Point cloud data recognition validity detection method, device, medium and electronic equipment

Abstract

The present application relates to the field of autonomous driving technology, and provides a method, device, medium and electronic device for detecting the effectiveness of point cloud data recognition. The method obtains preset point cloud data; the point cloud data includes an area to be corrected, wherein the area to be corrected is an inaccurately identified point cloud area, and then the point cloud data is identified to obtain a perception result area, and based on the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected is determined, and then the similarity parameter between the target perception result area and the area to be corrected is determined; the effectiveness of the point cloud data recognition is judged based on the similarity parameter. This method can save the time spent on detecting the effectiveness of point cloud data recognition and improve the efficiency of detecting the effectiveness of point cloud data recognition.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of autonomous driving technology, and in particular to a method, device, medium and electronic device for detecting the validity of point cloud data recognition. Background Art

[0002] At present, perception algorithms are usually used to identify point cloud data to achieve object recognition, such as the laser detection and ranging (LIDAR) algorithm. If the data in some areas of the point cloud data is not sufficient to support the perception algorithm to make judgments, it will lead to inaccurate recognition, such as missed detection and false detection. Therefore, the perception algorithm often needs to be upgraded in a targeted manner.

[0003] In related technologies, the upgraded perception algorithm needs to be tested in real scenes to detect whether the problem has been solved. The detection of the effectiveness of point cloud data recognition often involves many detection links and takes a long time. Therefore, when testing the effectiveness of point cloud data recognition, the detection efficiency is relatively low. Summary of the invention

[0004] The embodiments of the present application provide a method, device, medium and electronic device for detecting the effectiveness of point cloud data recognition, which can improve the detection efficiency of the effectiveness of point cloud data recognition.

[0005] In a first aspect, an embodiment of the present application provides a method for detecting the validity of point cloud data recognition, comprising:

[0006] Acquire preset point cloud data, wherein the point cloud data includes a to-be-corrected area, wherein the to-be-corrected area is a point cloud area that is not accurately identified;

[0007] Identifying the point cloud data to obtain a perception result area;

[0008] Determine, according to the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected;

[0009] Determining a similarity parameter between the target perception result area and the area to be corrected;

[0010] The validity of the point cloud data recognition is judged based on the similarity parameter.

[0011] The detection method for the effectiveness of point cloud data recognition provided in the embodiment of the present application first obtains preset point cloud data; the point cloud data includes an area to be corrected, wherein the area to be corrected is an inaccurately identified point cloud area, then the point cloud data is identified to obtain a perception result area, and based on the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected is determined, and then the similarity parameter between the target perception result area and the area to be corrected is determined; the effectiveness of the point cloud data recognition is judged based on the similarity parameter. The method identifies the preset point cloud data, and performs similarity calculation based on the recognition result, and judges the effectiveness of the point cloud data recognition based on the similarity parameter obtained by the similarity calculation. The detection process does not require real-life testing, and can more conveniently determine the effectiveness of the point cloud data recognition, saving the time spent on detecting the effectiveness of the point cloud data recognition, and improving the detection efficiency of the effectiveness of the point cloud data recognition.

[0012] In an optional embodiment, the point cloud data further includes a regular time range corresponding to the area to be corrected;

[0013] The step of identifying the point cloud data to obtain a perception result area includes:

[0014] Determining target point cloud data corresponding to the area to be corrected according to the rule time range and the point cloud data;

[0015] The target point cloud data is identified to obtain a perception result area.

[0016] In this embodiment, the point cloud data also includes a regular time range corresponding to the area to be corrected; based on the regular time range and the point cloud data, the target point cloud data corresponding to the area to be corrected is determined; the target point cloud data is identified to obtain the perception result area, and a target point cloud data acquisition mechanism is provided, so that the perception result area can be generated more efficiently, further saving the time spent on detecting the effectiveness of point cloud data recognition, and improving the detection efficiency of the effectiveness of point cloud data recognition.

[0017] In an optional embodiment, after identifying the point cloud data to obtain the perception result area, the method further includes:

[0018] If it is determined based on the area to be corrected and the perception result area that the area to be corrected and the perception result area do not overlap, an abnormal alarm for point cloud data recognition validity detection is performed.

[0019] In this embodiment, if it is determined that the area to be corrected and the perception result area do not overlap, then an abnormal alarm for point cloud data recognition validity detection is performed. This method provides an abnormal alarm mechanism by setting up monitoring for the situation where the area to be corrected and the perception result area do not overlap, so that the validity of point cloud data recognition can be determined more efficiently, saving the time spent on the detection of the validity of point cloud data recognition, and improving the detection efficiency of the validity of point cloud data recognition.

[0020] In an optional embodiment, the perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data;

[0021] The step of determining, based on the area to be corrected and the perception result area, a target perception result area overlapping with the area to be corrected includes:

[0022] If the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area.

[0023] In this embodiment, the perception result area has a corresponding perception result time. In the process of determining the target perception result area that overlaps with the area to be corrected, if the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area. By determining the target perception result area that overlaps with the area to be corrected based on the perception result time corresponding to the perception result area, the method can reduce the amount of calculation when determining the target perception result area, further save the time consumption of detecting the effectiveness of point cloud data recognition, and improve the detection efficiency of the effectiveness of point cloud data recognition.

[0024] In an optional embodiment, the determining of the similarity parameter between the target perception result area and the area to be corrected includes:

[0025] Determine a feature perception result area according to the target perception result area and the corresponding target perception result time range; the feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range;

[0026] Determine, according to the feature perception result area and the area to be corrected, a target feature perception result area that overlaps with the area to be corrected;

[0027] The similarity parameter is obtained according to the target feature perception result area and the area to be corrected.

[0028] In this embodiment, the trajectory of each frame of the target perception result area within the corresponding target perception result time range is spliced ​​to obtain a feature perception result area; then, based on the feature perception result area and the area to be corrected, a target feature perception result area that overlaps with the area to be corrected is determined, and the similarity parameter is obtained based on the target feature perception result area and the area to be corrected. This method provides a mechanism for evaluating the association between the target perception result area and the area to be corrected, and can efficiently determine the similarity parameter between the target perception result area and the area to be corrected based on the target perception result time range and the target perception result area, further improving the detection efficiency of the effectiveness of point cloud data recognition.

[0029] In an optional embodiment, obtaining the similarity parameter according to the target feature perception result area and the area to be corrected includes:

[0030] Determine a first area of ​​the target feature perception result area and a second area of ​​the area to be corrected according to the target feature perception result area and the area to be corrected;

[0031] The ratio of the first area to the second area is used as the similarity parameter.

[0032] In this embodiment, the first area of ​​the target feature perception result area and the second area of ​​the area to be corrected are determined according to the target feature perception result area and the area to be corrected; and the ratio of the first area to the second area is used as the similarity parameter. The similarity parameter determined by this method has a high evaluation scale recognition, can accurately reflect the similarity between the target feature perception result area and the area to be corrected, and the amount of calculation required to determine the similarity parameter is small, which can improve the accuracy of evaluating the upgraded results of the upgraded perception algorithm and further improve the detection efficiency of the effectiveness of point cloud data recognition.

[0033] In an optional embodiment, judging the validity of the point cloud data recognition based on the similarity parameter includes:

[0034] When the similarity parameter is greater than a preset similarity threshold, it is determined that the point cloud data recognition is valid.

[0035] In this embodiment, if the similarity parameter is greater than a preset similarity threshold, the point cloud data recognition is judged to be effective. By setting a preset similarity threshold, the method can conveniently judge the effectiveness of point cloud data recognition based on the similarity parameter, further improving the detection efficiency of the effectiveness of point cloud data recognition.

[0036] In a second aspect, the embodiment of the present application further provides a device for detecting the validity of point cloud data recognition, comprising:

[0037] A data acquisition unit, used to acquire preset point cloud data, wherein the point cloud data includes a region to be corrected, wherein the region to be corrected is a point cloud region that is not accurately identified;

[0038] A correction recognition unit is used to recognize the point cloud data to obtain a perception result area;

[0039] an overlap determination unit, configured to determine, based on the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected;

[0040] A similarity determination unit, used to determine a similarity parameter between the target perception result area and the area to be corrected;

[0041] The upgrade recognition unit is used to determine the validity of the point cloud data recognition based on the similarity parameter.

[0042] In an optional embodiment, the point cloud data further includes a regular time range corresponding to the area to be corrected; and the correction identification unit is specifically configured to:

[0043] Determining target point cloud data corresponding to the area to be corrected according to the rule time range and the point cloud data;

[0044] The target point cloud data is identified to obtain a perception result area.

[0045] In an optional embodiment, the device further includes:

[0046] The abnormality alarm unit is used to issue an abnormality alarm for point cloud data recognition validity detection if it is determined that the area to be corrected and the perception result area do not overlap based on the area to be corrected and the perception result area.

[0047] In an optional embodiment, the perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data; the overlap determination unit is specifically used to:

[0048] If the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area.

[0049] In an optional embodiment, the similarity determination unit is specifically used to:

[0050] Determine a feature perception result area according to the target perception result area and the corresponding target perception result time range; the feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range;

[0051] Determine, according to the feature perception result area and the area to be corrected, a target feature perception result area that overlaps with the area to be corrected;

[0052] The similarity parameter is obtained according to the target feature perception result area and the area to be corrected.

[0053] In an optional embodiment, the similarity determination unit is specifically used to:

[0054] Determine a first area of ​​the target feature perception result area and a second area of ​​the area to be corrected according to the target feature perception result area and the area to be corrected;

[0055] The ratio of the first area to the second area is used as the similarity parameter.

[0056] In an optional embodiment, the upgrade identification unit is specifically used to:

[0057] When the similarity parameter is greater than a preset similarity threshold, it is determined that the point cloud data recognition is valid.

[0058] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for detecting the validity of point cloud data recognition according to the first aspect is implemented.

[0059] In a fourth aspect, an embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the computer program is executed by the processor, the method for detecting the effectiveness of point cloud data recognition of the first aspect is implemented.

[0060] In a fifth aspect, an embodiment of the present application further provides a computer program product, wherein the computer program product comprises computer instructions. When the computer instructions are executed by a computing device, the computing device may execute a method as described in any one of the first aspects.

[0061] The technical effects brought about by any one of the implementation methods in the second to fifth aspects can refer to the technical effects brought about by the corresponding implementation method in the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of a flow chart of a method for detecting the validity of point cloud data recognition provided in an embodiment of the present application;

[0063] Figure 2 A schematic diagram of a flow chart of obtaining a perception result area of ​​a method for detecting the effectiveness of point cloud data recognition provided in an embodiment of the present application;

[0064] Figure 3 A schematic diagram of a process for determining a similarity parameter of a method for detecting the effectiveness of point cloud data recognition provided in an embodiment of the present application;

[0065] Figure 4 A schematic diagram of a feature perception result area obtained by splicing a method for detecting the effectiveness of point cloud data recognition provided in an embodiment of the present application;

[0066] Figure 5 A schematic flow chart of another method for detecting the effectiveness of point cloud data recognition provided in an embodiment of the present application;

[0067] Figure 6 A schematic diagram of the structure of a device for detecting the validity of point cloud data recognition provided in an embodiment of the present application;

[0068] Figure 7 A schematic diagram of the structure of another device for detecting the validity of point cloud data recognition provided in an embodiment of the present application;

[0069] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0071] It should be noted that the terms "including" and "having" and their variations involved in the documents of this application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0072] The following are explanations of some of the terms that appear in the text:

[0073] (1) Point cloud data: Point cloud data refers to a set of sampling points with spatial coordinates obtained by lidar.

[0074] (2) LIDAR (Light Detection and Ranging) algorithm: The LIDAR algorithm is a program algorithm for object recognition based on the point cloud data of the laser radar. The laser radar uses a laser as the light source and adopts photoelectric detection technology to obtain point cloud data. The point cloud data is then processed by the LIDAR algorithm to identify the object.

[0075] At present, perception algorithms are usually used to identify point cloud data to achieve object recognition, such as the laser detection and ranging (LIDAR) algorithm. If the data in some areas of the point cloud data is not sufficient to support the perception algorithm to make judgments, it will lead to inaccurate recognition, such as missed detection and false detection. Therefore, the perception algorithm often needs to be upgraded in a targeted manner.

[0076] In related technologies, the upgraded perception algorithm needs to be tested in real scenes to detect whether the problem has been solved. The detection of the effectiveness of point cloud data recognition often involves many detection links and takes a long time. Therefore, when testing the effectiveness of point cloud data recognition, the detection efficiency is relatively low.

[0077] In order to solve the above problems, the embodiment of the present application provides a method for detecting the effectiveness of point cloud data recognition, firstly obtaining preset point cloud data; the point cloud data includes an area to be corrected, wherein the area to be corrected is an inaccurately identified point cloud area, then identifying the point cloud data to obtain a perception result area, and determining a target perception result area that overlaps with the area to be corrected based on the area to be corrected and the perception result area, and then determining the similarity parameter between the target perception result area and the area to be corrected; and judging the effectiveness of point cloud data recognition based on the similarity parameter. The method recognizes the preset point cloud data, and performs similarity calculation based on the recognition result, and judges the effectiveness of point cloud data recognition based on the similarity parameter obtained by the similarity calculation. The detection process does not require real-scene testing, and can more conveniently determine the effectiveness of point cloud data recognition, saving the time spent on detecting the effectiveness of point cloud data recognition, and improving the detection efficiency of the effectiveness of point cloud data recognition.

[0078] The technical solution provided in the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0079] Figure 1 The figure shows a flow chart of the effectiveness detection of point cloud data recognition provided by an embodiment of the present application. Figure 1 As shown, the method may include the following steps:

[0080] Step S101, obtaining preset point cloud data; the point cloud data includes the area to be corrected.

[0081] Among them, the area to be corrected is an area that is not accurately identified.

[0082] In specific implementation, the point cloud data includes the area to be corrected and the normal area. The normal area is the area that has been accurately identified.

[0083] In some embodiments of the present application, the preset point cloud data can be obtained by configuring through the following process: determining the point cloud data that is not accurately recognized by the perception algorithm; combining the point cloud data and the video image corresponding to the point cloud data to determine the inaccurate recognition time period and object activity area of ​​an inaccurately recognized object; when reading the point cloud data with a player and displaying the point cloud image corresponding to the point cloud data frame by frame on the display, an inaccurately recognized area frame is drawn within the object activity area of ​​the inaccurately recognized object. The interior of the inaccurately recognized area frame is used as the area to be corrected, and the inaccurately recognized area frame is used as the normal area.

[0084] In specific implementation, the point cloud data that is not accurately recognized by the perception algorithm can be determined by comparing the video image with the object recognition result of the perception algorithm. The inaccurately recognized area box is usually a rectangular box.

[0085] For example, it is assumed that the preset point cloud data Data_1 includes a region to be corrected Scope_Tar and a normal region Scope_fine. The region to be corrected Scope_Tar may be an inner region of a rectangular frame Rec_Tar, and the normal region Scope_fine may be an outer region of the rectangular frame Rec_Tar.

[0086] Step S102: identify the point cloud data to obtain a perception result area.

[0087] In specific implementation, the point cloud data is identified through the upgraded perception algorithm to obtain the perception result area. There may be multiple perception result areas obtained by identifying the point cloud data.

[0088] Exemplarily, the point cloud data Data_1 is identified through the upgraded perception algorithm to obtain multiple perception result areas (Scope_Test_1, Scope_Test_2, ..., Scope_Test_I, ..., Scope_Test_n).

[0089] In some embodiments of the present application, the point cloud data also includes a regular time range corresponding to the area to be corrected.

[0090] In some embodiments of the present application, the preset point cloud data can be obtained by configuring through the following process: determining the point cloud data that is not accurately identified by the perception algorithm; combining the point cloud data and the video image corresponding to the point cloud data to determine the inaccurate identification time period and object activity area of ​​an inaccurately identified object; when reading the point cloud data with a player and displaying the point cloud image corresponding to the point cloud data frame by frame on the display, drawing an inaccurately identified area frame in the object activity area of ​​the inaccurately identified object, and selecting a certain time range in the missed detection time period as the regular time range corresponding to the inaccurately identified area frame. The inside of the inaccurately identified area frame is used as the area to be corrected, and the inaccurately identified area frame is used as the normal area.

[0091] It is understandable that the regular time range may be part or all of the inaccurately identified time period.

[0092] In some embodiments of the present application, the preset point cloud data may further include multiple areas to be corrected, each of which is an inner area of ​​a corresponding inaccurately identified area frame. The regular time ranges corresponding to the inaccurately identified area frames of the areas to be corrected do not overlap.

[0093] Figure 2 A schematic diagram of a flow chart of obtaining a perception result area of ​​a method for detecting the effectiveness of point cloud data recognition provided in an embodiment of the present application. Figure 2 As shown, the point cloud data can be identified and the perception result area can be obtained by the following steps:

[0094] Step S201, determining target point cloud data corresponding to the area to be corrected according to the rule time range and the point cloud data.

[0095] In specific implementation, the regular time range includes the start time and end time for determining the point cloud data frame in the point cloud data. For example, assuming that the start time of the point cloud data frame is the 2s time point of the point cloud data, and the end time of the point cloud data frame is the 4s time point of the point cloud data, then the regular time range is a time range of 2s to 4s. Through this regular time range, the point cloud data frame between the 2s time point and the 4s time point of the point cloud data can be determined to obtain the target point cloud data corresponding to the area to be corrected.

[0096] Step S202: Identify the target point cloud data to obtain a perception result area.

[0097] In specific implementation, when the upgraded perception algorithm recognizes point cloud data, it recognizes each point cloud data frame within a regular time range.

[0098] The method of this embodiment provides a mechanism for acquiring target point cloud data, which only determines the perception result area corresponding to the target point cloud data, thereby making the process of generating the perception result area more efficient, further saving the time spent on detecting the effectiveness of point cloud data recognition, and improving the detection efficiency of the effectiveness of point cloud data recognition.

[0099] Step S103: determining a target perception result region that overlaps with the area to be corrected according to the area to be corrected and the perception result region.

[0100] In some embodiments of the present application, the perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data; according to the area to be corrected and the perception result area, determining the target perception result area that overlaps with the area to be corrected can be achieved through the following process: obtaining the perception result areas one by one, and each time a perception result area is obtained, performing the following operations: if the current perception result area overlaps with the area to be corrected within the perception result time range corresponding to the current perception result area, then the current perception result area is used as the target perception result area.

[0101] Exemplarily, it is assumed that the perception result time corresponding to the perception result area Scope_Test_I is (2.5s~3s), and the point cloud data Data_1 has 2 point cloud data frames in the time range of 2.5s~3s, including the starting point cloud data frame PC_i and the ending point cloud data frame PC_j. According to the area to be corrected and the perception result area, the target perception result area overlapping with the area to be corrected is determined, and specifically, the perception result areas (Scope_Test_1, Scope_Test_2, ..., Scope_Test_I, ..., Scope_Test_n) are obtained one by one, and each time a perception result area is obtained, the following repeated operations are performed: Only the current perception result area is Scope_Test_I as an example for explanation, if the current perception result area Scope_Test_I overlaps with the area to be corrected Scope_Tar within the perception result time range (2.5s~3s) corresponding to the current perception result area Scope_Test_I, then the current perception result area Scope_Test_I is used as the target perception result area.

[0102] Step S104, determining a similarity parameter between the target perception result area and the area to be corrected.

[0103] In some embodiments of the present application, the similarity parameter between the target perception result area and the area to be corrected is determined, such as Figure 3 As shown, the following steps are included:

[0104] Step S301, determining a feature perception result area according to the target perception result area and the corresponding target perception result time range.

[0105] The feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range.

[0106] For example, the target perception result area is Scope_Test_I, and the target perception result time range corresponding to the target perception result area is (2.5s~3s), and the point cloud data Data_1 has 2 point cloud data frames in the time range of 2.5s~3s. Figure 4 As shown, the trajectories 401 and 402 of each frame of the target perception result area Scope_Test_I within the corresponding target perception result time range (2.5s to 3s) are spliced ​​to obtain the area Scope_Test_Joint, and the area Scope_Test_Joint is used as the feature perception result area.

[0107] Step S302: determining a target feature perception result area that overlaps with the area to be corrected based on the feature perception result area and the area to be corrected.

[0108] For example, see Figure 4 According to the feature perception result area Scope_Test_Joint and the area to be corrected 400, a target feature perception result area 404 overlapping with the area to be corrected 400 is determined.

[0109] Step S303, obtaining a similarity parameter according to the target feature perception result area and the area to be corrected.

[0110] Exemplarily, a similarity parameter Similar is obtained according to the target feature perception result area 404 and the area to be corrected 400 .

[0111] The method of this embodiment splices the trajectory of each frame of the target perception result area within the corresponding target perception result time range to obtain a feature perception result area; then, based on the feature perception result area and the area to be corrected, a target feature perception result area overlapping with the area to be corrected is determined, and a similarity parameter is obtained based on the target feature perception result area and the area to be corrected. This method provides a mechanism for evaluating the association between the target perception result area and the area to be corrected, and can efficiently determine the similarity parameter between the target perception result area and the area to be corrected based on the target perception result time range and the target perception result area, further saving the time spent on detecting the effectiveness of point cloud data recognition and improving the detection efficiency of the effectiveness of point cloud data recognition.

[0112] In some embodiments of the present application, a similarity parameter is obtained based on the target feature perception result area and the area to be corrected, specifically: based on the target feature perception result area and the area to be corrected, a first area of ​​the target feature perception result area and a second area of ​​the area to be corrected are determined; and the ratio of the first area to the second area is used as the similarity parameter.

[0113] For example, Figure 4 Taking the target feature perception result area 404 and the area to be corrected 400 as an example, according to the target feature perception result area 404 and the area to be corrected 400, the first area S1 of the target feature perception result area 404 and the second area S2 of the area to be corrected 400 are determined, and the ratio of the first area S1 to the second area S2 is used as the similarity parameter Similar.

[0114] Step S105: judging the validity of point cloud data recognition based on the similarity parameter.

[0115] In some embodiments of the present application, the validity of point cloud data recognition is judged based on a similarity parameter, which may be specifically: when the similarity parameter is greater than a preset similarity threshold, the point cloud data recognition is judged to be valid.

[0116] In the embodiment of the present application, the similarity threshold can be set to different values ​​according to the specific needs of the perception algorithm upgrade. For example, the similarity threshold can be set to 85%, and if the similarity parameter Similar is greater than 85%, it is determined that the point cloud data recognition is valid.

[0117] The detection method for the effectiveness of point cloud data recognition provided in the embodiment of the present application first obtains preset point cloud data; the point cloud data includes an area to be corrected, wherein the area to be corrected is an inaccurately identified point cloud area, then the point cloud data is identified to obtain a perception result area, and based on the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected is determined, and then the similarity parameter between the target perception result area and the area to be corrected is determined; the effectiveness of point cloud data recognition is judged based on the similarity parameter. The method identifies the preset point cloud data and calculates the similarity based on the recognition result, thereby judging the effectiveness of point cloud data recognition based on the similarity parameter obtained by the similarity calculation. The detection process does not require real-life testing, and can more conveniently determine the effectiveness of point cloud data recognition, saving the time spent on detecting the effectiveness of point cloud data recognition, and improving the detection efficiency of the effectiveness of point cloud data recognition.

[0118] In some embodiments of the present application, a step of performing an abnormal alarm for point cloud data recognition validity detection is also included. After the point cloud data is recognized and the perception result area is obtained, if it is determined that the area to be corrected and the perception result area do not overlap, then an abnormal alarm for point cloud data recognition validity detection is performed.

[0119] In specific implementation, in the process of configuring and obtaining preset point cloud data, the area to be corrected avoids selecting point cloud data frames with two objects that overlap in position. When the upgraded perception algorithm recognizes point cloud data, it outputs a group of perception result areas for the point cloud data frames within the regular time range. Each group of perception result areas has at most one target perception result area, and the target perception result area has an overlapping part with the area to be corrected within its corresponding perception result time. If it is determined that the area to be corrected and the perception result area do not overlap based on the area to be corrected and the perception result area, an abnormal alarm for point cloud data recognition validity detection is performed.

[0120] In the method of this embodiment, if it is determined that the area to be corrected and the perception result area do not overlap, then an abnormal alarm for point cloud data recognition validity detection is performed. This method provides an abnormal alarm mechanism by setting up monitoring for the situation where the area to be corrected and the perception result area do not overlap, so that the validity of point cloud data recognition can be determined more efficiently, saving the time spent on the detection of the validity of point cloud data recognition, and improving the detection efficiency of the validity of point cloud data recognition.

[0121] Although the embodiments of the present application provide the operating steps of the method as shown in the above embodiments or the accompanying drawings, more or fewer operating steps may be included in the above method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided in the embodiments of the present application. The above method may be executed in the order of the method shown in the embodiments or the accompanying drawings or in parallel during the actual processing process or when the device is executed.

[0122] Figure 5 A flow chart of another method for detecting the validity of point cloud data recognition provided by an embodiment of the present application. Figure 5 As shown, the following steps may be included:

[0123] Step S501, obtaining preset point cloud data; the point cloud data includes a region to be corrected and a regular time range corresponding to the region to be corrected.

[0124] Among them, the area to be corrected is an area that is not accurately identified.

[0125] In the point cloud data, the point cloud area excluding the area to be corrected is the normal area. The normal area is the area that has been accurately identified.

[0126] Step S502: determining target point cloud data corresponding to the area to be corrected according to the rule time range and the point cloud data.

[0127] Step S503: Identify the target point cloud data to obtain a perception result area.

[0128] The perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when the point cloud data is recognized. When the target point cloud data is recognized, there may be multiple perception result areas.

[0129] Step S504: determining a target perception result region that overlaps with the area to be corrected according to the area to be corrected and the perception result region.

[0130] During specific implementation, the process of determining the target perception result area that overlaps with the area to be corrected is based on the area to be corrected and the perception result area. Specifically, it can be as follows: if the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, then the current perception result area is used as the target perception result area.

[0131] In an embodiment of the present application, perception result areas are acquired one by one. Each time a perception result area is acquired, if it is determined that the acquired perception result area overlaps with the area to be corrected within the perception result time range corresponding to the current perception result area, the acquired perception result area is used as the target perception result area.

[0132] Step S505, determining a feature perception result area according to the target perception result area and the corresponding target perception result time range.

[0133] The feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range.

[0134] Step S506, determining a target feature perception result area that overlaps with the area to be corrected based on the feature perception result area and the area to be corrected.

[0135] Step S507, obtaining a similarity parameter according to the target feature perception result area and the area to be corrected.

[0136] In some embodiments, a similarity parameter is obtained based on the target feature perception result area and the area to be corrected. The first area of ​​the target feature perception result area and the second area of ​​the area to be corrected can be determined based on the target feature perception result area and the area to be corrected; and the ratio of the first area to the second area is used as the similarity parameter.

[0137] Step S508: When the similarity parameter is greater than a preset similarity threshold, it is determined that the point cloud data recognition is valid.

[0138] The method of this embodiment does not require real-scene testing after the perception algorithm is upgraded. After the algorithm is upgraded, a new target feature perception result area is obtained by recognizing preset point cloud data, and the target feature perception result area is compared with the problem area of ​​the perception algorithm before the algorithm upgrade to judge the effectiveness of the point cloud data recognition. It can quickly discover whether there is a problem with the effectiveness of the point cloud data recognition, achieve efficient acquisition of the effectiveness of the point cloud data recognition, reduce time and material consumption, improve the detection efficiency of the effectiveness of the point cloud data recognition, and reduce monitoring costs.

[0139] Based on the same inventive concept, the embodiment of the present application also provides a device for detecting the effectiveness of point cloud data recognition. Since the device is a device corresponding to the method for detecting the effectiveness of point cloud data recognition in the embodiment of the present application, and the principle of solving the problem by the device is similar to that of the method, the implementation of the device can refer to the implementation process of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0140] Figure 6 FIG. 1 is a schematic diagram showing a structure of a device for detecting the validity of point cloud data recognition provided by an embodiment of the present application. The device for detecting the validity of point cloud data recognition, such as Figure 6 As shown, it includes: a data acquisition unit 601, a correction identification unit 602, an overlap determination unit 603, a similarity determination unit 604 and an upgrade identification unit 605; wherein,

[0141] The data acquisition unit 601 is used to acquire preset point cloud data, where the point cloud data includes a region to be corrected, wherein the region to be corrected is a point cloud region that is not accurately identified;

[0142] A correction identification unit 602 is used to identify the point cloud data and obtain a perception result area;

[0143] An overlap determination unit 603, configured to determine, based on the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected;

[0144] A similarity determination unit 604 is used to determine a similarity parameter between the target perception result area and the area to be corrected;

[0145] The upgrade identification unit 605 is used to determine the validity of the point cloud data identification based on the similarity parameter.

[0146] In an optional embodiment, the point cloud data further includes a regular time range corresponding to the area to be corrected; the correction identification unit 602 is specifically used to:

[0147] According to the rule time range and point cloud data, determine the target point cloud data corresponding to the area to be corrected;

[0148] Identify the target point cloud data and obtain the perception result area.

[0149] In an optional embodiment, the perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data; the overlap determination unit 603 is specifically used to:

[0150] If the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area.

[0151] In an optional embodiment, the similarity determination unit 604 is specifically configured to:

[0152] Determine a feature perception result area according to the target perception result area and the corresponding target perception result time range; the feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range;

[0153] According to the feature perception result area and the area to be corrected, determining a target feature perception result area that overlaps with the area to be corrected;

[0154] According to the target feature perception result area and the area to be corrected, a similarity parameter is obtained.

[0155] In an optional embodiment, the similarity determination unit 604 is specifically configured to:

[0156] Determine a first area of ​​the target feature perception result area and a second area of ​​the area to be corrected according to the target feature perception result area and the area to be corrected;

[0157] The ratio of the first area to the second area is used as a similarity parameter.

[0158] In an optional embodiment, the upgrade identification unit 605 is specifically used to:

[0159] When the similarity parameter is greater than a preset similarity threshold, it is determined that the point cloud data recognition is valid.

[0160] In an optional embodiment, if Figure 7 As shown, the device also includes:

[0161] The abnormal alarm unit 701 is used to identify the point cloud data. After obtaining the perception result area, if it is determined that the area to be corrected and the perception result area do not overlap, then an abnormal alarm for point cloud data recognition validity detection is performed.

[0162] Based on the same inventive concept as the above method embodiment, an electronic device is also provided in the embodiment of the present application. The electronic device can be used to detect the effectiveness of point cloud data recognition. In one embodiment, the electronic device can be a server, or a terminal device or other electronic device. In this embodiment, the structure of the electronic device can be as follows: Figure 8 As shown, it includes a memory 801 , a communication module 803 and one or more processors 802 .

[0163] The memory 801 is used to store computer programs executed by the processor 802. The memory 801 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and programs required for running the instant messaging function, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0164] The memory 801 may be a volatile memory, such as a random-access memory (RAM); the memory 801 may also be a non-volatile memory, such as a read-only memory, a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), or the memory 801 may be any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 801 may be a combination of the above memories.

[0165] The processor 802 may include one or more central processing units (CPU) or a digital processing unit, etc. The processor 802 is used to implement the above-mentioned point cloud data recognition validity detection method when calling the computer program stored in the memory 801.

[0166] The communication module 803 is used to communicate with terminal devices and other servers.

[0167] The specific connection medium between the memory 801, the communication module 803 and the processor 802 is not limited in the embodiment of the present application. Figure 8 In the embodiment, the memory 801 and the processor 802 are connected via a bus 804. The bus 804 is Figure 8 The connections between other components are shown in bold lines, and are not intended to be limiting. Bus 804 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0168] An embodiment of the present application also provides a computer storage medium, in which computer executable instructions are stored. The computer executable instructions are used to implement a method for detecting the validity of point cloud data recognition in any embodiment of the present application.

[0169] According to one aspect of the present application, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the detection method for the effectiveness of point cloud data recognition in the above-mentioned embodiment. The program product can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, - but not limited to - an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0170] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for detecting the effectiveness of point cloud data recognition, characterized in that: include: Acquire preset point cloud data, wherein the point cloud data includes a to-be-corrected area, wherein the to-be-corrected area is a point cloud area that is not accurately identified; Identifying the point cloud data to obtain a perception result area; Determine, according to the area to be corrected and the perception result area, a target perception result area that overlaps with the area to be corrected; Determining a similarity parameter between the target perception result area and the area to be corrected; Determining the validity of the point cloud data recognition based on the similarity parameter; The perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data; The step of determining, based on the area to be corrected and the perception result area, a target perception result area overlapping with the area to be corrected includes: If the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area; The determining of the similarity parameter between the target perception result area and the area to be corrected includes: Determine a feature perception result area according to the target perception result area and the corresponding target perception result time range; the feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range; Determine, according to the feature perception result area and the area to be corrected, a target feature perception result area that overlaps with the area to be corrected; The similarity parameter is obtained according to the target feature perception result area and the area to be corrected.

2. The method according to claim 1, characterized in that The point cloud data also includes a regular time range corresponding to the area to be corrected; The step of identifying the point cloud data to obtain a perception result area includes: Determining target point cloud data corresponding to the area to be corrected according to the rule time range and the point cloud data; The target point cloud data is identified to obtain a perception result area.

3. The method according to claim 1, characterized in that After identifying the point cloud data and obtaining the perception result area, the method further includes: If it is determined based on the area to be corrected and the perception result area that the area to be corrected and the perception result area do not overlap, an abnormal alarm for point cloud data recognition validity detection is performed.

4. The method according to claim 1, characterized in that: The obtaining of the similarity parameter according to the target feature perception result area and the area to be corrected includes: Determine a first area of ​​the target feature perception result area and a second area of ​​the area to be corrected according to the target feature perception result area and the area to be corrected; The ratio of the first area to the second area is used as the similarity parameter.

5. The method according to claim 4, characterized in that The determining the validity of the point cloud data recognition based on the similarity parameter includes: When the similarity parameter is greater than a preset similarity threshold, it is determined that the point cloud data recognition is valid.

6. A device for detecting the validity of point cloud data recognition, characterized in that: include: A data acquisition unit, used for acquiring preset point cloud data; The point cloud data includes a to-be-corrected area, wherein the to-be-corrected area is an inaccurately identified point cloud area; A correction recognition unit is used to recognize the point cloud data to obtain a perception result area; an overlap determination unit, configured to determine, based on the to-be-corrected area and the perception result area, a target perception result area that overlaps with the to-be-corrected area; A similarity determination unit, used to determine a similarity parameter between the target perception result area and the area to be corrected; An upgraded identification unit is used to determine the validity of the point cloud data identification based on the similarity parameter; The perception result area has a corresponding perception result time; the perception result time is used to determine the starting point cloud data frame and the ending point cloud data frame corresponding to the perception result area when identifying the point cloud data; the overlap determination unit is specifically used to: If the current perception result area and the area to be corrected overlap within the perception result time range corresponding to the current perception result area, the current perception result area is used as the target perception result area; The similarity determination unit is specifically used to: Determining a feature perception result area according to the target perception result area and the corresponding target perception result time range; The feature perception result area is an area obtained by splicing the trajectory of each frame of the target perception result area within the corresponding target perception result time range; Determine, according to the feature perception result area and the area to be corrected, a target feature perception result area that overlaps with the area to be corrected; The similarity parameter is obtained according to the target feature perception result area and the area to be corrected.

7. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the computer program is executed by the processor, the method according to any one of claims 1 to 5 is implemented.

Images