Traffic Image Recognition Result Discrimination Method, Device, Electronic Device and Program Product

Abstract

Embodiments of the present disclosure disclose a method, apparatus, electronic device, and program product for discriminating traffic image recognition results. The method includes: obtaining at least two traffic images, recognizing the at least two traffic images to obtain a traffic element data set composed of traffic element data included in the at least two traffic images; processing the traffic element data set to obtain a high-precision and / or standard-precision traffic element data set; performing non-noise discrimination on the high-precision and / or standard-precision traffic element data; and performing noise discrimination on the remaining high-precision and / or standard-precision traffic element data except for the non-noise data to obtain a discrimination result. This technical solution can improve the accuracy of high-precision and / or standard-precision traffic element data, thereby avoiding problems such as image recognition errors, position parsing errors, high-precision map data creation errors, and high-precision map data creation redundancy caused by directly using the collected data, thus greatly improving the data quality of high-precision map data.

Description

Technical Field

[0001] The present disclosure relates to the technical field of traffic data processing, and particularly relates to a method, an apparatus, an electronic device, and a program product for discriminating traffic image recognition results. Background Art

[0002] With the development of society and the progress of science and technology, the more secure and efficient autonomous driving technology has been recognized as one of the future directions of the automotive travel industry, and many autonomous driving behaviors rely on high-precision map data. In order to improve the update frequency of high-precision map data and reduce the update cost of high-precision map data, in the prior art, the data collected by traffic image acquisition vehicles is used to update high-precision map data. However, due to the low accuracy of the collected data, errors may occur in image recognition and position parsing. If the collected data is directly used to update high-precision map data, problems such as incorrect creation and redundant creation of high-precision map data will occur, thereby reducing the quality of high-precision map data. Summary of the Invention

[0003] Embodiments of the present disclosure provide a method, an apparatus, an electronic device, and a program product for discriminating traffic image recognition results.

[0004] In a first aspect, embodiments of the present disclosure provide a method for discriminating traffic image recognition results.

[0005] Specifically, the method for discriminating traffic image recognition results includes:

[0006] Obtain at least two traffic images, perform recognition on the at least two traffic images, and obtain a traffic element data set composed of traffic element data included in the at least two traffic images, where the traffic element data includes: traffic element category, traffic element content, and traffic element position;

[0007] Perform processing on the traffic element data set to obtain a high-precision and / or standard-precision traffic element data set;

[0008] Perform non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set to obtain non-noise data in the high-precision and / or standard-precision traffic element data set;

[0009] Perform noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data to obtain a discrimination result of traffic image recognition results.

[0010] In an implementation manner of the present disclosure, the performing non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set includes:

[0011] Determine the matching degree between the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element dataset, where the matching degree includes the traffic element category matching degree, the traffic element content matching degree, and the traffic element position matching degree;

[0012] Determine whether there is high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element dataset with a matching degree higher than a preset matching degree threshold;

[0013] If there is, determine that the traffic element data with a matching degree higher than the preset matching degree threshold is non-noise data.

[0014] In one implementation manner of the present disclosure, the noise discrimination for the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element dataset except the non-noise data includes:

[0015] Determine whether the remaining high-precision and / or standard-precision traffic element data is one of the traffic element category recognition error noise, the traffic element content recognition error noise, and the traffic element position recognition error noise;

[0016] If it is determined that the remaining high-precision and / or standard-precision traffic element data is one of the traffic element category recognition error noise, the traffic element content recognition error noise, and the traffic element position recognition error noise, then determine that the high-precision and / or standard-precision traffic element data is noise data.

[0017] In one implementation manner of the present disclosure, the determination of whether the remaining high-precision and / or standard-precision traffic element data is traffic element category recognition error noise includes:

[0018] Determine whether the high-precision and / or standard-precision traffic element data meets the preset corresponding relationship requirements between the traffic element category and the traffic element content;

[0019] Determine whether the high-precision and / or standard-precision traffic element data meets the first near data consistency requirement;

[0020] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the preset corresponding relationship requirements, and / or does not meet the first near data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element category recognition error noise.

[0021] In one implementation manner of the present disclosure, the determination of whether the high-precision and / or standard-precision traffic element data meets the first near data consistency requirement includes:

[0022] Determine whether the proportion of data with the same category in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than the first preset proportion threshold, and whether the average collection frequency is higher than the first preset frequency threshold;

[0023] If the proportion of data with the same category in the traffic element data is lower than the first preset proportion threshold, and / or the average collection frequency is lower than the first preset frequency threshold, then determine that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement;

[0024] Determine whether there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within the first preset range;

[0025] If there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within the first preset range, then determine that the high-precision and / or standard-precision traffic element data does not meet the second consistency requirement;

[0026] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement and / or the second consistency requirement, then determine that the high-precision and / or standard-precision traffic element data does not meet the first near-data consistency requirement.

[0027] In one implementation of the present disclosure, determining whether the remaining high-precision and / or standard-precision traffic element data is traffic element content recognition error noise includes:

[0028] Determine whether the high-precision and / or standard-precision traffic element data meets the content rationality requirement;

[0029] Determine whether the high-precision and / or standard-precision traffic element data meets the second near-data consistency requirement;

[0030] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the content rationality requirement, and / or does not meet the second near-data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element content recognition error noise.

[0031] In one implementation of the present disclosure, the determining whether the high-precision and / or standard-precision traffic element data meets the second near-data consistency requirement includes:

[0032] Determine whether the proportion of data with the same content in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than a second preset proportion threshold, and whether the average acquisition frequency is higher than a second preset frequency threshold. If the proportion of data with the same category in the traffic element data is lower than the second preset proportion threshold, and / or the average acquisition frequency is lower than the second preset frequency threshold, then determine that the high-precision and / or standard-precision traffic element data does not meet the second nearby data consistency requirement.

[0033] In an implementation manner of the present disclosure, determining whether the remaining high-precision and / or standard-precision traffic element data is traffic element position recognition error noise includes:

[0034] Determine whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement;

[0035] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element position recognition error noise.

[0036] In an implementation manner of the present disclosure, the determining whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement includes:

[0037] Cluster the traffic element data corresponding to the high-precision and / or standard-precision traffic element data according to the distance between them to obtain two or more clustering groups, and determine whether the traffic element data meets the third consistency requirement according to the number of data included in the clustering groups and the distance between different clustering groups;

[0038] Determine the distance between high-precision and / or standard-precision traffic element data with the same category within a second preset range. If the distance is higher than a preset distance threshold, then determine that the traffic element data does not meet the fourth consistency requirement;

[0039] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third consistency requirement and the fourth consistency requirement, then determine that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement.

[0040] In a second aspect, an apparatus for discriminating traffic image recognition results is provided in an embodiment of the present disclosure.

[0041] Specifically, the apparatus for discriminating traffic image recognition results includes:

[0042] An identification module, configured to obtain at least two traffic images, identify the at least two traffic images, and obtain a traffic element data set composed of traffic element data included in the at least two traffic images, where the traffic element data includes: traffic element category, traffic element content, and traffic element location;

[0043] A processing module, configured to process the traffic element data set to obtain a high-precision and / or standard-precision traffic element data set;

[0044] A first discrimination module, configured to perform non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set to obtain non-noise data in the high-precision and / or standard-precision traffic element data set;

[0045] A second discrimination module, configured to perform noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data to obtain a discrimination result of the traffic image recognition result.

[0046] In a third aspect, an embodiment of the present disclosure provides an electronic device, including a memory and at least one processor, where the memory is used to store one or more computer instructions, and the one or more computer instructions are executed by the at least one processor to implement the method steps of the above traffic image recognition result discrimination method.

[0047] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium for storing computer instructions used by the traffic image recognition result discrimination device, which includes computer instructions for executing the above traffic image recognition result discrimination method for the traffic image recognition result discrimination device.

[0048] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program / instructions, where the computer program / instructions implement the method steps of the above traffic image recognition result discrimination method when executed by a processor.

[0049] The technical solution provided by the embodiment of the present disclosure may include the following beneficial effects:

[0050] The above technical solution performs noise recognition on the data collected by the traffic image acquisition vehicle and the high-precision and / or standard-precision traffic element data obtained after processing. This technical solution can improve the accuracy of the high-precision and / or standard-precision traffic element data, and thus avoid problems such as image recognition errors, position parsing errors, high-precision map data creation errors, and high-precision map data creation redundancy caused by directly using the collected data, thereby greatly improving the data quality of the high-precision map data.

[0051] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In conjunction with the accompanying drawings, other features, objects, and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments. In the drawings:

[0053] Figure 1 A flowchart showing a method for discriminating traffic image recognition results according to an embodiment of the present disclosure;

[0054] Figure 2 An overall flowchart showing a method for discriminating traffic image recognition results according to an embodiment of the present disclosure;

[0055] Figure 3 A structural block diagram showing a device for discriminating traffic image recognition results according to an embodiment of the present disclosure;

[0056] Figure 4 A structural block diagram showing an electronic device according to an embodiment of the present disclosure;

[0057] Figure 5 A schematic structural diagram of a computer system suitable for implementing a method for discriminating traffic image recognition results according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] In the following, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings, so that those skilled in the art can easily implement them. In addition, for clarity, parts irrelevant to the description of the exemplary embodiments are omitted in the drawings.

[0059] In the present disclosure, it should be understood that terms such as "including" or "having" are intended to indicate the existence of features, numbers, steps, actions, components, parts, or combinations thereof disclosed in this specification, and do not intend to exclude the possibility of the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0060] It should also be noted that, without conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other. The present disclosure will be described in detail below with reference to the drawings and in conjunction with the embodiments.

[0061] The technical solution provided by the embodiments of the present disclosure performs noise recognition by means of the data collected by a traffic image acquisition vehicle and the high-precision and / or standard-precision traffic element data obtained after processing. This technical solution can improve the accuracy of high-precision and / or standard-precision traffic element data, thereby avoiding problems such as image recognition errors, position parsing errors, incorrect creation of high-precision map data, and redundant creation of high-precision map data caused by directly using the collected data, thus greatly improving the data quality of high-precision map data.

[0062] Figure 1 The flowchart showing a method for discriminating traffic image recognition results according to an embodiment of the present disclosure is as Figure 1 shown. The traffic image recognition result discrimination method includes the following steps S101 - S104:

[0063] In step S101, at least two traffic images are acquired, and the at least two traffic images are recognized to obtain a traffic element data set composed of the traffic element data included in the at least two traffic images. Among them, the traffic element data includes: traffic element category, traffic element content, and traffic element position;

[0064] In step S102, the traffic element data set is processed to obtain a high-precision and / or standard-precision traffic element data set;

[0065] In step S103, non-noise discrimination is performed on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set to obtain the non-noise data in the high-precision and / or standard-precision traffic element data set;

[0066] In step S104, noise discrimination is performed on the high-precision and / or standard-precision traffic element data remaining in the high-precision and / or standard-precision traffic element data set except for the non-noise data to obtain the traffic image recognition result discrimination result.

[0067] As mentioned above, with the development of society and the progress of science and technology, more secure and efficient autonomous driving technology has been recognized as one of the future directions of the automotive travel industry, and many autonomous driving behaviors rely on high-precision map data. In order to improve the update frequency of high-precision map data and reduce the update cost of high-precision map data, the prior art uses the data collected by traffic image acquisition vehicles to update high-precision map data. However, due to the low accuracy of the collected data, there may be errors in image recognition and position parsing. If the collected data is directly used to update high-precision map data, problems such as incorrect creation of high-precision map data and redundant creation will occur, thereby reducing the quality of high-precision map data.

[0068] In view of the above problems, in this embodiment, a method for discriminating traffic image recognition results is proposed. This method performs noise recognition by means of the data collected by a traffic image acquisition vehicle and the high-precision and / or standard-precision traffic element data obtained after processing. This technical solution can improve the accuracy of high-precision and / or standard-precision traffic element data, and thus avoid problems such as image recognition errors, position parsing errors, high-precision map data creation errors, and high-precision map data creation redundancy caused by directly using the collected data, thereby greatly improving the data quality of high-precision map data.

[0069] In an embodiment of the present disclosure, the method for discriminating traffic image recognition results can be applied to computers, computing devices, electronic devices, servers, service clusters, etc. for reconstructing traffic sign vector models.

[0070] In an embodiment of the present disclosure, the traffic image refers to an image of a traffic scene collected by a traffic image acquisition vehicle and containing one or more traffic elements. Among them, the traffic image acquisition vehicle can be a dedicated traffic image acquisition vehicle or a crowdsourcing traffic image acquisition vehicle with other functions; the traffic elements can be, for example: ground arrows, lane lines, speed limit strips, zebra crossings, traffic signs, traffic lights, etc.

[0071] In an embodiment of the present disclosure, the traffic element data refers to data related to the traffic elements. The traffic element data can include, for example, one or more of the following data: traffic element category, traffic element content, and traffic element position. Among them, the traffic element category refers to the category to which the traffic element belongs. The traffic element category can be, for example: ground arrows, solid lane lines, dashed lane lines, speed limit strips, zebra crossings, traffic lights, turning and straight-ahead signs, speed limit signs, no-entry signs, etc. Among them, the traffic element content refers to the specific content contained in the traffic element. For example, if the category of the traffic element is a speed limit sign, the content of the traffic element can be the specific speed limit marked on the speed limit sign. Among them, the traffic element position refers to the physical space position where the traffic is located. For example, the traffic element position can be represented by the longitude, latitude, and altitude data where the traffic element is located. If the traffic element is a point, its position can be represented by the longitude, latitude, and altitude data of the point. If the traffic element is an object with a certain projected area, its position can be represented by the average longitude, latitude, and altitude data of all the points that make up the object.

[0072] In an embodiment of the present disclosure, considering that the traffic element data recognized from the traffic images directly collected by the vehicle may have low accuracy and low accuracy rate, therefore, the traffic element data included in the traffic element data set can be processed, processed, and produced to obtain a traffic element data set composed of high-precision, high-precision and / or standard-precision traffic element data. Among them, the high-precision traffic element data and the standard-precision traffic element data have different accuracies, different granularities, and different data collection frequencies. Among them, processing, processing, and producing high-precision and / or standard-precision traffic element data from the source data belongs to the technology that those skilled in the art should widely master, and will not be elaborated here. For example, the processing, processing, and production may include data denoising, data interpolation, data filling, data format standardization, extraction of traffic element data, and the like.

[0073] In the above embodiment, first, at least two traffic images are obtained, and then the at least two traffic images are recognized to obtain a traffic element data set composed of the traffic element data included in the at least two traffic images. Among them, each traffic image may include at least one traffic element, so there will be multiple traffic elements in at least two traffic images. All the traffic element data obtained by recognizing at least two traffic images can form a traffic element data set; then the traffic element data set is processed to obtain a high-precision and / or standard-precision traffic element data set; then the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set are first subjected to non-noise discrimination to obtain the high-precision and / or standard-precision traffic element data belonging to non-noise in the high-precision and / or standard-precision traffic element data set; then the high-precision and / or standard-precision traffic element data remaining in the high-precision and / or standard-precision traffic element data set except the non-noise data are subjected to noise discrimination again, and finally the discrimination result of the traffic image recognition result is obtained, that is, which of the traffic image recognition results are misrecognized noise data and which are correctly recognized non-noise data.

[0074] In an embodiment of the present disclosure, step S103, that is, the step of performing non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set, may include the following steps:

[0075] Determine the matching degree between the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set, where the matching degree includes traffic element category matching degree, traffic element content matching degree, and traffic element position matching degree;

[0076] Determine whether there is high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set whose matching degree is higher than a preset matching degree threshold;

[0077] If it exists, determine that the traffic element data with a matching degree higher than the preset matching degree threshold is non-noise data.

[0078] In this embodiment, by determining whether there is consistent high-precision and / or standard-precision traffic element data in the high-precision traffic element data set and / or the standard-precision traffic element data set, the non-noise discrimination of the high-precision and / or standard-precision traffic element data is performed. Specifically, the matching degree between each high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set can be calculated first. It should be noted that the calculation of the matching degree is within the scope of the same-precision traffic element data set, that is, the matching degree between each high-precision traffic element data in the high-precision traffic element data set and the matching degree between each standard-precision traffic element data in the standard-precision traffic element data set are calculated respectively. That is, the matching degree can include the traffic element category matching degree, the traffic element content matching degree, and the traffic element position matching degree between high-precision traffic element data and / or between standard-precision traffic element data. Among them, the calculation of the matching degree can be implemented by using the data matching degree algorithm in the prior art, which will not be elaborated here; then the calculated matching degree is compared with the preset matching degree threshold. The data with a matching degree higher than the preset matching degree threshold can be considered not to be isolated individual data in the high-precision and / or standard-precision traffic element data set, and thus it can be considered that the data is not noise data. Among them, the preset matching degree threshold can be set according to the actual application needs, and the present disclosure does not make special limitations on it.

[0079] In an embodiment of the present disclosure, in step S104, the step of performing noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data may include the following steps:

[0080] Determine whether the remaining high-precision and / or standard-precision traffic element data is one of the traffic element category recognition error noise, the traffic element content recognition error noise, and the traffic element position recognition error noise;

[0081] If it is determined that the remaining high-precision and / or standard-precision traffic element data is one of the traffic element category recognition error noise, the traffic element content recognition error noise, and the traffic element position recognition error noise, then determine that the high-precision and / or standard-precision traffic element data is noise data.

[0082] In this embodiment, noise discrimination is performed on the data after non-noise discrimination to obtain the final traffic image recognition result discrimination result. Specifically, for the high-precision and / or standard-precision traffic element data remaining in the high-precision and / or standard-precision traffic element data set except for the non-noise data, discrimination of traffic element category recognition error noise, traffic element content recognition error noise, or traffic element position recognition error noise is performed respectively. If the remaining high-precision and / or standard-precision traffic element data is discriminated as one of traffic element category recognition error noise, traffic element content recognition error noise, and traffic element position recognition error noise, it can be determined that the high-precision and / or standard-precision traffic element data is noise data.

[0083] In an embodiment of the present disclosure, the step of determining whether the remaining high-precision and / or standard-precision traffic element data is traffic element category recognition error noise may include the following steps:

[0084] Determine whether the high-precision and / or standard-precision traffic element data meets the preset correspondence requirement between the traffic element category and the traffic element content;

[0085] Determine whether the high-precision and / or standard-precision traffic element data meets the first nearby data consistency requirement;

[0086] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the preset correspondence requirement, and / or does not meet the first nearby data consistency requirement, it is determined that the high-precision and / or standard-precision traffic element data is traffic element category recognition error noise.

[0087] In this embodiment, the traffic element category recognition error noise is recognized by means of preset rule requirements and nearby data consistency requirements. That is, determine whether the high-precision and / or standard-precision traffic element data meets the preset correspondence requirement between the traffic element category and the traffic element content, that is, the preset rule requirement. For example, the preset correspondence requirement between the traffic element category and the traffic element content may be: the category of a traffic element with content containing words such as "ramp" should be "speed limit sign", the category of a traffic element with content containing words such as "exit" and "km" should be "exit sign", etc. If the category of a traffic element with content containing the word "exit" is "speed limit belt", it can be determined that the category of this traffic element is incorrect and does not meet the preset correspondence requirement; then determine whether the high-precision and / or standard-precision traffic element data meets the first nearby data consistency requirement; if the high-precision and / or standard-precision traffic element data does not meet the preset correspondence requirement, and / or does not meet the first nearby data consistency requirement, it can be determined that the high-precision and / or standard-precision traffic element data is traffic element category recognition error noise.

[0088] Further, the step of determining whether the high-precision and / or standard-precision traffic element data meets the first nearby data consistency requirement may include the following steps:

[0089] Determine whether the proportion of data with the same category in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than a first preset proportion threshold, and whether the average acquisition frequency is higher than a first preset frequency threshold;

[0090] If the proportion of data with the same category in the traffic element data is lower than the first preset proportion threshold, and / or the average acquisition frequency is lower than the first preset frequency threshold, it is determined that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement;

[0091] Determine whether there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within a first preset range;

[0092] If there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within the first preset range, it is determined that the high-precision and / or standard-precision traffic element data does not meet the second consistency requirement;

[0093] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement and / or the second consistency requirement, it is determined that the high-precision and / or standard-precision traffic element data does not meet the first nearby data consistency requirement.

[0094] In this embodiment, the consistency requirements are respectively discriminated for the traffic element data before processing and the high-precision and / or standard-precision traffic element data obtained after processing. When the two do not meet at least one consistency requirement, it can be determined that the high-precision and / or standard-precision traffic element data does not meet the first nearby data consistency requirement.

[0095] Specifically:

[0096] For the traffic element data before processing, that is, the traffic element data corresponding to the high-precision and / or standard-precision traffic element data, calculate the proportion of data with the same category in the traffic element data and the average data collection frequency. Then compare the proportion of data with the same category with a first preset proportion threshold and compare the average data collection frequency with a first preset frequency threshold. If the proportion of data with the same category is lower than the first preset proportion threshold, it is considered that there are fewer data of the same category in the traffic element data and there may be an error in traffic element category recognition. If the average data collection frequency is lower than the first preset frequency threshold, it indicates that there may be errors in data collection. Therefore, if the proportion of data with the same category is lower than the first preset proportion threshold, and / or the average data collection frequency is lower than the first preset frequency threshold, it is considered that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement. Among them, the first preset proportion threshold and the first preset frequency threshold can be set according to the actual application needs, and the present disclosure does not make special limitations on them.

[0097] For the high-precision and / or standard-precision traffic element data obtained after processing, determine whether there are data with the same traffic element content but different traffic element categories among the high-precision and / or standard-precision traffic element data within a first preset range. If so, it is considered that there may be errors in the recognition of these data categories, that is, it can be determined that the high-precision and / or standard-precision traffic element data does not meet the second consistency requirement. Among them, the first preset range can be set according to the actual application needs, and the present disclosure does not make special limitations on it.

[0098] Finally, if it is determined that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement and / or the second consistency requirement, it can be considered that the high-precision and / or standard-precision traffic element data does not meet the first adjacent data consistency requirement.

[0099] In an embodiment of the present disclosure, the step of determining whether the remaining high-precision and / or standard-precision traffic element data is noise with incorrect traffic element content recognition may include the following steps:

[0100] Determine whether the high-precision and / or standard-precision traffic element data meets the content rationality requirement;

[0101] Determine whether the high-precision and / or standard-precision traffic element data meets the second adjacent data consistency requirement;

[0102] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the content rationality requirement and / or does not meet the second adjacent data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is noise with incorrect traffic element content recognition.

[0103] In this embodiment, the recognition of traffic element content recognition errors and noises is performed by means of preset content rationality requirements and nearby data consistency requirements. That is, it is determined whether the high-precision and / or standard-precision traffic element data meets the content rationality requirements. For example, the content rationality requirements may be, for example, that the content of traffic elements with categories of "speed limit sign", "height limit sign", "width limit sign", and "weight limit sign" should be a number within a preset numerical range. For example, if the content of a traffic element with the category of "speed limit sign" is "2.6m", it can be considered that the content recognition of this traffic element is incorrect and does not meet the content rationality requirements. It should be noted that for traffic elements with the category of "electronic speed limit", the judgment of whether they meet the content rationality requirements is not performed.

[0104] Then it is determined whether the high-precision and / or standard-precision traffic element data meets the second nearby data consistency requirements; if the high-precision and / or standard-precision traffic element data does not meet the content rationality requirements and / or does not meet the second nearby data consistency requirements, it can be determined that the high-precision and / or standard-precision traffic element data is traffic element content recognition error noise.

[0105] Further, the step of determining whether the high-precision and / or standard-precision traffic element data meets the second nearby data consistency requirements may include the following steps:

[0106] Determine whether the proportion of data with the same content in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than the second preset proportion threshold, and whether the average acquisition frequency is higher than the second preset frequency threshold. If the proportion of data with the same category in the traffic element data is lower than the second preset proportion threshold and / or the average acquisition frequency is lower than the second preset frequency threshold, it is determined that the high-precision and / or standard-precision traffic element data does not meet the second nearby data consistency requirements.

[0107] In this embodiment, the consistency requirements are discriminated for the traffic element data before processing. If the traffic element data before processing does not meet the corresponding consistency requirements, it can be determined that the high-precision and / or standard-precision traffic element data does not meet the second nearby data consistency requirements.

[0108] Specifically:

[0109] For the traffic element data before processing, that is, the traffic element data corresponding to the high-precision and / or standard-precision traffic element data, calculate the proportion of data with the same content in the traffic element data and the average data collection frequency. Then compare the proportion of data with the same category with a second preset proportion threshold, and compare the average data collection frequency with a second preset frequency threshold. If the proportion of data with the same category is lower than the second preset proportion threshold, it is considered that there is less data with the same content in the traffic element data, and there may be an error in traffic element content recognition. If the average data collection frequency is lower than the second preset frequency threshold, it indicates that there may be errors in data collection. Therefore, if the proportion of data with the same content is lower than the second preset proportion threshold, and / or the average data collection frequency is lower than the second preset frequency threshold, it is considered that the high-precision and / or standard-precision traffic element data does not meet the second nearby data consistency requirement.

[0110] In an embodiment of the present disclosure, the step of determining whether the remaining high-precision and / or standard-precision traffic element data is traffic element position recognition error noise may include the following steps:

[0111] Determine whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement;

[0112] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element position recognition error noise.

[0113] In this embodiment, the traffic element position recognition error noise is recognized by means of the nearby data consistency requirement. That is, determine whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement. If the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement, then it can be determined that the high-precision and / or standard-precision traffic element data is traffic element position recognition error noise.

[0114] Further, the step of determining whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement may include the following steps:

[0115] Cluster the traffic element data corresponding to the high-precision and / or standard-precision traffic element data according to the distance between them to obtain two or more clustering groups. Determine whether the traffic element data meets the third consistency requirement according to the number of data included in the clustering groups and the distance between different clustering groups;

[0116] Determine the distance between high-precision and / or standard-precision traffic element data with the same category within a second preset range. If the distance is higher than a preset distance threshold, determine that the traffic element data does not meet the fourth consistency requirement;

[0117] If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third consistency requirement and the fourth consistency requirement, determine that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement.

[0118] In this embodiment, the distance between traffic element data is used to determine whether the high-precision and / or standard-precision traffic element data meets the third nearby data consistency requirement. Specifically, first calculate the distance between the traffic element data corresponding to the high-precision and / or standard-precision traffic element data, then cluster the distances to obtain two or more clustering groups. According to the number of data contained in the clustering groups and the distance between different clustering groups, it can be determined whether the traffic element data meets the third consistency requirement. For example, if the number of data contained in a certain clustering group is small, lower than a preset number threshold, or the distance between a certain clustering group and other clustering groups is far, farther than the preset distance threshold, it indicates that the data in this clustering group may have a position recognition error, that is, it does not meet the third consistency requirement; then calculate the distance between high-precision and / or standard-precision traffic element data with the same category within a second preset range. If the distance is higher than the preset distance threshold, it indicates that these data may have a position recognition error, and it can be determined that the traffic element data does not meet the fourth consistency requirement; if it is finally determined that the high-precision and / or standard-precision traffic element data does not meet the third consistency requirement and the fourth consistency requirement, it can be determined that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement.

[0119] Figure 2 Show the overall flowchart of the traffic image recognition result discrimination method according to an embodiment of the present disclosure, as Figure 2As shown, when discriminating the traffic image recognition result, the obtained traffic images are recognized to obtain a traffic element dataset, and then the traffic element dataset is processed to obtain a high-precision and / or standard-precision traffic element dataset. Then, non-noise discrimination is performed on the high-precision and / or standard-precision traffic element data, that is, it is determined whether there is high-precision and / or standard-precision traffic element data with a matching degree higher than a preset matching degree threshold. If so, the data is determined to be non-noise data. Then, noise discrimination is performed on the remaining high-precision and / or standard-precision traffic element data except the non-noise data, that is, it is determined whether the remaining high-precision and / or standard-precision traffic element data is one of the traffic element category recognition error noise, traffic element content recognition error noise, and traffic element position recognition error noise. If so, the high-precision and / or standard-precision traffic element data is determined to be noise data, and finally the traffic image recognition result discrimination result is obtained.

[0120] The following is an embodiment of the apparatus of the present disclosure, which can be used to execute the embodiment of the method of the present disclosure.

[0121] Figure 3 The structural block diagram of a traffic image recognition result discrimination device according to an embodiment of the present disclosure is shown. The device can be implemented as part or all of an electronic device through software, hardware, or a combination of both. As Figure 3 shown, the traffic image recognition result discrimination device includes:

[0122] An identification module 301, configured to obtain at least two traffic images, and perform identification on the at least two traffic images to obtain a traffic element dataset composed of traffic element data included in the at least two traffic images, where the traffic element data includes: traffic element category, traffic element content, and traffic element position;

[0123] A processing module 302, configured to process the traffic element dataset to obtain a high-precision and / or standard-precision traffic element dataset;

[0124] A first discrimination module 303, configured to perform non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element dataset to obtain non-noise data in the high-precision and / or standard-precision traffic element dataset;

[0125] A second discrimination module 304, configured to perform noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element dataset except the non-noise data to obtain a traffic image recognition result discrimination result.

[0126] As mentioned above, with the development of society and the progress of science and technology, autonomous driving technology with higher safety and efficiency has been recognized as one of the future directions of the automotive travel industry, and many autonomous driving behaviors rely on high-precision map data. In order to improve the update frequency of high-precision map data and reduce the update cost of high-precision map data, in the prior art, the acquisition data collected by traffic image acquisition vehicles is used to update high-precision map data. However, due to the low precision of the acquisition data, errors may occur in image recognition and position parsing. If the acquisition data is directly used to update high-precision map data, problems such as incorrect creation of high-precision map data and redundant creation will occur, thus reducing the quality of high-precision map data.

[0127] Considering the above problems, in this embodiment, a traffic image recognition result discrimination device is proposed. This device performs noise recognition by using the data collected by traffic image acquisition vehicles and the high-precision and / or standard-precision traffic element data obtained after processing. This technical solution can improve the accuracy of high-precision and / or standard-precision traffic element data, and thus avoid problems such as image recognition errors, position parsing errors, incorrect creation of high-precision map data, and redundant creation of high-precision map data caused by directly using acquisition data, thereby greatly improving the data quality of high-precision map data.

[0128] In an embodiment of the present disclosure, the traffic image recognition result discrimination device can be implemented as a computer, computing device, electronic device, server, service cluster, etc. for reconstructing a traffic sign vector model.

[0129] The technical terms and technical features involved in the related embodiments of the above device are the same as or similar to the technical terms and technical features mentioned in the related embodiments of the above method. For the explanations and descriptions of the technical terms and technical features involved in the related embodiments of the above device, reference can be made to the explanations and descriptions of the related embodiments of the above method, which will not be elaborated here.

[0130] The present disclosure also discloses an electronic device, Figure 4 showing a structural block diagram of an electronic device according to an embodiment of the present disclosure, as Figure 4 shown, the electronic device 400 includes a memory 401 and a processor 402; wherein,

[0131] The memory 401 is used to store one or more computer instructions, and the one or more computer instructions are executed by the processor 402 to implement the above method steps.

[0132] Figure 5 is a schematic structural diagram of a computer system suitable for implementing a traffic image recognition result discrimination method according to an embodiment of the present disclosure.

[0133] AsFigure 5 As shown in Figure 5 , computer system 500 includes a processing unit 501, which can perform various processes in the above-described embodiments according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage section 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the computer system 500 are also stored. The processing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0134] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed so that a computer program read from it can be installed into the storage section 508 as needed. Among them, the processing unit 501 can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, an NPU, etc.

[0135] Specifically, according to an embodiment of the present disclosure, the method described above can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program tangibly contained on a computer-readable medium, and the computer program includes program code for executing the method. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 509, and / or installed from the removable medium 511.

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0137] The units or modules involved in the embodiments described in the present disclosure can be implemented in software or in hardware. The units or modules described can also be provided in a processor, and the names of these units or modules do not, in some cases, constitute a limitation on the units or modules themselves.

[0138] As another aspect, the present disclosure also provides a computer-readable storage medium, which can be the computer-readable storage medium included in the device described in the above embodiments; or it can exist separately and be a computer-readable storage medium not assembled into the device. The computer-readable storage medium stores one or more programs, and the programs are used by one or more processors to execute the methods described in the present disclosure.

[0139] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of the invention involved in the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) having similar functions disclosed in the present disclosure.

Claims

1. A method for discriminating traffic image recognition results, comprising: Obtaining at least two traffic images, recognizing the at least two traffic images to obtain a traffic element data set composed of traffic element data included in the at least two traffic images, wherein the traffic element data includes: traffic element category, traffic element content, and traffic element location; Processing the traffic element data set to obtain a high-precision and / or standard-precision traffic element data set, wherein the high-precision traffic element data and the standard-precision traffic element data have different precisions, different granularities, and different data collection frequencies; Performing non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set to obtain non-noise data in the high-precision and / or standard-precision traffic element data set; Performing noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data to obtain a discrimination result of traffic image recognition results.

2. The method according to claim 1, wherein The performing non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set includes: Determining the matching degree between the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set, wherein the matching degree includes traffic element category matching degree, traffic element content matching degree, and traffic element location matching degree; Determining whether there is high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set with a matching degree higher than a preset matching degree threshold; If so, determining the traffic element data with a matching degree higher than the preset matching degree threshold as non-noise data.

3. The method according to claim 1 or 2, wherein The performing noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data includes: Determining whether the remaining high-precision and / or standard-precision traffic element data is one of traffic element category recognition error noise, traffic element content recognition error noise, and traffic element location recognition error noise; If it is determined that the remaining high-precision and / or standard-precision traffic element data is one of traffic element category recognition error noise, traffic element content recognition error noise, and traffic element location recognition error noise, then determining the high-precision and / or standard-precision traffic element data as noise data.

4. The method according to claim 3, wherein, The determining whether the remaining high-precision and / or standard-precision traffic element data is traffic element category recognition error noise includes: Determining whether the high-precision and / or standard-precision traffic element data meets the preset corresponding relationship requirements between the traffic element category and the traffic element content; Determining whether the high-precision and / or standard-precision traffic element data meets the first nearby data consistency requirements; If it is determined that the high-precision and / or standard-precision traffic element data does not meet the preset corresponding relationship requirements, and / or does not meet the first nearby data consistency requirements, then determining the high-precision and / or standard-precision traffic element data as traffic element category recognition error noise.

5. The method according to claim 4, wherein, The determining whether the high-precision and / or standard-precision traffic element data meets the first nearby data consistency requirements includes: Determine whether the proportion of data with the same category in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than the first preset proportion threshold, and whether the average acquisition frequency is higher than the first preset frequency threshold; If the proportion of data with the same category in the traffic element data is lower than the first preset proportion threshold, and / or the average acquisition frequency is lower than the first preset frequency threshold, then determine that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement; Determine whether there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within the first preset range; If there is data with the same traffic element content but different traffic element categories in the high-precision and / or standard-precision traffic element data within the first preset range, then determine that the high-precision and / or standard-precision traffic element data does not meet the second consistency requirement; If it is determined that the high-precision and / or standard-precision traffic element data does not meet the first consistency requirement and / or the second consistency requirement, then determine that the high-precision and / or standard-precision traffic element data does not meet the first adjacent data consistency requirement.

6. The method according to claim 3, wherein Determine whether the remaining high-precision and / or standard-precision traffic element data is traffic element content recognition error noise, including: Determine whether the high-precision and / or standard-precision traffic element data meets the content rationality requirement; Determine whether the high-precision and / or standard-precision traffic element data meets the second adjacent data consistency requirement; If it is determined that the high-precision and / or standard-precision traffic element data does not meet the content rationality requirement, and / or does not meet the second adjacent data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element content recognition error noise.

7. The method according to claim 6, wherein, The determination of whether the high-precision and / or standard-precision traffic element data meets the second adjacent data consistency requirement includes: Determine whether the proportion of data with the same content in the traffic element data corresponding to the high-precision and / or standard-precision traffic element data is higher than the second preset proportion threshold, and whether the average acquisition frequency is higher than the second preset frequency threshold. If the proportion of data with the same category in the traffic element data is lower than the second preset proportion threshold, and / or the average acquisition frequency is lower than the second preset frequency threshold, then determine that the high-precision and / or standard-precision traffic element data does not meet the second adjacent data consistency requirement.

8. The method according to claim 3, wherein, Determine whether the remaining high-precision and / or standard-precision traffic element data is traffic element position recognition error noise, including: Determine whether the high-precision and / or standard-precision traffic element data meets the third adjacent data consistency requirement; If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third adjacent data consistency requirement, then determine that the high-precision and / or standard-precision traffic element data is traffic element position recognition error noise.

9. The method according to claim 8, wherein, The determination of whether the high-precision and / or standard-precision traffic element data meets the third adjacent data consistency requirement includes: Cluster the traffic element data corresponding to the high-precision and / or standard-precision traffic element data according to the distance between them to obtain two or more clustering groups, and determine whether the traffic element data meets the third consistency requirement according to the number of data included in the clustering groups and the distance between different clustering groups; Determine the distance between high-precision and / or standard-precision traffic element data with the same category within the second preset range. If the distance is higher than the preset distance threshold, determine that the traffic element data does not meet the fourth consistency requirement; If it is determined that the high-precision and / or standard-precision traffic element data does not meet the third consistency requirement and the fourth consistency requirement, determine that the high-precision and / or standard-precision traffic element data does not meet the third nearby data consistency requirement.

10. A traffic image recognition result discrimination device, comprising: An identification module configured to obtain at least two traffic images, identify the at least two traffic images, and obtain a traffic element data set composed of the traffic element data included in the at least two traffic images, wherein the traffic element data includes: traffic element category, traffic element content, and traffic element location; A processing module configured to process the traffic element data set to obtain a high-precision and / or standard-precision traffic element data set, wherein the high-precision traffic element data and the standard-precision traffic element data have different precisions, different granularities, and different data acquisition frequencies; A first discrimination module configured to perform non-noise discrimination on the high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set to obtain the non-noise data in the high-precision and / or standard-precision traffic element data set; A second discrimination module configured to perform noise discrimination on the remaining high-precision and / or standard-precision traffic element data in the high-precision and / or standard-precision traffic element data set except the non-noise data to obtain a traffic image recognition result discrimination result.

11. An electronic device, comprising a memory and at least one processor; wherein, The memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the at least one processor to implement the method steps described in any one of claims 1-9.

12. A computer program product, comprising a computer program / instructions, wherein, When the computer program / instruction is executed by the processor, it implements the method steps described in any one of claims 1-9.

Images