Vehicle identification result error detection method, electronic device and computer readable storage medium

Abstract

The present application discloses a vehicle recognition result error detection method, an electronic device and a computer-readable storage medium, the method comprising: obtaining a current captured image through a target acquisition device set in a preset area, the current captured image including target vehicle recognition information; verifying the target vehicle recognition information in the current captured image according to the acquired historical vehicle recognition information, and obtaining a first recognition confidence of the target vehicle recognition information; obtaining a second recognition confidence of the target acquisition device, the second recognition confidence representing the probability that the vehicle recognition information in the historical captured image captured by the target acquisition device is correctly recognized, and the acquisition time of the historical captured image is earlier than the acquisition time of the current captured image; determining whether the target vehicle recognition information is recognized incorrectly according to the first recognition confidence and the second recognition confidence. In this way, the accuracy of vehicle recognition result error detection can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle identification, and in particular to a vehicle identification result error detection method, electronic equipment and computer-readable storage medium. Background Art

[0002] With the continuous advancement of transportation system technology, the use of vehicle identification has become more and more extensive, including vehicle trajectory detection, traffic flow analysis, and vehicle set construction. Vehicle identification mainly obtains key information about the vehicle in the image through image analysis. However, due to the influence of environmental factors and vehicle specifications, the results of vehicle identification are still prone to errors.

[0003] At present, the error detection of vehicle recognition results is mainly carried out through manual comparison, which compares the recognition results of the collected images with the registration information to determine whether there are errors in the current vehicle recognition results. If there are errors, they are corrected. However, the error detection process of manual comparison is inefficient and time-consuming.

[0004] Therefore, there is an urgent need for a vehicle identification error detection method to analyze and manage the vehicle identification results. Summary of the invention

[0005] The main technical problem solved by the present application is to provide a vehicle identification result error detection method, an electronic device and a computer-readable storage medium, which can improve the error detection accuracy of vehicle identification results.

[0006] To solve the above technical problems, a technical solution adopted in the present application is: a vehicle recognition result error detection method is provided, and the vehicle recognition result error detection method is applied to an error detection device, and the method includes: obtaining a current captured image through a target acquisition device set in a preset area, and the current captured image includes target vehicle identification information; verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain a first recognition confidence of the target vehicle identification information; obtaining a second recognition confidence of the target acquisition device, and the second recognition confidence represents the probability that the vehicle identification information in the historical captured image captured by the target acquisition device is correctly identified, and the acquisition time of the historical captured image is earlier than the acquisition time of the current captured image; determining the target recognition confidence of the target vehicle identification information according to the first recognition confidence and the second recognition confidence; in response to the target recognition confidence being less than a confidence threshold, determining that the target vehicle identification information is recognized incorrectly.

[0007] In order to solve the above technical problems, another technical solution adopted in the present application is: to provide an electronic device, including a memory and a processor, the memory storing program instructions, and the processor calling the program instructions from the memory to execute the above vehicle identification result error detection method.

[0008] In order to solve the above technical problems, another technical solution adopted by the present application is: providing a computer-readable storage medium including program data stored therein, and the program data is used to implement the above vehicle identification result error detection method when executed by a processor.

[0009] Compared with the above scheme, the conventional method of determining whether there is a recognition error in the target vehicle identification information of the current captured image by the law of time and space transfer, which leads to the problem of low error detection accuracy, the present application proposes a vehicle identification result error detection method, which includes: obtaining the current captured image by a target acquisition device set in a preset area, and the current captured image includes the target vehicle identification information; verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information, and obtaining the first recognition confidence of the target vehicle identification information; obtaining the second recognition confidence of the target acquisition device, and the second recognition confidence represents the probability that the vehicle identification information in the historical captured image captured by the target acquisition device is correctly identified, and the acquisition time of the historical captured image is earlier than the acquisition time of the current captured image; determining the target recognition confidence of the target vehicle identification information according to the first recognition confidence and the second recognition confidence; in response to the target recognition confidence being less than the confidence threshold, determining that the target vehicle identification information is recognized incorrectly. Thus, the target acquisition device combining the historical vehicle identification information and the current captured image respectively verifies the target vehicle identification information of the current captured image, and performs error detection processing on the target vehicle identification information in a multi-dimensional verification manner, thereby improving the accuracy of vehicle recognition result error detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0011] Figure 1 It is a flowchart of an exemplary embodiment of a vehicle identification result error detection method shown in the present application;

[0012] Figure 2 yes Figure 1 A flow chart of an exemplary embodiment of step S130 in the vehicle recognition result error detection method is shown;

[0013] Figure 3 yes Figure 1 A flow chart of an exemplary embodiment of step S120 in the vehicle recognition result error detection method is shown;

[0014] Figure 4 yes Figure 1 A flow chart of an exemplary embodiment of the vehicle recognition result error detection method after step S150 is shown;

[0015] Figure 5 is a schematic diagram of an interactive interface shown in this application;

[0016] Figure 6 is another schematic diagram of the interactive interface shown in the present application;

[0017] Figure 7 is another schematic diagram of the interactive interface shown in the present application;

[0018] Figure 8 is another schematic diagram of the interactive interface shown in the present application;

[0019] Fig. 9 is another schematic diagram of the interactive interface shown in the present application;

[0020] Fig.10 is another schematic diagram of the interactive interface shown in the present application;

[0021] Fig.11 It is a specific flow chart of an exemplary embodiment of a vehicle identification result error detection method shown in the present application;

[0022] Fig.12 yes Fig.11 A flow chart of an exemplary embodiment of step S530 in the vehicle recognition result error detection method is shown;

[0023] Fig.13 is a structural schematic diagram of an exemplary embodiment of an error detection device shown in the present application;

[0024] Fig.14 It is a structural schematic diagram of an embodiment of an electronic device provided by the present application;

[0025] Fig.15 It is a structural schematic diagram of an embodiment of a computer-readable storage medium provided by the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It will be appreciated that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some but not all structures related to the present application are shown in the drawings. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the art without making creative work are within the scope of protection of the present application.

[0027] First of all, it should be noted that in order to better manage vehicles, the images including vehicles collected by the acquisition equipment can be analyzed to obtain information about each vehicle, and aggregated in the same set to form a vehicle set. Vehicle sets can help analyze vehicle trajectories and accompanying relationships, so the quality of vehicle sets directly affects the good support for subsequent applications. The vehicle set mainly performs image analysis on the collected images, and aggregates the collected images of the same license plate obtained by analysis into a set. However, due to problems such as acquisition angle and clarity, there will be recognition deviations in the same set, resulting in multiple different vehicles in the same set, and the same vehicle may exist in multiple sets. Therefore, analyzing and managing abnormal sets is an urgent problem to be solved.

[0028] At present, vehicle collection management is mainly based on whether the average speed of vehicles passing through adjacent collection devices is reasonable, or whether the vehicles passing through adjacent collection devices are reasonable based on the feasible roads between the collection devices, thereby determining whether there are errors in the vehicle identification results. However, there are many similar vehicles in the preset area, and the accuracy of error detection based on a single spatiotemporal law is low.

[0029] Based on this, the present application proposes a vehicle identification result error detection method, an electronic device and a computer-readable storage medium, which perform error detection of vehicle identification results from multiple dimensions, greatly improving the accuracy of vehicle identification error detection.

[0030] The execution subject of the vehicle identification result error detection method may be a terminal device or a server or other processing device, wherein the terminal device may be a user equipment (UE), a computer, a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. The execution subject of the vehicle identification result error detection method may also be an error detection device. In some possible implementations, the vehicle identification result error detection method may be implemented by a processor calling a computer-readable instruction stored in a memory.

[0031] Specifically, see Figure 1 , Figure 1 It is a flowchart of an exemplary embodiment of a vehicle identification result error detection method shown in the present application.

[0032] S110: Acquire a current captured image through a target capture device set in a preset area, wherein the current captured image includes target vehicle identification information.

[0033] The preset area may be a fixed geographical area. For example, the preset area may be a pre-divided area, a city, etc. In practical applications, the preset area may be divided according to the moving speed and moving time of the vehicle, or may be divided according to actual needs.

[0034] The target acquisition device can be used to acquire images including the target vehicle. Exemplarily, the target acquisition device can be an image acquisition device, a video acquisition device, etc. The target acquisition device is set in a preset area. It can be understood that the error detection device can perform overall control and management on the acquisition device in the preset area.

[0035] The current acquired image may be an acquired image that needs to be error-checked. Exemplarily, the current acquired image may be acquired by a target acquisition device. In addition, an image whose image quality is greater than a preset quality threshold among multiple images acquired by the current acquired image may be used as the current acquired image.

[0036] The target vehicle identification information can perform vehicle identification processing on the current collected image to obtain the target vehicle identification information, and mark the target vehicle identification information in the current collected image. If the current collected image includes vehicle identification information of multiple vehicles, one of them can be determined in turn as the target vehicle identification information. In other embodiments, when analyzing and managing the vehicle set of the target vehicle, the target acquisition device is the acquisition device in each subset of the vehicle set of the target vehicle, the current acquired image is the acquisition image in each subset of the vehicle set, and the target vehicle identification information is the identification information of the target vehicle determined from each acquired image in the vehicle set.

[0037] The error detection device acquires the current captured image through a target capture device arranged in a preset area, and performs vehicle recognition processing on the current captured image to obtain target vehicle recognition information in the current captured image.

[0038] S120: Verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain a first identification confidence level of the target vehicle identification information.

[0039] The historical vehicle identification information includes process information and result information from vehicle identification to vehicle restoration from the collected images collected in the preset area within the preset time period. The process information includes the process of restoring from the wrong identification result to the correct identification result, etc.; the result information includes the identification result of the vehicle identification and the restoration result of the vehicle restoration, etc. Exemplarily, the historical vehicle identification information is obtained by performing vehicle identification processing on the historical collected images collected by each collection device in the preset area and restoring the vehicle identification information with errors. Furthermore, the historical vehicle identification information also includes vehicle trajectory information obtained by processing the vehicle identification information of each vehicle, trajectory information of vehicle-related objects, etc.

[0040] The first recognition confidence can be the probability that the target vehicle recognition information is correct obtained by verifying the target vehicle recognition information in the current captured image through the historical vehicle recognition information. For example, the first recognition confidence of the target vehicle recognition information of the current captured image can be determined through the vehicle trajectory information in the historical vehicle recognition information, or the first recognition confidence of the target vehicle recognition information of the current captured image can be determined through the trajectory information of the vehicle-related object, or the first recognition confidence of the target vehicle recognition information of the current captured image can be determined through the association relationship between the target vehicle and each acquisition device in the preset area; or the first recognition confidence of the target vehicle recognition information of the current captured image can be determined through the historical wrong recognition information and the historical correct recognition information of the target vehicle, or the first recognition confidence of the target vehicle recognition information can be determined through the frequency of occurrence of the target vehicle, etc. Of course, the above methods can also be combined in any way to determine the first recognition confidence, where the combination method can be a weighted combination. As an example, the first sub-recognition confidence of the target vehicle identification information of the current acquired image is determined by the vehicle trajectory information in the historical vehicle identification information, the second sub-recognition confidence of the target vehicle identification information of the current acquired image is determined by the trajectory information of the vehicle-associated object, and the third sub-recognition confidence of the target vehicle identification information of the current acquired image is determined by the association between the target vehicle and each acquisition device in the preset area; the fourth sub-recognition confidence of the target vehicle identification information of the current acquired image is determined by the historical incorrect identification information and correct identification information of the target vehicle, and the fifth sub-recognition confidence of the target vehicle identification information is determined by the frequency of occurrence of the target vehicle. Finally, the first sub-recognition confidence, the second sub-recognition confidence, the third sub-recognition confidence, the fourth sub-recognition confidence and the fifth sub-recognition confidence are weighted to obtain the first recognition confidence.

[0041] S130: Obtain a second recognition confidence of the target acquisition device, where the second recognition confidence represents the probability that the vehicle recognition information in the historical acquisition image acquired by the target acquisition device is correctly identified, and the acquisition time of the historical acquisition image is earlier than the acquisition time of the current acquisition image.

[0042] The second recognition confidence may be the probability of identifying the correct target vehicle identification information through the current acquisition image acquired by the target acquisition device. Specifically, the second recognition confidence of the target acquisition device may be determined by the vehicle identification information of the historical acquisition image acquired by the target acquisition device. The acquisition time of the historical acquisition image is earlier than the acquisition time of the current acquisition image.

[0043] The historical collected images are collected by the target collection device. The vehicle identification information in the historical collected images can be the process information and result information of the historical collected images from vehicle identification to vehicle restoration obtained by identifying and processing the collected images collected by the target collection device within a preset time period.

[0044] The error detection device obtains the vehicle identification information of the historical acquisition image acquired by the target acquisition device, and determines the second recognition confidence of the target acquisition device according to the probability of correct recognition in the vehicle identification information.

[0045] S140: Determine a target recognition confidence of the target vehicle recognition information according to the first recognition confidence and the second recognition confidence.

[0046] The target recognition confidence represents the probability that the target vehicle recognition information is a correct recognition result. For example, the sum of the first recognition confidence and the second recognition confidence can be determined as the target recognition confidence. The first recognition confidence and the second recognition confidence can also be weighted and combined to obtain the target recognition confidence. The weighting method can be preset or continuously adjusted and determined according to historical recognition situations.

[0047] After the error detection device obtains the first recognition confidence and the second recognition confidence, it performs fusion processing on the first recognition confidence and the second recognition confidence to obtain the target recognition confidence of the target vehicle recognition information.

[0048] S150: In response to the target recognition confidence being less than the confidence threshold, determining that the target vehicle recognition information is incorrectly recognized.

[0049] The confidence threshold can be preset. For example, the confidence threshold can be set to 0.6, 0.7, 0.8 or 0.9, etc. When the target recognition confidence of the target vehicle recognition information is less than the confidence threshold, it means that the probability of the target vehicle recognition information being misidentified is high and the probability of being correctly identified is low. At this time, it can be determined that the target vehicle recognition information is misidentified.

[0050] It can be seen that the vehicle recognition result error detection method of the embodiment of the present application obtains the current captured image through a target acquisition device set in a preset area, and the current captured image includes the target vehicle recognition information; the target vehicle recognition information in the current captured image is verified and processed according to the acquired historical vehicle recognition information to obtain the first recognition confidence of the target vehicle recognition information; the second recognition confidence of the target acquisition device is obtained, and the second recognition confidence represents the probability that the vehicle recognition information in the historical captured image captured by the target acquisition device is correctly recognized, and the acquisition time of the historical captured image is earlier than the acquisition time of the current captured image; the target recognition confidence of the target vehicle recognition information is determined according to the first recognition confidence and the second recognition confidence; in response to the target recognition confidence being less than the confidence threshold, it is determined that the target vehicle recognition information is recognized incorrectly. In this way, the target acquisition device that combines the historical vehicle recognition information and the current captured image respectively verifies the target vehicle recognition information of the current captured image, and performs error detection processing on the target vehicle recognition information in a multi-dimensional verification manner, thereby improving the accuracy of vehicle recognition result error detection.

[0051] In other embodiments, a current captured image is acquired by a target capture device disposed in a preset area, the current captured image including target vehicle identification information; the target vehicle identification information in the current captured image is verified based on the acquired historical vehicle identification information to obtain a first recognition confidence of the target vehicle identification information; a second recognition confidence of the target capture device is acquired, the second recognition confidence representing the probability that the vehicle identification information in the historical captured images captured by the target capture device is correctly identified, the acquisition time of the historical captured images being earlier than the acquisition time of the current captured image; the target vehicle identification information in the current captured image is verified based on the acquired vehicle registration information to obtain a third recognition confidence of the target vehicle identification information; the target recognition confidence of the target vehicle identification information is determined based on the first recognition confidence, the second recognition confidence and the third recognition confidence; in response to the target recognition confidence being less than a confidence threshold, it is determined that the target vehicle identification information is incorrectly recognized.

[0052] Among them, the registration information may include the information of the vehicle registered at the vehicle administration office, ETC registration information, vehicle violation information, vehicle accident handling information, etc. The method of determining the third recognition confidence of the target vehicle identification information based on the registration information includes but is not limited to: verifying whether the target vehicle identification information meets the license plate making rules according to the license plate making rules issued by the vehicle administration office to determine the third recognition confidence of the target vehicle identification information; or determining the third recognition confidence of the target vehicle identification information according to whether the license plate number information registered by the vehicle at the vehicle administration office matches the corresponding license plate type; or determining whether the target vehicle identification information is consistent according to the relationship between the target vehicle registered at the vehicle administration office and its owner to determine the third recognition confidence of the target vehicle identification information; or determining whether the vehicle attributes of the target vehicle are consistent with the registered vehicle attributes according to the information registered at the vehicle administration office to determine the third recognition confidence; or determining the third recognition confidence according to whether the license plate number in the target vehicle identification information exists in the registration information of the vehicle administration office, etc. In other embodiments, the above determination methods can also be combined arbitrarily to obtain the third recognition confidence. As an example, multiple initial confidence levels can be determined based on the license plate making rules issued by the vehicle management office, the license plate number information registered with the vehicle management office and the corresponding license plate type, the relationship between the target vehicle registered with the vehicle management office and its owner, and whether the license plate number in the target vehicle identification information exists in the information registered with the vehicle management office. The multiple initial confidence levels can be combined to obtain a third recognition confidence level.

[0053] The first recognition confidence can be found in step S120 and will not be described in detail here.

[0054] The registration information is thus added to further determine whether there is an error in the target vehicle identification information, thereby making the accuracy of the vehicle identification result error detection higher.

[0055] Based on the above embodiments, the embodiments of the present application adopt Figure 2 The flowchart details how to obtain the second recognition confidence of the target acquisition device. Figure 2 , Figure 2 yes Figure 1 The flowchart of an exemplary embodiment of step S130 in the vehicle recognition result error detection method is shown. Specifically, step S130 obtains the second recognition confidence of the target acquisition device, and the second recognition confidence represents the probability that the vehicle recognition information in the historical acquisition image acquired by the target acquisition device is correctly identified. The process specifically includes the following steps:

[0056] S210: Analyze and process the factors of recognition error based on the vehicle recognition information in the historically collected images to determine the target recognition error factors of the target collection device.

[0057] The target recognition error factor may be an objective reason for the target acquisition device to cause the recognition error. Exemplarily, the vehicle recognition information of the historical acquisition image of the target acquisition device may be analyzed and processed to determine the target recognition error factor of the target acquisition device. The target recognition error factor includes at least one. For example, the target recognition error factor may include but is not limited to: the target acquisition device does not support the recognition of new energy license plates (only recognizes 7 digits for new energy license plates or cannot recognize the type of new energy license plates), the target acquisition device does not have enough fill light or is too exposed (the misrecognition rate of passing vehicle data at night is high), the construction problem of the target acquisition device (the overall misrecognition rate of passing vehicle data of the target acquisition device is high), the target acquisition device has a low recognition accuracy rate for easily misrecognized characters (for example, the misrecognition rate of character pairs such as 1-T, 2-Z is high), etc.

[0058] In some embodiments, the error recognition information of the vehicle identification information in the historical acquisition image is obtained; the initial factor analysis is performed on the error recognition information to obtain the recognition error factors in each error recognition information; in response to the recognition error factors including program error factors and device error factors, the probability of correct recognition under the program error factors in the historical vehicle identification information and the probability of correct recognition under the device error factors by the target acquisition device are determined; if the probability of correct recognition under the device error factors is less than the probability of correct recognition under the program error factors, the device error factors in the corresponding recognition error factors are determined as the target device error factors. In other embodiments, if the recognition error factors include program error factors, vehicle error factors and device error factors, the above method can also be used to determine the probability of correct recognition under each error factor, thereby determining whether the device error factor is the target recognition error factor of the target acquisition device. For example, in an erroneous recognition information, the recognition error of the captured image is caused by both insufficient fill light of the target acquisition device and the excessive height of the vehicle. In this case, the probability of a vehicle with the same height as the vehicle being correctly identified in the historical vehicle identification information can be determined. At this time, it is not required to collect the target acquisition device, and it can be determined by the images collected by each acquisition device in the preset area; and the probability that the image collected by the target acquisition device under the fill light condition is correctly identified; according to the size of the two probabilities, determine whether insufficient fill light of the target acquisition device is a factor of target recognition error of the target acquisition device.

[0059] In other embodiments, it is also possible to obtain the erroneous recognition information of the vehicle recognition information in the historically collected images; perform initial factor analysis on the erroneous recognition information to determine the recognition error factors in each erroneous recognition information; in response to the recognition error factor belonging to the device error factor, the corresponding recognition error factor is determined as the target recognition error factor of the target collection device, and the device error factor represents the recognition error of the target collection device. In this case, it is only necessary that the recognition error factor includes the device error factor, and the device error factor is determined as the target recognition error factor of the target collection device.

[0060] In order to further ensure the accuracy of the target recognition error factors, the error recognition information can also be subjected to initial factor analysis and processing to determine the recognition error factors in each error recognition information; obtain the number of the same device error factors included in each recognition error factor; if the number is greater than the preset number, the corresponding recognition error factor is determined as the target recognition error factor of the target acquisition device. That is, only when the number of recognition errors caused by a device error factor exceeds the preset value, it will be used as the target recognition error factor of the target acquisition device. This can prevent accidental recognition errors from mistakenly classifying the corresponding recognition error factor as the target recognition error factor of the target acquisition device. For example, in the vehicle recognition information of the historical acquisition image, the number of recognition errors in the historical acquisition image collected at night due to insufficient fill light of the target acquisition device is 100, then the insufficient fill light at night can be determined as the target recognition error factor of the target acquisition device.

[0061] S220: Determine a second recognition confidence level of the current captured image captured by the target capture device according to the target recognition error factor of the target capture device.

[0062] After determining the target recognition error factor of the target acquisition device, the error detection device determines whether the target acquisition device has the target recognition error factor when acquiring the current acquisition image. If so, the error detection device determines the target recognition error factor based on the probability of correct recognition of the target acquisition device under the target recognition error factor. Among them, there may be one or more target recognition error factors of the target acquisition device. In some embodiments, based on the probability of correct recognition of the target acquisition device under each target recognition error factor; the probability of correct recognition under each target recognition error factor is weighted to obtain the second recognition confidence of the target acquisition device.

[0063] In other embodiments, in response to the target acquisition device satisfying the target recognition error factor when acquiring the current acquisition image, the correct recognition probability of the target acquisition device when the target recognition error factor exists is determined based on the vehicle recognition information in the historical acquisition image; the second recognition confidence of the target acquisition device acquiring the current acquisition image is determined based on the correct recognition probability of the target acquisition device when the target recognition error factor exists. Among them, it is judged whether the target acquisition device satisfies each target recognition error factor when acquiring the current acquisition image, and the correct recognition probability of the target acquisition device under each satisfied target recognition error factor is determined based on the vehicle recognition information in the historical acquisition image; the second recognition confidence of the target acquisition device acquiring the current acquisition image is determined based on the correct recognition probability under each satisfied target recognition error factor. Thus, the objective reasons of the target acquisition device are analyzed through the vehicle recognition information of the historical acquisition image of the target acquisition device. Compared with the traditional analysis that only relies on the current acquisition image, the present application increases the confidence of the target acquisition device's device factor in judging the target vehicle recognition information, and improves the accuracy of vehicle recognition result error detection.

[0064] As an example, target recognition error factors of the target acquisition device include insufficient fill light at night, high misrecognition rate of 1-T character pairs, etc. At this time, if the target acquisition device only acquires the current acquired image at night, it means that the current acquired image meets the target recognition error factors of insufficient fill light at night; the number of correct vehicle identifications is determined from the vehicle identification information of the historical acquired images acquired by the target acquisition device at night, and the ratio between the number of correct vehicle identifications and the total number of historical acquired images acquired by the target acquisition device at night is determined as the correct recognition probability of the target acquisition device when the fill light at night is insufficient; finally, the correct recognition probability is determined as the second recognition confidence of the target acquisition device acquiring the current acquired image.

[0065] As another example, when the target recognition error factors of the target acquisition device include insufficient fill light at night, high misrecognition rate of 1-T character pairs, etc., if the target acquisition device not only acquires the current acquisition image at night, but also the license plate number in the target vehicle identification information includes 1 or T, it means that the current acquisition image satisfies the target recognition error factors of insufficient fill light at night and high misrecognition rate of 1-T character pairs; the number of correct vehicle identifications is determined from the vehicle identification information of the historical acquisition images acquired by the target acquisition device at night, and the number of correct vehicle identifications is divided by the total number of historical acquisition images acquired by the target acquisition device at night. The ratio between the number of acquired images that correctly identify 1 and T and the total number of acquired images that exist 1 or T is determined as the first recognition correct probability of the target acquisition device when the fill light is insufficient at night; the recognition results of the vehicle identification information that exist 1 or T in the historical acquired images of the target acquisition device are obtained, and the number of acquired images that correctly identify 1 and T and the total number of acquired images that exist 1 or T are obtained; the ratio between the number of acquired images that correctly identify 1 and T and the total number of acquired images that exist 1 or T is determined as the second recognition correct probability of the target acquisition device when recognizing character 1 or character T; the second recognition confidence of the target acquisition device acquiring the current acquired image is determined based on the first recognition correct probability and the second recognition correct probability.

[0066] In other embodiments, the step of obtaining the second recognition confidence of the target acquisition device may further include: obtaining vehicle identification information of historical acquisition images of adjacent acquisition devices of the target acquisition device; determining the number of spatiotemporal contradictions in the target acquisition device based on the vehicle identification information of the historical acquisition images of the adjacent acquisition devices; and determining the second recognition confidence of the target acquisition device based on the number of spatiotemporal contradictions in the target acquisition device and the total number of historical acquisition images of the target acquisition device. The ratio between the number of spatiotemporal contradictions and the total number of historical acquisition images may be determined as the second recognition confidence of the target acquisition device.

[0067] Based on the above embodiments, the embodiments of the present application adopt Figure 3 The flowchart details how to determine the first recognition confidence of the target vehicle identification information based on the historical vehicle identification information. Figure 3 , Figure 3 yes Figure 1 The flowchart of an exemplary embodiment of step S120 in the vehicle recognition result error detection method is shown. Specifically, step S120 verifies the target vehicle recognition information in the current captured image according to the acquired historical vehicle recognition information, and the process of obtaining the first recognition confidence of the target vehicle recognition information specifically includes the following steps:

[0068] S310: Analyze and process the historical vehicle identification information to obtain the collection equipment that has collected the target vehicle in the preset area.

[0069] The error detection device analyzes and processes the historical vehicle identification information of each collection device in the preset area, obtains the vehicles collected by each collection device, and determines the collection device that has collected the target vehicle. Specifically, the error detection device can count the frequency of each collection device collecting the target vehicle to obtain the association relationship between the collection device and the target vehicle. The collection devices in the preset area can also be clustered according to sub-areas, and the frequency of the target vehicle being collected by the collection devices in each sub-area is counted to obtain the association relationship between the target vehicle and the sub-area. The frequency can be counted by the number of collection times and / or the number of collection days, and the division of the sub-area includes but is not limited to geohash blocks.

[0070] S320: In response to the existence of the target acquisition device among the acquisition devices that have acquired the target vehicle, determining a first preset recognition confidence as a first recognition confidence of the target vehicle identification information.

[0071] The first preset recognition confidence represents the probability that the target vehicle identification information identified in the current collection image collected by the target collection device is correct when the target collection device exists in the historical vehicle identification information of the target vehicle. Exemplarily, the vehicle identification information in the historical collection image of the target collection device can be analyzed and processed to determine the association between the target vehicle and the target collection device, and the first preset recognition confidence can be set according to the association. Specifically, the first preset recognition confidence can be determined according to the number of times and / or days the target vehicle is collected under the target collection device. Exemplarily, the ratio between the number of times the target collection device collects the target vehicle and the total number of times the target collection device collects each vehicle is determined as the first preset recognition confidence. In other embodiments, the frequency region can be divided, and the first preset recognition confidence can be determined according to the frequency region. For example, when the frequency of the target vehicle being collected by the target collection device is less than 10, the first preset recognition confidence is set to 0.3; when the frequency of the target vehicle being collected by the target collection device is greater than or equal to 10 and less than or equal to 100, the first preset recognition confidence is set to 0.5; when the frequency of the target vehicle being collected by the target collection device is greater than 100, the first preset recognition confidence is set to 0.8.

[0072] In addition, the confidence of the association between the target vehicle and the sub-region can be calculated to obtain the regional confidence of the target vehicle in the sub-region where the target acquisition device is located, and the regional confidence is determined as the first recognition confidence of the target vehicle identification information. In other embodiments, the confidence between the target vehicle and the target acquisition device, and the confidence between the target vehicle and the sub-region where the target acquisition device is located can be obtained respectively, and then the confidences of the two are weighted and summed to obtain the first recognition confidence.

[0073] Furthermore, the historical vehicle identification information of the target vehicle can be further divided into time periods and dates for segmented analysis to obtain the association relationship between the target vehicle and the various collection devices in the preset area in different time periods, as well as the association relationship between the target vehicle and the various sub-areas in the preset area in different time periods; and then determine the first identification confidence of the target vehicle identification information based on the association relationship.

[0074] The error detection device can also perform a comprehensive analysis of the historical vehicle identification information and the target vehicle identification information of the target vehicle. If the frequency of the target vehicle under the target acquisition device is less than the frequency threshold, and the frequency under the previous acquisition device of the target acquisition device is greater than the frequency threshold, the confidence level of the target vehicle identification information is set. At this time, a smaller confidence level can be set.

[0075] In other embodiments, after determining that the target acquisition device exists among the acquisition devices that have acquired images of the target vehicle, the error detection device determines the first recognition confidence by combining the correct probability of the target acquisition device under the target recognition error factor.

[0076] Therefore, the probability of the target vehicle being collected by the target collection device is determined through the association relationship between the target vehicle and the collection device, combined with the vehicle's driving habits.

[0077] S330: In response to the target acquisition device not existing in the acquisition devices that have acquired the target vehicle, determining the second preset recognition confidence as the first recognition confidence of the target vehicle identification information; wherein the first preset recognition confidence is greater than the second preset recognition confidence.

[0078] The second preset recognition confidence represents the probability that the target vehicle identification information identified in the current collection image collected by the target collection device is correct when the target collection device does not exist in the historical vehicle identification information of the target vehicle. For example, if the target collection device has not collected the target vehicle, it means that the probability of the target vehicle appearing under the target collection device is small, and a smaller confidence can be set at this time. For example, the second preset recognition confidence can be set to less than 0.5, 0.2, 0.3 or 0.4, etc.

[0079] In other embodiments, the first recognition correct probability of the target vehicle recognition information is determined based on the historical vehicle recognition information of the target vehicle collected by the collection device in the historical vehicle recognition information; the second recognition correct probability under the target collection device is determined based on the historical vehicle recognition information of the target vehicle collected by the target collection device in the historical vehicle recognition information; the first recognition confidence of the target vehicle recognition information is determined based on the first recognition correct probability and the second recognition correct probability. Thus, the recognition status of the target vehicle itself is determined through the historical vehicle recognition information of the target vehicle, and the probability of the target vehicle being misidentified is judged, thereby determining the first recognition confidence of the target vehicle recognition information.

[0080] The first correct recognition probability is obtained by analyzing and processing the historical vehicle identification information of the target vehicle collected by all collection devices in the preset area to obtain the first correct recognition probability of the target vehicle identification information. For example, the number of correct recognitions in the historical vehicle identification information of the target vehicle collected by each collection device and the total number of historical vehicle identification information can be counted, and the ratio between the number of correct recognitions and the total number is determined as the first correct recognition probability.

[0081] The second recognition correct probability is obtained by analyzing and processing the historical vehicle identification information of the target vehicle collected by the target acquisition device to obtain the second recognition correct probability of the target vehicle identification information under the target acquisition device. For example, the number of correct identifications in the historical vehicle identification information of the target vehicle collected by the target acquisition device and the total number of historical vehicle identification information can be counted, and the ratio between the number of correct identifications and the total number is determined as the second recognition correct probability.

[0082] As an example, if the target vehicle is collected by each collection device in the preset area times, including The first recognition probability of the target vehicle is ; The target vehicle is collected by the target collection device , among which The target vehicle is misidentified, then the probability of correct identification of the second target vehicle is ) / ; The correct probability of the first recognition is and the second recognition correct probability The first recognition confidence of the target vehicle recognition information is obtained by combining the information. The combination can be performed in a weighted combination manner.

[0083] In addition, the preset area can also be divided into sub-areas, and the third recognition correct probability of the sub-area where the target collection device is located can be determined based on the historical vehicle identification information about the target vehicle collected by the collection device in each sub-area in the historical vehicle identification information; the first recognition confidence of the target vehicle identification information can be determined based on the first recognition correct probability, the second recognition correct probability and the third recognition correct probability.

[0084] Furthermore, the historical vehicle identification information of the target vehicle can be further divided into different time periods and dates for segmented analysis to obtain the fourth correct identification probability of the target vehicle regarding the historical vehicle identification information of the target vehicle collected by the collection device in different time periods; the fifth correct identification probability of the target vehicle regarding the historical vehicle identification information of the target vehicle collected by the target collection device in different time periods; the sixth correct identification probability of the target vehicle regarding the historical vehicle identification information of the target vehicle collected by the collection device in each sub-area in different time periods. The fourth correct identification probability, the fifth correct identification probability and the sixth correct identification probability are combined to obtain the first identification confidence. The first identification confidence can also be obtained by combining the first to sixth correct identification probabilities.

[0085] It should be noted that the confidence determination method of this embodiment can determine the first recognition confidence alone or in any combination. As an example of combination, the historical vehicle identification information is analyzed and processed to obtain the collection devices that have collected the target vehicle in the preset area; if the target collection device exists in the collection devices that have collected the target vehicle, the first preset recognition confidence is determined; if the target collection device does not exist in the collection devices that have collected the target vehicle, the second preset recognition confidence is determined; the first recognition correct probability of the target vehicle identification information is determined based on the historical vehicle identification information about the target vehicle collected by the collection device in the historical vehicle identification information; the second recognition correct probability under the target collection device is determined based on the historical vehicle identification information about the target vehicle collected by the target collection device in the historical vehicle identification information; the first recognition confidence is determined based on the first preset recognition confidence, the second preset recognition confidence, the first recognition correct probability and the second recognition correct probability.

[0086] Based on the above embodiments, the embodiments of the present application adopt Figure 4 The flowchart details how to restore the error recognition results. Figure 4 , Figure 4 yes Figure 1 The flowchart of an exemplary embodiment after step S150 in the vehicle recognition result error detection method is shown. Specifically, after the process of determining that the target vehicle recognition information is incorrectly recognized in response to the target recognition confidence being less than the confidence threshold in step S150, the following steps are specifically included:

[0087] S410: Restoring the target vehicle identification information according to the historical vehicle identification information to obtain a first restoration result and a first restoration confidence.

[0088] The first restoration result is mainly obtained by restoring the target vehicle identification information based on the historical vehicle identification information. Exemplarily, the target vehicle identification information can be restored based on the vehicle trajectory information in the historical vehicle identification information to obtain the first restoration result; or the target vehicle identification information can be restored based on the trajectory information of the vehicle-related object of the target vehicle to obtain the first restoration result; or the target vehicle identification information can be restored based on the historical wrong identification information and the historical correct identification information of each vehicle to obtain the first restoration result. Among them, the method of restoring the target vehicle identification information based on the vehicle trajectory information specifically includes: analyzing the necessary points of each vehicle's trajectory, and determining the vehicle whose necessary points of the trajectory exist in the target collection device as the first restoration result; or restoring the target vehicle identification information based on the vehicle identification information in the nearby time and space to obtain the first restoration result of the target vehicle identification information.

[0089] The method of restoring the target vehicle identification information by using the historical misidentification information and the historical correct identification information of each vehicle specifically includes: determining the number of misidentifications of each vehicle as the target vehicle, and determining the first restoration result of the target vehicle according to the number of misidentifications. As an example, if the number of times vehicle N is misidentified as the target vehicle is 2, and the number of times vehicle V is misidentified as the target vehicle is 1, then vehicle N and vehicle V are determined as the first restoration results, and the confidence of vehicle N and vehicle V are calculated respectively, and the first restoration confidence of vehicle N can be determined as 2 / 3, and the first restoration confidence of vehicle V can be determined as 1 / 3.

[0090] It should be noted that the above methods may be combined in any way to obtain the first restoration result.

[0091] The first restoration confidence represents the probability that the first restoration result obtained by restoring the target vehicle identification information according to the historical vehicle identification information is a correct restoration result. The first restoration confidence can be pre-set according to the historical vehicle identification information. The first restoration result and the first restoration confidence can be one or more.

[0092] S420: Restoring the target vehicle identification information according to the acquired vehicle registration information to obtain a second restoration result and a second restoration confidence.

[0093] The second restoration result is mainly obtained by restoring the target vehicle identification information through the vehicle registration information. The vehicle registration information includes the registration information of the vehicle management office. For example, the target vehicle identification information can be restored according to the special license plate number to obtain the second restoration result; or the target vehicle identification information can be restored according to the relationship between the vehicle and the owner in the vehicle registration information to obtain the second restoration result, etc.

[0094] It should be noted that the above methods may be combined arbitrarily to obtain the second restoration result.

[0095] The second restoration confidence represents the probability that the first restoration result obtained by restoring the target vehicle identification information according to the acquired vehicle registration information is a correct restoration result. The second restoration confidence can be pre-set according to the vehicle registration information. The second restoration result and the second restoration confidence can be one or more.

[0096] S430: Determine a target restoration confidence of the target vehicle identification information according to the first restoration confidence and the second restoration confidence.

[0097] The error detection device combines the first restoration confidence and the second restoration confidence to obtain the target restoration confidence of the target vehicle identification information. In some embodiments, the first restoration confidence and the second restoration confidence can be weightedly combined regardless of whether the restoration results are consistent to obtain the target restoration confidence of the target vehicle identification information. In other embodiments, the restoration confidences of the same restoration results can also be weightedly combined to obtain the target restoration confidence of the target vehicle identification information under different restoration results.

[0098] S440: In response to the target restoration confidence being greater than a preset restoration confidence threshold, determining target vehicle restoration information of the target vehicle identification information according to the first restoration result and the second restoration result.

[0099] In some embodiments, regardless of whether the restoration results are consistent, the error detection device performs a weighted combination of the first restoration confidence and the second restoration confidence to obtain a target restoration confidence that is greater than a preset restoration confidence threshold. In this case, the restoration result with the highest confidence is selected from the first restoration result and the second restoration result as the target vehicle restoration information of the target vehicle identification information.

[0100] In other embodiments, when the error detection device performs weighted combination of the restoration confidences of the same restoration results to obtain the target restoration confidence of the target vehicle identification information under different restoration results, the error detection device then determines whether each target restoration confidence is greater than a preset restoration confidence threshold. If there is a target restoration confidence greater than the preset restoration confidence threshold, the restoration result corresponding to the maximum target restoration confidence is determined as the target vehicle restoration information of the target vehicle identification information.

[0101] Therefore, the target vehicle identification information is restored through a variety of restoration methods, which greatly improves the restoration accuracy.

[0102] The error detection device also provides an interactive interface, see Figure 5-Figure 7The user can select the target vehicle and the analysis time period in the interactive interface. In response to the user's selection operation, the error detection device displays the target recognition confidence of the target vehicle at each track point in the analysis time period in the interactive interface. The target recognition confidence of each track point is also the target recognition confidence of the target vehicle recognition information in the acquisition image acquired by each acquisition device. The user can adjust the displayed target recognition confidence in the interactive interface. The adjustment method can be to click on the acquisition image corresponding to each track point, and select whether the corresponding target vehicle recognition information is recognized incorrectly according to the acquisition image.

[0103] See also Figure 8-Figure 10 The interactive interface can also display other vehicle identification information where the target vehicle is mistakenly identified as other vehicles, that is, display other vehicle identification information where the restoration result is the target vehicle in the historical misidentification information of other vehicles. The target restoration confidence of other vehicles can also be displayed. The user can adjust the target restoration confidence of other vehicles displayed by clicking on the track point where other vehicles are restored to the target vehicle, opening the corresponding collected image, and determining whether the restoration result is correct based on the corresponding collected image. It also supports checking the correct restoration result and removing the wrong restoration result. Fig.10 ,The user can also select a specified area or a specified acquisition device, and select a specified time to display the restored data analysis of other vehicles, and finally determine the target trajectory of the target vehicle.

[0104] At the same time, the interactive interface can also display abnormal license plate data statistics, including the total number of abnormal license plates, the total number of audits, the total number of confirmed abnormalities, the total number of abnormal license plates yesterday, the total number of audits yesterday, and the total number of confirmed abnormalities yesterday, etc. Therefore, the target recognition confidence of each track point of the target vehicle can be displayed through the interactive interface, and the target restoration confidence of each track point of the restoration result of the suspected target vehicle can also be displayed, and users are supported to manually verify and add the original track.

[0105] In order to elaborate on the vehicle recognition result error detection method of the present application, Fig.11 The flowchart and Fig.12 The flowchart shown further illustrates it, and the details are as follows:

[0106] S510: Acquire the sensing related information and registration information of vehicles, vehicle-related objects, etc. in the preset area within the preset time period; including the set registration information of all vehicles and vehicle-related objects; wherein the trajectory information includes the vehicle front-end collection information, the trajectory information of the vehicle-related objects, the vehicle collection equipment information, the vehicle back-end secondary analysis information, the road network information, the ETC collection information, the IMSI collection information, the associated object collection information, etc. The full amount of vehicle collection information includes the vehicle registration information of the vehicle management office, the ETC registration information, the vehicle violation information, the vehicle accident handling information, the collection information, etc.

[0107] S520: Analyze the target vehicle identification information in the current captured image according to the registration information, and the analysis methods include but are not limited to: whether the license plate number in the target vehicle identification information meets the vehicle management office's license plate making rules, whether the license plate number matches the license plate type, the license plate number is not registered, the vehicle attributes in the target vehicle identification information are inconsistent with the vehicle registration information of the vehicle management office, whether the front-end and back-end parsing information are consistent, and the relationship between the target vehicle and the associated object is inconsistent with the relationship between the target vehicle and the owner registered by the vehicle management office. Different confidence levels can be assigned to the above analysis methods to obtain a third recognition confidence level;

[0108] S530: Analyze the target vehicle identification information in the current captured image according to the perception information, the perception information includes the historical vehicle identification information and the vehicle identification information in the historical captured images of the target acquisition device, and analyze the target vehicle identification information according to the historical vehicle identification information and the vehicle identification information in the historical captured images of the target acquisition device. The historical vehicle identification information includes the trajectory information, historical wrong identification information and historical correct identification information of the target vehicle, and determine the association relationship between the target vehicle and each sub-area in the preset area and assign an association relationship score based on the trajectory information of the target vehicle, the association relationship between the target vehicle and each acquisition device in the preset area and assign an association relationship score, and determine the first sub-identification confidence of the target acquisition device according to the association relationship score between the adjacent acquisition device and the target vehicle. Furthermore, segmented analysis can be performed according to time periods, dates, etc., to analyze the integral of the association relationship between the target vehicle and the target acquisition device, as well as the integral of the association relationship between the target vehicle and the sub-area where the target acquisition device is located; determine whether the target vehicle identification information has a time-space contradiction based on the trajectory information of the target vehicle to determine the second sub-identification confidence, determine the third sub-identification confidence based on whether the number of license plate appearances is abnormal, and determine the fourth sub-identification confidence based on whether the trajectory of the target vehicle is inconsistent with the trajectory of the associated object. According to historical erroneous identification information and historical correct identification information, determine the first identification correct probability of the target vehicle under each acquisition device in the preset area, the second identification correct probability of the target vehicle under the target acquisition device, and the third identification correct probability of the target vehicle in the sub-area where the target acquisition device is located. Finally, determine the first identification confidence of the target vehicle identification information based on the first sub-identification confidence, the second sub-identification confidence, the third sub-identification confidence, the fourth sub-identification confidence, the first identification correct probability, the second identification correct probability, and the third identification correct probability.

[0109] Analyze the vehicle identification information in the historical collected images of the target acquisition device, and determine the second recognition confidence of the target vehicle identification information in the current collected image collected by the target acquisition device. Specifically, determine the target recognition error factor of the target acquisition device based on the vehicle identification information in the historical collected images of the target acquisition device, and when the current collected image meets the target recognition error factor of the target acquisition device, analyze the recognition correct probability of the target acquisition device under the target recognition error factor; determine the second recognition confidence of the target vehicle identification information based on the recognition correct probability.

[0110] S540: Comprehensively analyze the first recognition confidence, the second recognition confidence and the third recognition confidence to obtain the target recognition confidence of the target vehicle recognition information, and restore the target vehicle recognition information whose target recognition confidence is less than the confidence threshold. Among them, the restoration method includes: correcting the license plate type based on high-confidence vehicle recognition information, correcting the license plate type based on special license plate numbers, correcting the license plate based on the relationship between the target vehicle and the associated object, correcting the license plate based on vehicle collection data, correcting the license plate based on nearby time and space, correcting the license plate based on the necessary points of the trajectory, and correcting the license plate based on the historical misidentification information of the vehicle and the restoration information of the misidentification, etc. The above restoration method can be calculated for confidence to obtain the target restoration confidence. If the target restoration confidence is greater than the preset restoration confidence threshold, it is considered that the target vehicle recognition information can be restored by the license plate, and the restoration result is output.

[0111] See also Fig.13 , Fig.13 It is a structural diagram of an exemplary embodiment of an error detection device shown in the present application. The error detection device 1300 includes an acquisition module 1301, a verification module 1302, a confidence acquisition module 1303, a confidence determination module 1304 and a judgment module 1305; the acquisition module 1301 is used to acquire a current acquisition image through a target acquisition device set in a preset area, and the current acquisition image includes target vehicle identification information; the verification module 1302 is used to verify the target vehicle identification information in the current acquisition image according to the acquired historical vehicle identification information, and obtain a first recognition confidence of the target vehicle identification information; the confidence acquisition module 1303 is used to acquire a second recognition confidence of the target acquisition device, and the second recognition confidence represents the probability that the vehicle identification information in the historical acquisition image acquired by the target acquisition device is correctly identified, and the acquisition time of the historical acquisition image is earlier than the acquisition time of the current acquisition image; the confidence determination module 1304 is used to determine the target recognition confidence of the target vehicle identification information according to the first recognition confidence and the second recognition confidence; the judgment module 1305 is used to determine that the target vehicle identification information is incorrectly identified in response to the target recognition confidence being less than the confidence threshold.

[0112] In the above scheme, the error detection device of the embodiment of the present application obtains the current captured image through the target acquisition device set in the preset area, and the current captured image includes the target vehicle identification information; verifies the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain the first recognition confidence of the target vehicle identification information; obtains the second recognition confidence of the target acquisition device, and the second recognition confidence represents the probability that the vehicle identification information in the historical captured image captured by the target acquisition device is correctly identified, and the acquisition time of the historical captured image is earlier than the acquisition time of the current captured image; determines the target recognition confidence of the target vehicle identification information according to the first recognition confidence and the second recognition confidence; in response to the target recognition confidence being less than the confidence threshold, it is determined that the target vehicle identification information is incorrectly identified. In this way, the target acquisition device combined with the historical vehicle identification information and the current captured image respectively verifies the target vehicle identification information of the current captured image, and performs error detection processing on the target vehicle identification information in a multi-dimensional verification manner, thereby improving the accuracy of error detection of vehicle identification results.

[0113] Among them, the functions of each module can be found in the embodiment of the vehicle identification result error detection method, which will not be repeated here.

[0114] In order to implement the vehicle recognition result error detection method of the above embodiment, the present application proposes another electronic device, which is specifically referred to as Fig.14 , Fig.14 It is a structural schematic diagram of an embodiment of an electronic device provided by the present application.

[0115] The electronic device 1400 includes a memory 1401 and a processor 1402 , wherein the memory 1401 and the processor 1402 are coupled.

[0116] The memory 1401 is used to store program data, and the processor 1402 is used to execute the program data to implement the vehicle identification result error detection method of the above embodiment.

[0117] In this embodiment, the processor 1402 may also be referred to as a CPU (Central Processing Unit). The processor 1402 may be an integrated circuit chip having signal processing capabilities. The processor 1402 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. A general-purpose processor may be a microprocessor or the processor 1402 may also be any conventional processor, etc.

[0118] The present application also provides a computer readable storage medium 1500, such as Fig.15As shown, the computer-readable storage medium 1500 is used to store program data 1501. When the program data 1501 is executed by the processor, it is used to implement the vehicle identification result error detection method in the method embodiment of the present application.

[0119] The method involved in the embodiment of the vehicle identification result error detection method of the present application, when implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a device, such as a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0120] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A vehicle recognition result error detection method, characterized in that: The vehicle recognition result error detection method is applied to an error detection device, and the method comprises: Acquire a current captured image by a target capture device disposed in a preset area, wherein the current captured image includes target vehicle identification information; Verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain a first identification confidence of the target vehicle identification information; The step of verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain a first recognition confidence of the target vehicle identification information includes: Analyze and process the historical vehicle identification information to obtain a collection device that has collected the target vehicle in the preset area; In response to the target acquisition device existing in the acquisition devices that have acquired the target vehicle, determining a first preset recognition confidence as a first recognition confidence of the target vehicle identification information; In response to the target acquisition device not existing in the acquisition devices that have acquired the target vehicle, determining the second preset recognition confidence as the first recognition confidence of the target vehicle identification information; Wherein, the first preset recognition confidence is greater than the second preset recognition confidence; Acquire a second recognition confidence of the target acquisition device, the second recognition confidence representing a probability that the vehicle recognition information in the historical acquisition image acquired by the target acquisition device is correctly identified, where the acquisition time of the historical acquisition image is earlier than the acquisition time of the current acquisition image; Determining a target recognition confidence of the target vehicle recognition information according to the first recognition confidence and the second recognition confidence; In response to the target recognition confidence being less than a confidence threshold, it is determined that the target vehicle recognition information is incorrectly recognized.

2. The vehicle recognition result error detection method according to claim 1, characterized in that: The step of obtaining a second recognition confidence of the target acquisition device, wherein the second recognition confidence represents the probability that the vehicle recognition information in the historical acquisition image acquired by the target acquisition device is correctly identified, includes: Performing an analysis of factors of recognition error based on the vehicle recognition information in the historically collected images to determine target recognition error factors of the target acquisition device; A second recognition confidence level of the current acquisition image acquired by the target acquisition device is determined according to a target recognition error factor of the target acquisition device.

3. The vehicle recognition result error detection method according to claim 2, characterized in that: The step of analyzing and processing the factors of recognition errors according to the vehicle recognition information in the historically collected images to determine the target recognition error factors of the target acquisition device comprises: Obtaining erroneous identification information of vehicle identification information in the historically collected image; Performing initial factor analysis on the error recognition information to determine the recognition error factors in each error recognition information; In response to the recognition error factor belonging to a device error factor, the corresponding recognition error factor is determined as a target recognition error factor of the target acquisition device, and the device error factor represents the recognition error of the target acquisition device.

4. The vehicle recognition result error detection method according to claim 2, characterized in that: The step of determining a second recognition confidence level of the current captured image captured by the target capture device according to a target recognition error factor of the target capture device comprises: In response to the target acquisition device satisfying the target recognition error factor when acquiring the current acquisition image, determining the recognition correctness probability of the target acquisition device when the target recognition error factor exists according to the vehicle recognition information in the historical acquisition image; The second recognition confidence of the current acquisition image acquired by the target acquisition device is determined according to the recognition correct probability when the target acquisition device has the target recognition error factor.

5. The vehicle recognition result error detection method according to claim 1, characterized in that: The step of verifying the target vehicle identification information in the current captured image according to the acquired historical vehicle identification information to obtain a first recognition confidence of the target vehicle identification information includes: Determine a first recognition correctness probability of the target vehicle identification information according to the historical vehicle identification information about the target vehicle collected by the collection device in the historical vehicle identification information; Determine a second recognition correct probability under the target acquisition device according to the historical vehicle identification information about the target vehicle acquired by the target acquisition device in the historical vehicle identification information; A first recognition confidence level of the target vehicle recognition information is determined according to the first recognition correct probability and the second recognition correct probability.

6. The vehicle recognition result error detection method according to claim 1, characterized in that: After the step of obtaining a second recognition confidence of the target acquisition device, wherein the second recognition confidence represents the probability that the vehicle recognition information in the historical acquisition image acquired by the target acquisition device is correctly identified, and the acquisition time of the historical acquisition image is earlier than the acquisition time of the current acquisition image, the method further includes: Verifying the target vehicle identification information in the current captured image according to the acquired vehicle registration information to obtain a third identification confidence level of the target vehicle identification information; The step of determining the target recognition confidence of the target vehicle recognition information according to the first recognition confidence and the second recognition confidence comprises: Determining a target recognition confidence of the target vehicle recognition information according to the first recognition confidence, the second recognition confidence, and the third recognition confidence; In response to the target recognition confidence being less than a confidence threshold, it is determined that the target vehicle recognition information is incorrectly recognized.

7. The vehicle recognition result error detection method according to claim 1, characterized in that: After the step of determining that the target vehicle identification information is incorrectly identified in response to the target identification confidence being less than a confidence threshold, the method further includes: Restoring the target vehicle identification information according to the historical vehicle identification information to obtain a first restoration result and a first restoration confidence; Performing restoration processing on the target vehicle identification information according to the acquired vehicle registration information to obtain a second restoration result and a second restoration confidence; Determining a target restoration confidence of the target vehicle identification information according to the first restoration confidence and the second restoration confidence; In response to the target restoration confidence being greater than a preset restoration confidence threshold, target vehicle restoration information of the target vehicle identification information is determined according to the first restoration result and the second restoration result.

8. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores program instructions, and the processor retrieves the program instructions from the memory to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: include: Program data is stored, and when the program data is executed by a processor, it is used to implement the method according to any one of claims 1 to 7.

Images