Evaluation methods, devices, and equipment for autonomous driving models

Abstract

The present application provides a method, apparatus, and device for evaluating an autonomous driving model. The method includes: obtaining a pre-processed file and processing the pre-processed file based on a preset first autonomous driving model to obtain a first output file; sending the pre-processed file to at least one domain controller based on a network file system (NFS) sharing service; receiving a second output file sent by at least one domain controller based on the NFS sharing service; and determining an evaluation result based on the first output file and the second output file sent by at least one domain controller. The method of the present application can reduce the model evaluation gap between different types of devices, obtain more accurate evaluation results, and improve the efficiency of autonomous driving model evaluation.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to an evaluation method, device, and equipment for an autonomous driving model. Background Art

[0002] Deploy autonomous driving models in smart vehicles for model prediction. In order to determine whether the autonomous driving models meet application requirements, the trained autonomous driving models need to be evaluated in different domain controllers.

[0003] In existing technologies, algorithm engineers usually train and evaluate autonomous driving models in cloud clusters to obtain evaluation results; at the same time, they evaluate the autonomous driving models in domain controllers to obtain evaluation results. Based on the two evaluation results, they analyze the processing performance of the autonomous driving models.

[0004] However, in the above method, the cloud cluster and the domain controller use different programming languages ​​and processing technology architectures, resulting in large differences in the evaluation results obtained by the two. This makes it difficult to locate performance problems of the autonomous driving model, affecting the performance optimization of the autonomous driving model. Summary of the Invention

[0005] The present application provides an evaluation method, apparatus, and device for an autonomous driving model, which are used to solve the problem of difficulty in locating performance issues of the autonomous driving model due to the different programming languages ​​and processing technology architectures used by the cloud cluster and the domain controller.

[0006] In a first aspect, the present application provides a method for evaluating an autonomous driving model, the method being applied to a high-performance computing (HPC) server, the method comprising:

[0007] Obtaining a pre-processing file, and processing the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained by processing a pre-stored image, and the pre-processing file has a first preset file format; and the first output file represents a prediction processing process of the first autonomous driving model;

[0008] Sending the pre-processed file to at least one domain controller based on a network file system (NFS) sharing service; wherein the domain controller is configured to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents a prediction processing process of the second autonomous driving model, and the second output file has a second preset file format;

[0009] Based on the NFS sharing service, receiving a second output file sent by the at least one domain controller;

[0010] An evaluation result is determined based on the first output file and the second output file sent by the at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each of the second autonomous driving models.

[0011] In one example, determining an evaluation result based on the first output file and the second output file sent by the at least one domain controller includes:

[0012] Processing the first output file to obtain a first post-processing file; wherein the first post-processing file represents a prediction processing process of the first autonomous driving model;

[0013] Processing the second output file sent by the at least one domain controller to obtain a second post-processed file of the second output file sent by the at least one domain controller; wherein the second post-processed file represents a prediction processing process of the second autonomous driving model;

[0014] The evaluation result is determined according to the first post-processing file and the at least one second post-processing file.

[0015] In one example, the evaluation result includes a first result, where the first result represents a difference in accuracy between prediction results of the second autonomous driving models;

[0016] Determining the evaluation result according to the first post-processing file and the at least one second post-processing file includes:

[0017] Performing evaluation processing on the first post-processing file to obtain a first evaluation accuracy of the first post-processing file; wherein the first evaluation accuracy represents the accuracy of the prediction result of the first autonomous driving model;

[0018] performing evaluation processing on the at least one second post-processing file to obtain a second evaluation accuracy of the at least one second post-processing file; wherein the second evaluation accuracy represents the accuracy of the prediction result of the second autonomous driving model;

[0019] The first result is determined according to the first evaluation accuracy and each of the second evaluation accuracies.

[0020] In one example, the second autonomous driving model is deployed with an optimization method, the optimization method being used to optimize a prediction processing process of the second autonomous driving model; the evaluation result further includes a second result, the second result representing the advantages and disadvantages of the optimization methods deployed by each of the second autonomous driving models;

[0021] Determining the evaluation result according to the first post-processing file and the at least one second post-processing file includes:

[0022] performing evaluation processing on the first post-processing file to obtain a first prediction efficiency of the first post-processing file; wherein the first prediction efficiency represents a speed of a prediction processing process of the first autonomous driving model;

[0023] performing evaluation processing on the at least one second post-processing file to obtain a second prediction efficiency of the at least one second post-processing file; wherein the second prediction efficiency represents a speed of a prediction processing process of the second autonomous driving model;

[0024] The second result is determined according to the first prediction efficiency and each of the second prediction efficiencies.

[0025] In one example, before obtaining the pre-processing file, include:

[0026] The to-be-processed file is obtained from a preset database, and the pre-processed file is determined according to the to-be-processed file; wherein the to-be-processed file includes a picture pre-stored in the preset database.

[0027] In one example, determining the pre-processing file according to the to-be-processed file includes:

[0028] Parsing the file to be processed to obtain a parsed file of the file to be processed; wherein the parsed file includes data of a pre-stored image;

[0029] The format of the parsed file of the file to be processed is converted to obtain the pre-processed file.

[0030] In one example, the method further includes:

[0031] Configure an NFS file sharing service; wherein the NFS file sharing service is used to mount on the domain controller for file sharing.

[0032] In an example, configuring the NFS file sharing service includes:

[0033] Receive an NFS software package sent by an NFS server and create a shared directory; wherein the NFS software package includes a configuration file for providing an NFS file sharing service; the shared directory includes a storage path of files to be shared;

[0034] Add the shared directory to the configuration file in the NFS software package;

[0035] In response to the validation instruction of the NFS software package, the shared directory is mounted to the domain controller for file sharing; wherein the validation instruction is used to indicate that the NFS file sharing service in the NFS software package is valid.

[0036] In a second aspect, the present application provides an evaluation device for an autonomous driving model, the device being applied to a high-performance computing (HPC) server; the device comprising:

[0037] A first acquiring unit, configured to acquire a pre-processing file;

[0038] a first processing unit, configured to process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained by processing a pre-stored image, and the pre-processing file has a first preset file format; and the first output file represents a prediction processing process of the first autonomous driving model;

[0039] a sending unit, configured to send the pre-processed file to at least one domain controller based on a network file system (NFS) sharing service; wherein the domain controller is configured to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents a prediction processing process of the second autonomous driving model, and the second output file has a second preset file format;

[0040] a receiving unit, configured to receive, based on the NFS sharing service, a second output file sent by the at least one domain controller;

[0041] The first determination unit is used to determine an evaluation result based on the first output file and the second output file sent by the at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each of the second autonomous driving models.

[0042] In one example, the first determining unit includes:

[0043] a first processing module, configured to process the first output file to obtain a first post-processing file; wherein the first post-processing file represents a prediction processing process of the first autonomous driving model;

[0044] a second processing module, configured to process the second output file sent by the at least one domain controller to obtain a second post-processed file of the second output file sent by the at least one domain controller; wherein the second post-processed file represents a prediction processing process of the second autonomous driving model;

[0045] A determination module is used to determine the evaluation result according to the first post-processing file and the at least one second post-processing file.

[0046] In one example, the evaluation result includes a first result, where the first result represents a difference in accuracy between prediction results of the second autonomous driving models;

[0047] The determining module is specifically configured to:

[0048] Performing evaluation processing on the first post-processing file to obtain a first evaluation accuracy of the first post-processing file; wherein the first evaluation accuracy represents the accuracy of the prediction result of the first autonomous driving model;

[0049] performing evaluation processing on the at least one second post-processing file to obtain a second evaluation accuracy of the at least one second post-processing file; wherein the second evaluation accuracy represents the accuracy of the prediction result of the second autonomous driving model;

[0050] The first result is determined according to the first evaluation accuracy and each of the second evaluation accuracies.

[0051] In one example, the second autonomous driving model is deployed with an optimization method, the optimization method being used to optimize a prediction processing process of the second autonomous driving model; the evaluation result further includes a second result, the second result representing the advantages and disadvantages of the optimization methods deployed by each of the second autonomous driving models;

[0052] The determining module is further specifically configured to:

[0053] performing evaluation processing on the first post-processing file to obtain a first prediction efficiency of the first post-processing file; wherein the first prediction efficiency represents a speed of a prediction processing process of the first autonomous driving model;

[0054] performing evaluation processing on the at least one second post-processing file to obtain a second prediction efficiency of the at least one second post-processing file; wherein the second prediction efficiency represents a speed of a prediction processing process of the second autonomous driving model;

[0055] The second result is determined according to the first prediction efficiency and each of the second prediction efficiencies.

[0056] In one example, before the first obtaining unit is used to obtain the pre-processing file, the process includes:

[0057] A second acquiring unit, configured to acquire a file to be processed from a preset database;

[0058] The second determining unit is configured to determine the pre-processing file according to the file to be processed; wherein the file to be processed includes pictures pre-stored in the preset database.

[0059] In one example, the second determining unit includes:

[0060] A parsing module, configured to parse the file to be processed to obtain a parsed file of the file to be processed; wherein the parsed file includes data of pre-stored images;

[0061] The conversion module is used to convert the format of the parsed file of the file to be processed to obtain the pre-processing file.

[0062] In one example, the apparatus further includes:

[0063] A configuration unit is used to configure an NFS file sharing service; wherein the NFS file sharing service is used to be mounted on the domain controller for file sharing.

[0064] In one example, the configuration unit includes:

[0065] A receiving module, configured to receive an NFS software package sent by an NFS server and create a shared directory; wherein the NFS software package includes a configuration file for providing an NFS file sharing service; and the shared directory includes a storage path for files to be shared;

[0066] An adding module, used for adding the shared directory to the configuration file in the NFS software package;

[0067] The mounting module is used to mount the shared directory to the domain controller in response to the effectiveness instruction of the NFS software package to perform file sharing; wherein the effectiveness instruction is used to indicate that the NFS file sharing service in the NFS software package is effective.

[0068] In a third aspect, the present application provides a high performance computing (HPC) server, comprising: a processor, and a memory communicatively connected to the processor;

[0069] The memory stores computer-executable instructions;

[0070] The processor executes the computer-executable instructions stored in the memory to implement the method described in the first aspect.

[0071] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described in the first aspect.

[0072] In a fifth aspect, the present application provides a computer program product, comprising: a computer program, the computer program being stored in a readable storage medium, at least one processor of an electronic device being able to read the computer program from the readable storage medium, and the at least one processor executing the computer program so that the electronic device executes the method described in the first aspect.

[0073] The evaluation method, apparatus, and device for the autonomous driving model provided in the present application use a network file system NFS sharing service set up on a high-performance computing (HPC) server. During the model prediction process on the HPC server and the domain controller, the HPC server uniformly sets the model input file format and output file format of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller. Based on the HPC server, the output files of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller are evaluated and compared to obtain the evaluation results of the autonomous driving model. Furthermore, the model evaluation gap between different types of devices can be reduced, and more accurate evaluation results can be obtained to improve the efficiency of autonomous driving model evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0075] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0076] Figure 2 A flowchart of an evaluation method for an autonomous driving model provided in an embodiment of the present application;

[0077] Figure 3 A flowchart of another autonomous driving model evaluation method provided in an embodiment of the present application;

[0078] Figure 4 A schematic diagram of the structure of an evaluation device for an autonomous driving model provided in an embodiment of the present application;

[0079] Figure 5 A schematic diagram of the structure of another autonomous driving model evaluation device provided in an embodiment of the present application;

[0080] Figure 6 A schematic diagram of the structure of a high-performance computing server provided in an embodiment of the present application;

[0081] Figure 7 The figure is a block diagram of a high performance computing server according to an exemplary embodiment.

[0082] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0083] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0084] Figure 1 This is a schematic diagram of an application scenario provided by this application. Figure 1 As shown, this scenario includes a high-performance computing (HPC) server 101 and at least one domain controller 102. A user deploys a Network File System (NFS) file sharing service on HPC server 101. HPC server 101 mounts the NFS file sharing service on each domain controller 102 to enable file sharing between the HPC server 101 and each domain controller 102, as well as between domain controllers 102. Autonomous driving model predictions are performed on the HPC server 101 and each domain controller 102, respectively. Based on the HPC server 101, the model prediction processes of the HPC server 101 and each domain controller 102 are evaluated and compared to obtain evaluation results. Furthermore, through the HPC server 101, using the remote connection tool SSh (Secure Shell), not only can user management be performed for all domain controller 102 users, but also the work processes of all domain controllers 102 can be controlled. Furthermore, when performing model predictions on the domain controller 102, the accuracy and performance differences caused by multiple work processes running simultaneously can be reduced.

[0085] Autonomous driving models are deployed in smart vehicles for model prediction. In order to determine whether the autonomous driving models meet the application requirements, the trained autonomous driving models need to be evaluated in different domain controllers.

[0086] In one example, algorithm engineers typically train and evaluate autonomous driving models in a cloud cluster to obtain evaluation results; at the same time, they evaluate the autonomous driving model in a domain controller to obtain evaluation results. Based on the two evaluation results, they analyze the processing performance of the autonomous driving model.

[0087] However, in the above method, the cloud cluster and each domain controller use different programming languages ​​and processing technology architectures, resulting in significant differences in the evaluation results obtained between the two. This makes it difficult to locate performance problems of the autonomous driving model, affecting the performance optimization of the autonomous driving model.

[0088] The evaluation method, device and equipment of the autonomous driving model provided in this application are intended to solve the above technical problems of the prior art.

[0089] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0090] It should be noted that the evaluation method, device and equipment of the autonomous driving model of the present application can be used in the field of autonomous driving technology, and can also be used in any field other than the field of autonomous driving technology. The application field of the evaluation method, device and equipment of the autonomous driving model of the present application is not limited.

[0091] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0092] Figure 2 A flow chart of an evaluation method for an autonomous driving model provided in an embodiment of the present application is shown as follows: Figure 2 As shown, the method is applied to a high performance computing (HPC) server, and the method includes:

[0093] S101. Obtain a pre-processing file, and process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained after processing a pre-stored image, and the pre-processing file has a first preset file format; the first output file represents the prediction processing process of the first autonomous driving model.

[0094] Exemplarily, the execution subject of this embodiment can be a high-performance computing (HPC) server. First, based on the high-performance computing (HPC) server, the acquired image is pre-processed to obtain image data, and the image data is organized into a file, namely a pre-processing file, such as a tensor file. The pre-processing file has a pre-set first preset file format, which is determined in advance based on negotiation between the algorithm training engineer and the domain controller deployment engineer. The HPC server inputs the obtained pre-processing file into the first autonomous driving model originally deployed on the HPC server. The first autonomous driving model performs prediction processing on the pre-processing file to obtain a first output file corresponding to the first autonomous driving model. The first output file represents the prediction processing process of the first autonomous driving model, such as the prediction efficiency, prediction processing time, and prediction results of the prediction processing process of the first autonomous driving model.

[0095] S102. Based on the network file system NFS sharing service, the pre-processing file is sent to at least one domain controller; wherein the domain controller is used to process the pre-processing file based on a preset second autonomous driving model to obtain a second output file; the second output file represents the prediction processing process of the second autonomous driving model, and the second output file has a second preset file format.

[0096] In an exemplary embodiment, a network file system (NFS) sharing service is pre-deployed on a high-performance computing (HPC) server and mounted on at least one domain controller. The HPC server sends the obtained pre-processed file to each domain controller via the NFS sharing service. A second autonomous driving model trained on the HPC server is pre-deployed on each domain controller. After receiving the pre-processed file, each domain controller inputs the pre-processed file into the pre-set second autonomous driving model. For example, the pre-processed file (Tensor file) is read in via deployed C++ code. The second autonomous driving model then performs prediction processing on the pre-processed file to produce a second output file, including the prediction process of the second autonomous driving model, such as the prediction efficiency, prediction processing time, and prediction results of the second autonomous driving model. The second output file has a second pre-set file format, which is pre-determined based on negotiation between algorithm training engineers and domain controller deployment engineers, such as a TensorOutput file. This ensures that the model input files on the HPC server and each domain controller are consistent, reducing the impact of the model prediction environment on the prediction process and improving the accuracy of the evaluation results.

[0097] S103: Based on the NFS sharing service, a second output file sent by at least one domain controller is received.

[0098] For example, each domain controller sends the obtained prediction processing process of the second autonomous driving model, i.e., the second output file, to the HPC server based on the NFS shared service mounted on each domain controller. The prediction processing process of the second autonomous driving model sent by each domain controller is received based on the NFS shared service deployed on the HPC server.

[0099] S104. Determine an evaluation result based on the first output file and the second output file sent by at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each second autonomous driving model.

[0100] Exemplarily, based on the HPC server, according to a pre-set evaluation method, the prediction processing process of the first autonomous driving model, that is, the first output file, and the prediction processing process of each second autonomous driving model, that is, the second output file, are evaluated and analyzed to obtain evaluation results; that is, using the different operation methods of the cloud and the domain controller in the HPC server, the HPC server respectively calls the model prediction process in the cloud in the HPC server and the model prediction process in the domain controller, and manages them to achieve unified evaluation and comparison between the model prediction results output by the cloud and multiple domain controllers in the HPC server, that is, to obtain the pros and cons of the prediction performance of each second autonomous driving model; at the same time, by deploying some automated operation script tools, the tool chain can be uniformly managed, so that large-scale automated evaluation can be carried out when multiple domain controllers are deployed, without the need for repeated manual operation by domain controller engineers and repeated operation by algorithm training engineers, thereby improving the evaluation efficiency of the autonomous driving model.

[0101] In this embodiment, a method for evaluating an autonomous driving model is provided. During the process of performing model prediction on the HPC server and each domain controller, the HPC server uniformly sets the model input file format and output file format of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller through a network file system NFS sharing service set on a high-performance computing (HPC) server. Based on the HPC server, the output files of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller are evaluated and compared to obtain an evaluation result of the autonomous driving model. Furthermore, the model evaluation gap between different types of device ends can be reduced, and more accurate evaluation results can be obtained, thereby improving the efficiency of autonomous driving model evaluation.

[0102] Figure 3 A flow chart of another method for evaluating an autonomous driving model provided in an embodiment of the present application is shown as follows: Figure 3 As shown, the method is applied to a high performance computing (HPC) server, and the method includes:

[0103] S201. Configure an NFS file sharing service. The NFS file sharing service is used to mount on a domain controller for file sharing.

[0104] Exemplarily, the NFS file sharing service is configured on the HDC server, ie, the NFS file sharing service is deployed, and the deployed NFS file sharing service is mounted on each domain controller to perform file sharing between the HDC server and each domain controller.

[0105] In one example, step S201 includes the following steps:

[0106] The first step of step S201 is to receive an NFS software package sent by an NFS server and create a shared directory; wherein the NFS software package includes a configuration file for providing an NFS file sharing service; and the shared directory includes a storage path of files to be shared.

[0107] The second step of step S201 is to add a shared directory to the configuration file in the NFS software package.

[0108] The third step of step S201 is to mount the shared directory to the domain controller in response to the activation instruction of the NFS software package to perform file sharing; wherein the activation instruction is used to indicate that the NFS file sharing service in the NFS software package is activated.

[0109] Specifically, the NFS server is pre-installed and configured, ensuring that the NFS server has installed the NFS software package, wherein the NFS software package includes a configuration file for providing NFS file sharing services, and the NFS server sends the installed NFS software package to the HPC server. Based on the HPC server, the NFS software package sent by the NFS server is received, and after the NFS software package is installed, a shared directory is created, and the storage path of the files to be shared is placed in the shared directory. The HPC server adds the configuration of the shared directory to the configuration file in the NFS software package. The user inputs the effectiveness instruction of the NFS software package, that is, the NFS file sharing service in the NFS software package is effective. In response to the effectiveness instruction of the NFS software package, the HPC server mounts the shared directory to the domain controller for file sharing.

[0110] S202: Obtain a file to be processed from a preset database, and determine a pre-processing file based on the file to be processed; wherein the file to be processed includes pictures pre-stored in the preset database.

[0111] Exemplarily, based on the HPC server, pre-stored images are obtained from a pre-deployed preset database and organized into files to be processed; based on a preset data processing method, the obtained files to be processed are processed to obtain pre-processing files, which have a first preset file format.

[0112] In one example, determining the pre-processing file based on the file to be processed in step S202 includes the following steps:

[0113] The first step of step S202 is to parse the file to be processed to obtain a parsed file of the file to be processed; wherein the parsed file includes data of pre-stored pictures.

[0114] The second step of step S202 is to convert the format of the parsed file of the file to be processed to obtain a pre-processed file.

[0115] For example, on an HPC server, algorithm engineers write preprocessing code (preprocessing) to parse the acquired file to be processed, generating a parsed file of the file to be processed. The parsed file includes pre-stored image data to meet the data input requirements of the model in the HPC server and each domain controller. The HPC server converts the parsed file of the file to be processed and saves it as a preprocessed file in a preset file format.

[0116] S203. Obtain a pre-processing file, and process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained after processing a pre-stored image, and the pre-processing file has a first preset file format; the first output file represents the prediction processing process of the first autonomous driving model.

[0117] For example, this step refers to the content of step S101 and will not be repeated here.

[0118] S204. Based on the network file system NFS sharing service, the pre-processing file is sent to at least one domain controller; wherein the domain controller is used to process the pre-processing file based on a preset second autonomous driving model to obtain a second output file; the second output file represents the prediction processing process of the second autonomous driving model, and the second output file has a second preset file format.

[0119] For example, this step refers to the content of step S102 and will not be repeated here.

[0120] S205: Based on the NFS sharing service, a second output file sent by at least one domain controller is received.

[0121] For example, this step refers to the content of step S103 and will not be repeated here.

[0122] S206. Process the first output file to obtain a first post-processing file; wherein the first post-processing file represents a prediction processing process of the first autonomous driving model.

[0123] Exemplarily, based on the HPC server, a post-processing code written by an algorithm engineer is pre-set, such as in Python code. The post-processing code is used to perform necessary post-processing on the data of the prediction processing process of the first autonomous driving model, that is, the first output file, to obtain a first post-processing file; wherein, the first post-processing file represents the prediction processing process of the first autonomous driving model.

[0124] S207. Process the second output file sent by the at least one domain controller to obtain a second post-processing file of the second output file sent by the at least one domain controller; wherein the second post-processing file represents a prediction processing process of the second autonomous driving model.

[0125] Exemplarily, based on the HPC server, a post-processing code written by an algorithm engineer is pre-set to pre-process the first processing file representing the prediction processing process of the first autonomous driving model. For example, in the Python code, the post-processing code is used to perform necessary post-processing on the second output file sent by each domain controller to obtain a second post-processing file; wherein the second post-processing file represents the prediction processing process of the first autonomous driving model; thereby, the execution code of the post-processing operations on different files is made consistent to improve the accuracy of the evaluation results.

[0126] S208: Determine an evaluation result according to the first post-processing file and at least one second post-processing file.

[0127] Exemplarily, based on the HPC server, after the operations corresponding to the same post-processing code are processed, the first post-processing file and multiple second post-processing files obtained can obtain evaluation results based on a preset evaluation method, and then obtain the pros and cons of the prediction performance of each second autonomous driving model; thereby improving the evaluation efficiency of the autonomous driving model.

[0128] In one example, the evaluation result includes a first result, and the first result represents the difference between the accuracy of the prediction results of each second autonomous driving model; step S208 includes the following steps:

[0129] The first step of step S208 is to perform evaluation processing on the first post-processing file to obtain a first evaluation accuracy of the first post-processing file; wherein the first evaluation accuracy represents the accuracy of the prediction result of the first autonomous driving model.

[0130] The second step of step S208 is to perform evaluation processing on at least one second post-processing file to obtain a second evaluation accuracy of at least one second post-processing file; wherein the second evaluation accuracy represents the accuracy of the prediction result of the second autonomous driving model.

[0131] In the third step of step S208 , a first result is determined according to the first evaluation accuracy and each second evaluation accuracy.

[0132] Specifically, to understand the differences in the accuracy of the prediction results of the second autonomous driving model in each domain controller, the HPC server executes a preset evaluation code for evaluating the accuracy of the model prediction results, evaluates the obtained first post-processed file, and obtains a first evaluation accuracy for the first post-processed file, thereby representing the accuracy of the prediction results of the first autonomous driving model in the HPC server. Simultaneously, the preset evaluation code for evaluating the accuracy of the model prediction results is executed again, evaluating each second post-processed file, thereby obtaining a second evaluation accuracy for each second post-processed file, thereby representing the accuracy of the prediction results of the second autonomous driving model in each domain controller. The HPC server compares and analyzes the first evaluation accuracy and each second evaluation accuracy to obtain a first result, thereby understanding the differences in the accuracy of the prediction results of the second autonomous driving model in each domain controller. Furthermore, if the autonomous driving model deployed in each domain controller is consistent, the impact of the deployment environment in each domain controller on the prediction accuracy of the autonomous driving model can be determined, thereby improving the efficiency of autonomous driving model evaluation.

[0133] In one example, the second autonomous driving model is deployed with an optimization method, and the optimization method is used to optimize the prediction processing process of the second autonomous driving model; the evaluation results also include a second result, and the second result represents the advantages and disadvantages of each optimization method deployed by the second autonomous driving model; step S208 includes the following steps:

[0134] The first step of step S208 is to evaluate the first post-processing file to obtain a first prediction efficiency of the first post-processing file; wherein the first prediction efficiency represents the speed of the prediction processing process of the first autonomous driving model.

[0135] The second step of step S208 is to evaluate and process at least one second post-processing file to obtain a second prediction efficiency of at least one second post-processing file; wherein the second prediction efficiency represents the speed of the prediction processing process of the second autonomous driving model.

[0136] In the third step of step S208, a second result is determined according to the first prediction efficiency and each second prediction efficiency.

[0137] Specifically, the second autonomous driving model in each domain controller deploys an optimization method to optimize the prediction processing process of the second autonomous driving model and improve the efficiency of model prediction, such as FP16 algorithm optimization, INT8 quantization method, sparse matrix optimization method, etc.; for the first post-processing file, and the model prediction results in each post-processing file, in order to understand the advantages and disadvantages of the optimization methods deployed by the second autonomous driving model in each domain controller; based on the HPC server, execute the preset evaluation code for the model prediction efficiency of the evaluation processing, evaluate the obtained first post-processing file, and obtain the first prediction efficiency of the first post-processing file to characterize the speed of the prediction processing process of the first autonomous driving model of the HPC server. At the same time, execute the preset evaluation code for the model prediction efficiency of the evaluation processing again, evaluate each second post-processing file, and obtain the second prediction efficiency of each second post-processing file to characterize the speed of the prediction processing process of the second autonomous driving model of each domain controller. The HPC server compares and analyzes the first prediction efficiency and each second prediction efficiency to obtain a second result, that is, it can be used to know the advantages and disadvantages of the optimization methods deployed by the second autonomous driving model in each domain controller; further, when the autonomous driving models deployed by each domain controller are inconsistent; or when the model engineer deploys the tool chain iteratively on the same domain controller to deploy multiple different autonomous driving models, the advantages and disadvantages of the optimization methods deployed by each second autonomous driving model can be obtained, such as the situation in which the INT8 quantization loss in different domain controller platforms affects the accuracy of the final autonomous driving model, and the optimization efficiency differences of the optimization methods on the domain controller side such as different pruning and sparse matrices, and then corresponding optimization suggestions are put forward to improve the performance of the autonomous driving model.

[0138] In this embodiment, based on the above embodiment, on the one hand, through the network file system NFS sharing service set up on the high-performance computing HPC server, the HPC server uniformly sets the model input file format and output file format of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller, and evaluates and compares the output files of the autonomous driving model in the HPC server and the autonomous driving model in the domain controller to obtain the evaluation results of the autonomous driving model; thereby, the model evaluation gap between different types of device ends can be reduced, and more accurate evaluation results can be obtained to improve the efficiency of the autonomous driving model evaluation; on the other hand, by using the code of the same processing operation to process the output files of different device ends, and using the code of the same evaluation operation to evaluate the output files of different device ends to obtain the evaluation results, the model evaluation efficiency of different device ends is greatly improved, and the performance of the autonomous driving model can be further improved.

[0139] Figure 4A schematic diagram of the structure of an evaluation device for an autonomous driving model provided in an embodiment of the present application is shown as follows: Figure 4 As shown, the device 300 is applied to a high-performance computing (HPC) server, and the device 300 includes:

[0140] The first acquiring unit 301 is configured to acquire a pre-processing file.

[0141] The first processing unit 302 is used to process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained after processing a pre-stored image, and the pre-processing file has a first preset file format; the first output file represents the prediction processing process of the first autonomous driving model.

[0142] The sending unit 303 is used to send the pre-processed file to at least one domain controller based on the network file system NFS sharing service; wherein the domain controller is used to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents the prediction processing process of the second autonomous driving model, and the second output file has a second preset file format.

[0143] The receiving unit 304 is configured to receive a second output file sent by at least one domain controller based on the NFS sharing service.

[0144] The first determination unit 305 is used to determine an evaluation result based on the first output file and the second output file sent by at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each second autonomous driving model.

[0145] The device of this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principles are the same and will not be repeated here.

[0146] Figure 5 A schematic diagram of the structure of another automatic driving model evaluation device provided in an embodiment of the present application is shown as follows: Figure 5 As shown, the device 400 is applied to a high-performance computing (HPC) server, and the device 400 includes:

[0147] The first acquiring unit 401 is configured to acquire a pre-processing file.

[0148] The first processing unit 402 is used to process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained after processing a pre-stored image, and the pre-processing file has a first preset file format; the first output file represents the prediction processing process of the first autonomous driving model.

[0149] A sending unit 403 is configured to send a pre-processed file to at least one domain controller based on a network file system NFS sharing service; wherein the domain controller is configured to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents a prediction processing process of the second autonomous driving model, and the second output file has a second preset file format.

[0150] The receiving unit 404 is configured to receive a second output file sent by at least one domain controller based on the NFS sharing service.

[0151] The first determination unit 405 is used to determine an evaluation result based on the first output file and the second output file sent by at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each second autonomous driving model.

[0152] In one example, the first determining unit 405 includes:

[0153] The first processing module 4051 is used to process the first output file to obtain a first post-processing file; wherein the first post-processing file represents the prediction processing process of the first autonomous driving model.

[0154] The second processing module 4052 is used to process the second output file sent by the at least one domain controller to obtain a second post-processing file of the second output file sent by the at least one domain controller; wherein the second post-processing file represents the prediction processing process of the second autonomous driving model.

[0155] The determination module 4053 is configured to determine an evaluation result based on the first post-processing file and at least one second post-processing file.

[0156] In one example, the evaluation results include a first result, which represents the difference in accuracy between the prediction results of each second autonomous driving model.

[0157] The determination module 4053 is specifically configured to:

[0158] The first post-processing file is evaluated and processed to obtain a first evaluation accuracy of the first post-processing file; wherein the first evaluation accuracy represents the accuracy of the prediction result of the first autonomous driving model.

[0159] An evaluation process is performed on the at least one second post-processing file to obtain a second evaluation accuracy of the at least one second post-processing file; wherein the second evaluation accuracy represents the accuracy of the prediction result of the second autonomous driving model.

[0160] A first result is determined based on the first evaluation accuracy and each second evaluation accuracy.

[0161] In one example, the second autonomous driving model is deployed with an optimization method, which is used to optimize the prediction processing process of the second autonomous driving model; the evaluation results also include a second result, which characterizes the advantages and disadvantages of the optimization methods deployed by each second autonomous driving model.

[0162] The determination module 4053 is further specifically configured to:

[0163] The first post-processing file is evaluated and processed to obtain a first prediction efficiency of the first post-processing file; wherein the first prediction efficiency represents the speed of the prediction processing process of the first autonomous driving model.

[0164] The at least one second post-processing file is evaluated and processed to obtain a second prediction efficiency of the at least one second post-processing file; wherein the second prediction efficiency represents the speed of the prediction processing process of the second autonomous driving model.

[0165] A second result is determined based on the first prediction efficiency and each second prediction efficiency.

[0166] In one example, before the first obtaining unit 401 is used to obtain the pre-processing file, the following steps are included:

[0167] The second acquiring unit 406 is configured to acquire the file to be processed from a preset database.

[0168] The second determining unit 407 is configured to determine a pre-processing file according to the file to be processed; wherein the file to be processed includes pictures pre-stored in a preset database.

[0169] In one example, the second determining unit 407 includes:

[0170] The parsing module 4071 is used to parse the file to be processed to obtain a parsed file of the file to be processed; wherein the parsed file includes data of pre-stored pictures.

[0171] The conversion module 4072 is used to convert the format of the parsed file of the file to be processed to obtain a pre-processed file.

[0172] In one example, the apparatus 400 further includes:

[0173] The configuration unit 408 is used to configure the NFS file sharing service; wherein the NFS file sharing service is used to be mounted on the domain controller for file sharing.

[0174] In one example, the configuration unit 408 includes:

[0175] The receiving module 4081 is used to receive the NFS software package sent by the NFS server and create a shared directory; wherein the NFS software package includes a configuration file for providing NFS file sharing service; and the shared directory includes a storage path of files to be shared.

[0176] Added module 4082 for adding shared directories to the configuration files in the NFS package.

[0177] The mounting module 4083 is used to mount the shared directory to the domain controller in response to the activation instruction of the NFS software package for file sharing; wherein the activation instruction is used to indicate that the NFS file sharing service in the NFS software package is effective.

[0178] The device of this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principles are the same and will not be repeated here.

[0179] Figure 6 A schematic diagram of the structure of a high performance computing server provided in an embodiment of the present application is shown as follows: Figure 6 As shown, the HPC server 500 includes: a memory 501 and a processor 502; the memory 501 is used to store instructions executable by the processor 502.

[0180] The processor 502 is configured to execute the method provided in the above embodiment.

[0181] The HPC server 500 further includes a receiver 503 and a transmitter 504. The receiver 503 is used to receive instructions and data sent by other devices, and the transmitter 504 is used to send instructions and data to external devices.

[0182] Figure 7 1 is a block diagram of a high-performance computing server according to an exemplary embodiment. The high-performance computing HPC server 800 can be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0183] HPC server 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output interface 812 , a sensor component 814 , and a communication component 816 .

[0184] The processing component 802 generally controls the overall operation of the HPC server 800, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 802 may include one or more modules to facilitate interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

[0185] The memory 804 is configured to store various types of data to support operations on the HPC server 800. Examples of such data include instructions for any application or method operating on the HPC server 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory, electrically erasable programmable read-only memory, erasable programmable read-only memory, programmable read-only memory, read-only memory, magnetic memory, flash memory, magnetic disk, or optical disk.

[0186] The power supply component 806 provides power to the various components of the HPC server 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the HPC server 800.

[0187] The multimedia component 808 includes a screen that provides an output interface between the HPC server 800 and the user. In some embodiments, the screen may include a liquid crystal display and a touch panel. If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensor can not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the HPC server 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0188] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone, and when the HPC server 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive external audio signals. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 88 also includes a speaker for outputting audio signals.

[0189] The input / output interface 812 provides an interface between the processing component 802 and peripheral interface modules, such as a keyboard, a click wheel, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0190] The sensor assembly 814 includes one or more sensors for providing various aspects of status information for the HPC server 800. For example, the sensor assembly 814 can detect the on / off state of the HPC server 800, the relative positioning of components, such as the display and keypad of the HPC server 800. The sensor assembly 814 can also detect changes in the position of the HPC server 800 or a component in the HPC server 800, the presence or absence of user contact with the HPC server 800, the orientation or acceleration / deceleration of the HPC server 800, and temperature changes of the HPC server 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include a light sensor for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0191] The communication component 816 is configured to facilitate wired or wireless communication between the HPC server 800 and other devices. The HPC server 800 can access a wireless network based on a communication standard. In an exemplary embodiment, the communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near-field communication module to facilitate short-range communication. For example, the near-field communication module can be implemented based on radio frequency identification technology, infrared data association technology, ultra-wideband technology, and other technologies.

[0192] In an exemplary embodiment, the HPC server 800 may be implemented by one or more application-specific integrated circuits, digital signal processors, digital signal processing devices, programmable logic devices, field programmable gate arrays, controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0193] According to an embodiment of the present application, the present application further provides a non-transitory computer-readable storage medium including instructions, such as a memory 804 including instructions, which can be executed by a processor 820 of an HPC server 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a random access memory, a magnetic tape, a floppy disk, an optical disk, or other data storage device.

[0194] According to an embodiment of the present application, the present application also provides a non-transitory computer-readable storage medium, which, when instructions in the storage medium are executed by a processor of an HPC server, enables the HPC server to execute the above method.

[0195] The present application also provides a computer program product, comprising a computer program stored in a computer-readable storage medium. At least one processor of a device can read the computer program from the computer-readable storage medium, and at least one processor executes the computer program so that the device implements the methods provided in the various embodiments described above.

[0196] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0197] The modules may be physically separate, for example, installed in different locations on a single device, or installed on different devices, or distributed across multiple network units, or distributed across multiple processors. The modules may also be integrated, for example, installed in the same device, or integrated into a set of codes. The modules may exist in the form of hardware, or in the form of software, or may be implemented in the form of software plus hardware. The present application may select some or all of the modules according to actual needs to achieve the purpose of the present embodiment.

[0198] When each module is implemented as an integrated module in the form of a software function module, it can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to perform some steps of the methods of each embodiment of the present application.

[0199] It should be understood that, although the various steps in the flowcharts of the above embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they may be performed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but may be performed at different times, and their execution order is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of sub-steps or stages of other steps.

[0200] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0201] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for evaluating an autonomous driving model, characterized in that: The method is applied to a high-performance computing (HPC) server; the method comprises: Obtaining a pre-processing file, and processing the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained by processing a pre-stored image, and the pre-processing file has a first preset file format; and the first output file represents a prediction processing process of the first autonomous driving model; Sending the pre-processed file to at least one domain controller based on a network file system (NFS) sharing service; wherein the domain controller is configured to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents a prediction processing process of the second autonomous driving model, and the second output file has a second preset file format; Based on the NFS sharing service, receiving a second output file sent by the at least one domain controller; An evaluation result is determined based on the first output file and the second output file sent by the at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each of the second autonomous driving models.

2. The method according to claim 1, characterized in that Determining an evaluation result according to the first output file and the second output file sent by the at least one domain controller includes: Processing the first output file to obtain a first post-processing file; wherein the first post-processing file represents a prediction processing process of the first autonomous driving model; Processing the second output file sent by the at least one domain controller to obtain a second post-processed file of the second output file sent by the at least one domain controller; wherein the second post-processed file represents a prediction processing process of the second autonomous driving model; The evaluation result is determined according to the first post-processing file and the at least one second post-processing file.

3. The method according to claim 2, characterized in that The evaluation result includes a first result, where the first result represents a difference in accuracy between prediction results of the second autonomous driving models; Determining the evaluation result according to the first post-processing file and the at least one second post-processing file includes: Performing evaluation processing on the first post-processing file to obtain a first evaluation accuracy of the first post-processing file; wherein the first evaluation accuracy represents the accuracy of the prediction result of the first autonomous driving model; performing evaluation processing on the at least one second post-processing file to obtain a second evaluation accuracy of the at least one second post-processing file; wherein the second evaluation accuracy represents the accuracy of the prediction result of the second autonomous driving model; The first result is determined according to the first evaluation accuracy and each of the second evaluation accuracies.

4. The method according to claim 2, characterized in that The second autonomous driving model is deployed with an optimization method, the optimization method being used to optimize a prediction processing process of the second autonomous driving model; the evaluation result further includes a second result, the second result representing the relative merits of the optimization methods deployed by each of the second autonomous driving models; Determining the evaluation result according to the first post-processing file and the at least one second post-processing file includes: performing evaluation processing on the first post-processing file to obtain a first prediction efficiency of the first post-processing file; wherein the first prediction efficiency represents a speed of a prediction processing process of the first autonomous driving model; performing evaluation processing on the at least one second post-processing file to obtain a second prediction efficiency of the at least one second post-processing file; wherein the second prediction efficiency represents a speed of a prediction processing process of the second autonomous driving model; The second result is determined according to the first prediction efficiency and each of the second prediction efficiencies.

5. The method according to any one of claims 1 to 4, characterized in that Before obtaining the pre-processing file, include: The to-be-processed file is obtained from a preset database, and the pre-processed file is determined according to the to-be-processed file; wherein the to-be-processed file includes a picture pre-stored in the preset database.

6. The method according to claim 5, characterized in that Determining the pre-processing file according to the to-be-processed file includes: Parsing the file to be processed to obtain a parsed file of the file to be processed; wherein the parsed file includes data of a pre-stored image; The format of the parsed file of the file to be processed is converted to obtain the pre-processed file.

7. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Configure an NFS file sharing service; wherein the NFS file sharing service is used to mount on the domain controller for file sharing.

8. The method according to claim 7, characterized in that Configure the NFS file sharing service, including: Receive an NFS software package sent by an NFS server and create a shared directory; wherein the NFS software package includes a configuration file for providing an NFS file sharing service; the shared directory includes a storage path of files to be shared; Add the shared directory to the configuration file in the NFS software package; In response to the validation instruction of the NFS software package, the shared directory is mounted to the domain controller for file sharing; wherein the validation instruction is used to indicate that the NFS file sharing service in the NFS software package is valid.

9. An evaluation device for an autonomous driving model, characterized in that: The device is applied to a high-performance computing (HPC) server; the device comprises: A first acquiring unit, configured to acquire a pre-processing file; a first processing unit, configured to process the pre-processing file based on a preset first autonomous driving model to obtain a first output file; wherein the pre-processing file includes image data obtained by processing a pre-stored image, and the pre-processing file has a first preset file format; and the first output file represents a prediction processing process of the first autonomous driving model; a sending unit, configured to send the pre-processed file to at least one domain controller based on a network file system (NFS) sharing service; wherein the domain controller is configured to process the pre-processed file based on a preset second autonomous driving model to obtain a second output file; the second output file represents a prediction processing process of the second autonomous driving model, and the second output file has a second preset file format; a receiving unit, configured to receive, based on the NFS sharing service, a second output file sent by the at least one domain controller; The first determination unit is used to determine an evaluation result based on the first output file and the second output file sent by the at least one domain controller; wherein the evaluation result represents the quality of the prediction performance of each of the second autonomous driving models.

10. A high performance computing (HPC) server, characterized in that: include: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 8 when executed by a processor.

12. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 8 when the computer program is executed by a processor.

Images