METHOD AND APPARATUS FOR TARGET DETECTION IN VEHICLE RACE, DEVICE AND MEANS
Patent Information
- Authority / Receiving Office
- MX · MX
- Patent Type
- Patents
- Current Assignee / Owner
- CHONGQING CHANGAN AUTOMOBILE CO LTD
- Filing Date
- 2023-03-30
- Publication Date
- 2026-06-12
Smart Images
Figure MX434852B0
Abstract
Description
This description relates to the technical field of safety of the intended functions of automatic driving, in particular to a method and apparatus for detecting targets in moving vehicles, a device and a means. Background of the invention In recent years, with the rise of deep learning technology, the field of computer vision has made rapid progress. As a typical application, target detection is widely used in facial recognition, intelligent driving, and other areas. In the field of intelligent driving, target detection primarily focuses on the detection of road users, traffic signs, and general obstacles. This type of scenario requires a large amount of road test data collected by real vehicles. However, during collection, due to pedestrians, vehicles, non-motorized vehicles, or other obstacles, a large number of samples are hidden in the road test data.During training and prediction of such samples, due to the lack of a large information area, features are incomplete or missing, and a detector's performance will decline, easily leading to inaccurate recognition results and a failure to recall. Therefore, improving the recognition accuracy of a protected target has become an urgent technical problem. Brief description of the invention The present disclosure provides a vehicle moving target detection method and apparatus, a device and a means, which solve the problem existing in the prior art of low recognition accuracy of a protected target, which easily leads to erroneous recognition or missed detection. An embodiment of a first aspect of the present disclosure provides a method for detecting targets in moving vehicles, including the steps of: acquiring an image to be detected from a target vehicle in a moving process; inputting the image to be detected to a first preconfigured detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; inputting the image to be detected to a second preconfigured detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image, each target to be detected and the reference target having a correspondence relationship; comparing the first recognition result with the second recognition result;and correcting the second recognition result according to a comparison result to obtain a target recognition result.; According to the foregoing technical means, the embodiment of the present disclosure acquires an image to be detected from a target vehicle in a running process; inputs the image to be detected to a first preset detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; and inputs the image to be detected to a second preset detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship. Accordingly, the first recognition result and the second recognition result also have a certain correspondence relationship.The first recognition result is compared with the second recognition result for mutual verification, such that the second recognition result is corrected according to a comparison result to obtain the accuracy of a recognition result for the target. Thus, by comparing the first recognition result corresponding to the reference target with the second recognition result corresponding to the target to be detected, and correcting them, misrecognition or non-detection is avoided. Furthermore, the second recognition result includes a predicted category corresponding to each target to be detected and a first confidence coefficient; the first recognition result includes a reference category corresponding to each target to be detected; correcting the second recognition result according to a comparison result to obtain a target recognition result includes determining that the predicted category corresponding to the current target to be detected is a correct result if the first recognition result is equal to the predicted category corresponding to a current target to be detected;if the first recognition result is different from the predicted category corresponding to the current target to be detected, determining that the predicted category corresponding to the current target to be detected is a confused result, and determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined to be correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, correcting the predicted category corresponding to the current target to be detected to be the corresponding reference category in the first recognition result.; According to the above technical means, the embodiment of the present disclosure can improve the accuracy of the recognition result of a target to be detected. Furthermore, determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result includes averaging the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results to obtain that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result. According to the above technical means, the modality of the present description can ensure the rationality of the correction. Furthermore, determining, according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined to be the correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result includes: performing a weighted average operation according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined to be the correct results, and the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined to be the confusing results,and determining that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result, where the weight of the first confidence coefficient of each predicted category determined as the correct result is greater than the weight of the first confidence coefficient of each predicted category determined as the confusing result., According to the above technical means, in the embodiment of the present disclosure, during correction, the confidence coefficients of the prediction categories that are determined as the correct results and the confusing results can be comprehensively considered, thereby ensuring rationality. Furthermore, the method also includes: acquiring first sample data containing the reference target and labeled data corresponding to the first sample data from the road test data; configuring a first pre-built detector according to the first sample data and labeled data corresponding to the first sample data, so that the first detector correctly outputs the first reference target recognition result; acquiring second sample data containing the target to be detected and labeled data corresponding to the second sample data from the road test data;and configure a pre-constructed second detector according to the second sample data and the labeled data corresponding to the second sample data, so as to enable the second detector to correctly output the second recognition result for the target to be detected.; According to the above technical means, the first detector and the second detector are respectively configured according to the data of the first sample and the data of the second sample, as well as the labeled data corresponding to the data of the first sample and the data of the second sample in the road test data, so that the accuracy of the configured first detector and the second detector can be ensured. In addition, the reference target is a lane guidance marking, and the target to be detected is a lane arrow marking. Based on the above technical means, the current lane information of a vehicle can be accurately recognized, providing a reliable basis for intelligent driving. Furthermore, comparing the first recognition result with the second recognition result includes: if the image to be detected does not simultaneously contain the reference target and the target to be detected, comparing the first recognition result or the second recognition result acquired first with a second recognition result or a first recognition result acquired later within a predetermined time interval. According to the above technical means, when the image to be detected does not simultaneously contain the reference target and the target to be detected, the accuracy of the target recognition result can also be improved. An embodiment of a second aspect of the present disclosure provides a vehicle moving target detection apparatus, including: an acquisition module, configured to acquire an image to be detected from a moving target vehicle; a first detection module, configured to input the image to be detected to a first preset detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; a second detection module, configured to input the image to be detected to a second preset detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship;a comparison module, configured to compare the first recognition result with the second recognition result; and a processing module, configured to correct the second recognition result according to a comparison result to obtain a target recognition result. Furthermore, the second recognition result includes a predicted category corresponding to each target to be detected, and a first confidence coefficient; the first recognition result includes a reference category corresponding to each target to be detected; and the processing module is configured to: if the first recognition result is equal to the predicted category corresponding to a current target to be detected, determine that the predicted category corresponding to the current target to be detected is a correct result;if the first recognition result is different from the predicted category corresponding to the current target to be detected, determining that the predicted category corresponding to the current target to be detected is a confusing result, and determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined to be correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, correcting the predicted category corresponding to the current target to be detected to be the corresponding reference category in the first recognition result.; In addition, the processing module is configured to: average the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results to obtain that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result. Furthermore, the processing module is configured to: perform the weighted average operation according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results and the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the confusing results, and determine that the prediction category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result, wherein the weight of the first confidence coefficient of each prediction category determined as a correct result is greater than the weight of the first confidence coefficient of each prediction category determined as a confusing result. Furthermore, the processing module is also configured to: acquire first sample data containing the reference target and labeled data corresponding to the first sample data from the road test data; configure a first pre-built detector based on the first sample data and the labeled data corresponding to the first sample data, so that the first detector correctly outputs the first recognition result of the reference target; acquire a second data sample containing the target to be detected and labeled data corresponding to the second sample of road test data;and configure a second pre-constructed detector according to the second data sample and the labeled data corresponding to the second data sample, so as to enable the second detector to correctly output the second recognition result for the target to be detected.; In addition, the reference target is a lane guidance marking, and the target to be detected is a lane arrow marking. Furthermore, the comparison module is configured to: if the image to be detected does not simultaneously contain the reference target and the target to be detected, compare the first recognition result or the second recognition result acquired first with a second recognition result or a first recognition result acquired subsequently within a predetermined time interval. An embodiment of a third aspect of the present disclosure provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and operable on the processor, wherein the processor executes the program to implement the method of detecting moving vehicle targets of the above embodiments. An embodiment of a fourth aspect of the present disclosure provides a computer-readable storage medium. The computer-readable storage medium stores a computer instruction. The computer-readable instruction QQPPPn / cznz / zi / υιλι computing is set to make a computer triple the method of detecting targets in moving vehicles of the previous modalities. Embodiments of the present disclosure may acquire an image to be detected from a target vehicle in a running process; input the image to be detected to a first preconfigured detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; and input the image to be detected to a second preconfigured detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship. Thus, the first recognition result and the second recognition result also have a certain correspondence relationship.The first recognition result is compared with the second recognition result for mutual verification, such that the second recognition result is corrected according to a comparison result to obtain the accuracy of a recognition result for the target. Thus, by comparing the first recognition result corresponding to the reference target with the second recognition result corresponding to the target to be detected, and correcting them, misrecognition or non-detection is avoided. Additional aspects and advantages of the present application will be provided in the following descriptions, part of which will become apparent from the following descriptions or will be learned through practice of the present description. Brief description of the drawings The above and / or additional aspects and advantages of the present description will become apparent and easily understandable from the following descriptions of the embodiments with reference to the attached drawings. Figure 1 is a flow diagram of a vehicle moving target detection method provided according to an embodiment of the present disclosure; Figure 2 is a block diagram of a system capable of applying the method of detecting targets in vehicle circulation provided according to an embodiment of the present description; Figure 3 is a flowchart of the actual labeling, training, and prediction algorithm of a vehicle moving target detection method provided according to an embodiment of the present disclosure; Figure 4 is a flowchart of a method for detecting targets in moving vehicles according to an embodiment of the present disclosure; Figure 5 is a schematic diagram of a reference map of a vehicle moving target detection method according to an embodiment of the present disclosure; Figure 6 is a flowchart of an arbitration method in the vehicle moving target detection method provided according to an embodiment of the present disclosure; Figure 7 is a schematic block diagram of a vehicle moving target detection apparatus according to an embodiment of the present disclosure; and Figure 8 is a sample diagram of an electronic device according to an embodiment of the present disclosure. Detailed description of the embodiments of the invention The embodiments of the present disclosure are described in detail below. Examples of the embodiments are shown in the accompanying drawings. Identical or similar reference numerals represent identical or similar elements or elements having identical or similar functions throughout the entire assembly. The embodiments described below with reference to the drawings are exemplary and intended to explain the present disclosure, and should not be construed as limiting it. Figure 1 is a flowchart of a method for detecting moving vehicle targets provided according to one embodiment of the present disclosure. The method may be applied to a terminal. The terminal may include, but is not limited to, a smartphone, a tablet, a laptop, or a device installed in a vehicle. In other embodiments, the method may also be applied to a server, which will not be particularly limited. Referring to Figure 1, the method for detecting targets in moving vehicles includes at least steps P110 to P150. The following example shows the application of the method to a terminal, which is described in detail below: In step P110, an image to be detected is acquired from a moving target vehicle. In one embodiment, an image acquisition device may be arranged on a target vehicle and is configured to acquire, in a driving process of the target vehicle, an image in front of the target vehicle in real time as an image to be detected. In step P120, the image to be detected is input to a first preset detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected. The reference target can be used to help recognize the target of a target to be detected. For example, if the target to be detected is a lane arrow marking, the reference target can be a double-lane arrow marking. In this embodiment, the terminal may input the image to be detected into the preconfigured first detector. The first detector may have target detection and recognition functions, and may recognize a target in the image to be detected to output the first recognition result of the reference target contained in the image to be detected. For example, if the reference target is the lane guidance sign, the first recognition result may include a driving direction of each lane on the lane guidance sign, a relative distance and relative positions between lanes, and the like. In step P130, the image to be detected is input to a second preset detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship. The target to be detected can be a real object to be detected, such as a lane arrow sign, a traffic light, and a traffic sign. In this embodiment, the terminal may also input the image to be detected into the preconfigured second detector. The second detector also has target detection and recognition functions. It should be noted that one difference between the first and second detectors is that they recognize different targets. The second detector may recognize the image to be detected to output the second recognition result of at least one target to be detected contained in the image to be detected. The second recognition result may include a predicted category corresponding to each detected target and a corresponding first confidence coefficient. It should be noted that since there is a correspondence relationship between the reference target and the target to be detected, there must also be a correspondence relationship between the first recognition result and the second recognition result. In one example, one or more target detection algorithms from the YOLO (You Only Look Once) collection, the R-CNN (Region-based Convolutional Neural Network) collection, or the SSD (Single Shot Multi-Frame Detector) collection may be used to configure the first detector and the second detector. In one embodiment of the present disclosure, the method also includes: the first sample data containing the reference target and the labeled data corresponding to the first sample data are acquired from the road test data; A first pre-built detector is configured based on the first sample data and the labeled data corresponding to the first sample data, so that the first detector can correctly output the first recognition result of the reference target; Second sample data containing the target to be detected and labeled data corresponding to the second sample data are acquired from the road test data; and a pre-built second detector is configured based on the second sample data and the labeled data corresponding to the second sample data, so that the second detector can correctly output the second recognition result of the target to be detected. In this embodiment, the road test data may be road data taken by a real vehicle, and may be a video or an image. The first sample data containing the reference target may be sifted from the road test data, and the first sample data is labeled. For example, if the reference target is a lane guidance sign, a lane type of each lane, the relative positions between lanes, and the like may be labeled. A pre-built first detector is then configured based on the first sample data and the labeled data corresponding to the first sample data, so that the first detector can correctly output the first recognition result of the reference target. The second sample data containing the target to be detected can be sifted from the road test data, and the second sample data labeled. For example, if the target to be detected is a lane arrow marking, a lane type corresponding to each lane arrow marking, the relative positions between lanes, and the like can be labeled. A pre-built second detector is then configured based on the second sample data and the labeled data corresponding to the second sample data, so as to enable the second detector to correctly output the second recognition result of the target to be detected. Therefore, the accuracy of the recognition results of the first and second detectors is guaranteed through preconfiguration, and the accuracy of target detection is improved after comparison. In step P140, the first recognition result is compared with the second recognition result. In step P150, the second recognition result is corrected according to a comparison result to obtain a target recognition result. The correction may be a process of correcting the second recognition result based on the first recognition result. In one embodiment, since the first recognition result is recognized according to the reference target, the first recognition result can be used as a priori information for the second recognition result, that is, the first recognition result can be used as a judgment basis for determining whether the second recognition result is correct. By comparing the first recognition result with the second recognition result, it is possible, on the one hand, to verify whether the second recognition result is correct, and, on the other hand, to provide a correct predicted category for an erroneous second recognition result. Even if the target to be detected is protected, the accuracy of the target recognition result obtained after correction can be guaranteed. Based on the embodiment shown in Figure 1, in one embodiment of the present disclosure, the second recognition result includes a predicted category corresponding to each target to be detected and a first confidence coefficient, and the first recognition result includes a reference category corresponding to each target to be detected. Step P150 specifically includes: The second recognition result is corrected based on a comparison result to obtain a target recognition result, which includes: If the first recognition result is the same as the predicted category corresponding to a current target to be detected, the predicted category corresponding to the current target to be detected is determined to be a correct result; If the first recognition result is different from the predicted category corresponding to the current target to be detected, it is determined that the predicted category corresponding to the current target to be detected is a confusing result, and it is determined, according to the first confidence coefficients of the predicted categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, the predicted category corresponding to the current target to be detected is corrected to be the same predicted category as the first recognition result. In this embodiment, the first recognition result includes the base category corresponding to each target to be detected. It should be understood that the first recognition result is a recognition result corresponding to the reference target, so the first recognition result can be used as a judgment basis to determine whether the second recognition result is correct. Therefore, the category provided in the first recognition result can be used as the reference category for each target to be detected. A lane guidance sign used as a reference target is an example.The first recognition result includes the lane type of each lane in the lane marking, such as a left-turn lane, a straight-through lane, and a right-turn lane, and the relative positions between the lanes, such as the left-turn lane being located on the left side of the straight-through lane, and the like. Therefore, the lane type of each lane contained in the first recognition result can be used as the base category of each arrow in the arrow marking (i.e., the target to be detected) in the second recognition result. According to the comparison result, if the first recognition result is the same as the predicted category corresponding to the current target to be detected, it means that the predicted category corresponding to the current target to be detected is a correct recognition result and can be saved. It should be noted that the image to be detected may contain more than one target to be detected. For a case where there are multiple targets to be detected, the second recognition result may include the predicted categories corresponding to multiple (two or more) targets to be detected, the confidence coefficients, and the relative positions between the targets to be detected. Thus, when comparing the first recognition result with the second, the comparison may be carried out according to the relative positions in the sequence. If the first recognition result is different from the predicted category corresponding to the current target to be detected, that is, if the reference category, corresponding to the target to be detected, in the first recognition result is different from the predicted category of the target to be detected in the second recognition result, the predicted category corresponding to the current target to be detected may be determined as the confusing result, that is, the second recognition result must be corrected.For the current target to be detected, whose predicted category is determined to be a confusing result, it can be determined, according to the first confidence coefficients of the predicted categories corresponding to other targets to be detected adjacent to the current target to be detected and which are determined to be correct results, that the predicted category corresponding to the current target to be detected will be the second confidence coefficient of the first recognition result. Specifically, for the current target to be detected, whose prediction category is determined to be a confusing result, it may be determined whether the comparison results of other targets to be detected adjacent to the current target to be detected have the prediction categories determined as correct results. If so, it may be determined, based on the first confidence coefficients of the prediction categories corresponding to other targets to be detected and determined as correct results, that the predicted category corresponding to the current target to be detected will be the second confidence coefficient of the first recognition result. It should be understood that the second confidence coefficient may be the credibility to indicate that the first recognition result is correct.If the second confidence coefficient is higher, the credibility indicating that the predicted category will be the base category is higher, and vice versa. For example, three arrows of the lane guidance sign (i.e., the targets to be detected) 1, 2, and 3 are arranged in sequence. The prediction categories of arrow 1 and arrow 3 match the reference category of the corresponding first recognition result and are determined as correct results, while the prediction category corresponding to arrow 2 is different from the reference category of the corresponding first recognition result. In this way, it can be determined, based on the first confidence coefficients of the prediction categories of arrow 1 and arrow 3, that the prediction category corresponding to arrow 2 will be the second confidence coefficient of the base category in the first recognition result. It should be understood that if the similarity between the first recognition result and the second recognition result is greater, it indicates that the first recognition result is more reliable.Therefore, the credibility indicating that prediction categories determined as confusing outcomes will be the corresponding reference category is higher. In an example, for the current target to be detected, whose prediction category is determined to be the confusing result, the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined to be the correct results may be averaged to obtain the second confidence coefficient of the current target to be detected. Once the second confidence coefficient corresponding to the QQPPPn / cznz / zi / υιλι current target to be detected, the second confidence coefficient may be compared with a pre-set predetermined threshold. If the second confidence coefficient reaches the predetermined threshold, that is, if the second confidence coefficient is greater than or equal to the predetermined threshold, the predicted category corresponding to the current target to be detected may be corrected to the corresponding reference category; otherwise, the predicted category may be saved, and the second recognition result that has undergone the above correction is output as the target recognition result. Therefore, in the above embodiment, by determining, according to the first confidence coefficients of prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and which are determined as correct results, that the predicted category of the current target to be detected will be the second confidence coefficient of the reference category, and correcting, according to the second confidence coefficient, the predicted category corresponding to the current one to be detected, the rationality of the correction can be guaranteed, thereby ensuring the accuracy of the obtained target recognition result. Based on the above embodiments, in an embodiment of the present disclosure, the step (determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results, that the predicted category corresponding to the current target will be the second confidence coefficient of the first recognition result) includes: The first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined as correct results are averaged to obtain that the predicted category corresponding to the current target to be detected will be the second confidence coefficient of the first recognition result. In one embodiment of the present disclosure, the step (determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results, that the predicted category corresponding to the current target will be the QQPPPn / cznz / zi / υιλι second confidence coefficient of the first recognition result) includes: performing the weighted average operation according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results and the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the confusing results, and it is determined that the prediction category corresponding to the current target to be detected will be the second confidence coefficient of the first recognition result, wherein the weight of the first confidence coefficient of each prediction category determined as the correct result is greater than the weight of the first confidence coefficient of each prediction category determined as the confusing result. In this embodiment, not only can it be determined, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results, that the predicted category of the current target to be detected is the second confidence coefficient of the base category, but also the first confidence coefficients of the prediction categories that correspond to other adjacent targets to be detected are determined and are determined as the confusing results. Specifically, the weighted average operation may be performed on the first confidence coefficients of the prediction categories determined as correct results and the first confidence coefficients determined as confusing results, to determine that the predicted category corresponding to the current target to be detected will be the second confidence coefficient of the base category in the first recognition result. Those skilled in the art may adjust the corresponding weights according to prior experience. The weights of the first confidence coefficients of the prediction categories determined as correct results are greater than the weights of the first confidence coefficients of the prediction categories determined as confusing results, so as to ensure the reasonableness of the weight adjustment. Based on the above embodiments, in one embodiment of the present disclosure, the step (that the first recognition result is compared with the QQPPPn / cznz / zi / υιλι second recognition result) includes: If the image to be detected does not simultaneously contain the reference target and the target to be detected, the first recognition result or the second recognition result acquired first is compared with a second recognition result or a first recognition result acquired subsequently within a predetermined time interval. In this embodiment, when the image to be detected does not simultaneously contain the reference target and the target to be detected, the first recognition result or the second recognition result acquired first can be compared with the second recognition result or the first recognition result acquired subsequently within the predetermined time interval, so that it can be ensured that the comparison can also be carried out even when the reference target and the target to be detected do not appear simultaneously in the image to be detected.Meanwhile, by setting the predetermined time interval, the validity of the first recognition result or the second recognition result acquired first can be guaranteed, thereby avoiding a failure of the first recognition result or the second recognition result caused by a vehicle traveling over a long distance due to an extremely long time interval. It should be noted that those skilled in the art can set the predetermined time interval based on prior experience, such as 5 s and 10 s. Based on the technical solutions of the previous modality, a specific application scenario of the modality of the present description is presented below: Figure 2 is a block diagram of a system capable of applying the method of detecting targets in vehicle circulation provided according to an embodiment of the present description. Referring to Figure 2, the system includes a prior information extraction module, a target detection module, and a target arbitration module. The prior information extraction module is configured to recognize the reference target contained in the image to be detected and output the information. QQPPPn / cznz / zi / υιλι corresponding prior (i.e., the first recognition result). The target detection module is configured to recognize the target to be detected contained in the image to be detected and output a corresponding recognition result (i.e., the second recognition result). The target recognition module is configured to compare the prior information with the recognition result, so as to arbitrate (i.e., correct) the recognition result, and finally output a recognition result after arbitration, which is the target recognition result. Figure 3 is a flowchart of the labeling, training, and actual prediction algorithm for a method for detecting targets in moving vehicles according to one embodiment of the present disclosure. A description will now be provided, using the application of the method to a lane arrow marking scenario as an example. As shown in Figure 3, prior information refers to the information obtained before the algorithm is trained or predicted. In a guide arrow recognition scenario, prior information can come from multiple sources, including, but not limited to, information fused by one or more traffic lights, point cloud data, and lane indicators. Preferably, information including, but not limited to, the category, shape, position, and the like of a lane is extracted from the lane guidance markings of each road. The advantages are as follows: (1) The lane markings are usually located at a high location and are not easy to shield. (2) The lane markings have a high contrast with the background and are easily recognizable accurately. In the algorithm's labeling and training phase, two detectors must be configured, respectively, for lane markings (i.e., the reference target) and lane arrow markings (i.e., the target to be detected). Training is carried out in two rounds. In the first round, a lane guidance sign detector (detector 1) is obtained. In the labeling stage, a sample containing a lane guidance sign, which may be a video or an image, is examined according to road test data collected by a real vehicle, and then the lane guidance sign and each lane within the lane guidance sign are labeled, including the lane position and category, thereby obtaining a certain amount of labeled data. In the configuration stage, the sample and the corresponding labeled data are used for configuration. In one example, one or more target detection algorithms from the YOLO collection, R-CNN collection, or SSD collection may be used for configuration to obtain the lane guidance sign detector that can recognize various lane types. The second round refers to the first round, in which a lane arrow marking detector (detector 2) is obtained. The difference is that the labeled data from the second round corresponds to a lane sample from the road test data. Once the two detectors are obtained, they serve two purposes: In application 1, the two detectors can be used for automatic labeling. In application 2, the two detectors can be deployed in mass-produced vehicles to support underlying intelligent driving services. In the algorithm prediction phase, application II is taken as an example. The detectors are deployed inside a mass-produced vehicle. After the mass-produced vehicle drives on a road, when the system performs road lane recognition, the system inputs the collected image information into the two detectors respectively: The lane marking detector detects whether a lane guidance marking is present. If so, the detector simultaneously captures lane guidance information, including the number of lanes and a lane type (i.e., the baseline category). The lane arrow marking detector detects whether an arrow marking appears in the image. If so, the detector detects a category (i.e., the predicted category) of the arrow marking. It should be noted that there will be two cases in the image to be detected: (1) The image simultaneously contains a lane marking and the direction of travel sign on the road, which is called a synchronous information scenario. (2) The lane marking and the lane arrow marking do not appear in the same image. For example, the image of a lane marking appears first and the image of a lane guidance arrow appears later on the road after several seconds, or the image of the lane guidance arrow appears first and the image of the lane marking appears later after several seconds, which is called an asynchronous information scenario. In case (1), the information output by the two detectors is highly coherent in theory and they can echo each other.In case (2), the information emitted by the two detectors may have a variation, that is, the information does not correspond to the same road, so it must be analyzed thoroughly. The target recognition and arbitration processes are described using case (1) as an example. The difference between case (2) and case (1) is the order in which the prior information is obtained. The output information of the two detectors is aligned by setting a time window (i.e., a predetermined time interval). Within the time window, the obtained information first waits for the acquisition of other information. If no data is obtained beyond the time window, a time window will be recreated with the data obtained the next time. The entire detection process of case (1) is shown in Figure 4. In the following description, the lane marking detector is called the first detector, and the lane arrow marking detector is called the second detector. First, the first detector recognizes the position of the lane marking in the image and the lane information (the number of lanes, the relative positions of the lanes on the entire road, and the relative distance between lanes) in the lane guidance sign. As shown in Figure 5, information for five lanes is contained, and there are four types of guidance categories (i.e., A, B, C, and D). There is a relative distance between every two lanes, and the lanes are connected to form a reference map. The nodes in the reference map represent the lane categories, and the edge weights (connecting lines between nodes) represent the relative distances, thus forming the prior information of the road. Second, the second detector recognizes the roadway guidance arrow in the image, including information from the lane arrow markings of one or more lanes, and obtains the predicted category of each lane arrow marking and the relative positions and distances between the lane arrow markings. Using perspective restoration technology, a subgraph similar to that in Figure 5 is constructed to form the predicted roadway information. Third, the subgraph of predicted information is matched with the reference map of prior information, and two matching methods (i.e., subgraph matching algorithm) are provided for this scenario: (1) If a target vehicle's current driving lane arrow marking is not shielded by other vehicles, the lane arrow marking information can be used as a matching preference because the road guidance arrow of the current lane is in the front right, with a small degree of perspective modality, has a large proportion in the image, and is more easily recognized accurately. The left and right lanes of the current lane are matched, and then the lanes in the next layer are matched. Weighted summation is performed on each matching result, with the weights decreasing layer by layer from the current lane.Finally, the matching shape with the highest score is selected as the lane matching result. (2) If the current driving lane is protected, unprotected guide arrow markings are matched preferentially, which are searched for from the left and right lanes. Matching starts from the first unprotected lane found, which is the matching preference, and skips over protected lanes. Fourth, the base category of each node (lane) is obtained based on an isomorphic subgraph. Finally, the reference category and the predicted category are determined using a specific arbitration method to obtain a final category for each predicted target. An embodiment of the present disclosure also provides an arbitration method, as shown in Figure 6: (1) Traverse a set of detection targets (i.e., the first recognition result corresponding to at least one target to be detected); for each target (i.e., the target to be detected, which is a lane here): determine whether the predicted category is consistent with the base category: if the predicted category is consistent with the base category, consider the predicted category as a correct recognition, and consider the node as a correct node; and if not, consider the node as a confusing node that needs to be arbitrated. (2) Traverse each confusing node: Acquire a set of nodes (i.e., other targets to be detected adjacent to the current target to be detected) directly connected to the confusing node; conduct a vote at each node determined to be correct to determine whether the node's predicted category is correct. Specifically, the confidence coefficient (i.e., the first confidence coefficient) of each node is summed and averaged to obtain a confidence coefficient (i.e., the second confidence coefficient) that is regarded as the base category of the confusing node. (3) Set a threshold for the confidence coefficient, such as 0.8; when the confidence coefficient is greater than the threshold, determine that the category is the base category; otherwise, determine that the category is the predicted category. At the same time, a node with an updated category is considered a correct node, and an average confidence coefficient of the neighboring nodes after voting is used as the node's confidence coefficient. (4) When the determination has been completed at all confusing nodes, ending the recognition. Optionally, the voting nodes in (2) may include confusing nodes that are not arbitrated or confusing nodes that are being arbitrated. At this point, the weighted average method is used to assign a higher weight to the correct nodes and a lower weight to the confusing nodes. Simply put, for the road sign recognition scenario, guidance information for other lanes can be accurately calculated based on the relative positions and distances between the current lane and a vehicle's guidance information. Based on this, the intelligent driving system can develop a decision plan based on the accurately recognized lane guidance to ensure the efficient implementation of its downstream applications. Next, a vehicle moving target detection apparatus provided according to an embodiment of the present disclosure will be described with reference to the accompanying drawings. Figure 7 is a schematic block diagram of a vehicle moving target detection apparatus according to an embodiment of the present disclosure. As shown in Figure 7, the vehicle moving target detection apparatus includes: an acquisition module 710, configured to acquire an image to be detected from a target vehicle in a traffic process; a first detection module 720, configured to input the image to be detected into a first preconfigured detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; a second detection module 730, configured to input the image to be detected into a second preconfigured detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship; a comparison module 740, configured to compare the first recognition result with the second recognition result; and a processing module 750, configured to correct the second recognition result according to a comparison result to obtain a target recognition result. Furthermore, the second recognition result includes a predicted category corresponding to each target to be detected, and a first confidence coefficient; the first recognition result includes a reference category corresponding to each target to be detected; and the processing module 750 is configured to: if the first recognition result is equal to the predicted category corresponding to a current target to be detected, determine that the predicted category corresponding to the current target to be detected is a correct result;if the first recognition result is different from the predicted category corresponding to the current target to be detected, determining that the predicted category corresponding to the current target to be detected is a confused result, and determining, according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined to be correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, correcting the predicted category corresponding to the current target to be detected to be the same predicted category as the first recognition result.; In addition, the processing module 750 is configured to: average the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and that are determined as correct results to obtain that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result. Furthermore, the processing module 750 is configured to: perform the weighted summation operation according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results and the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the confusing results, and determine that the prediction category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result, wherein a weight of the first confidence coefficient of each prediction category determined as a correct result is greater than the weight of the first confidence coefficient of each prediction category determined as a confusing result. Furthermore, the processing module 750 is also configured to: acquire first sample data containing the reference target and labeled data corresponding to the first sample data from the road test data; configure a pre-constructed first detector according to the first sample data and the labeled data corresponding to the first sample data, so as to enable the first detector to correctly output the first recognition result for the reference target; acquire a second data sample containing the target to be detected and labeled data corresponding to the second sample of road test data;and configure a second pre-constructed detector according to the second data sample and the labeled data corresponding to the second data sample, so as to enable the second detector to correctly output the second recognition result for the target to be detected.; In addition, the reference target is a lane guidance marking, and the target to be detected is a lane arrow marking. QQPPPn / cznz / zi / υιλι Furthermore, the comparison module 740 is configured to: if the image to be detected does not simultaneously contain the reference target and the target to be detected, compare the first recognition result or the second recognition result acquired first with a second recognition result or a first recognition result acquired subsequently within a predetermined time interval. It should be noted that the above explanation of the mode of the vehicle-moving target detection method is also applicable to the vehicle-moving target detection apparatus of this mode, and repeated descriptions will be omitted here. The traveling vehicle target detection apparatus provided according to the embodiment of the present disclosure may acquire an image to be detected from a target vehicle in a traveling process; input the image to be detected to a first preset detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; and input the image to be detected to a second preset detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship. Thus, the first recognition result and the second recognition result also have a certain correspondence relationship.The first recognition result is compared with the second recognition result for mutual verification, such that the second recognition result is corrected according to a comparison result to obtain the accuracy of a recognition result for the target. Thus, by comparing the first recognition result corresponding to the reference target with the second recognition result corresponding to the target to be detected, and correcting them, misrecognition or non-detection is avoided. Figure 8 is a schematic structural diagram of an electronic device provided by one embodiment of the present disclosure. The electronic device may include: a memory 601, a processor 602, and a computer program stored in the memory 601 and executable on the processor 602. The processor 602, when executing the programs, implements the method of detecting moving targets of the vehicles provided for in the previous mode. In addition, the electronic device also includes: a communication interface 603, configured to achieve communication between the memory 601 and the processor 602. Memory 601 is configured to store computer programs executable on processor 602. Memory 601 may include high-speed random access memory (RAM), or non-volatile memory, for example, at least one disk memory. If the memory 601, the processor 602, and the communication interface 603 are implemented independently, the communication interface 603, the memory 601, and the processor 602 may be connected to each other via a bus and complete the communication between them. The bus may be an Industrial Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industrial Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only a thick line is used in Figure 6, but this does not mean that there is only one bus or one type of bus. Optionally, in terms of specific implementation, if the memory 601, the processor 602 and the communication interface 603 are integrated on a chip, the memory 601, the processor 602 and the communication interface 603 may communicate with each other through internal interfaces. The processor 602 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIO), or is configured to implement one or more integrated circuits of the embodiment of the present disclosure. An embodiment of the present disclosure further provides a computer-readable storage medium that stores a computer program. The program, when executed by a processor, implements the above method of target detection in vehicle traffic. In the description of the present specification, reference terms such as one embodiment, some embodiments, examples, specific examples, or some examples mean that the specific features, structures, materials, or characteristics described in combination with the embodiments or examples are included in at least one embodiment or example of the present description. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined into one or N embodiments or examples in a suitable manner. Furthermore, those skilled in the art can connect and combine the different embodiments or examples and the characteristics of the different embodiments or examples described in this specification without contradicting each other. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply a relative importance or number of technical features indicated. Therefore, the features defined by "first" and "second" may explicitly instruct or implicitly include at least one feature. In the description of the present invention, unless expressly specified otherwise, the meaning of "N" is "at least two," as in "two" and "three." Any process or method description in the flowchart or otherwise described herein may be understood as a module, segment, or portion of code that includes one or more executable instructions for implementing specific functions or logical steps of the process. The scope of preferred embodiments of the present disclosure includes additional implementations, which may not be in the order shown or discussed, including the embodiment of functions substantially simultaneously or in reverse order depending on the functions involved. This should be understood by those skilled in the art to which the embodiments of the present disclosure pertain. The logic and / or steps depicted in flowcharts or otherwise described herein, for example, may be thought of as an ordered list of executable instructions for implementing logical functions, may be specifically implemented on any computer-readable medium for use with, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system capable of acquiring instructions from the instruction execution system, apparatus, or device and executing them). In terms of this specification, a computer-readable medium may be any apparatus capable of containing, storing, communicating, propagating, or transporting the program for use by or in conjunction with an instruction execution system, apparatus, or device.More specific examples (non-exhaustive list) of computer-readable media include: an electrical connecting piece (an electronic apparatus) with one or N wires, a portable computer disk cartridge (a magnetic apparatus), a random access memory (RAM), a read-only memory (ROM), an editable erasable read-only memory (EPROM or flash memory), a fiber optic apparatus, and a portable compact disc read-only memory (CDROM).Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program can be printed, since optical scanning can be performed, for example, through the paper or other medium, and then editing and interpretation are carried out; if necessary, other suitable means are used to carry out processing, in order to obtain the program electronically; and then the program is stored in a computer memory. It should be understood that each part of the present description may be implemented by hardware, software, firmware, or a combination thereof. In the above implementation modes, N steps or methods may be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another implementation, it may be implemented by any one or a combination of the following technologies known in the art: discrete logic circuits with logic gate circuits used to perform logic functions for digital signals, application-specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. Those of ordinary skill in the art can understand that the implementation of all or part of the method steps of the above embodiments can be completed by a program that instructs the relevant hardware. The program can be stored on a computer-readable storage medium. The program can include one or a combination of the steps of the method embodiment. Furthermore, all functional units in all embodiments of the present invention may be integrated into a processing module, or each unit may exist physically alone, or two or more units may be integrated into a module. The above integrated modules may be implemented in the form of hardware, or they may be implemented in the form of software functional modules. The integrated module, if implemented as a software functional unit and sold or used as a stand-alone product, may be stored on a computer-readable storage medium. The aforementioned storage medium may be a read-only memory, a magnetic disk, an optical disk, and the like. Although embodiments of the present disclosure have been shown and described above, it may be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Changes, modifications, substitutions, and variations to the aforementioned embodiments may be made by those skilled in the art within the scope of the present disclosure.
Claims
1. A method for detecting targets in vehicle traffic, comprising the steps of: acquiring an image to be detected from a target vehicle in a traffic process; inputting the image to be detected to a pre-configured first detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; inputting the image to be detected to a pre-configured second detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship; comparing the first recognition result with the second recognition result; and correcting the second recognition result according to a comparison result to obtain a target recognition result.
2. The method according to claim 1, wherein the second recognition result comprises a predicted category corresponding to each target to be detected and a first confidence coefficient, and the first recognition result comprises a reference category corresponding to each target to be detected; correcting the second recognition result according to a comparison result to obtain a recognition result of the target comprises: if the first recognition result is the same as the predicted category corresponding to a current target to be detected, determining that the predicted category corresponding to the current target to be detected is a correct result;if the first recognition result is different from the predicted category corresponding to the current target to be detected, determining that the predicted category corresponding to the current target to be detected is a confused result, and determining, according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and which are determined to be the correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, correcting the predicted category corresponding to the current target to be detected to be the corresponding reference category in the first recognition result.; 3. The method according to claim 2, wherein determining, according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined as correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result comprises: averaging the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined as correct results to obtain that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result.
4. The method according to claim 2, wherein determining, according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined as correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result comprises: performing the weighted average operation according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined as correct results and the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and determined as confusing results,and determining that the prediction category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result, wherein a weight of the first confidence coefficient of each prediction category determined as a correct result is greater than a weight of the first confidence coefficient of each prediction category determined as an unclear result.
5. The method according to claim 2, further comprising: acquiring first sample data containing the reference target and labeled data corresponding to the first sample data from the road test data; configuring a first pre-built detector based on the first sample data and the labeled data corresponding to the first sample data, so that the first detector correctly outputs the first recognition result of the reference target; acquiring second sample data containing the target to be detected and labeled data corresponding to the second sample data from the road test data;and configure a second pre-built detector based on the second sample data and the labeled data corresponding to the second sample data, so that the second detector can correctly output the second recognition result of the target to be detected.; 6. The method according to any one of claims 1 to 5, wherein the reference target is a lane marking, and the target to be detected is a lane arrow marking.
7. The method according to any one of claims 1 to 5, wherein comparing the first recognition result with the second recognition result comprises: if the image to be detected does not simultaneously contain the reference target and the target to be detected, comparing the first recognition result or the second recognition result acquired first with a second recognition result or a first recognition result acquired subsequently within a predetermined time interval.
8. A vehicle-moving target detection apparatus, comprising: an acquisition module configured to acquire an image to be detected from a target vehicle in a moving process; a first detection module configured to input the image to be detected to a first preconfigured detector, such that the first detector outputs a first recognition result of a reference target contained in the image to be detected; a second detection module configured to input the image to be detected to a second preconfigured detector, such that the second detector outputs a second recognition result of at least one target to be detected contained in the image to be detected, each target to be detected and the reference target having a correspondence relationship; a comparison module configured to compare the first recognition result with the second recognition result;and a processing module, configured to correct the second recognition result according to a comparison result to obtain a target recognition result.; 9. The apparatus according to claim 8, wherein the second recognition result comprises a predicted category corresponding to each target to be detected and a first confidence coefficient, and the first recognition result comprises a reference category corresponding to each target to be detected; the processing module is configured to: if the first recognition result is the same as the predicted category corresponding to a current target to be detected, determine that the predicted category corresponding to the current target to be detected is a correct result;if the first recognition result is different from the predicted category corresponding to the current target to be detected, determining that the predicted category corresponding to the current target to be detected is a confused result, and determining, according to the first confidence coefficients of the prediction categories corresponding to other targets to be detected adjacent to the current target to be detected and which are determined to be correct results, that the predicted category corresponding to the current target to be detected is a second confidence coefficient of the first recognition result; and if the second confidence coefficient reaches a predetermined threshold, correcting the predicted category corresponding to the current target to be detected so that QQPPPn / cznz / zi / υιλι is the corresponding reference category in the first recognition result.; 10. The apparatus according to claim 9, wherein the processing module is configured to: average the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as correct results to obtain that the predicted category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result.
11. The apparatus according to claim 9, wherein the processing module is configured to: perform the weighted average operation according to the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the correct results and the first confidence coefficients of the prediction categories that correspond to other targets to be detected adjacent to the current target to be detected and are determined as the confusing results, and determine that the prediction category corresponding to the current target to be detected is the second confidence coefficient of the first recognition result, wherein a weight of the first confidence coefficient of each prediction category determined as a correct result is greater than a weight of the first confidence coefficient of each prediction category determined as a confusing result.
12. The apparatus according to claim 9, wherein the processing module is configured to: acquire first sample data containing the reference target and labeled data corresponding to the first sample data from the road test data; configure a first pre-constructed detector based on the first sample data and the labeled data corresponding to the first sample data, so that the first detector can correctly output the first recognition result of the reference target; acquire second sample data containing the target to be detected and labeled data corresponding to the second sample data from the road test data;and configure a second pre-built detector based on the data of the second sample and the labeled data corresponding to the data of the second sample, so that the second detector can correctly output the second recognition result of the target to be detected.; 13. The apparatus according to any one of claims 8 to 12, wherein the reference target is a lane marking, and the target to be detected is a lane arrow marking.
14. The apparatus according to any one of claims 8 to 12, wherein the comparison module is configured to: if the image to be detected does not simultaneously contain the reference target and the target to be detected, compare the first recognition result or the second recognition result acquired at first with a second recognition result or a first recognition result acquired subsequently within a predetermined time interval.
15. An electronic device, comprising a memory, a processor and a computer program stored in the memory and operable on the processor, wherein the processor executes the program to implement the method of detecting moving vehicle targets according to any one of claims 1 to 7.
16. A computer-readable storage medium storing a computer program, wherein the program, when executed by a processor, implements the method of detecting moving vehicle targets according to any one of claims 1 to 7.