the accuracy and reliability of the memory parking process by improving the detection method of the vehicle operation domain. The technical solution described in the embodiment of this application can enhance the accuracy of the vehicle's environmental recognition when performing memory parking.

[0027] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0028] Before introducing this application solution, the relevant technologies are first introduced:

[0029] Memory parking technology uses ultrasonic sensors, surround-view cameras, controllers and other devices to automatically memorize the parking route and reproduce the path when needed to complete the automatic parking operation, thereby simplifying the driver's parking process and improving parking efficiency and safety.

[0030] As a form of low-speed autonomous driving, memory parking requires working within a specific operational design domain (ODD) to ensure its safety and feasibility under given conditions. ODD defines the environmental boundaries in which the system can operate safely, covering multiple factors such as geographic location, weather conditions, and traffic flow.

[0031] Generally speaking, the memory parking function is limited to outdoor parking lots and indoor parking lots. Therefore, accurately judging whether the vehicle is currently in a parking lot environment is a key link in the operation domain detection. Currently, the two common detection methods are:

[0032] Based on GPS positioning and map data: This method relies on accurate global positioning system (GPS) information and a detailed geographic information system (GIS) database. When a vehicle enters a known parking area, the system compares the real-time location coordinates with pre-stored data to confirm the vehicle's current location. However, in underground or other enclosed spaces, such as tunnels, due to the lack of sufficient satellite signal support, this method is prone to positioning errors and even misjudges all places where GPS signals are lost as underground parking lots.

[0033] Using sensors to detect specific targets: Another common method is to capture the surrounding environment features through sensing devices such as lidar or cameras, such as parking space markings, signboards, entrance and exit signs, and toll gates. Although this method can provide immediate and intuitive information feedback, in some special cases (such as channel conversion areas or ramps between multi-storey parking lots), it is easy to cause false detection due to the lack of typical parking lot features, affecting the user experience.

[0034] In summary, the relevant technologies have obvious defects in accurately and reliably identifying parking lot environments: on the one hand, the GPS-based method is subject to signal coverage issues; on the other hand, the sensor-based method is limited by the presence or absence of the target object and its visibility. Both are difficult to achieve ideal detection effects.

[0035] In view of this, the embodiments of the present application are dedicated to providing a vehicle special scene detection method, system and vehicle, which detects the special scene of the vehicle in the indoor parking lot, thereby determining whether the vehicle is located in the indoor parking lot, so as to improve the detection accuracy of the vehicle operation domain. The following embodiments are described in detail one by one.

[0036] Exemplary Methods

[0037] Figure 1 A flow chart of a vehicle special scene detection method provided in an embodiment of the present application. Figure 1 As shown, the vehicle special scene detection method provided in this embodiment includes the following steps S101 and S102:

[0038] S101. When the first M detection results of the scene where the vehicle is located all indicate that the vehicle is located in an indoor parking lot, and the current detection result indicates that the vehicle is located in a non-parking lot, obtain N detection results after the current detection result.

[0039] In this step, M and N are both positive integers. The previous M detection results, the current detection result, and the N detection results after the current detection result of the scene where the vehicle is located can be determined by respectively collecting the previous M frames of images, the current frame image, and the N frames of images after the current frame image from the vehicle's surrounding environment. The specific implementation method is as follows:

[0040] The current detection result can be obtained by the following steps: obtaining a current frame image collected from the vehicle's surrounding environment; performing feature extraction on the current frame image to obtain image features corresponding to the current frame image; and determining the current detection result according to the image features corresponding to the current frame image. The current detection result includes indoor parking lots, outdoor parking lots, or non-parking lots.

[0041] The previous M detection results before the current detection result can be obtained by the following steps: obtaining the previous M frames of images collected from the vehicle surrounding environment before the current frame image; performing feature extraction on each of the previous M frames of images to obtain image features corresponding to each of the previous M frames of images; and determining the previous M detection results according to the image features corresponding to each of the previous M frames of images. Each detection result of the previous M detection results includes an indoor parking lot, an outdoor parking lot, or a non-parking lot.

[0042] The N detection results after the current detection result can be obtained by the following steps: obtaining N frames of images collected after the current frame of image of the surrounding environment of the vehicle; performing feature extraction on the N frames of image to obtain image features corresponding to each of the N frames of image; and determining the N detection results according to the image features corresponding to each of the N frames of image. Each detection result of the N detection results includes an indoor parking lot, an outdoor parking lot, or a non-parking lot.

[0043] This embodiment can realize real-time detection of the scene where the vehicle is located. If the detection result corresponding to the M-th frame image obtained at the current shooting moment indicates that the vehicle is located in an indoor parking lot, the detection results of the M-1 consecutive frame images before the M-th frame image are obtained, and it is determined whether the M-1 consecutive detection results all indicate that the scene where the vehicle is located is an indoor parking lot, and whether the position of the vehicle in the M-frame image has changed. If the M-1 consecutive detection results all indicate that the scene where the vehicle is located is an indoor parking lot, and the position of the vehicle in the M-frame image has changed, it is determined that the vehicle is currently in an indoor parking lot. That is, at this time, the conditions of the vehicle operation domain have been met, and the automatic parking function can be turned on.

[0044] In order to take both detection accuracy and efficiency into consideration, N can be any integer between 5 and 15, such as 9, 10 or 11.

[0045] Whether the position of the vehicle in the M-frame image changes can be determined by comparing the distance change between the position of the vehicle in the Mth frame image and the position of the vehicle in the first frame image in the continuous M-frame image.

[0046] In some embodiments, in order to facilitate dynamic recording and identification of the real-time status of the vehicle, a flag bit park_odd may be set for the indoor parking lot. The flag bit park_odd includes two states: True and False. True represents that the vehicle is located in the indoor parking lot. False represents that the vehicle is not located in the indoor parking lot. That is, when it is detected that the vehicle is located in the indoor parking lot, the flag bit park_odd is set to True; when it is detected that the vehicle is not located in the indoor parking lot, the flag bit park_odd is set to False.

[0047] S102: Determine whether the vehicle is located in a special scene of an indoor parking lot based on N detection results after the current detection result.

[0048] Through the first M detection results, it is possible to detect that the vehicle starts to enter the indoor parking lot and moves in it. If the current detection result indicates that the vehicle is not in the parking lot, it means that the vehicle may be currently in a special scene in the indoor parking lot where there is no GPS signal or no parking lot-specific features. In order to improve the recognition accuracy of the vehicle in the special scene of the indoor parking lot, it is necessary to combine the N detection results after the current detection result to further determine the scene where the vehicle is located.

[0049] Among them, special scenarios include areas in indoor parking lots where there is no GPS signal or where there are no parking lot-specific features. For example, in indoor parking lots, due to the building structure, the GPS signal may be shielded, resulting in the inability to detect that the vehicle is in the indoor parking lot. Or, channel conversion or cross-floor ramps in indoor parking lots. Channel conversion refers to the road between two interconnected and close underground parking lots. Cross-floor ramps refer to the connecting roads between parking lots on different floors, such as the ramp between the underground first floor parking lot and the underground second floor parking lot in a shopping mall.

[0050] In some embodiments, the N detection results include multiple detection results. Then step S102 specifically includes: if the number of detection results indicating that the vehicle is located in a non-parking lot in the multiple detection results after the current detection result is greater than or equal to a preset number, it is determined that the vehicle is located in a special scene of an indoor parking lot.

[0051] In order to take both detection accuracy and efficiency into consideration, N can be any integer between 5 and 15, such as 9, 10 or 11.

[0052] That is, if the current detection result indicates that the vehicle is currently located in a non-parking lot, and more than a preset number of the following multiple detection results also indicate that the vehicle is located in a non-parking lot, it can be determined that the vehicle is currently located in a special scene of an indoor parking lot, and the vehicle location corresponding to the current detection result is recorded as location 1. The preset number can be set to half of the number of multiple detection results.

[0053] In some embodiments, in order to facilitate dynamic recording and identification of the real-time status of the vehicle. A flag bit nopark_odd may also be set for special scenarios in indoor parking lots. The flag bit nopark_odd includes two states: True and False. True represents a special scenario in which the vehicle is located in an indoor parking lot. False represents a special scenario in which the vehicle is not located in an indoor parking lot. That is, when a special scenario in which the vehicle is located in an indoor parking lot is detected, the flag bit nopark_odd is set to True; when a special scenario in which the vehicle is not located in an indoor parking lot is detected, the flag bit nopark_odd is set to False.

[0054] Furthermore, when it is determined that the vehicle is located in a special scenario of an indoor parking lot, the memory parking function can be activated.

[0055] In this embodiment, when it is detected that the vehicle is located in a special scenario of an indoor parking lot, it is considered that the vehicle meets the conditions of the operating domain and the memory parking function can be turned on.

[0056] In some embodiments, Figure 2 As shown, after determining that the vehicle is located in a special scene of an indoor parking lot, the vehicle special scene detection method further includes the following steps S201, S202 and S203:

[0057] S201. When a detection result after multiple detection results indicates that the vehicle is located in a special scene of an indoor parking lot, obtain position change information of the vehicle in the special scene of the indoor parking lot.

[0058] Specifically, if the detection results after multiple detection results indicate that the vehicle is located in the special scene of the indoor parking lot, that is, the indoor parking lot flag bit park_odd and the special scene flag bit nopark_odd are both in the true state, the vehicle's location information location2 is continuously recorded, and the location change information between location2 and location1 is iteratively determined, and according to the location change information of the vehicle in the special scene of the indoor parking lot, it is determined whether the vehicle is still located in the special scene of the indoor parking lot. Specifically, it may include step S202 and step S203.

[0059] S202: If the distance change corresponding to the position change information is greater than or equal to the preset distance, it is determined that the vehicle is outside the special scene of the indoor parking lot.

[0060] S203: If the distance change corresponding to the position change information is less than a preset distance, it is determined that the vehicle is located in a special scenario of an indoor parking lot.

[0061] In some embodiments, the location change information can be determined by the distance change between location2 and location1. If the distance change is less than a preset distance, it means that the vehicle has been in the special scene of the indoor parking lot; if the distance change is greater than or equal to the preset distance, it means that the vehicle has left the special scene of the indoor parking lot, and the indoor parking lot flag bit park_odd and the special scene flag bit nopark_odd are both set to false.

[0062] The above-mentioned embodiment introduces a specific implementation method of extracting image features from the current frame image and determining the current detection result.

[0063] In some embodiments, the current frame image may include: images of the vehicle surrounding environment collected from different angles at the current frame image collection moment. Feature extraction is performed on the current frame image to obtain image features corresponding to the current frame image; and the current detection result is determined based on the image features, including: extracting image features from each image corresponding to the current frame image collection moment, and fusing the image features corresponding to each image to obtain fused image features; and determining the current detection result based on the fused image features.

[0064] Among them, different angles include four angles of the vehicle's front and rear views, namely, front, rear, left front and right front angles. That is to say, four cameras are set at the front, rear, left front and right front of the vehicle to collect four images of the front and rear views, such as the vehicle's front environment image, rear environment image, left front environment image and right front environment image, and the four images of the front and rear views are feature spliced ​​on the RGB channel to fuse the four images to obtain fused image features. Afterwards, it is convenient to detect the scene where the vehicle is currently located based on the fused image features to obtain the current detection result.

[0065] In some embodiments, the scene detection model can also be used to extract features of the current frame image to obtain image features corresponding to the current frame image; and the current detection result is determined based on the image features. The scene detection model can be trained based on each frame image sample collected from the vehicle's surrounding environment and the corresponding annotated scene category.

[0066] Figure 3 The schematic diagram of the framework of the scene detection model provided in the embodiment of the present application is as follows. Figure 3 As shown, the scene detection model includes a feature extraction network 31 and a classifier 32; wherein the feature extraction network 31 is used to extract features from the current frame image to obtain image features corresponding to the current frame image; the classifier 32 is used to determine the current detection result according to the image features. The feature extraction network may include FPN (Feature Pyramid Network), VGGNet, ResNet or InceptionNet, etc. The classifier 32 may be a classification head.

[0067] The following is an introduction to the training process of the scene detection model with the help of the attached figure:

[0068] Figure 4 The following is a flow chart of the training process of the scene detection model provided in the embodiment of the present application. Figure 4 As shown, the scene detection model can be trained through the following steps:

[0069] S401, acquiring each frame image sample collected from the vehicle surrounding environment, each of the frame image samples corresponding to a labeled scene category.

[0070] In some embodiments, each frame of image samples includes multiple image samples taken from different angles, such as a vehicle front environment image sample, a rear environment image sample, a left front environment image sample, and a right front environment image sample, and the multiple image samples correspond to the same annotated scene category.

[0071] In this step, each frame of image samples should cover various possible scenes, including but not limited to indoor parking lots, outdoor parking lots, special scenes of indoor parking lots, etc. Each frame of image samples corresponds to a labeled scene category, such as "indoor parking lot", "outdoor parking lot", "non-parking lot", etc. These labeled data are the basis for training the model to ensure that the model can learn the characteristics of different scenes.

[0072] In some embodiments, after acquiring the four images of the front perimeter, first determine whether the four images are aligned. If not, delete them. If aligned, determine whether the four images can determine their scene categories. If not, send them to a pending folder and add them to the corner case scene. Then align them and send them to the folder of the corresponding scene category. If yes, send them directly to the folder of the corresponding scene category. Afterwards, data acceptance can be performed on the images in the folders of each scene category. If they are unqualified, they are re-labeled. If they are qualified, they are used as image sample sets for training.

[0073] S402 , extracting features from each frame image sample using the scene detection model to be trained, and obtaining image feature samples corresponding to each frame image sample.

[0074] Specifically, each frame image sample is input into the scene detection model to be trained, and the feature extraction network 31 in the model extracts the image features of each frame image sample. The image features may include information such as color, texture, edge, shape, etc.

[0075] In some embodiments, S402 includes: extracting a plurality of image feature samples from a plurality of image samples respectively through the scene detection model to be trained, and fusing the plurality of image features to obtain a fused feature sample.

[0076] Specifically, multiple image samples can be input into the scene detection model to be trained, and the feature extraction network 31 in the model will extract the image features of each image sample. Afterwards, the image features extracted from the multiple image samples are fused to generate fused feature samples. Feature fusion can be achieved in a variety of ways, such as weighted averaging, maximum pooling, feature splicing, etc. The fused feature samples can more comprehensively reflect the information of the vehicle's surrounding environment.

[0077] S403 . Predict the category of the scene where the vehicle is located based on the image feature samples to obtain a predicted scene category.

[0078] The classifier 32 in the scene detection model to be trained will output a predicted scene category based on the image feature samples. For example, the model may predict that the vehicle is currently located in an "indoor parking lot", "outdoor parking lot" or "non-parking lot". The predicted scene category output by the model is a probability distribution, which indicates the probability of the vehicle being in each scene category.

[0079] In some embodiments, step S403 includes: predicting the category of the scene where the vehicle is located based on the fused feature samples to obtain a predicted scene category. For example, step S403 may predict the category of the scene where the vehicle is located by a classifier.

[0080] S404: Determine the scene category prediction loss based on the difference between the predicted scene category and the labeled scene label, and converge the scene detection model to be trained based on the scene category prediction loss to obtain a scene detection model.

[0081] By comparing the predicted scene category of the model with the actually labeled scene category, the difference between the two is calculated, which is the scene category prediction loss. The scene category prediction loss can adopt a loss function such as cross-entropy loss.

[0082] Based on the calculated scene category prediction loss, an optimization algorithm (such as gradient descent) is used to adjust the model parameters to minimize the loss. This process is called model convergence. Through multiple iterations of training, the model gradually learns to more accurately predict the scene category in which the vehicle is located.

[0083] Through the above steps, this embodiment can train a high-precision scene detection model for real-time identification of the scene category that the vehicle is in. This can not only improve the accuracy of scene recognition, but also enhance the robustness of the model in complex environments.

[0084] In some embodiments, after determining that the vehicle is located in a special scene of an indoor parking lot, the special scene detection method further includes: in the P detection results after the multiple detection results, if the number of detection results indicating that the vehicle is located in an outdoor parking lot is greater than a preset number, then determining that the vehicle is located in an outdoor parking lot. P is a positive integer. In order to take into account both detection accuracy and efficiency, P can be any integer between 10 and 20. For example, 14, 15 or 16.

[0085] The following takes model detection as an example to illustrate the vehicle special scene detection method of the present application. Figure 5 As shown, the example specifically includes:

[0086] S501, obtaining a detection result corresponding to the Mth frame image output by the model according to the Mth frame image of the vehicle surrounding environment collected, where M is a positive integer.

[0087] S502: If the detection result corresponding to the M-th frame image is an indoor parking lot, obtain the M-1 frame image before the M-th frame image, and determine whether the detection results of the consecutive M frame images all indicate that the vehicle is located in an indoor parking lot, and whether the vehicle position in the M frame images changes.

[0088] On the other hand, if the detection result corresponding to the Mth frame image is an outdoor parking lot or a non-parking lot, the outdoor parking lot or the non-parking lot is output as the actual detection result.

[0089] S503. If the detection results of the continuous M frame images all indicate that the vehicle is located in an indoor parking lot, and the vehicle position in the M frame images changes, it means that the vehicle has entered the indoor parking lot. The indoor parking lot flag park_odd is set to true, and the scene where the vehicle is located is determined based on the detection result corresponding to the M+1 frame image corresponding to the M+1 frame image.

[0090] On the other hand, if not all of the detection results of the continuous M frame images indicate that the vehicle is located in an indoor parking lot, and at least one of the following conditions is satisfied: the vehicle position in the M frame images has not changed, the detection result corresponding to the Mth frame image output by the model is taken as the actual detection result.

[0091] S504. If the detection result corresponding to the M+1th frame image indicates that the vehicle is located outside the parking lot, obtain the detection results corresponding to the next N frames of images output by the model, and determine the number of detection results under each scene category based on the detection results corresponding to the N frames of images, and determine the actual detection result based on the scene category with the largest number of detection results.

[0092] For example, N frames of images are 10 frames of images, and 6 of the 10 frames of images indicate that the vehicle is located in an outdoor parking lot or an indoor parking lot, then the outdoor parking lot or the indoor parking lot is taken as the actual detection result; and when 6 of the 10 frames of images indicate that the vehicle is located in a non-parking lot, then the 10 frames of images are all corrected to an indoor parking lot, and the indoor parking lot is output as the actual detection result of the 10 frames of images.

[0093] S505. If the scene category with the largest number of detection results is non-parking lot, the indoor parking lot is taken as the actual output result, and the special scene flag nopark_odd is set to true, and the current location of the vehicle is recorded as location_start.

[0094] On the other hand, if the scene category with the largest number of detection results is indoor parking lot or outdoor parking lot, indoor parking lot or outdoor parking lot is used as the actual output result.

[0095] S506: Make a judgment based on the output result of the next frame of the model.

[0096] Specifically, the detection result corresponding to the next frame image after N frames of images output by the model is obtained. If the detection result corresponding to the next frame image indicates that it is not a parking lot, it is determined whether it is necessary to rebuild the map.

[0097] On the other hand, if the detection result corresponding to the next frame image indicates that the vehicle is located in an outdoor parking lot, the detection results corresponding to the consecutive P frame images after the next frame image are obtained, and it is determined whether the consecutive P frame images are an outdoor parking lot. If so, the outdoor parking lot is output, and the indoor parking lot flag park_odd and the special scene flag nopark_odd are both set to false; otherwise, the detection result output by the model is used as the actual detection result.

[0098] On the other hand, if the detection result corresponding to the next frame image indicates that the vehicle is located in an indoor parking lot, it is determined whether the detection result corresponding to the K frame image after the next frame image is an indoor parking lot; if so, the indoor parking lot flag park_odd and the special scene flag nopark_odd are both set to false.

[0099] S507: If re-building the map is not required, the vehicle position in each frame after N frames is recorded as location_end, and it is determined whether the distance change between location_end and location_start is greater than or equal to a preset distance.

[0100] On the other hand, if the map needs to be re-built, the vehicle position after the map is re-built is recorded as location_start, and the process returns to step 6).

[0101] S508: If it is determined that the distance change between location_end and location_start is greater than or equal to the preset distance, it is determined that the vehicle has exited the indoor parking lot, and the indoor parking lot flag park_odd is updated to false, and the special scene flag nopark_odd is updated to false.

[0102] The embodiments of the present application provide a vehicle special scene detection method, system and vehicle, which determines that the vehicle is located in an indoor parking lot according to the previous M detection results of the scene where the vehicle is located, and the current detection result indicates that the vehicle is located in a non-parking lot, obtains N detection results after the current detection result, and determines whether the vehicle is located in a special scene of the indoor parking lot according to the N detection results after the current detection result. Since the previous M detection results indicate that the vehicle has entered the indoor parking lot, after the vehicle enters the indoor parking lot, if the next detection result indicates that the vehicle is located in a non-parking lot, it means that the vehicle is likely to have entered the special scene of the indoor parking lot at this time. By combining the next N detection results, it can be more accurately determined whether the vehicle is located in the special scene of the indoor parking lot. In this way, the special situation of the vehicle in the indoor parking lot can be effectively identified, thereby improving the detection accuracy of the vehicle operation domain. In addition, by using a visual sensor to obtain the front and peripheral four-way images of the vehicle, and fusing the semantic information of the four-way camera, multi-scale feature extraction can be performed, the entire scene information can be obtained, and the recognition accuracy of the scene category where the vehicle is located can be further improved.

[0103] Exemplary Devices

[0104] Corresponding to the above-mentioned vehicle special scene detection method, an embodiment of the present application also provides a vehicle special scene detection device. Figure 6 Schematic diagram of a vehicle special scene detection device provided in an embodiment of the present application. Figure 6As shown, the vehicle special scene detection device provided by the embodiment of the present application includes: an acquisition unit 61 and a determination unit 62; the acquisition unit 61 is used to obtain N detection results after the current detection result when the previous M detection results of the scene where the vehicle is located all indicate that the vehicle is located in an indoor parking lot, and when the current detection result indicates that the vehicle is located in a non-parking lot, the M and N are both positive integers; the determination unit 62 is used to determine whether the vehicle is located in the special scene of the indoor parking lot based on the N detection results after the current detection result.

[0105] In some embodiments, the N detection results include multiple detection results, and the determination unit 62 determines whether the vehicle is located in the special scene of the indoor parking lot based on the N detection results after the current detection result, including: if, among the multiple detection results after the current detection result, the number of detection results indicating that the vehicle is located in the non-parking lot is greater than or equal to a preset number, then it is determined that the vehicle is located in the special scene of the indoor parking lot.

[0106] In some embodiments, after the determination unit 62 determines that the vehicle is located in the special scene of the indoor parking lot, the acquisition unit 61 is used to obtain the position change information of the vehicle in the special scene of the indoor parking lot when the detection result after the multiple detection results indicates that the vehicle is located in the special scene of the indoor parking lot; the determination unit 62 is also used to determine that the vehicle is outside the special scene of the indoor parking lot when the distance change corresponding to the position change information is greater than or equal to a preset distance; and determine that the vehicle is located in the special scene of the indoor parking lot when the distance change corresponding to the position change information is less than a preset distance.

[0107] In some embodiments, after the determination unit 62 determines that the vehicle is located in the special scenario of the indoor parking lot, it is also used to: in the P detection results after the multiple detection results, if the number of detection results indicating that the vehicle is located in an outdoor parking lot is greater than a preset number, then it is determined that the vehicle is located in the outdoor parking lot, and P is a positive integer.

[0108] In some embodiments, the acquisition unit 61 acquires the current detection result, including: acquiring a current frame image captured of the vehicle's surrounding environment; performing feature extraction on the current frame image to obtain image features corresponding to the current frame image; and determining the current detection result based on the image features corresponding to the current frame image.

[0109] In some embodiments, the current frame image includes: various images of the vehicle surrounding environment collected from different angles at the moment of current frame image acquisition; wherein the acquisition unit 61 performs feature extraction on the current frame image to obtain image features corresponding to the current frame image; based on the image features, the current detection result is determined, including: extracting image features of each image corresponding to the moment of current frame image acquisition, and fusing the image features corresponding to each image to obtain fused image features; based on the fused image features, the current detection result is determined.

[0110] In some embodiments, the acquisition unit 61 performs feature extraction on the current frame image to obtain image features corresponding to the current frame image; and determines the current detection result based on the image features, including: inputting the current frame image into a scene detection model, and performing feature extraction on the current frame image through the scene detection model to obtain image features corresponding to the current frame image; and determining the current detection result based on the image features corresponding to the current frame image; wherein the scene detection model is trained based on each frame image sample collected from the vehicle's surrounding environment and the corresponding annotated scene categories.

[0111] In some embodiments, the device further includes: a control unit 63, configured to start a memory parking function when determining that the vehicle is located in a special scenario of the indoor parking lot.

[0112] The vehicle special scene detection device provided in this embodiment belongs to the same application concept as the vehicle special scene detection method provided in the above embodiments of this application, and can execute the vehicle special scene detection method provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the vehicle special scene detection method. For technical details not fully described in this embodiment, please refer to the specific processing content of the vehicle special scene detection method provided in the above embodiments of this application, and will not be repeated here.

[0113] The functions implemented by the above acquisition unit 61, determination unit 62 and control unit 63 can be implemented by the same or different processors respectively, and the embodiment of the present application is not limited thereto.

[0114] It should be understood that the units in the above devices can be implemented in the form of a processor calling software. For example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or realize the functions of each unit of the device, wherein the processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory in the device or a memory outside the device. Alternatively, the units in the device can be implemented in the form of hardware circuits, and the functions of some or all units can be realized by designing the hardware circuits. The hardware circuit can be understood as one or more processors; for example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are realized by designing the logical relationship of the components in the circuit; for another example, in another implementation, the hardware circuit can be implemented by PLD, taking FPGA as an example, which can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured by the configuration file, so as to realize the functions of some or all of the above units. All units of the above devices can be implemented in the form of a processor calling software, or in the form of hardware circuits, or in part by a processor calling software, and the remaining part is implemented in the form of hardware circuits.

[0115] In an embodiment of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor may be a circuit with the ability to read and run instructions, such as a CPU, a microprocessor, a GPU, or a DSP; in another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or reconfigurable, such as a hardware circuit implemented by an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the hardware circuit configuration can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, DPU, etc.

[0116] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0117] In addition, all or part of the units in the above device can be integrated together, or can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a SOC. The SOC may include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The type of the at least one processor may be different, for example, including a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0118] Exemplary Systems

[0119] The present application embodiment proposes a vehicle special scene detection system, see Figure 7 As shown, the vehicle special scene detection system includes: a detection device, the detection device includes a memory 200 and a processor 210;

[0120] The memory 200 is connected to the processor 210 and is used to store programs;

[0121] The processor 210 is used to implement the vehicle special scene detection method disclosed in any of the above embodiments by running the program stored in the memory 200.

[0122] Specifically, the above system may further include: a bus, a communication interface 220 , an input device 230 and an output device 240 .

[0123] The processor 210, the memory 200, the communication interface 220, the input device 230 and the output device 240 are connected to each other via a bus.

[0124] A bus may include a pathway that transfers information between components of a computer system.

[0125] The processor 210 may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the scheme of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0126] The processor 210 may include a main processor, and may also include a baseband chip, a modem, and the like.

[0127] The memory 200 stores a program for executing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include a program code, and the program code includes a computer operation instruction. More specifically, the memory 200 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk storage, a flash, and the like.

[0128] The input device 230 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor.

[0129] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0130] The communication interface 220 may include any transceiver or the like to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0131] The processor 210 executes the program stored in the memory 200 and calls other devices, which can be used to implement each step of any vehicle special scene detection method provided in the above embodiments of the present application.

[0132] An embodiment of the present application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the vehicle special scene detection method introduced in any of the above embodiments. The specific processing process and its beneficial effects can be found in the introduction to the embodiment of the above-mentioned vehicle special scene detection method.

[0133] An embodiment of the present application also proposes a vehicle, which is equipped with the above-mentioned vehicle special scene detection system and is used to execute the steps of the above-mentioned vehicle special scene detection method.

[0134] Exemplary computer program products and storage media

[0135] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the vehicle special scene detection method according to various embodiments of the present application described in any of the above embodiments of this specification.

[0136] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0137] In addition, an embodiment of the present application may also be a storage medium on which a computer program is stored. The computer program is executed by a processor to execute the steps of the vehicle special scene detection method according to various embodiments of the present application described in any of the above embodiments of this specification, and specifically can implement the steps of the vehicle special scene detection method as described above.

[0138] For the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0139] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0140] The steps in the methods of each embodiment of the present application can be adjusted in order, combined and deleted according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0141] The modules and sub-modules in the devices and terminals of the various embodiments of the present application can be combined, divided and deleted according to actual needs.

[0142] In the several embodiments provided in the present application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the terminal embodiments described above are only schematic, for example, the division of modules or submodules is only a logical function division, and there may be other division methods in actual implementation, for example, multiple submodules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0143] The modules or submodules described as separate components may or may not be physically separated, and the components of the modules or submodules may or may not be physical modules or submodules, that is, they may be located in one place, or they may be distributed on multiple network modules or submodules. Some or all of the modules or submodules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0144] In addition, each functional module or submodule in each embodiment of the present application may be integrated into one processing module, or each module or submodule may exist physically separately, or two or more modules or submodules may be integrated into one module. The above-mentioned integrated modules or submodules may be implemented in the form of hardware or in the form of software functional modules or submodules.

[0145] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0146] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly by hardware, software units executed by a processor, or a combination of the two. The software units may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0147] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0148] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle special scene detection method, characterized in that: include: When the first M detection results of the scene where the vehicle is located all indicate that the vehicle is located in an indoor parking lot, and the current detection result indicates that the vehicle is located in a non-parking lot, obtaining N detection results after the current detection result, where both M and N are positive integers; According to N detection results after the current detection result, it is determined whether the vehicle is located in a special scene of the indoor parking lot.

2. The method according to claim 1, characterized in that The N detection results include multiple detection results, and determining whether the vehicle is located in a special scene of the indoor parking lot according to the N detection results after the current detection result includes: If, among the multiple detection results after the current detection result, the number of detection results indicating that the vehicle is located in the non-parking lot is greater than or equal to a preset number, it is determined that the vehicle is located in a special scenario of the indoor parking lot.

3. The method according to claim 2, characterized in that After determining that the vehicle is located in the special scene of the indoor parking lot, the method further includes: When a detection result after the multiple detection results indicates that the vehicle is located in a special scene of the indoor parking lot, obtaining position change information of the vehicle in the special scene of the indoor parking lot; If the distance change corresponding to the position change information is greater than or equal to the preset distance, it is determined that the vehicle is outside the special scene of the indoor parking lot; If the distance change corresponding to the position change information is less than a preset distance, it is determined that the vehicle is located in a special scene of the indoor parking lot.

4. The method according to claim 2, characterized in that: After determining that the vehicle is located in the special scene of the indoor parking lot, the method further includes: If, in P detection results after the multiple detection results, the number of detection results indicating that the vehicle is located in an outdoor parking lot is greater than a preset number, it is determined that the vehicle is located in the outdoor parking lot, and P is a positive integer.

5. The method according to any one of claims 1 to 4, characterized in that The current detection result is obtained by the following steps: Acquire a current frame image captured of the vehicle's surrounding environment; Performing feature extraction on the current frame image to obtain image features corresponding to the current frame image; The current detection result is determined according to the image features corresponding to the current frame image.

6. The method according to claim 5, characterized in that The current frame image includes: images of the vehicle surrounding environment collected from different angles at the current frame image collection moment; The step of extracting features from the current frame image to obtain image features corresponding to the current frame image; and determining the current detection result according to the image features includes: Extracting image features from each image corresponding to the current frame image acquisition moment, and fusing the image features corresponding to each image to obtain a fused image feature; The current detection result is determined according to the fused image feature.

7. The method according to claim 5, characterized in that The step of extracting features from the current frame image to obtain image features corresponding to the current frame image; and determining the current detection result according to the image features includes: Inputting the current frame image into a scene detection model, performing feature extraction on the current frame image through the scene detection model to obtain image features corresponding to the current frame image; determining the current detection result according to the image features corresponding to the current frame image; The scene detection model is trained based on each frame of image samples collected from the vehicle's surrounding environment and the corresponding annotated scene categories.

8. The method according to claim 2, characterized in that: After determining that the vehicle is located in the special scenario of the indoor parking lot, the method further includes: When it is determined that the vehicle is located in a special scenario of the indoor parking lot, a memory parking function is started.

9. A vehicle special scene detection system, characterized in that: include: A detection device, the detection device includes a memory and a processor, the memory stores a detection program, and the detection program, when executed by the processor, is used to implement the vehicle special scene detection method according to any one of claims 1-8.

10. A vehicle, characterized in that: The vehicle is equipped with the vehicle special scene detection system as claimed in claim 9.

Images