Vehicle searching method and device, medium and product

Abstract

The invention provides a vehicle searching method and device, a medium and a program product. The invention provides a method for searching a vehicle, which comprises the following steps of: in response to a vehicle searching request of a user side, reading video data recorded by a vehicle camera; extracting candidate target objects in each video frame based on the video data; performing character and / or symbol recognition on each candidate target object to obtain a recognition result; and determining the priority of each identification result according to a preset weight, and determining the position of the user vehicle based on the identification result with the highest priority.

Description

Technical Field

[0001] This disclosure relates to the field of vehicles, and more specifically to methods, apparatus, media, and products for vehicle location. Background Technology

[0002] With the development of the economy and society, vehicles have become an important tool for people's daily travel.

[0003] To cope with the ever-increasing number of vehicles, parking lots are constantly expanding in size, and the paths within them are becoming increasingly complex. When users return to the parking lot to retrieve their vehicles after completing their tasks, they often find it difficult to locate their own car among the numerous parked vehicles.

[0004] Furthermore, when vehicles are parked in indoor parking lots, the poor quality of GPS signals, or even the near absence of GPS signals, makes it difficult for users to locate their vehicles using GPS signals. Summary of the Invention

[0005] According to a first aspect of this disclosure, a method for locating a vehicle is provided, comprising: in response to a vehicle locating request from a user terminal, reading video data recorded by a vehicle camera; extracting candidate target objects from each video frame based on the video data; performing text and / or symbol recognition on each candidate target object to obtain recognition results; determining the priority of each recognition result according to a pre-set weight, and determining the location of the user's vehicle based on the recognition result with the highest priority.

[0006] According to embodiments of this disclosure, the video data read corresponds to video data between a predetermined time point before parking is completed and the completion of parking, wherein the predetermined time point is a predetermined time point or a predetermined distance position.

[0007] According to embodiments of this disclosure, extracting candidate target objects from each video frame based on the video data includes: performing target detection on each video frame to generate one or more prediction boxes containing candidate target objects; determining the confidence level of the one or more prediction boxes; determining the video frame where the highest confidence level of the prediction box exceeds a threshold as a valid video frame; and extracting candidate target objects within the prediction boxes in each valid video frame.

[0008] According to embodiments of this disclosure, extracting candidate target objects within prediction boxes in each valid video frame includes: determining the prediction box with the highest confidence in the valid video frame as a valid prediction box; selecting a subset of valid prediction boxes and extracting candidate target objects from them.

[0009] According to embodiments of this disclosure, the pre-set weights include at least one or more of the following: weights assigned based on the timestamps of the video frames corresponding to the recognition results, weights assigned based on the confidence of the valid prediction boxes associated with the recognition results, and weights assigned based on the repetition frequency of the recognition results among all recognition results.

[0010] According to embodiments of this disclosure, the step of extracting candidate target objects from each video frame based on the video data further includes performing the following preprocessing on each video frame: converting each video frame in the video data into a tensor; adjusting the tensor of the video frame to reduce the precision to a predetermined range; and standardizing the pixel values ​​in the tensor to a predetermined range.

[0011] According to a second aspect of this disclosure, an apparatus for vehicle locating is provided, comprising: at least one processing unit; and a memory storing computer-readable instructions and communicatively connected to the at least one processing unit, the at least one processing unit being configured to execute the method according to the first aspect when executing the computer-readable instructions stored in the memory.

[0012] According to a third aspect of this disclosure, an apparatus for vehicle location is provided, comprising: a camera for recording video data of the environment surrounding the vehicle; a data acquisition unit for reading the video data recorded by the camera in response to a vehicle location request from a user; a target detection unit for extracting candidate target objects from each video frame based on the video data; an image recognition unit for performing text and / or symbol recognition on each candidate target object to obtain a recognition result; and a location determination unit for determining the priority of each recognition result according to a preset weight, and determining the location of the user's vehicle based on the recognition result with the highest priority.

[0013] According to embodiments of this disclosure, the camera includes one or more cameras located on the front, rear, left, and right sides of the vehicle.

[0014] According to embodiments of this disclosure, the step of extracting candidate target objects from each video frame based on the video data includes: performing target detection on each video frame to generate one or more prediction boxes containing candidate target objects; determining the confidence level of the one or more prediction boxes; determining the video frame in which the highest confidence level of the prediction box exceeds a threshold as a valid video frame, and extracting candidate target objects within the prediction boxes in each valid video frame.

[0015] According to embodiments of this disclosure, the step of extracting candidate target objects within prediction boxes in each valid video frame includes: determining the prediction box with the highest confidence in the valid video frame as a valid prediction box; selecting a subset of valid prediction boxes and extracting candidate target objects from them.

[0016] According to embodiments of this disclosure, the pre-set weights include at least one or more of the following: weights assigned based on the timestamps of the video frames corresponding to the recognition results, weights assigned based on the confidence of the valid prediction boxes associated with the recognition results, and weights assigned based on the repetition frequency of the recognition results among all recognition results.

[0017] According to embodiments of this disclosure, the apparatus may further include: a result output component configured to process the location of the user vehicle into a standardized format and transmit it.

[0018] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided, on which a computer program / instructions are stored, which, when executed by a processor, implement the method according to the first aspect.

[0019] According to a fifth aspect of this disclosure, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the method according to the first aspect.

[0020] By utilizing the methods, apparatus, media, and program products provided in this disclosure, intelligent reverse vehicle finding functions with high computational efficiency and high recognition accuracy can be achieved while reducing the hardware requirements and costs for function implementation. Attached Figure Description

[0021] These and / or other aspects and advantages of this disclosure will become clearer and more readily understood from the following detailed description of embodiments of this disclosure taken in conjunction with the accompanying drawings, wherein:

[0022] Figure 1 A schematic diagram of a vehicle-finding system according to an embodiment of the present disclosure is shown.

[0023] Figure 2 A flowchart of a method for locating a vehicle according to an embodiment of the present disclosure is shown;

[0024] Figure 3 An example process for extracting candidate target objects in a vehicle-finding method according to embodiments of the present disclosure is illustrated; and

[0025] Figure 4 A schematic diagram of a vehicle-finding device according to an embodiment of the present disclosure is shown.

[0026] Figure 5 A schematic diagram of a vehicle-finding device according to yet another embodiment of the present disclosure is shown.

[0027] It should be understood that these accompanying drawings are used to provide a further understanding of the embodiments of the present invention and form part of the specification. They are used together with the embodiments of the present invention to explain the invention and do not constitute a limitation thereof. Furthermore, in the drawings, the same reference numerals generally represent the same components or steps. Detailed Implementation

[0028] To better illustrate the technical solutions of this disclosure, the following detailed description is provided in conjunction with the accompanying drawings and specific embodiments. It should be understood that all other embodiments obtained by those skilled in the art based on the embodiments described herein without creative effort should fall within the protection scope of this invention. Furthermore, the embodiments described herein are only a part of the embodiments of this invention, not all of them. These embodiments are merely illustrative and exemplary, and therefore should not be construed as limiting the scope of this invention.

[0029] In parking lots where GPS signals may be absent or weak, one possible method for users to locate their vehicles is to use the parking lot's license plate recognition (LPR) system to capture license plate information and identify and track vehicles. However, this approach, relying on parking lot facilities for license plate recognition to support vehicle identification and tracking, has high deployment and maintenance costs. Furthermore, since parking lot facilities typically need to be installed in fixed locations, this method requires a large number of facilities and significant computing resources. In addition, parking lots without these facilities cannot provide vehicle location services. Moreover, the complex and dimly lit environments in parking lots present challenges in obtaining accurate identification and analysis results for vehicle location.

[0030] In view of the above-mentioned problems in the prior art, this disclosure provides a method and apparatus for vehicle location.

[0031] First, combine Figure 1 A vehicle-finding system 100 according to an embodiment of the present disclosure will be described. For example... Figure 1 As shown, system 100 may include user-side device 110 and vehicle-side device 120, wherein vehicle-side device 120 may be communicatively coupled to user-side device 110.

[0032] User-side device 110 can be a user's portable device (such as a mobile phone, smart bracelet, etc.), and the user can interact with vehicle-side device 120 through, for example, an application installed on it to perform the vehicle search process.

[0033] The vehicle-side device 120 is mounted on a vehicle and can be configured to perform a vehicle-finding method 200 according to embodiments of the present disclosure. The vehicle-side device may include a camera assembly 121, a data processing unit 122, and a data storage unit 123. The data processing unit 122 is configured to read and process the data buffer of video data from the camera assembly 121 in response to a vehicle-finding request from a user. The data storage unit 123 is configured to store the video data from the camera assembly 121. The data processing unit 122 may be coupled to the camera assembly 121, for example, via a cable; and the data processing unit 122 may be coupled to the data storage unit 123, for example, via a USB interface. The camera assembly 121 can be any camera-enabled device mounted on the vehicle for recording the vehicle's driving process, such as a dashcam.

[0034] In embodiments according to this disclosure, for example, camera group 121 may include one or more of a plurality of cameras arranged on the front, rear, left, and right sides of the vehicle. For example, it may include cameras in the aforementioned directions to form a surround-view camera system, thereby enabling the acquisition of multi-view video of the vehicle's driving process for vehicle location.

[0035] In embodiments according to this disclosure, when a user needs to return to the parking lot to find their car, they can remotely send a car-finding request to the vehicle-side device 120 via the user-side device 110. Upon receiving the car-finding request from the user-side device, the vehicle-side device 120 can be awakened from sleep mode and begin operation. For example, the data processing device 122 can be activated in response to the car-finding request from the user-side device, read video data from the data buffer of the camera group 121, process and recognize the video data to obtain a recognition result that can be used to help the user find their car, and finally send the recognition result back to the user-side device 110 for the user to find their car.

[0036] In embodiments according to this disclosure, system 100 may further include cloud server 130. For example... Figure 1 As shown, the user-side device 110 and the vehicle-side device 120 can be communicatively coupled to the cloud server 130, and can communicate via the cloud server 130. As a non-limiting example, a user-initiated remote wake-up of the vehicle-side device 120 can be executed via the cloud server 130. Specifically, as... Figure 1As shown, when a user needs to return to the parking lot to find their car, they can send a car-finding request through the user-side device 110. The cloud server 130 can receive the car-finding request initiated by the user and, in response to the request, remotely wake up the vehicle-side device 120, which then powers on and begins operation. After the vehicle-side device 120 obtains the identification result for the user to find their car, it sends the identification result back to the user-side device 110 through the cloud server 220 for the user to find their car.

[0037] Next, we will combine Figure 2 The present disclosure provides a detailed description of a method for locating a vehicle according to embodiments thereof.

[0038] Figure 2 A flowchart of a vehicle-finding method 200 according to an embodiment of the present disclosure is shown, which can be performed on the vehicle side.

[0039] like Figure 2 As shown, in step S201, in response to a vehicle-finding request from the user, video data recorded by the vehicle camera can be read. The vehicle camera can be any device with video recording function installed on the vehicle for recording the vehicle's driving process, such as a dashcam.

[0040] In the vehicle location method according to this embodiment, the read video data may correspond to video data between a predetermined time point before parking is completed and the completion of parking. This predetermined time point may be, for example, a predetermined time point (e.g., 5 minutes before parking is completed) or a predetermined distance location (e.g., 100 meters before parking is completed), etc., and this disclosure does not impose any limitations on this. In this way, the read video data can at least partially reflect factors such as the time, distance traveled, and speed during the parking process, thereby enabling these factors to be comprehensively considered in subsequent video data processing and analysis to more accurately assist the user in locating the vehicle's position.

[0041] In step S202, candidate target objects are extracted from each video frame based on the video data. In this disclosure, a candidate target object refers to a target object in a video frame that possesses specific characteristics and can be used to assist in vehicle location. This can include text or symbols, such as parking space signs, navigation signs, parking area division information, or access information in a parking lot. Alternatively or additionally, candidate target objects may also include objects with landmark characteristics, such as parking lot pillars or toll booths.

[0042] In embodiments according to this disclosure, as a non-limiting example, the order in which candidate target objects are extracted from each video frame can be processed in reverse order, starting from the last frame of the video data, thereby enabling more efficient identification of candidate target objects located closer to the user's vehicle. It should be understood that the selection strategy for the order in which video frames are processed to extract candidate target objects can be dynamically adjusted according to actual needs and scenarios, and this disclosure does not impose any limitations on this.

[0043] In the vehicle-finding method according to this embodiment, the processing in step S202 can be performed in various suitable ways. Optionally, step S202 further includes: performing target detection on each video frame to generate one or more predicted bounding boxes containing candidate target objects; determining the confidence level of the one or more predicted bounding boxes; determining video frames in which the highest confidence level of a predicted bounding box exceeds a threshold as valid video frames; selecting a subset of valid predicted bounding boxes and extracting candidate target objects from them.

[0044] The following will refer to further Figure 3 A detailed explanation of an example implementation of step S202 is provided.

[0045] Figure 3 An example process 300 for extracting candidate target objects according to an embodiment of the present disclosure is shown.

[0046] like Figure 3 As shown, in step S301, object detection can be performed on each video frame to generate one or more prediction boxes containing candidate object objects. For example, if an expected object is detected in the current frame, a prediction box corresponding to the expected object is generated; if no expected object is detected in the current frame, no prediction box is generated. In this disclosure, various object detection algorithms can be used to perform frame-by-frame detection on video data to extract candidate object objects. As a non-limiting example, R-CNN series algorithms or YOLO series algorithms can be used.

[0047] At step S302, the confidence level of one or more prediction boxes in each frame can be determined. The confidence level of a prediction box can characterize the probability that the target object in the prediction box is the expected target object, and various appropriate methods can be used in this disclosure to determine the confidence level of a video frame. Optionally, solutions for conventional target detection algorithms with different target objects can be optimized accordingly. For example, as a non-limiting example, in this disclosure, the target detection algorithm can be optimized so that the prediction box only obtains a high confidence level when text / symbols are detected, thus accurately filtering out candidate target objects with high-quality text / symbol classes, thereby reducing the amount of data processing and improving the overall computational efficiency and accuracy.

[0048] In step S303, the video frames in which the highest confidence of the predicted bounding box exceeds the threshold are determined as valid video frames.

[0049] The predicted bounding box with the highest confidence indicates that the candidate target object included in that predicted bounding box has the highest probability of being the expected target object. For example, in this step, the confidence of each predicted bounding box contained in each video frame is determined, and video frames with the highest confidence of predicted bounding boxes exceeding a threshold are considered valid video frames for subsequent processing. In this disclosure, different confidence thresholds can be trained for different types of object detection algorithms, and this disclosure does not limit this. By performing the above processing on frames with predicted bounding boxes to obtain valid predicted frames, high-quality candidate target objects can be further selected, and the accuracy of recognition can be improved while reducing the amount of data processing in subsequent recognition processes.

[0050] As a non-limiting example, in this disclosure, video frames other than valid predicted frames, such as frames that do not generate prediction boxes, or video frames in which the highest confidence of the prediction boxes does not exceed a threshold, may be removed from the video frame set as invalid frames.

[0051] In step S304, candidate target objects within the prediction bounding boxes of each valid video frame can be extracted. In this step, various appropriate methods can be used to extract candidate target objects from the prediction bounding boxes (e.g., the YOLO algorithm can be used to extract images containing the candidate target objects from the valid video frames), and this disclosure does not impose any limitations on this. It should be noted that when extracting images containing candidate target objects, the timestamp of the frame containing the valid prediction bounding box needs to be retained for processing in subsequent steps.

[0052] Optionally, in this step, predicted bounding boxes with a confidence level exceeding a threshold in the valid video frames can be used as valid predicted bounding boxes, and candidate target objects can be extracted from these valid predicted bounding boxes. Alternatively, only the predicted bounding boxes with the highest confidence level in the valid frames can be used as valid predicted bounding boxes for subsequent processing to reduce the amount of data processing. Furthermore, a subset of valid predicted bounding boxes can be selected as needed to extract candidate target objects from them to further reduce the amount of data processing. Those skilled in the art can use various appropriate methods to select an appropriate subset of valid predicted bounding boxes, and this disclosure does not limit this. For example, a predetermined number of valid predicted bounding boxes with high confidence levels can be selected from the valid predicted bounding boxes, or valid predicted bounding boxes with a confidence level exceeding a certain predetermined threshold can be selected, or predicted bounding boxes at predetermined time points can be selected based on the timestamp of the frame in which the predicted bounding box is located, and so on.

[0053] The above has been combined Figure 3An example process for extracting candidate target objects in an embodiment of this disclosure is described. As an optional example, before step S301, the video data can be preprocessed to reduce the computational burden, improve the efficiency of data processing, and ensure the accuracy of the information. Specifically, the data frames of the read video data can be converted into tensors frame by frame; then, the tensors of the video frames are adjusted to reduce the precision to a predetermined range (e.g., to half precision); finally, the pixel values ​​in the tensors are standardized to a specified range.

[0054] As another optional example, image quality enhancement can be performed on the candidate target objects extracted according to step S304 (e.g., the captured image set including the candidate target objects). For example, the pixel value intensity distribution of the extracted images including the candidate target objects can be adjusted to enhance overall clarity. For example, histogram equalization can be performed on the pixel value intensity distribution of the image, thereby making the grayscale value distribution of the image more uniform, enhancing contrast, and making details clearer. This improves the clarity of the candidate target objects to be recognized, thereby reducing the recognition difficulty caused by poor image quality due to dim lighting in indoor parking lots and improving recognition accuracy.

[0055] Return to reference Figure 2 In step S203, text and / or symbol recognition can be performed on each of the candidate target objects to obtain recognition results. For example, in embodiments according to this disclosure, a convolutional recurrent neural network (CRNN) model can be used to perform batch prediction on the extracted image set of candidate target objects to perform text and symbol recognition. It should be understood that various image recognition algorithms can be used as needed to perform frame-by-frame detection on the extracted candidate target objects to obtain recognition results.

[0056] In step S204, the priority of each recognition result can be determined according to the preset weight, so as to determine the location of the user vehicle based on the recognition result with the highest priority.

[0057] In embodiments according to this disclosure, the weights of each factor in the recognition result can be set as needed, thereby determining the priority of each recognition result. As a non-limiting example, in this disclosure, the pre-set weights may include at least one or more of the following: weights assigned based on the timestamps of the video frames corresponding to the recognition result; weights assigned based on the confidence level of the valid predicted bounding boxes associated with the recognition result; and weights assigned based on the repetition frequency of the recognition result among all recognition results. For example, if the timestamp of the video frame corresponding to the recognition result is closer to the parking time, then the recognition result is assigned a larger weight; or, if the confidence level of the valid predicted bounding boxes associated with the recognition result is higher, then the recognition result is assigned a larger weight; or, if the repetition frequency (or number of repetitions) of the recognition result is higher among all recognition results, then the recognition result is assigned a larger weight.

[0058] Furthermore, as an alternative or additional scheme to timestamps, in embodiments according to this disclosure, weights can also be assigned based on distance parameters of video frames associated with the recognition results. For example, the data processing device 122 can read distance parameters corresponding to video frames from the vehicle's inertial measurement unit (IMU) while reading video data from the data buffer of the vehicle's camera array, and assign greater weights to video frames closer to the parking position.

[0059] By considering the above factors to assign weights and determine the priority of each recognition result, and then determining the location of the user's vehicle based on the recognition result with the highest priority, the data can be strategically filtered and processed, thereby improving the accuracy of the final output recognition result and its relevance to the user's vehicle location, thus helping to determine the location of the user's vehicle more efficiently and accurately.

[0060] It should be understood that, based on the above combination Figures 2-3 The method described according to embodiments of the present disclosure, in optional implementations, may include preprocessing, extraction of candidate target objects, and other steps to reduce data processing volume and lower hardware requirements and costs for function implementation. This enables efficient local edge computing without sending video data to a remote server (e.g., cloud server 220) for processing. However, the method described above according to embodiments of the present disclosure is not limited to execution only on the vehicle-side device 120, but may be executed on other sides (e.g., cloud server 220) depending on the specific circumstances.

[0061] Figure 4 A schematic diagram of a vehicle-finding device 400 according to an embodiment of the present disclosure is shown. Figure 4 As shown, the device may include a camera 401, a data acquisition component 402, a target detection component 403, an image recognition component 404, and a location determination component 405.

[0062] For example, in an embodiment according to this disclosure, camera 401 may include one or more cameras on the front, rear, left, and right sides of the vehicle.

[0063] Specifically, in embodiments according to this disclosure, the data acquisition unit 402 may be configured to read video data recorded by the camera 401 in response to a vehicle-finding request from a user. The target detection unit 403 may be configured to extract candidate target objects from each video frame based on the video data. Optionally, the target detection unit 403 may perform target detection on each video frame to generate one or more prediction boxes containing candidate target objects; determine the confidence level of the one or more prediction boxes; determine video frames where the highest confidence level of the prediction boxes exceeds a threshold as valid video frames; determine the prediction box with the highest confidence level among the valid video frames as a valid prediction box; select a subset of valid prediction boxes and extract candidate target objects from them. The image recognition unit 404 may be configured to perform text and / or symbol recognition on each of the candidate target objects to obtain recognition results. The location determination unit 405 may be configured to determine the priority of each recognition result according to a preset weight, and determine the location of the user's vehicle based on the recognition result with the highest priority. In embodiments according to this disclosure, the device 400 may further include a result output unit. The result output component can be configured to send the determined location of the user's vehicle to the user-side device 110. Optionally, the result output component can process the user's vehicle location into a standardized format before sending it, thereby maximizing the standardization of information transmission and enabling functional scalability. For example, the standardized format can be represented as, for instance, parking lot building + floor + area + parking space number. In the case where system 100 includes a cloud server 130, the result output component can be communicatively coupled to an information acquisition component in the cloud, and the user-side device 110 can receive the determined location of the user's vehicle from the cloud (e.g., communicatively coupled to the information acquisition component in the cloud) through an information push component installed thereon.

[0064] Optionally, after the user-side device 110 receives the determined location of the user's vehicle and ends the vehicle search, the user can provide feedback on the accuracy of the result (e.g., submit text or image feedback) to optimize the vehicle search function. For example, if the determined location of the user's vehicle differs from the actual parking location, and the user provides image feedback of the actual parking location, this image feedback can be collected and used to optimize the object detection algorithm and image recognition algorithm in steps S202-S203, thereby continuously improving the user experience and promoting functional advancement.

[0065] Figure 5A schematic diagram of a vehicle-finding device 500 according to an embodiment of the present disclosure is shown. Figure 5 As shown, the device 500 may include a processor 501 and a memory 502.

[0066] The processor 501 can be any processing-capable device capable of implementing the functions of the various embodiments of this disclosure. For example, it can be a general-purpose processor, digital signal processor (DSP), ASIC, field programmable gate array (FPGA) or other programmable logic device (PLD), discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein.

[0067] The memory 502 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, or other removable / non-removable, volatile / non-volatile computer system memory, such as hard disk drives, floppy disks, CD-ROMs, DVD-ROMs, or other optical storage media.

[0068] In this embodiment, the memory 502 stores computer program instructions, and the processor 501 can execute the instructions stored in the memory 502. When the computer program instructions are executed by the processor, the processor performs the vehicle locating method of this embodiment. The vehicle locating method is similar to that described above. Figures 2 to 3 The descriptions are essentially the same, so for the sake of brevity, they will not be repeated.

[0069] The vehicle locating method / apparatus according to this disclosure can also be implemented by providing a computer program product containing program code implementing the method or apparatus, or by any storage medium storing such a computer program product.

[0070] As a non-limiting example, the vehicle-finding method and apparatus according to this disclosure use data cache from vehicle cameras as a video data source, thus eliminating reliance on parking lot infrastructure and reducing vehicle-finding costs; strategic data preprocessing and probabilistic filtering mechanisms improve computational efficiency, thereby enhancing the predictive performance of the function; the use of lightweight visual models for local edge computing reduces the hardware requirements and costs of function implementation, improving versatility; locally inferred text recognition information can be transmitted to the user's mobile application, continuously improving the user experience and promoting functional advancement through user interaction; and the standardization of information transmission can be maximized, making the function scalable and laying the foundation for potential enhancements to parking lot navigation functions.

[0071] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the embodiments of this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0072] The block diagrams of devices, apparatuses, devices, and systems involved in the embodiments of this disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0073] Additionally, as used herein, the “or” used in a list of items beginning with “at least one” indicates a separate list, such that a list of, for example, “at least one of A, B, or C” means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word “exemplary” does not imply that the described example is preferred or better than other examples.

[0074] It should also be noted that in the apparatus and methods of this disclosure, the components or steps can be disassembled and / or recombined. Such disassembly and / or recombination should be considered as equivalent solutions to this disclosure.

[0075] It will be understood by one of ordinary skill in the art that all or any part of the methods and apparatus of this disclosure can be implemented in hardware, firmware, software, or a combination thereof in any computing device (including processors, storage media, etc.) or network of computing devices. The hardware may be a general-purpose processor, digital signal processor (DSP), ASIC, field-programmable gate array (FPGA) or other programmable logic device (PLD), discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The general-purpose processor may be a microprocessor, but alternatively, it may be any commercially available processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors cooperating with a DSP core, or any other such configuration. The software may reside in any form of computer-readable tangible storage medium. By way of example and not limitation, such computer-readable tangible storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other tangible medium that can be used to carry or store desired program code in the form of instructions or data structures and is accessible by a computer. If used herein, the disks include compact discs (CDs), laser discs, optical discs, digital universal discs (DVDs), floppy disks, and Blu-ray discs.

[0076] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0077] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0078] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for locating a vehicle, comprising: In response to a user's vehicle location request, the system reads video data recorded by the vehicle's camera. Based on the video data, candidate target objects are extracted from each video frame; The candidate target objects are subjected to text and / or symbol recognition to obtain recognition results; The priority of each recognition result is determined according to a pre-set weight, and the location of the user's vehicle is determined based on the recognition result with the highest priority.

2. The method according to claim 1, wherein, The video data read corresponds to the video data between a predetermined time point before parking is completed and the completion of parking, where the predetermined time point is a predetermined time point or a predetermined distance position.

3. The method according to claim 1, wherein, Extracting candidate target objects from each video frame based on the video data includes: Perform object detection on each video frame to generate one or more prediction boxes containing candidate object objects; Determine the confidence level of the one or more prediction boxes; The video frames whose highest confidence level of the predicted bounding box exceeds a threshold are identified as valid video frames. Extract candidate target objects within the prediction bounding boxes in each valid video frame.

4. The method according to claim 3, wherein, Extracting candidate target objects from each video frame based on the video data includes processing them in reverse order, starting from the last frame of the video data.

5. The method according to claim 3, wherein, Extracting candidate target objects within the predicted bounding boxes in each valid video frame includes: The prediction box with the highest confidence in the valid video frame is identified as the valid prediction box; Select a subset of valid prediction boxes and extract candidate target objects from them.

6. The method according to claim 5, wherein, The pre-defined weights include at least one or more of the following: weights assigned based on the timestamps of the video frames corresponding to the recognition results, weights assigned based on the confidence of the valid prediction boxes associated with the recognition results, and weights assigned based on the repetition frequency of the recognition results among all recognition results.

7. The method according to claim 3, wherein, The step of extracting candidate target objects from each video frame based on the video data also includes performing the following preprocessing on the video data: Convert each video frame in the video data into a tensor; Adjust the tensor of the video frame to reduce the precision to a predetermined range; The pixel values ​​in the tensor are standardized to a predetermined range.

8. The method of claim 1, further comprising: The determined location of the user's vehicle is processed into a standardized format and sent to the user's terminal.

9. A device for locating a vehicle, comprising: At least one processing unit; and The memory stores computer-readable instructions and is communicatively connected to the at least one processing unit. The at least one processing unit is configured to execute the method according to any one of claims 1-8 when running the computer-readable instructions stored in the memory.

10. A device for locating a vehicle, comprising: Cameras record video data of the environment surrounding the vehicle; The data acquisition component, in response to the user's vehicle location request, reads the video data recorded by the camera; The target detection component extracts candidate target objects from each video frame based on the video data; An image recognition component performs text and / or symbol recognition on each of the candidate target objects to obtain recognition results; The location determination component determines the priority of each recognition result according to a pre-set weight, and determines the location of the user's vehicle based on the recognition result with the highest priority.

11. The apparatus of claim 10, wherein the camera comprises one or more cameras located on the front, rear, left, and right sides of the vehicle.

12. The apparatus of claim 10, wherein, The step of extracting candidate target objects from each video frame based on the video data includes: Perform object detection on each video frame to generate one or more prediction boxes containing candidate object objects; Determine the confidence level of the one or more prediction boxes; The video frame with the highest confidence score of the predicted bounding box exceeding the threshold is identified as a valid video frame. Extract candidate target objects within the prediction bounding boxes in each valid video frame.

13. The apparatus according to claim 12, wherein, The extraction of candidate target objects within the prediction bounding boxes in each valid video frame includes: The prediction box with the highest confidence in the valid video frame is identified as the valid prediction box; Select a subset of valid prediction boxes and extract candidate target objects from them.

14. The apparatus according to claim 13, wherein, The pre-defined weights include at least one or more of the following: weights assigned based on the timestamps of the video frames corresponding to the recognition results, weights assigned based on the confidence of the valid prediction boxes associated with the recognition results, and weights assigned based on the repetition frequency of the recognition results among all recognition results.

15. A non-transitory computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implement the method according to any one of claims 1-8.

16. A computer program product comprising a computer program / instructions that, when executed by a processor, implement the method according to any one of claims 1-8.

Images