Vehicle damage assessment method, device, equipment and storage medium

Abstract

The present invention relates to the field of artificial intelligence technology, and in particular to a vehicle damage assessment method, device, equipment, and storage medium. The method comprises obtaining raw vehicle damage assessment data; performing component detection based on the raw vehicle damage assessment data using a pre-trained component detection model to obtain component detection results; generating corresponding raw data supplementary instructions based on the component detection results, and outputting the raw data supplementary instructions in accordance with a preset instruction output method; pre-processing the raw vehicle damage assessment data to obtain vehicle damage assessment information; performing damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain damage detection results; determining a corresponding repair plan based on the damage detection results, and outputting the damage detection results and the corresponding repair plan. This method solves the problem of inaccurate vehicle damage assessment judgment caused by incomplete raw data upload in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a vehicle damage assessment method, device, equipment and storage medium. Background Art

[0002] When a driver is involved in a traffic accident, not only does their right to life and health suffer, but their vehicle also often sustains varying degrees of damage from collisions or scrapes. Insurance companies typically dispatch professional auto damage assessors to the scene, or conduct a comprehensive online survey based on photos or videos taken by the vehicle owner or salesperson, ultimately providing appropriate vehicle repair and claims solutions.

[0003] This type of traditional solution relies on manual review and judgment. On the one hand, it is inefficient and requires a huge workload. On the other hand, in order to avoid situations where insurance companies suffer unnecessary losses due to errors in judgment caused by incomplete image and video information, different understandings, and lack of experience, the professionalism required of adjusters is extremely high, which will result in large labor costs.

[0004] Existing solutions for vehicle damage assessment using artificial intelligence usually detect vehicle damage based on raw data such as pictures and videos uploaded manually at one time. These solutions are unable to provide corresponding feedback when the raw data is not uploaded completely, resulting in inaccurate assessments of vehicle damage.

[0005] In summary, the existing technology has the problem of inaccurate judgment of vehicle damage caused by incomplete uploading of original data. Summary of the Invention

[0006] The main purpose of this application is to provide a vehicle damage assessment method, device, equipment and storage medium to solve the problem of inaccurate judgment of vehicle damage caused by incomplete uploading of original data in the prior art.

[0007] A first aspect of the present invention provides a vehicle damage assessment method, which includes: obtaining original vehicle damage assessment data; performing component detection based on the original vehicle damage assessment data through a pre-trained component detection model to obtain component detection results; generating corresponding original data supplementary instructions based on the component detection results, and outputting the original data supplementary instructions in accordance with a preset instruction output method; pre-processing the original vehicle damage assessment data to obtain vehicle damage assessment information; performing damage detection based on the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; determining a corresponding maintenance plan based on the damage detection results, and outputting the damage detection results and the corresponding maintenance plan.

[0008] Optionally, in a first implementation manner of the first aspect of the present invention, generating corresponding raw data supplement instructions based on the component detection results includes: comparing the component detection results with a preset complete component table to obtain a comparison result; parsing the comparison result to obtain detected components and undetected components; and generating corresponding raw data supplement instructions based on the undetected components.

[0009] Optionally, in a second implementation manner of the first aspect of the present invention, the vehicle damage assessment original data includes at least one of digital image data, laser detection data and acoustic wave detection data, and the preprocessing of the vehicle damage assessment original data to obtain vehicle damage assessment information includes: parsing the vehicle damage assessment original data, and splitting the vehicle damage assessment original data into the digital image data, the laser detection data and the acoustic wave detection data according to a preset splitting method; parsing the laser detection data and the acoustic wave detection data respectively to obtain laser detection values and acoustic wave detection values corresponding to each position point in the detection space; converting the laser detection values and acoustic wave detection values corresponding to each position point in the detection space into corresponding image feature values to obtain a laser detection map and an acoustic wave detection map; parsing the digital image data, the laser detection map and the acoustic wave detection map to obtain a correspondence between the digital image data, the laser detection map and the acoustic wave detection map; and constructing the vehicle damage assessment information based on the digital image data, the laser detection map and the acoustic wave detection map according to the correspondence.

[0010] Optionally, in a third implementation of the first aspect of the present invention, determining a corresponding maintenance plan based on the damage detection result includes: parsing the damage detection result to obtain a damage category and a corresponding damage degree; constructing a corresponding damage index based on the damage category and the corresponding damage degree; and searching for a corresponding maintenance plan in a preset maintenance plan table based on the damage index.

[0011] Optionally, in a fourth implementation method of the first aspect of the present invention, after the vehicle damage assessment original data is detected by a pre-trained component detection model to obtain a component detection result, it also includes: judging whether the component detection result meets the component detection condition; if so, outputting a judgment result that the component detection condition is met; if not, generating a corresponding original data supplementary instruction based on the component detection result, and outputting the original data supplementary instruction through a preset interactive platform; obtaining supplementary data corresponding to the original data supplementary instruction from the interactive platform; detecting the vehicle damage assessment original data by a pre-trained component detection model to obtain a second component detection result; judging whether the second component detection result meets the component detection condition; if so, outputting a judgment result that the component detection condition is met; if not, repeating the above steps until the corresponding component detection result meets the component detection condition.

[0012] Optionally, in a fifth implementation of the first aspect of the present invention, before performing component detection based on the vehicle damage assessment original data through a pre-trained component detection model and obtaining the component detection result, it also includes: acquiring vehicle component data and corresponding component labeling data, and constructing a vehicle component data set based on the vehicle component data and the corresponding component labeling data; inputting the data in the vehicle component data set into a preset neural network model for training in a preset manner to obtain a first training result; calculating the loss value through a cross entropy function based on the first training result, and recording the loss value and the parameter characteristics of the neural network model; inputting the data in the vehicle component data set into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards to obtain the component detection model.

[0013] Optionally, in a sixth implementation of the first aspect of the present invention, before performing damage detection according to the vehicle damage assessment information through a pre-trained damage detection model and obtaining a damage detection result, it also includes: acquiring vehicle damage data and corresponding damage annotation data, and constructing a vehicle damage data set based on the vehicle damage data and the corresponding damage annotation data; inputting the data in the vehicle damage data set into a preset second neural network model for training in a preset manner to obtain a second training result; calculating the loss value through a cross entropy function based on the second training result, and recording the loss value and parameter characteristics of the neural network model; inputting the data in the vehicle damage data set into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards to obtain the damage detection model.

[0014] The second aspect of the present invention provides a vehicle damage assessment device, comprising: an acquisition module for acquiring original vehicle damage assessment data; a component detection module for performing component detection based on the original vehicle damage assessment data through a pre-trained component detection model to obtain component detection results; an instruction generation module for generating corresponding original data supplementary instructions based on the component detection results, and outputting the original data supplementary instructions in accordance with a preset instruction output method; a preprocessing module for preprocessing the original vehicle damage assessment data to obtain vehicle damage assessment information; a damage detection module for performing damage detection based on the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; an output module for determining a corresponding maintenance plan based on the damage detection results, and outputting the damage detection results and the corresponding maintenance plan.

[0015] Optionally, in a first implementation manner of the second aspect of the present invention, the instruction generation module includes: a comparison unit, used to compare the component detection result with a preset complete component table to obtain a comparison result; a first parsing unit, used to parse the comparison result to obtain detected components and undetected components; and a first generation unit, used to generate corresponding original data supplementary instructions based on the undetected components.

[0016] Optionally, in a second implementation of the second aspect of the present invention, the preprocessing module includes: a second parsing unit, used to parse the original vehicle damage assessment data, and split the original vehicle damage assessment data into the digital image data, the laser detection data and the acoustic wave detection data according to a preset splitting method; a third parsing unit, used to parse the laser detection data and the acoustic wave detection data respectively, and obtain the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space; a conversion unit, used to convert the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space into corresponding image feature values, and obtain a laser detection map and an acoustic wave detection map; a fourth parsing unit, used to parse the digital image data, the laser detection map and the acoustic wave detection map, and obtain the correspondence between the digital image data, the laser detection map and the acoustic wave detection map; a first construction unit, used to construct the vehicle damage assessment information based on the digital image data, the laser detection map and the acoustic wave detection map according to the correspondence.

[0017] Optionally, in a third implementation of the second aspect of the present invention, the output module includes: a fifth parsing unit, used to parse the damage detection result to obtain the damage category and the corresponding damage degree; a second construction unit, used to construct a corresponding damage index according to the damage category and the corresponding damage degree; and a search unit, used to search for a corresponding maintenance plan in a preset maintenance plan table according to the damage index.

[0018] Optionally, in a fourth implementation of the second aspect of the present invention, the vehicle damage assessment device further includes: a judgment module for judging whether the component detection result satisfies the component detection condition; if so, outputting a judgment result that the component detection condition is satisfied; if not, generating a corresponding original data supplementary instruction based on the component detection result, and outputting the original data supplementary instruction through a preset interactive platform; obtaining supplementary data corresponding to the original data supplementary instruction from the interactive platform; detecting the vehicle damage assessment original data through a pre-trained component detection model to obtain a second component detection result; judging whether the second component detection result satisfies the component detection condition; if so, outputting a judgment result that the component detection condition is satisfied; if not, repeating the above steps until the corresponding component detection result satisfies the component detection condition.

[0019] Optionally, in a fifth implementation of the second aspect of the present invention, the vehicle damage assessment device further includes: a first training module, for acquiring vehicle component data and corresponding component labeling data, and constructing a vehicle component data set based on the vehicle component data and the corresponding component labeling data; inputting the data in the vehicle component data set into a preset neural network model for training in a preset manner to obtain a first training result; according to the first training result, calculating the loss value through a cross entropy function, and recording the loss value and parameter characteristics of the neural network model; inputting the data in the vehicle component data set into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards to obtain the component detection model.

[0020] Optionally, in a sixth implementation of the second aspect of the present invention, the vehicle damage assessment device further includes a second training module for acquiring vehicle damage data and corresponding damage annotation data, and constructing a vehicle damage data set based on the vehicle damage data and the corresponding damage annotation data; inputting the data in the vehicle damage data set into a preset second neural network model for training in a preset manner to obtain a second training result; according to the second training result, calculating the loss value through a cross entropy function, and recording the loss value and the parameter characteristics of the neural network model; inputting the data in the vehicle damage data set into the second neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards to obtain the damage detection model.

[0021] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes each step of the above-mentioned vehicle damage assessment method.

[0022] A fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute each step of the above-mentioned vehicle damage assessment method.

[0023] In the technical solution of the present invention, the method specifically obtains original vehicle damage assessment data; performs component detection based on the original vehicle damage assessment data through a pre-trained component detection model to obtain component detection results; generates corresponding original data supplementary instructions based on the component detection results, and outputs the original data supplementary instructions according to a preset instruction output method; pre-processes the original vehicle damage assessment data to obtain vehicle damage assessment information; performs damage detection based on the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; determines a corresponding maintenance plan based on the damage detection results, and outputs the damage detection results and the corresponding maintenance plan; thereby solving the problem of inaccurate judgment of vehicle damage caused by incomplete uploading of original data in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a schematic diagram of a first embodiment of a vehicle damage assessment method according to an embodiment of the present invention;

[0025] Figure 2 1 is a schematic diagram of a second embodiment of a vehicle damage assessment method according to an embodiment of the present invention;

[0026] Figure 3 1 is a schematic diagram of a third embodiment of a vehicle damage assessment method according to an embodiment of the present invention;

[0027] Figure 4 A schematic diagram of an embodiment of a vehicle damage assessment device according to an embodiment of the present invention;

[0028] Figure 5 A schematic diagram of another embodiment of a vehicle damage assessment device according to an embodiment of the present invention;

[0029] Figure 6 FIG. 1 is a schematic diagram of an embodiment of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] To address the problem of inaccurate vehicle damage assessment caused by incomplete raw data upload in the prior art, the present application provides a vehicle damage assessment method, apparatus, device, and storage medium. The method obtains raw vehicle damage assessment data; performs component detection based on the raw vehicle damage assessment data using a pre-trained component detection model to obtain component detection results; generates corresponding raw data supplementary instructions based on the component detection results, and outputs the raw data supplementary instructions according to a preset instruction output method; pre-processes the raw vehicle damage assessment data to obtain vehicle damage assessment information; performs damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain damage detection results; determines a corresponding repair plan based on the damage detection results, and outputs the damage detection results and the corresponding repair plan. This method thus addresses the problem of inaccurate vehicle damage assessment caused by incomplete raw data upload in the prior art.

[0031] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0032] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 The first embodiment of the vehicle damage assessment method in the embodiment of the present invention is implemented by the following steps:

[0033] 101. Obtain original data on vehicle damage assessment;

[0034] In this step, the vehicle damage assessment original data includes at least digital image data, wherein the digital image data includes video data and picture data;

[0035] In practical applications, the original vehicle damage assessment data also includes laser detection data and acoustic wave detection data;

[0036] This step can be implemented in the following ways:

[0037] Reading first original data from a preset interactive interface;

[0038] Reading supplementary original data from a supplementary data interface preset in the interactive interface;

[0039] The first original data and the supplementary original data are saved in a preset manner to obtain the vehicle damage assessment original data.

[0040] 102. Using the pre-trained component detection model, perform component detection based on the original vehicle damage assessment data to obtain component detection results;

[0041] In this step, the component detection model includes at least an input layer, an embedding layer, a hidden layer and an output layer;

[0042] This step can be implemented in the following ways:

[0043] Converting the raw vehicle damage assessment data into a feature vector matrix through the input layer and outputting it to the embedding layer;

[0044] Converting the feature vector matrix into a corresponding embedding expression matrix through the embedding layer and outputting it to the hidden layer;

[0045] Calculate the label vector corresponding to the embedded expression matrix through the hidden layer and output it to the output layer;

[0046] Calling the Softmax algorithm through the output layer to calculate the classification probability of the embedded expression matrix and its corresponding label vector, and outputting the prediction result;

[0047] The prediction result is analyzed to obtain the component detection result.

[0048] In practical applications, the component detection model can be trained in the following way:

[0049] Acquire vehicle component data and corresponding component annotation data, and construct a vehicle component dataset based on the vehicle component data and the corresponding component annotation data;

[0050] Inputting the data in the vehicle component data set into a preset neural network model for training in a preset manner to obtain a first training result;

[0051] Calculating a loss value using a cross entropy function based on the first training result, and recording the loss value and parameter characteristics of the neural network model;

[0052] The data in the vehicle component data set is input into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the component detection model.

[0053] 103. Generate corresponding raw data supplementary instructions based on the component detection results, and output the raw data supplementary instructions according to a preset instruction output method;

[0054] This step can be implemented in the following ways:

[0055] Comparing the component detection result with a preset complete component table to obtain a comparison result, wherein the comparison result at least includes the identification of the detected component and the identification of the undetected component;

[0056] searching for corresponding position information from the complete component table according to the detected component identifier and the undetected component identifier;

[0057] constructing relative position information between the detected component and the undetected component according to the undetected component identifier and the position information;

[0058] A corresponding raw data supplement instruction is generated according to the undetected component identification and the corresponding relative position information.

[0059] 104. Preprocess the original vehicle damage assessment data to obtain vehicle damage assessment information;

[0060] This step can be implemented in the following ways:

[0061] Determining whether the vehicle damage assessment raw data includes the laser detection data or the acoustic wave detection data;

[0062] If not included, the vehicle damage assessment raw data is framed according to a preset frame-drawing method to obtain vehicle damage assessment information. For example, if the preset frame-drawing method is to draw a frame every 0.1 seconds, the vehicle damage assessment raw data is framed every 0.1 seconds;

[0063] If included, the vehicle damage assessment raw data is parsed to obtain data category information, wherein the data category information at least includes digital image data category, laser detection data category, and acoustic wave detection data category;

[0064] Splitting the original vehicle damage assessment data according to the data category information to obtain classified original data;

[0065] According to a preset preprocessing method, the classified raw data is preprocessed to obtain the vehicle damage assessment information.

[0066] Furthermore, the process of preprocessing the classified raw data according to a preset preprocessing method to obtain the vehicle damage assessment information includes:

[0067] If the data category information corresponding to the classified original data includes a laser detection data category, extracting the laser detection data;

[0068] Analyzing the laser detection data to obtain the laser detection value corresponding to each position point in the detection space;

[0069] Converting the laser detection value corresponding to each position point in the detection space into a corresponding image feature value to obtain a laser detection map, for example, converting the laser detection value corresponding to each position point in the detection space into a corresponding grayscale value;

[0070] Analyzing the digital image data and the laser detection map to obtain a corresponding relationship between the digital image data and the laser detection map;

[0071] According to the corresponding relationship, constructing the vehicle damage assessment information based on the digital image data and the laser detection image;

[0072] If the data category information corresponding to the classified original data includes the acoustic wave detection data category, extracting the acoustic wave detection data;

[0073] Analyzing the acoustic wave detection data to obtain the acoustic wave detection value corresponding to each position point in the detection space;

[0074] Converting the acoustic wave detection value corresponding to each position point in the detection space into a corresponding image feature value to obtain an acoustic wave detection map, for example, converting the acoustic wave detection value corresponding to each position point in the detection space into a corresponding grayscale value;

[0075] Analyzing the digital image data and the acoustic wave detection image to obtain a corresponding relationship between the digital image data and the acoustic wave detection image;

[0076] According to the corresponding relationship, the vehicle damage assessment information is constructed based on the digital image data and the acoustic wave detection image.

[0077] 105. Using the pre-trained damage detection model, damage detection is performed based on the vehicle damage assessment information to obtain damage detection results;

[0078] In this step, the damage detection model at least includes a damage detection model input layer, a damage detection model embedding layer, a damage detection model hidden layer and a damage detection model output layer;

[0079] This step can be implemented in the following ways:

[0080] Converting the vehicle damage assessment information into a feature vector matrix through the damage detection model input layer and outputting it to the damage detection model embedding layer;

[0081] Converting the feature vector matrix into a corresponding embedding expression matrix through the damage detection model embedding layer and outputting it to the damage detection model hidden layer;

[0082] Calculating the damage type label vector corresponding to the embedded expression matrix through the hidden layer of the damage detection model and outputting it to the output layer of the damage detection model;

[0083] The Softmax algorithm is called through the output layer of the damage detection model to calculate the classification probability of the embedded expression matrix and its corresponding damage type label vector, and the damage degree prediction result corresponding to each damage type is output;

[0084] The damage type and the corresponding damage degree prediction result are analyzed to obtain the damage detection result.

[0085] In practical applications, the damage detection model can be trained in the following way:

[0086] Acquire vehicle damage data and corresponding damage annotation data, and construct a vehicle damage dataset based on the vehicle damage data and the corresponding damage annotation data;

[0087] Inputting the data in the vehicle damage data set into a preset second neural network model for training in a preset manner to obtain a second training result;

[0088] Calculating a loss value using a cross entropy function based on the second training result, and recording the loss value and parameter characteristics of the neural network model;

[0089] The data in the vehicle damage data set is input into the second neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the damage detection model.

[0090] 106. Determine a corresponding maintenance plan based on the damage detection result, and output the damage detection result and the corresponding maintenance plan.

[0091] This step can be implemented in the following ways:

[0092] Analyzing the damage detection results to obtain damage categories and corresponding damage degrees;

[0093] Constructing a corresponding damage index according to the damage category and the corresponding damage degree, for example, obtaining identification numbers corresponding to the damage category and the damage degree, and connecting them with a separator to obtain a corresponding damage index;

[0094] According to the damage index, a corresponding maintenance plan is searched in a preset maintenance plan table.

[0095] By implementing the above method, the original data of vehicle damage assessment is obtained; component detection is performed on the original data of vehicle damage assessment through a pre-trained component detection model to obtain component detection results; corresponding original data supplementary instructions are generated based on the component detection results, and the original data supplementary instructions are output according to a preset instruction output method; the original data of vehicle damage assessment is pre-processed to obtain vehicle damage assessment information; damage detection is performed on the original data of vehicle damage assessment through a pre-trained damage detection model to obtain damage detection results; a corresponding maintenance plan is determined based on the damage detection results, and the damage detection results and the corresponding maintenance plan are output; in the above, corresponding component detection is performed on the original data of vehicle damage assessment, and corresponding original data supplementary instructions are constructed. The process of generating instructions and outputting them can improve the completeness of the original data upload, perform damage detection through the pre-trained damage detection model, obtain damage detection results, and determine the corresponding maintenance plan, and can output the damage detection results to judge the damage condition of the vehicle, thereby solving the problem of inaccurate judgment of the damage condition of the vehicle caused by incomplete original data upload in the prior art.

[0096] See also Figure 2 The second embodiment of the vehicle damage assessment method in the embodiment of the present invention is implemented as follows:

[0097] 201. Obtain original data of vehicle damage assessment;

[0098] In this step, the original data of vehicle damage assessment includes at least vehicle video, vehicle photos and vehicle parameters;

[0099] This step can be implemented in the following ways:

[0100] Obtain the original data through the upload interface corresponding to the preset user interaction interface;

[0101] Determine whether there is an original data supplement instruction;

[0102] If there is an original data supplement instruction, the upload interface corresponding to the preset user interaction interface is polled;

[0103] If a data transmission event is polled, the supplementary data is obtained through the upload interface, and the original data and the supplementary data are saved in a preset format to obtain the original vehicle damage assessment information;

[0104] If no data transmission event is polled, continue polling;

[0105] If there is no original data supplement instruction, the original data and the supplementary data are saved in a preset format to obtain the original vehicle damage assessment data.

[0106] 202. Perform component inspection based on the original vehicle damage assessment data, obtain component inspection results, determine whether component inspection conditions are met, and construct corresponding original data supplement instructions;

[0107] This step can be implemented in the following ways:

[0108] Testing the vehicle damage assessment raw data using a pre-trained component detection model;

[0109] Determining whether the component detection result meets the component detection condition;

[0110] If it is satisfied, the judgment result is output as meeting the component detection conditions;

[0111] If not, generating a corresponding raw data supplement instruction based on the component detection result, and outputting the raw data supplement instruction through a preset user interaction interface;

[0112] Acquire the supplementary data corresponding to the original data supplementation instruction from the user interaction interface;

[0113] Detecting the supplementary data using a pre-trained component detection model to obtain a second component detection result;

[0114] determining whether the second component detection result satisfies the component detection condition;

[0115] If it is satisfied, the judgment result is output as meeting the component detection conditions;

[0116] If not, repeat the above steps until the corresponding component detection result meets the component detection condition.

[0117] Furthermore, the process of testing the vehicle damage assessment raw data using the pre-trained component detection model includes:

[0118] Using a pre-trained vehicle part detection model to identify vehicle parts on the original vehicle damage assessment data, obtaining a part identification result;

[0119] The part recognition results are analyzed to obtain identification information and position information of each part of the vehicle.

[0120] Furthermore, the process of determining whether the component detection result meets the component detection condition includes:

[0121] Extracting identification information and location information of each part of the vehicle, searching for the corresponding detection completeness in a preset detection completeness table, and obtaining a comprehensive detection completeness;

[0122] Determine whether the comprehensive detection completeness is greater than a preset threshold; if so, the component detection condition is met; if less than, the component detection condition is not met.

[0123] 203. Based on the original vehicle damage assessment data, damage detection is performed using a pre-trained damage detection model;

[0124] This step can be implemented in the following ways:

[0125] Analyzing the original vehicle damage assessment data to obtain color information and depth information of each position;

[0126] Converting the color information and depth information of each position into a data matrix, for example, converting the color information and depth information of each position into a four-channel data matrix;

[0127] The damage detection is performed on the data matrix using a pre-trained damage detection model to obtain a damage detection result.

[0128] 204. Determine a corresponding maintenance plan and maintenance amount based on the damage detection result, and output the damage detection result, maintenance plan, and maintenance amount.

[0129] This step can be implemented in the following ways:

[0130] Analyzing the damage detection result to obtain a damage index;

[0131] According to the damage index, searching for a corresponding maintenance plan in a preset maintenance plan table;

[0132] According to the maintenance plan, searching for the corresponding maintenance amount in the preset maintenance amount table;

[0133] According to the corresponding relationship among the damage detection result, the maintenance plan and the maintenance amount, the damage detection result, the maintenance plan and the maintenance amount are output on the user interaction interface.

[0134] By implementing the above method, the original data of vehicle damage assessment is obtained; component detection is performed according to the original data of vehicle damage assessment through a pre-trained component detection model to obtain component detection results; corresponding original data supplement instructions are generated based on the component detection results, and the original data supplement instructions are output according to a preset instruction output method; the original data of vehicle damage assessment is pre-processed to obtain vehicle damage assessment information; damage detection is performed according to the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; a corresponding maintenance plan is determined based on the damage detection results, and the damage detection results and the corresponding maintenance plan are output; the original data of vehicle damage assessment is obtained; component detection is performed according to the original data of vehicle damage assessment to obtain component detection results, and it is determined whether the component detection conditions are met, and the corresponding original data supplement instructions are constructed; damage detection is performed according to the original data of vehicle damage assessment through a pre-trained damage detection model; the corresponding maintenance plan and maintenance amount are determined based on the damage detection results, and the damage detection results, maintenance plan and maintenance amount are output; thereby solving the problem of inaccurate judgment of vehicle damage caused by incomplete original data uploading in the prior art.

[0135] See also Figure 3 The third embodiment of the vehicle damage assessment method in the embodiment of the present invention is implemented as follows:

[0136] 301. Interact with the user through a preset user interaction interface and read the original data uploaded by the user to obtain the original vehicle damage assessment data;

[0137] In this step, the user interaction interface is set on a terminal, wherein the terminal includes at least a mobile phone and a mobile computer device;

[0138] Outputting interactive instructions at a preset location on the user interaction interface, wherein the interactive instructions include voice instructions, image annotation instructions, and text instructions. For example, by guiding the user to upload the original data through an interactive page and voice prompts designed on the mobile phone to guide the user to take a circular photo;

[0139] Read the original data uploaded by the user, and perform component detection through the preset component detection model to obtain the component detection results;

[0140] Constructing upload completion information based on the component detection results, for example, judging whether the detection results corresponding to each component meet preset conditions based on the component detection results, and generating different color identifications to obtain upload completion information;

[0141] According to the upload completion information, corresponding raw data supplement instructions are constructed and output, for example, at a preset position on the user interaction interface, detected components and undetected components are marked with different colors, and the names and identifiers corresponding to the undetected components are output;

[0142] Read the supplementary original data uploaded by the user and repeat the above process until the upload completion information meets the preset requirements.

[0143] In actual applications, before performing component detection using a preset component detection model and obtaining component detection results, this step also includes:

[0144] Obtaining the original data and extracting video data therein;

[0145] The video data is periodically extracted at preset time intervals, for example, at time intervals of 0.05 seconds.

[0146] In practical applications, the component detection model can be obtained by converting the deep learning model trained under the TensorFlow framework using the TFLite tool;

[0147] The deep learning model adopts at least one algorithm among Faster RCNN, EAST, SSD and YOLO.

[0148] In practical applications, the component detection model includes a MobileNet network model.

[0149] 302. Preprocessing the original vehicle damage assessment data to obtain original vehicle damage assessment information, and transmitting the original vehicle damage assessment information to a preset vehicle damage assessment backend;

[0150] In this step, the vehicle damage assessment backend is deployed based on a GPU cluster;

[0151] In this step, the vehicle damage assessment backend has a backend interface;

[0152] This step can be implemented in the following ways:

[0153] Parsing the original vehicle damage assessment data and saving it in the form of a data object to obtain the original vehicle damage assessment information;

[0154] The original vehicle damage assessment information is transmitted to the vehicle damage assessment backend through the backend interface.

[0155] In practical applications, the vehicle damage assessment backend has a damage detection model, wherein the damage detection model can be obtained by training a transfer learning model based on PyTorch;

[0156] After the original vehicle damage assessment information is transmitted to the preset vehicle damage assessment backend, this step also includes:

[0157] The original vehicle damage assessment information is processed through the damage detection model to obtain corresponding damage results and repair plans.

[0158] 303. Read the return result of the vehicle damage assessment background from the background interface, obtain the damage result and the corresponding repair plan, and output the damage result and the corresponding repair plan through the user interaction interface.

[0159] In this step, the process of outputting the damage results and the corresponding repair plan through the user interaction interface includes:

[0160] Converting the damage result and the corresponding repair plan into natural language and outputting them in the natural language output position of the user interaction interface through text display and voice broadcast;

[0161] The damage result and the corresponding repair plan are converted into image information, and the image information is output at the image information output position of the user interaction interface. For example, a vehicle loss result schematic diagram is constructed according to the damage result, and the corresponding repair plan is displayed at the corresponding position.

[0162] By implementing the above method, raw vehicle damage assessment data is obtained; a pre-trained component detection model is used to perform component detection based on the raw vehicle damage assessment data to obtain component detection results; a corresponding raw data supplement instruction is generated based on the component detection results, and the raw data supplement instruction is output according to a preset instruction output mode; the raw vehicle damage assessment data is pre-processed to obtain vehicle damage assessment information; a pre-trained damage detection model is used to perform damage detection based on the vehicle damage assessment information to obtain damage detection results; a corresponding repair plan is determined based on the damage detection results, and the damage detection results and corresponding repair plan are output; the method interacts with a user through a preset user interaction interface and reads raw data uploaded by the user to obtain raw vehicle damage assessment data; the raw vehicle damage assessment data is pre-processed to obtain raw vehicle damage assessment information, and the raw vehicle damage assessment information is transmitted to a preset vehicle damage assessment backend; the return result of the vehicle damage assessment backend is read from the backend interface to obtain a damage result and a corresponding repair plan, and the damage result and the corresponding repair plan are output through the user interaction interface; thereby solving the problem of inaccurate judgment of vehicle damage caused by incomplete raw data uploading in the prior art.

[0163] The above describes the vehicle damage assessment method in the embodiment of the present invention. The following describes the vehicle damage assessment device in the embodiment of the present invention. Please refer to Figure 4An embodiment of a vehicle damage assessment device in an embodiment of the present invention includes:

[0164] Acquisition module 401, used to obtain original data of vehicle damage assessment;

[0165] A component detection module 402 is configured to perform component detection based on the vehicle damage assessment raw data using a pre-trained component detection model to obtain component detection results;

[0166] An instruction generation module 403 is configured to generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction in a preset instruction output mode;

[0167] A pre-processing module 404 is used to pre-process the raw vehicle damage assessment data to obtain vehicle damage assessment information;

[0168] The damage detection module 405 is configured to perform damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain a damage detection result;

[0169] The output module 406 is configured to determine a corresponding maintenance plan based on the damage detection result, and output the damage detection result and the corresponding maintenance plan.

[0170] By implementing the above-mentioned device, the original data of vehicle damage assessment is obtained; component detection is performed based on the original data of vehicle damage assessment through a pre-trained component detection model to obtain component detection results; corresponding original data supplementary instructions are generated based on the component detection results, and the original data supplementary instructions are output according to a preset instruction output method; the original data of vehicle damage assessment is pre-processed to obtain vehicle damage assessment information; damage detection is performed based on the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; the corresponding maintenance plan is determined based on the damage detection results, and the damage detection results and the corresponding maintenance plan are output; thereby solving the problem of inaccurate judgment of vehicle damage caused by incomplete original data uploading in the prior art.

[0171] See also Figure 5 Another embodiment of the vehicle damage assessment device in the embodiment of the present invention includes:

[0172] Acquisition module 401, used to obtain original data of vehicle damage assessment;

[0173] The first training module 408 is used to obtain vehicle component data and corresponding component annotation data, and construct a vehicle component data set based on the vehicle component data and the corresponding component annotation data; input the data in the vehicle component data set into a preset neural network model for training in a preset manner to obtain a first training result; based on the first training result, calculate the loss value through the cross entropy function, record the loss value and the parameter characteristics of the neural network model; input the data in the vehicle component data set into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the component detection model.

[0174] A component detection module 402 is configured to perform component detection based on the vehicle damage assessment raw data using a pre-trained component detection model to obtain component detection results;

[0175] The judgment module 407 is used to judge whether the component detection result satisfies the component detection condition; if so, output the judgment result that the component detection condition is satisfied; if not, generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction through a preset interactive platform; obtain supplementary data corresponding to the raw data supplement instruction from the interactive platform; detect the vehicle damage assessment raw data using a pre-trained component detection model to obtain a second component detection result; judge whether the second component detection result satisfies the component detection condition; if so, output the judgment result that the component detection condition is satisfied; if not, repeat the above steps until the corresponding component detection result satisfies the component detection condition;

[0176] An instruction generation module 403 is configured to generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction in a preset instruction output mode;

[0177] A pre-processing module 404 is used to pre-process the raw vehicle damage assessment data to obtain vehicle damage assessment information;

[0178] The second training module 409 is used to obtain vehicle damage data and corresponding damage annotation data, and construct a vehicle damage data set based on the vehicle damage data and the corresponding damage annotation data; input the data in the vehicle damage data set into a preset second neural network model for training in a preset manner to obtain a second training result; based on the second training result, calculate the loss value through the cross entropy function, and record the loss value and the parameter characteristics of the neural network model; input the data in the vehicle damage data set into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the damage detection model.

[0179] The damage detection module 405 is configured to perform damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain a damage detection result;

[0180] The output module 406 is configured to determine a corresponding maintenance plan based on the damage detection result, and output the damage detection result and the corresponding maintenance plan.

[0181] In this embodiment, the instruction generation module 403 includes:

[0182] A comparison unit 4031 is used to compare the component detection result with a preset complete component table to obtain a comparison result;

[0183] A first parsing unit 4032 is configured to parse the comparison result to obtain detected components and undetected components;

[0184] The first generating unit 4033 is configured to generate a corresponding raw data supplement instruction according to the undetected component.

[0185] In this embodiment, the pre-processing module 404 includes:

[0186] The second parsing unit 4041 is used to parse the original vehicle damage assessment data and split the original vehicle damage assessment data into the digital image data, the laser detection data, and the acoustic wave detection data according to a preset splitting method;

[0187] The third parsing unit 4042 is used to parse the laser detection data and the acoustic wave detection data respectively to obtain the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space;

[0188] A conversion unit 4043 is configured to convert the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space into a corresponding image feature value to obtain a laser detection image and an acoustic wave detection image;

[0189] a fourth parsing unit 4044 configured to parse the digital image data, the laser detection image, and the acoustic wave detection image to obtain a corresponding relationship between the digital image data, the laser detection image, and the acoustic wave detection image;

[0190] The first construction unit 4045 is used to construct the vehicle damage assessment information based on the digital image data, the laser detection image and the acoustic wave detection image according to the corresponding relationship.

[0191] In this embodiment, the output module 406 includes:

[0192] The fifth parsing unit 4061 is configured to parse the damage detection result to obtain a damage category and a corresponding damage degree;

[0193] The second construction unit 4062 is used to construct a corresponding damage index according to the damage type and the corresponding damage degree;

[0194] The search unit 4063 is configured to search for a corresponding maintenance plan in a preset maintenance plan table according to the damage index.

[0195] By implementing the above-mentioned device, the original data of vehicle damage assessment is obtained; component detection is performed based on the original data of vehicle damage assessment through a pre-trained component detection model to obtain component detection results; corresponding original data supplementary instructions are generated based on the component detection results, and the original data supplementary instructions are output according to a preset instruction output method; the original data of vehicle damage assessment is pre-processed to obtain vehicle damage assessment information; damage detection is performed based on the vehicle damage assessment information through a pre-trained damage detection model to obtain damage detection results; the corresponding maintenance plan is determined based on the damage detection results, and the damage detection results and the corresponding maintenance plan are output; thereby solving the problem of inaccurate judgment of vehicle damage caused by incomplete original data uploading in the prior art.

[0196] See also Figure 6 , an embodiment of the computer device in the embodiment of the present invention is described in detail below from the perspective of hardware processing.

[0197] Figure 6 6 is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. The computer device 600 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 610 (for example, one or more processors) and a memory 620, and one or more storage media 630 (for example, one or more mass storage devices) storing application programs 633 or data 632. Among them, the memory 620 and the storage medium 630 can be temporary storage or permanent storage. The program stored in the storage medium 630 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the computer device 600. Furthermore, the processor 610 can be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the computer device 600.

[0198] The computer device 600 may further include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input and output interfaces 660, and / or one or more operating systems 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 6 The computer device structure shown does not constitute a limitation on the computer device provided in this application, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently.

[0199] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the above-mentioned vehicle damage assessment method.

[0200] In practical applications, the above-mentioned methods can be implemented based on artificial intelligence technology, where artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results. Specifically, they can be executed based on a server, which can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0201] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0202] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0203] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle damage assessment method, characterized in that: The vehicle damage assessment method includes: Obtain original vehicle damage assessment data; Using a pre-trained component detection model, performing component detection based on the original vehicle damage assessment data to obtain component detection results; Determining whether the component detection result meets the component detection condition; If it is satisfied, the judgment result is output as meeting the component detection conditions; If not, generating a corresponding raw data supplement instruction based on the component detection result, and outputting the raw data supplement instruction through a preset interactive platform; Acquire supplementary data corresponding to the original data supplementation instruction from the interactive platform; Detecting the vehicle damage assessment raw data using a pre-trained component detection model to obtain a second component detection result; determining whether the second component detection result satisfies the component detection condition; If it is satisfied, the judgment result is output as meeting the component detection conditions; If not, repeat the above steps until the corresponding component detection result meets the component detection condition; Generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction according to a preset instruction output method; Preprocessing the original vehicle damage assessment data to obtain vehicle damage assessment information; Using a pre-trained damage detection model, damage detection is performed based on the vehicle damage assessment information to obtain a damage detection result; Determining a corresponding maintenance plan based on the damage detection result, and outputting the damage detection result and the corresponding maintenance plan; The generating of the corresponding raw data supplement instruction based on the component detection result includes: Comparing the component detection result with a preset complete component table to obtain a comparison result; Analyzing the comparison results to obtain detected components and undetected components; Generate corresponding raw data supplement instructions according to the undetected components.

2. The vehicle damage assessment method according to claim 1, characterized in that: The vehicle damage assessment raw data includes at least one of digital image data, laser detection data, and acoustic wave detection data. The preprocessing of the vehicle damage assessment raw data to obtain vehicle damage assessment information includes: Parsing the original vehicle damage assessment data and splitting the original vehicle damage assessment data into the digital image data, the laser detection data, and the acoustic wave detection data according to a preset splitting method; Analyzing the laser detection data and the acoustic wave detection data respectively to obtain the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space; Converting the laser detection value and the acoustic wave detection value corresponding to each position point in the detection space into corresponding image feature values to obtain a laser detection image and an acoustic wave detection image; parsing the digital image data, the laser detection image, and the acoustic wave detection image to obtain a corresponding relationship among the digital image data, the laser detection image, and the acoustic wave detection image; According to the corresponding relationship, the vehicle damage assessment information is constructed based on the digital image data, the laser detection image and the acoustic wave detection image.

3. The vehicle damage assessment method according to claim 1, characterized in that: Determining a corresponding maintenance plan based on the damage detection result includes: Analyzing the damage detection results to obtain damage categories and corresponding damage degrees; Constructing a corresponding damage index according to the damage category and the corresponding damage degree; According to the damage index, a corresponding maintenance plan is searched in a preset maintenance plan table.

4. The vehicle damage assessment method according to any one of claims 1 to 3, characterized in that: Before performing component detection based on the vehicle damage assessment raw data using a pre-trained component detection model to obtain component detection results, the method further includes: Acquire vehicle component data and corresponding component annotation data, and construct a vehicle component dataset based on the vehicle component data and the corresponding component annotation data; Inputting the data in the vehicle component data set into a preset neural network model for training in a preset manner to obtain a first training result; Calculating a loss value using a cross entropy function based on the first training result, and recording the loss value and parameter characteristics of the neural network model; The data in the vehicle component data set is input into the neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the component detection model.

5. The vehicle damage assessment method according to any one of claims 1 to 3, characterized in that: Before performing damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain a damage detection result, the method further includes: Acquire vehicle damage data and corresponding damage annotation data, and construct a vehicle damage dataset based on the vehicle damage data and the corresponding damage annotation data; Inputting the data in the vehicle damage data set into a preset second neural network model for training in a preset manner to obtain a second training result; Calculating a loss value using a cross entropy function based on the second training result, and recording the loss value and parameter characteristics of the neural network model; The data in the vehicle damage data set is input into the second neural network model for training until the corresponding loss value and the corresponding parameter characteristics meet the preset standards, thereby obtaining the damage detection model.

6. A vehicle damage assessment device, characterized in that: The device comprises: Acquisition module, used to obtain original data of vehicle damage assessment; A component detection module is used to perform component detection based on the vehicle damage assessment raw data using a pre-trained component detection model to obtain component detection results; A judgment module is used to judge whether the component detection result satisfies the component detection condition; if so, output a judgment result that the component detection condition is satisfied; if not, generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction through a preset interactive platform; obtain supplementary data corresponding to the raw data supplement instruction from the interactive platform; detect the vehicle damage assessment raw data through a pre-trained component detection model to obtain a second component detection result; judge whether the second component detection result satisfies the component detection condition; if so, output a judgment result that the component detection condition is satisfied; if not, repeat the above steps until the corresponding component detection result satisfies the component detection condition; An instruction generation module is used to generate a corresponding raw data supplement instruction based on the component detection result, and output the raw data supplement instruction according to a preset instruction output method; A preprocessing module, used to preprocess the raw data of vehicle damage assessment to obtain vehicle damage assessment information; A damage detection module is used to perform damage detection based on the vehicle damage assessment information using a pre-trained damage detection model to obtain a damage detection result; an output module, configured to determine a corresponding maintenance plan based on the damage detection result, and output the damage detection result and the corresponding maintenance plan; The instruction generation module includes: a comparison unit, used to compare the component detection result with a preset complete component table to obtain a comparison result; a first parsing unit, used to parse the comparison result to obtain detected components and undetected components; and a first generation unit, used to generate corresponding raw data supplementary instructions based on the undetected components.

7. The vehicle damage assessment device according to claim 6, characterized in that: The preprocessing module includes: a second parsing unit, used to parse the original vehicle damage assessment data, and split the original vehicle damage assessment data into digital image data, laser detection data and acoustic wave detection data according to a preset splitting method; a third parsing unit, used to parse the laser detection data and the acoustic wave detection data respectively, and obtain the laser detection value and acoustic wave detection value corresponding to each position point in the detection space; a conversion unit, used to convert the laser detection value and acoustic wave detection value corresponding to each position point in the detection space into corresponding image feature values, and obtain a laser detection map and an acoustic wave detection map; a fourth parsing unit, used to parse the digital image data, the laser detection map and the acoustic wave detection map, and obtain the correspondence between the digital image data, the laser detection map and the acoustic wave detection map; a first construction unit, used to construct the vehicle damage assessment information based on the digital image data, the laser detection map and the acoustic wave detection map according to the correspondence.

8. The vehicle damage assessment device according to claim 6, characterized in that: The output module includes: a fifth parsing unit, used to parse the damage detection result to obtain a damage category and a corresponding damage degree; a second construction unit, used to construct a corresponding damage index based on the damage category and the corresponding damage degree; and a search unit, used to search for a corresponding maintenance plan in a preset maintenance plan table based on the damage index.

9. A computer device, characterized in that: include: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the computer device to execute the various steps of the vehicle damage assessment method as described in any one of claims 1-5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the various steps of the vehicle damage assessment method as described in any one of claims 1 to 5 are implemented.

Images