Target fusion evaluation method and device, electronic equipment and storage medium

Abstract

The application discloses a target fusion evaluation method and device, electronic equipment and a storage medium. The target fusion evaluation method comprises the following steps: matching true value data meeting a preset condition with a fusion target; determining the scores of various indexes of the fusion target according to a preset evaluation standard; and determining a final score according to the mean value of the weighted sum of the scores of the various indexes. The application has the beneficial effects that the true value data meeting the preset condition is matched with the fusion target, the scores of the various indexes of the fusion target are determined according to the preset evaluation standard, the final score is determined according to the mean value of the weighted sum of the scores of the various indexes, the fusion algorithm is optimized, and the reliability of target detection is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automobile perception technology, and in particular to a target fusion evaluation method and device, electronic equipment and a storage medium. BACKGROUND

[0002] With the development of intelligent driving technology, the target detection technology of the surrounding environment of the vehicle is gradually mature. Due to the limitations of incomplete scene coverage and low accuracy of a single sensor, multi-sensor multi-target tracking has gradually developed. In order to analyze the tracking performance and optimize the fusion tracking algorithm, evaluation indicators need to be developed. Although there are various methods for implementing target tracking, there are few studies on evaluation methods for tracking results, and the most influential are the MOTA (Multiple Object Tracking Accuracy) and MOTP (Multiple Object Tracking Precision) evaluation indicators proposed by the MOT Challenge competition.

[0003] The MOT Challenge competition mainly detects the performance of multi-target tracking of pedestrians, and evaluates the tracking performance of various tracking algorithms for pedestrians by calculating the accuracy, precision, total number of false positives and total number of missed reports of multi-target tracking. Although these evaluation standards can better reflect the tracking performance of the algorithm, the input of this competition is the original camera image, and the sensor is single, which cannot be directly applied to vehicles that need to be equipped with multi-sensor fusion. In addition, intelligent driving focuses on objects other than pedestrians, and also needs to focus on car targets on the road in order to realize functions such as following a car and emergency braking.

[0004] Therefore, there are few evaluation standards for intelligent driving fusion target detection, and most evaluation standards use original sensor signals such as camera pictures and millimeter wave radar point clouds as input for evaluation. However, intelligent driving technology requires many sensors, and it is difficult to implement raw data fusion, so the problem of target-level fusion of the output of each sensor needs to be improved. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the purpose of the present application is to provide a target fusion evaluation method and device, electronic equipment and a storage medium.

[0006] To achieve this purpose, the present application adopts the following technical solutions:

[0007] In a first aspect, a target fusion evaluation method includes the following steps:

[0008] Matching the true value data and the fusion target that meet the preset conditions with each other;

[0009] The final score is determined by the mean of the weighted sum of the scores of each index of the fusion target according to the preset evaluation criteria.

[0010] In one embodiment, the step of matching the true value data satisfying the preset condition with the fusion target further comprises:

[0011] The parameters of the fusion target are determined, wherein the parameters at least include: target number, target longitudinal distance, target lateral distance, target longitudinal relative speed and target lateral relative speed.

[0012] In one embodiment, the step of matching the true value data satisfying the preset condition with the fusion target comprises:

[0013] The true value data and the fusion target are synchronized according to the sampling period of the true value system and the output period of the fusion target.

[0014] The synchronized true value data and the fusion target are matched according to the Euclidean distance.

[0015]

[0016]

[0017] wherein, x t is the true value target longitudinal distance, y t is the true value target lateral distance, gate is the threshold value, and the coefficients k and b are determined by the sensor error characteristics.

[0018] In one embodiment, the step of matching the true value data satisfying the preset condition with the fusion target further comprises:

[0019] If the fusion target matched with the true value target in the last frame still exists in the current frame and is within the threshold value gate, the fusion target in the last frame is still matched with the true value target.

[0020] In one embodiment, the step of matching the true value data satisfying the preset condition with the fusion target further comprises:

[0021] If the fusion target in the last frame disappears in the current frame or the fusion target in the last frame exceeds the threshold value gate in the current frame, the step of matching the true value data satisfying the preset condition with the fusion target is returned to.

[0022] In one embodiment, the preset evaluation criteria include the following indexes:

[0023] The first index is used to represent the proportion of the number of times of different fusion targets in two consecutive frames to the total number of frames.

[0024] The second index is used to represent the proportion of the total frame number of the times that two or more fusion targets appear near the true value target, wherein the appearance of a continuous frame is counted only once.

[0025] The third index is used to represent the proportion of the total frame number of the times of state jumping.

[0026] In an embodiment, the preset evaluation criteria further include the following indexes:

[0027] The fourth index is used to represent the longitudinal position deviation of the fusion target.

[0028] The fifth index is used to represent the lateral position deviation of the fusion target.

[0029] The sixth index is used to represent the longitudinal relative speed deviation of the fusion target.

[0030] The seventh index is used to represent the lateral relative speed deviation of the fusion target.

[0031] The eighth index is used to represent the root mean square error of the longitudinal position of the fusion target.

[0032] The ninth index is used to represent the root mean square error of the lateral position of the fusion target.

[0033] The tenth index is used to represent the root mean square error of the longitudinal relative speed of the fusion target.

[0034] The eleventh index is used to represent the root mean square error of the lateral relative speed of the fusion target.

[0035] In an embodiment, the step of determining the final score according to the mean value of the weighted sum of the scores of each index of the fusion target determined according to the preset evaluation criteria includes:

[0036] The score interval of each index is set respectively, and the score of each index is calculated by linear interpolation:

[0037]

[0038] Wherein, KPI represents each index, up represents the maximum allowable error, and low represents the expected error value.

[0039] According to the importance of each index, the weighting coefficient is set, and the final score is obtained according to the following format:

[0040]

[0041] Wherein, P is the final total score, ω i is the weighting coefficient of the i-th evaluation index, and the sum of all weighting coefficients is 1, point i is the score of the i-th evaluation index.

[0042] In a second aspect, a target fusion evaluation device comprises:

[0043] A first module is configured to match the true value data satisfying the preset condition with the fusion target;

[0044] A second module is configured to determine the index scores of the fusion target according to the preset evaluation standard, and determine the final score by averaging the weighted sum of the index scores.

[0045] In a third aspect, an electronic device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the target fusion evaluation method as described above when executing the computer program.

[0046] In a fourth aspect, a computer readable storage medium stores computer instructions, and the computer instructions make the computer execute the steps of the target fusion evaluation method as described above.

[0047] The present application has the following advantages: by matching the true value data satisfying the preset condition with the fusion target, determining the index scores of the fusion target according to the preset evaluation standard, and determining the final score by averaging the weighted sum of the index scores, the fusion algorithm is optimized, and the reliability of target detection is improved.

[0048] Additional aspects and advantages of the present application will be made apparent from the following description, which, taken together with the accompanying drawings, will provide a better understanding of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which:

[0050] Figure 1 is a structural schematic diagram of an exemplary system architecture suitable for the target fusion evaluation method of the present embodiment;

[0051] Figure 2 is a flowchart of the target fusion evaluation method provided by the present embodiment;

[0052] Figure 3 is a schematic diagram of the change of X error with longitudinal distance;

[0053] Figure 4 is a schematic diagram of the change of Y error with longitudinal distance;

[0054] Figure 5 is a schematic diagram of the change of V x error with longitudinal distance;

[0055] Figure 6 is V y a schematic diagram of error variation with longitudinal distance;

[0056] Figure 7 is a relationship between the front radar longitudinal distance deviation and the target relative velocity;

[0057] Figure 8 is a structural schematic diagram of the target fusion evaluation device provided by the embodiment;

[0058] Figure 9 is a structural schematic diagram of the electronic device provided by the embodiment. DETAILED DESCRIPTION

[0059] The application will be further described below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the application, and not to limit the application. In addition, it should be noted that, for the convenience of description, only the parts related to the application are shown in the drawings, not all the structures.

[0060] In the description of the application, unless otherwise explicitly specified and limited, the terms "connected", "connected", "fixed" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or it can be integrated; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the internal communication of two elements or the interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in the application can be understood according to the specific circumstances.

[0061] Those skilled in the art can understand that, unless specifically stated, the singular forms "a", "an" and "said" used herein also include the plural forms. It should be further understood that the phrase "comprising" used in the specification of the present application means that the features, integers, steps, operations, elements and / or components exist, but does not exclude the existence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0062] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as that generally understood by those skilled in the art to which the present application belongs. It should also be understood that terms such as those defined in a general dictionary should be understood as having meanings consistent with those in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as in the embodiments of the present application.

[0063] Figure 1An exemplary system architecture 100 of a target fusion evaluation method to which embodiments of the present disclosure can be applied is shown.

[0064] As shown in Figure 1 The system architecture 100 can include a terminal device 101, a network 102, a server 103, and an obstacle sensing device 104. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 can include various connection types, such as wired, wireless communication links, or fiber optic cables, and the like.

[0065] A user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, and the like. Various communication client applications can be installed on the terminal device 101, such as monitoring applications, image processing applications, instant messaging tools, and the like.

[0066] The terminal device 101 can be various electronic devices, including but not limited to mobile terminals such as in-vehicle terminals, mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet PCs), PMPs (Portable Multimedia Players), and the like, as well as fixed terminals such as digital TVs, desktop computers, and the like.

[0067] The server 103 can be a server that provides various services, such as a background data processing server that processes at least two target sensing location information uploaded by the terminal device 101. The background data processing server can perform target fusion based on the received at least two target sensing location information to obtain fusion location information, evaluation information of a target obstacle, and the like.

[0068] The obstacle sensing device 104 is used to collect sensing information for a target object. For example, the obstacle sensing device 104 can include a camera, a lidar, and the like, to collect images, point cloud data, and the like. The obstacle sensing device 104 can be connected to the terminal device 101, or connected to the server 103 through the network 102. The obstacle sensing device 104 can be disposed on a mobile device such as a vehicle, an aircraft, a ship, or the like, or can be disposed at a fixed location.

[0069] It should be noted that the target fusion evaluation method provided by the embodiments of the present disclosure can be executed by the server 103, or can be executed by the terminal device 101. Correspondingly, the target fusion evaluation apparatus can be disposed in the server 103, or can be disposed in the terminal device 101.

[0070] It should be understood that Figure 1The number of terminal devices, networks, servers and obstacle sensing devices in the system architecture is only illustrative. According to the implementation needs, there can be any number of terminal devices, networks, servers and obstacle sensing devices. For example, in the case where at least two target sensing position information does not need to be obtained remotely, the above system architecture can not include a network and a server, but only a terminal device and an obstacle sensing device.

[0071] The embodiment provides a target fusion evaluation method, which can be applied to various scenes, including but not limited to cloud technology, artificial intelligence, intelligent transportation, assisted driving, intelligent vehicles, etc.

[0072] In an embodiment, a plurality of obstacle sensing devices 104 can be installed on the vehicle body. The obstacle sensing devices 104 can include a combination of one or more of a laser sensor, a temperature sensor, an infrared sensor, an acceleration sensor, etc. Each obstacle sensing device 104 can be connected to a corresponding sensing processing unit. The obstacle sensing device 104 transmits the collected real-time data to the connected sensing processing unit for data preprocessing and data correlation. After the sensing processing unit completes data processing, the data is output to the data fusion module for sensing data fusion. Finally, the data fusion module outputs the fused data to the application end. The application end can visualize the fused data in the form of an image or other forms, or can control the vehicle to perform corresponding actions such as acceleration, bypassing obstacles, turning, etc. based on the fused data, or can perform fault prediction or troubleshooting based on the fused data. The application of specific fused data is not limited here.

[0073] Figure 2 is a flowchart of the target fusion evaluation method provided by the embodiment, as shown in Figure 2 , the method comprises steps S10-S30.

[0074] S10: Determine the true value data and the parameters of the fusion target.

[0075] The target data output by the obstacle sensing device 104 is collected in real time through the automobile CAN line, and at the same time, the true value system RT-range collects the real state of the target object to ensure spatiotemporal consistency. The sensor target data is input into the Simulnk fusion model built.

[0076] The parameters at least include: target number ID, target longitudinal distance x, target transverse distance y, target longitudinal relative speed v x and target transverse relative speed v y .

[0077] Step S20, match the true value data and the fusion target that meet the preset conditions with each other.

[0078] Specifically, step S20 comprises S201, synchronizing the true value data and the fusion target according to the true value system sampling period and the fusion target output period.

[0079] The true value system adopts RT-range, in the embodiment, the sampling period of the true value system is 20 ms, and the fusion target output period is 60 ms. In order to ensure the accuracy of the calculation result, it is necessary to find the true value data closest to the output time of the fusion target for processing. When the time difference between the fusion result and the true value is not more than 20 ms, the true value data at this time is output for the next calculation, otherwise the true value data is initialized and assigned as 0, and does not enter the target matching and index calculation.

[0080] Step S20 comprises S202, matching the synchronized true value data and the fusion target according to the Euclidean distance:

[0081]

[0082] Wherein, x t is the true value target longitudinal distance, y t is the true value target transverse distance, gate is the threshold value, and the coefficients k and b are determined by the sensor error characteristics.

[0083] Since the farther the distance, the greater the sensor measurement error, the threshold value should change with the distance, and the calculation formula is as follows:

[0084]

[0085] In the embodiment, k is 0.14 and b is 6.

[0086] It should be noted that, in order to avoid the existence of close-range clutter interference with the matching of the real target and the fusion target in the middle, the nearest neighbor algorithm is used to match at the beginning, and the fusion target ID matched with the true value target will be remembered in the subsequent. That is, if the fusion target matched with the true value target in the last frame still exists in the current frame, and is within the threshold gate, then even if there may be a fusion target closer to the true value in the current frame, the fusion target in the last frame is still matched with the true value target. If the fusion target in the last frame disappears in the current frame or the fusion target in the last frame exceeds the threshold gate in the current frame, the nearest neighbor algorithm is used to match again, and if there is no suitable fusion target to match finally, the following steps are skipped and the next frame is returned to step S201.

[0087] Step S30, determining the scores of each index of the fusion target according to the preset evaluation standard, and determining the final score by weighted sum of the scores of each index.

[0088] It should be noted that the preset evaluation standard includes the following indexes:

[0089] A first index KPI1 is used to represent a ratio of different times of fusion targets in front and back two frames to total frame numbers.

[0090] A second index KPI2 is used to represent a ratio of times of two or more fusion targets appearing near a true value target to total frame numbers, wherein the appearance of consecutive frames is counted only once.

[0091] A third index KPI3 is used to represent a ratio of times of state jumps to total frame numbers.

[0092] It should be noted that the KPI1-KPI3 can reflect the tracking stability of the fusion algorithm, and can also be used to judge whether the fusion algorithm has a logical error.

[0093] The KPI1 represents the frequent degree of ID jumps of the fusion target in the whole tracking process, which can be counted once by comparing whether the fusion target ID matched with the true value target in the front and back frames changes, and finally the ratio of the jump times to the total tracking frame numbers is counted.

[0094] The KPI2 calculates the ratio of the times of target splitting in the tracking process to the total tracking frame numbers, and judges whether there is another fusion target in the close range of the fusion target and maintains a certain frame number, and if there is such a target, the count is added by 1.

[0095] The KPI3 calculates the ratio of the times of target state jumps to the total tracking frame numbers, and the formula for judging the state jump is as follows:

[0096]

[0097] Value raw represents the state value of the fusion target in the current frame, including distance, relative speed, Value truedate represents the state value of the fusion target in the last frame, Value truedate represents the true value in the current frame, represents the true value in the last frame, and δ represents a tolerance coefficient. When the above conditions are met, the state jump count is added by 1, and finally the ratio of the state jump times to the total tracking frame numbers is calculated.

[0098] The preset evaluation standard further includes the following indexes:

[0099] A fourth index KPI4 is used to represent a longitudinal position deviation of the fusion target.

[0100] A fifth index KPI5 is used to represent a lateral position deviation of the fusion target.

[0101] A sixth index KPI6 is used to represent a longitudinal relative speed deviation of the fusion target.

[0102] A seventh index KPI7 is used to represent a lateral relative speed deviation of the fusion target.

[0103] an eighth index KPI 8 for representing a root mean square error of a longitudinal position of the fusion target;

[0104] a ninth index KPI 9 for representing a root mean square error of a lateral position of the fusion target;

[0105] a tenth index KPI 10 for representing a root mean square error of a longitudinal relative speed of the fusion target;

[0106] an eleventh index KPI 11 for representing a root mean square error of a lateral relative speed of the fusion target.

[0107] The KPI 4-KPI 11 indicates the error between the fusion target position and the true value, the bias reflects the accuracy of the fusion target state, and the root mean square error reflects the dispersion degree of the fusion target state, and the calculation formula is as follows:

[0108]

[0109]

[0110] wherein, KPI Bias is the bias value, KPI RMSE is the root mean square error value, and N represents the total number of frames.

[0111] All the above indexes reflect the stability and accuracy of the fusion algorithm, can judge whether there is a bug in the internal logic of the fusion model, can be used as a basis for adjusting the related calibration quantities of the fusion algorithm such as the noise in the Kalman filter, and can calculate the evaluation results of each working condition according to different working conditions and analyze the reasons for the differences between each working condition.

[0112] Figures 3-6 The trend of several errors can be seen, and it can be seen that the farther the longitudinal distance is, the greater the error is.

[0113] In addition, not only the error of the fusion result, but also the parameter error of each obstacle sensing device 104 can be calculated by the above method. If the error of a certain obstacle sensing device 104 is smaller, the weight of the obstacle sensing device 104 can be adjusted when the obstacle sensing device 104 detects the target. At the same time, the influence of the target distance, speed and other factors on the error distribution of each obstacle sensing device 104 can be analyzed, and the target data can be compensated before the target data is input into the fusion model, so as to reduce the error of the input end and further reduce the error of the fusion target,

[0114] The step S30 further comprises: setting a score interval for each index respectively, and calculating the score of each index by using linear interpolation:

[0115]

[0116] Wherein, KPI refers to each index, up represents the maximum allowable error, and low represents the expected error value.

[0117] Then, according to the importance of each index, the weighting coefficient is set, and the final score is obtained according to the following format:

[0118]

[0119] Wherein, P is the final total score, ω i is the weighting coefficient of the i-th evaluation index, and the sum of all weighting coefficients is 1, point i is the score of the i-th evaluation index.

[0120] Table 1 is the calculation result of the fusion evaluation index.

[0121]

[0122] Table 1

[0123] By analyzing the evaluation index results in Table 1, for the data with abnormal evaluation index, the vulnerabilities in the fusion model can be traced back to find, and the fusion model calibration value can be changed according to the results to optimize the fusion model. At the same time, the evaluation index under different working conditions can be calculated, the difference of the evaluation index is compared, and the poor working condition is processed separately according to the fusion algorithm.

[0124] It should be noted that in addition to improving the fusion algorithm by the fusion result KPI value, the position and speed error of each sensor under different working conditions can be calculated, the difference between the position and speed error of the obstacle sensing device 104 (such as radar and camera) is analyzed, the weight of the smaller error is increased, and the fusion target error is reduced. In addition, the error characteristics of the obstacle sensing device 104 at different speeds and positions can be analyzed, and one kind of error can be compensated separately, or the fusion target error can be reduced. As Figure 7 indicated, the longitudinal distance deviation of the front radar is negatively correlated with the target speed.

[0125] The target fusion evaluation method provided in the embodiment matches the true value data meeting the preset condition with the fusion target, determines the index score of the fusion target according to the preset evaluation standard, and determines the final score by weighted sum of the mean value of the index scores, thereby optimizing the fusion algorithm and improving the reliability of target detection.

[0126] The embodiment also provides a target fusion evaluation device, as Figure 8 indicated, the target fusion evaluation device includes a first module 31 and a second module 32.

[0127] The first module 31 is configured to match the true value data satisfying the preset condition with the fusion target.

[0128] The second module 32 is configured to determine an index score of the fusion target according to a preset evaluation standard, and determine a final score according to a mean value of weighted summation of the index scores.

[0129] It should be noted that the target fusion evaluation apparatus provided in the embodiment can also be a computer program (including program code) running in a computer device. For example, the target fusion evaluation apparatus is an application program, which can be used to execute corresponding steps in the above-mentioned method provided in the embodiment.

[0130] In some possible implementation manners, the target fusion evaluation apparatus provided in the embodiment can be implemented in a combination of software and hardware. For example, the target fusion evaluation apparatus in the embodiment can be a processor in the form of a hardware decoding processor, which is programmed to execute the target fusion evaluation method provided in the embodiment. For example, the processor in the form of a hardware decoding processor can be one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs) or other electronic elements.

[0131] In some possible implementation manners, the target fusion evaluation apparatus provided in the embodiment can be implemented in a software manner. The target fusion evaluation apparatus can be software in the form of a program and a plug-in, and includes a series of modules to implement the target fusion evaluation method provided in the embodiment.

[0132] The target fusion evaluation apparatus provided in the embodiment matches the true value data satisfying the preset condition with the fusion target, determines an index score of the fusion target according to a preset evaluation standard, and determines a final score according to a mean value of weighted summation of the index scores, thereby optimizing the fusion algorithm and improving the reliability of target detection.

[0133] The embodiment of the present application further provides an electronic device, Figure 9 is a structural schematic diagram of the electronic device of the embodiment of the present application, like Figure 9As shown, the electronic device 1000 in the embodiment can include a processor 1001, a network interface 1004 and a memory 1005, in addition, the above-mentioned electronic device 1000 can further include a user interface 1003, and at least one communication bus 1002. Wherein, the communication bus 1002 is used to realize the connection communication between the components. Wherein, the user interface 1003 can include a display screen (Display), a keyboard (Keyboard), and the optional user interface 1003 can further include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface). The memory 1004 can be a high-speed RAM memory, or a non-volatile memory, for example, at least one disk storage. The memory 1005 can optionally be at least one storage device located away from the aforementioned processor 1001. As Figure 9 As shown, the memory 1005 as a computer readable storage medium can include an operating system, a network communication module, a user interface module and a device control application.

[0134] As shown, the electronic device 1000, the network interface 1004 can provide network communication function; and the user interface 1003 is mainly used to provide the interface for the user to input; and the processor 1001 can be used to call the device control application stored in the memory 1005, to realize: Figure 9

[0135] Matching the true value data meeting the preset condition with the fusion target with each other;

[0136] According to the preset evaluation standard, the scores of each index of the fusion target are determined, and the mean value of the weighted sum of the scores of each index is determined as the final score.

[0137] It should be understood that in some possible embodiments, the above-mentioned processor 1001 can be a central processing unit (central processing unit, CPU), and the processor can also be other general-purpose processors, DSPs, ASICs, FPGAs or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The memory can include read-only memory and random access memory, and provide instructions and data for the processor. A part of the memory can also include a non-volatile random access memory. For example, the memory can also store device type information.

[0138] ​In specific implementation, the electronic device 1000 can execute the implementation manners provided by each step of the control method by each function module built-in the electronic device 1000, and the implementation manners provided by each step are specifically referable, and details are not described herein.

[0139] The electronic device provided in the embodiment can match the true value data satisfying the preset condition with the fusion target, determine the index score of the fusion target according to the preset evaluation standard, determine the final score according to the mean of the weighted sum of the index scores, optimize the fusion algorithm, and improve the reliability of target detection.

[0140] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement each step of the target fusion evaluation method in the above embodiment, and the implementation manners provided by each step are specifically referable, and details are not described herein.

[0141] The computer readable storage medium provided in the embodiment can match the true value data satisfying the preset condition with the fusion target, determine the index score of the fusion target according to the preset evaluation standard, determine the final score according to the mean of the weighted sum of the index scores, optimize the fusion algorithm, and improve the reliability of target detection.

[0142] It should be understood that, although each step in the flowchart of the accompanying drawings is displayed in sequence according to the indication of the arrow, these steps are not necessarily executed in sequence according to the indication of the arrow. Unless explicitly stated herein, the execution of these steps is not strictly limited in sequence, and they can be executed in other sequences. Moreover, at least part of the steps in the flowchart of the accompanying drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0143] The above only describes some embodiments of the present application, and it should be pointed out that, for those skilled in the art, without departing from the principles of the present application, some improvements and refinements can be made, and these improvements and refinements should be regarded as the protection scope of the present application.

Claims

1. A method for evaluating target fusion, characterized by, The method comprises the following steps: matching the true value data meeting the preset condition with the fusion target; determining the scores of various indexes of the fusion target according to the preset evaluation standard, and determining the final score by averaging the weighted sum of the scores of the various indexes; the preset evaluation standard comprises the following indexes: a first index for representing the ratio of the number of times of changes of the fusion target before and after two frames to the total number of frames; a second index for representing the ratio of the number of times of the appearance of two or more fusion targets near the true value target to the total number of frames, wherein the appearance of consecutive frames is counted only once; a third index for representing the ratio of the number of times of state jumps to the total number of frames; the first index is KPI 1, which is obtained by comparing whether the fusion target ID matching the true value target before and after the frames changes, counting once for each change, and finally counting the ratio of the number of jumps to the total number of tracking frames; the second index is KPI 2, which is obtained by judging whether there is another fusion target in the close range of the fusion target and maintaining a certain number of frames, and counting 1 if there is such a target; the third index is KPI 3, which is obtained by calculating the ratio of the number of times of the appearance of the target state jump to the total number of tracking frames, and the formula for judging the state jump is as follows: ， wherein, represents the fusion target state value of the current frame, including distance, relative speed, represents the fusion target state value of the previous frame, represents the true value of the current frame, represents the true value of the previous frame, represents the tolerance coefficient.

2. The target fusion evaluation method according to claim 1, characterized by, Before the step of matching the true value data meeting the preset condition with the fusion target, the method further comprises: determining the parameters of the fusion target, wherein the parameters at least include: target number, target longitudinal distance, target lateral distance, target longitudinal relative speed and target lateral relative speed.

3. The target fusion evaluation method according to claim 2, characterized by, The step of matching the true value data meeting the preset condition with the fusion target comprises: synchronizing the true value data and the fusion target according to the sampling period of the true value system and the output period of the fusion target; matching the synchronized true value data and the fusion target according to the Euclidean distance: wherein is a true value target longitudinal distance, is a true value target lateral distance, is a threshold value, the coefficient and are determined by sensor error characteristics.

4. The target fusion evaluation method according to claim 3, characterized by, The step of matching the true value data meeting the preset condition with the fusion target further comprises: if the fusion target matching the true value target in the last frame still exists in the current frame and is within the threshold gate, the fusion target in the last frame is still matched with the true value target.

5. The target fusion evaluation method according to claim 4, characterized by, The step of matching the true value data meeting the preset condition with the fusion target further comprises: if the fusion target in the last frame disappears in the current frame or the fusion target in the last frame exceeds the threshold gate in the current frame, return to the step of matching the true value data meeting the preset condition with the fusion target.

6. The target fusion evaluation method of claim 1, wherein, The preset evaluation standard further comprises the following indexes: a fourth index for representing the longitudinal position deviation of the fusion target; a fifth index for representing the lateral position deviation of the fusion target; a sixth index for representing the longitudinal relative speed deviation of the fusion target; a seventh index for representing the lateral relative speed deviation of the fusion target; an eighth index for representing the root mean square error of the longitudinal position of the fusion target; a ninth index for representing the root mean square error of the lateral position of the fusion target; a tenth index for representing the root mean square error of the longitudinal relative speed of the fusion target; an eleventh index for representing the root mean square error of the lateral relative speed of the fusion target.

7. The target fusion evaluation method according to claim 6, characterized by, The step of determining the scores of various indexes of the fusion target according to the preset evaluation standard, and determining the final score by averaging the weighted sum of the scores of the various indexes comprises: Score intervals are set for each index respectively, and scores of each index are calculated by linear interpolation: wherein, denotes each indicator, denotes the maximum allowable error, denotes the expected error value; According to the importance of each index, a weighting coefficient is set, and the final score is obtained according to the following format: wherein, is the final total score, is the weighting factor for the i-th evaluation criterion, with the sum of all weighting factors being 1, is the score for the i-th evaluation criterion.

8. An object fusion evaluation apparatus characterized by comprising: Comprise: The first module is used to match the true value data and the fusion target that meet the preset condition with each other; The second module is used to determine the index score of the fusion target according to the preset evaluation standard, and the final score is determined by the average value of the weighted sum of the index scores; The preset evaluation standard includes the following indexes: The first index is used to represent the ratio of the number of times that the fusion target before and after changes to the total frame number; The second index is used to represent the ratio of the number of times that two or more fusion targets appear near the true value target to the total frame number, wherein the appearance of consecutive frames is counted only once; The third index is used to represent the ratio of the number of times that the state jumps to the total frame number; The first index is KPI 1, which counts the number of times that the fusion target ID changes by comparing the matching fusion target ID of the previous frame and the next frame with the true value target, and counts once for each change, and finally counts the ratio of the number of times that the state jumps to the total tracking frame number; The second index is KPI 2, which judges whether there is another fusion target in the close range of the fusion target and maintains a certain number of frames, and counts 1 if there is such a target; The third index is KPI 3, which calculates the ratio of the number of times that the target appears state jump to the total tracking frame number, and the formula for judging state jump is as follows: ， wherein, denotes the fusion target state value of the current frame, including distance, relative speed, denotes the fusion target state value of the previous frame, denotes the true value of the current frame, denotes the true value of the previous frame, denotes the tolerance coefficient.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the steps of the target fusion evaluation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, and the computer instructions make the computer execute the steps of the target fusion evaluation method according to any one of claims 1 to 7.

Images