KR102125999B1 - Ai calculation device, method and computer program - Google Patents

Abstract

According to the present embodiment, receiving the entire image of the accident vehicle and the partial image of the accident site from the user terminal; Determining a vehicle type and year of the accident vehicle using the entire image and pre-stored standard vehicle images; Detecting damaged parts included in the entire image and the partial image using a partial detection algorithm; Determining a breakage level for each breakage region using a breakage determination algorithm; And using the pre-stored table, generating work information for the accident vehicle in consideration of the breakage level and the breakage level for each breakage section, the vehicle type, and the model year, obtaining detailed work related to the work information, and the detail Disclosing an AI automatic estimation method comprising; calculating labor cost reflecting the current wage level of jobs, generating an automatic quotation summing up the wage costs, and providing the automatic quotation to the user terminal.

Description

AI automatic estimator, method and computer program {AI CALCULATION DEVICE, METHOD AND COMPUTER PROGRAM}

Embodiments of the present invention relate to AI automatic estimating devices, methods and computer programs.

2. Description of the Related Art In recent years, automobiles have been increasingly used as vehicles for transportation, transportation, and living space. In the past, it was a form of a vehicle that functions as a simple transportation and transportation means, but in recent years, it has changed into a form of a vehicle that functions as a living space. Through this, it is possible to increase the utilization of urban space, and typical examples include camper cars and food trucks.

In addition, it is becoming more common to have one or more vehicles per household. Accordingly, the usage of automobiles is greatly increased, and the importance of accident compensation is increasing along with convenience.

When an automobile accident occurs, damage to the accident is generally calculated through confirmation of an industrial company repairing the vehicle. The damaged vehicle must be moved to the industrial company and the actual repair must be completed by the industrial company, but as the final loss is calculated, the damage cannot be immediately confirmed in time, and the repair cost may increase excessively depending on the experience of the industrial staff and the will of the owner. Can.

According to embodiments of the present invention, an AI automatic estimating apparatus, method, and computer program that automatically infers the damage level of an accident vehicle and generates an automatic quotation without receiving an industrial company may be provided.

In addition, according to embodiments of the present invention, an AI automatic estimating apparatus, method, and computer program that suppresses the occurrence of a repair process exceeding the degree of damage and reduces the error of damage estimation may be provided.

In addition, according to embodiments of the present invention, an AI automatic estimating apparatus, method, and computer program for automatically generating an estimate with only an input of an image that minimizes user input may be provided.

Further, according to embodiments of the present invention, an AI automatic estimating apparatus, method, and computer program for simply receiving an accident and providing a simple damage calculation through a terminal possessed by a user may be provided.

In addition, according to embodiments of the present invention, by updating the algorithm for calculating the damage level for each vehicle part and vehicle parts from an image, an AI automatic estimating apparatus, method, and computer program for improving the accuracy of the degree of damage can be provided.

AI automatic estimation method according to embodiments of the present invention comprises the steps of the AI automatic estimation device receiving the entire image of the accident vehicle and the partial image of the accident site from the user terminal; Determining a vehicle type and year of the accident vehicle using the entire image and pre-stored standard vehicle images; Detecting damaged parts included in the entire image and the partial image using a partial detection algorithm; Determining a breakage level for each breakage region using a breakage determination algorithm; And using the pre-stored table to generate work information for the accident vehicle in consideration of the breakage level and the breakage level for each breakage part, the car name, the vehicle model, the year, the paint coat, and the like. It may include; obtaining related detailed tasks, generating an automatic quotation by reflecting the standardized working hours of the detailed tasks and the number of maintenance contracts made by each company, and providing the automatic quotation to the user terminal; .

The automatic AI estimation method may further include the step of delivering the automatic quote to the industrial terminal where the accident vehicle has been moved.

The AI automatic quotation method includes receiving the suitability for the automatic quote, the suitability for the detailed work included in the automatic quote, and the suitability for the cost of the detailed work; And in consideration of the received suitability, modifying the logic to generate an automatic quote; may include.

The AI automatic estimation method acquires the owned vehicle information of the user terminal based on the identification information of the user terminal, determines whether the vehicle model and the model year acquired through the captured vehicle information and the acquired image match, and matches It can be characterized by controlling to generate an automatic quote only in one case.

The AI automatic estimation method requests additional verification of the first damaged part when the first damaged part found in the image taken before the accident is detected among the one or more damaged parts determined through the captured image. It can be characterized by.

The above additional verification is firstly delivered to the terminal of the vehicle owner or the terminal of the vehicle's insurance company employee, and when it is determined that the first damaged portion is caused by the owner's negligence, the automatic quotation for the repair estimate for the first damaged portion It can be characterized as to be excluded from.

The AI automatic estimating device may be characterized in that it provides a detailed list of work details to an industrial terminal so that repairs can be made according to the automatically recognized damage level.

The partial detection algorithm or the damage determination algorithm may be updated by an external management server and may be updated by a signal transmitted from the management server.

The automatic quotation may be determined by the standard acceptance criteria, and the standard acceptance criteria is set to be updated through a predetermined correction procedure when a plurality of correction requests are received from at least one terminal among user terminals. Can be done with

A method of automatically calculating the degree of damage according to embodiments of the present invention uses an algorithm learned through a set of one or more images that are labeled and input, and corresponds to one or more images of an accident vehicle with a damaged portion of the accident vehicle Detecting object regions; Determining and labeling labels of object areas detected in correspondence with each part of the accident vehicle; Filtering each labeled object area by a method adopted through a CNN layer, and generating a list of damage levels arranged in the order of the magnitude of the damage level and the matching probability inferred from the degree of damage based on the filtered images; It may include; determining the highest level of damage in the list of damage levels.

The step of detecting the object areas detects object areas to include areas recognized as bumpers, fenders, and doors of the accident vehicle in the one or more image sets, but to match irregular boundaries of the bumpers, fenders, and doors. It can be characterized by being.

The labeled object region may be characterized by being filtered by at least one of a Sobel mask or a canny edge detector.

The filtered object region may be characterized in that after filtering, the contrast is uniformized by a uniformity algorithm.

The damage level of each object area is set to one of normal, sheet metal 0.5, sheet metal 1, sheet metal 1.5, sheet metal 2, and exchange, but is not limited thereto. The breakage level list is arranged in the order of the matching probability for normal, sheet metal 0.5, sheet metal 1, sheet metal 1.5, sheet metal 2, exchange, and is characterized by providing the highest level of damage level with the highest matching probability. Can.

Accident vehicle image recognition according to embodiments of the present invention calculates whether or not an error in a damaged portion of one or more images is input, and when the error level exceeds a preset first value, labels the image of the accident vehicle again Learning; may further include.

Accident vehicle image recognition according to embodiments of the present invention receives an error level of a damage level and, when the error level exceeds a preset second value, re-learns an object area and a damage level corresponding to the damaged part It may further include;

The computer program according to the embodiment of the present invention may be stored in a medium to execute any one of the accident vehicle image recognition method according to the embodiment of the present invention using a computer.

In addition to this, another method for implementing the present invention, another system, and a computer-readable recording medium for recording a computer program for executing the method are further provided.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

According to an embodiment of the present invention, it is possible to automatically infer the damage level of the accident vehicle and generate an automatic quotation without receiving the industrial company.

In addition, according to the embodiment of the present invention, it is possible to suppress the occurrence of a repair process exceeding the degree of damage and reduce errors in damage estimation.

In addition, according to an embodiment of the present invention, an estimate can be automatically generated only by input of an image that minimizes user input.

Further, according to an embodiment of the present invention, it is possible to simply receive an accident and provide a simple damage calculation through a terminal carried by a user.

In addition, according to an embodiment of the present invention, by updating an algorithm for calculating a damage level for each vehicle part and a vehicle part from an image, the accuracy of the degree of damage can be increased.

1 is a block diagram showing the structure of an AI automatic quotation system according to embodiments of the present invention.
2 is a block diagram showing the structure of an AI automatic estimation device according to embodiments of the present invention.
3 is a block diagram showing the structure of the breakage level determining unit 240.
4 is a diagram showing the structure of a management server according to embodiments of the present invention.
5 is a flowchart of an AI automatic estimation method according to embodiments of the present invention.
6 is an exemplary diagram representing damaged parts recognized by the AI automatic estimator.
7 is an exemplary view of a damage level list provided by the AI automatic estimation device according to embodiments of the present invention.
8 and 9 are diagrams for explaining an image pre-processing process performed by the breakage level determination unit of the AI automatic estimator.
FIG. 10 is an exemplary view of a screen comparing an estimate automatically generated by the AI automatic quotation device and an actual quotation from an industrial company.
11 is an exemplary diagram of result data obtained through an image.
12A, 12B, and 12C are exemplary views of standard hydraulic acceptance criteria used in the AI automatic estimator.

The present invention can be applied to various transformations and can have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail in the detailed description. Effects and features of the present invention and methods for achieving them will be clarified with reference to embodiments described below in detail together with the drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various forms.

According to embodiments of the present invention, the AI automatic quotation system may generate an automatic estimate by analyzing the image of the accident vehicle and automatically calculating the repair cost. The AI automatic quotation system can prevent the addition of excessive repair details by sending an automatic quotation to the terminal of the industrial company that received the accident vehicle according to the image recognition technology and standard repair recognition standards. By providing standard repair details for the accident vehicle, the increase in auto insurance premiums can be limited. The AI auto-estimate system can accurately detect the damage level of each part of the vehicle and the vehicle by using only the image of the accident vehicle. The AI auto-quote system helps you handle accidents in conjunction with auto insurance companies. The AI automatic quotation system can automatically generate and provide estimates based on the degree of damage and damage that were determined by the naked eye through the image of the accident vehicle. A repair quote for a standardized accident vehicle can be generated to prevent unpredictable charges.

1 is a block diagram showing the structure of an AI automatic quotation system according to embodiments of the present invention.

The AI automatic quotation system 10 may include a user terminal 100, 400, an AI automatic quotation device 200, and a management server 300.

The AI automatic estimating apparatus 200 may control the function of photographing the accident vehicle to be automatically executed when the automatic quote request signal is received from the terminal 100 of the owner or the insurance company employee of the accident vehicle. The AI automatic quotation system 10 will take and receive the necessary images in the preview screen, in the process of scanning the accident vehicle.

The AI automatic quotation device 200 generates an automatic quotation for the accident vehicle according to an automatic quotation request signal. The AI automatic estimating device 200 calculates the damage level for each vehicle type, year, accident site, and vehicle part from the images of the accident vehicle. The generated automatic quotation is provided to the management server 300 for history management or to the terminal 400 of the industrial company in order to process the accident vehicle with proper damages.

The user terminals 100 and 400 are electronic devices provided by users related to the accident vehicle, such as a vehicle owner, an insurance company employee, a terminal of an industrial company, etc., and the AI automatic estimating device 200 and the management server 300 through a communication network. ). The user terminals 100 and 400 may be provided with information processed by the AI automatic estimating apparatus 200 and may input data to the management server 300.

When an accident is detected with a shock or the like by the black box or the sensor of the accident vehicle 101, an accident detection signal is generated by the user terminal 100, and the user terminal 100 receives the accident detection signal from the AI automatic estimation device 200 ). The AI automatic estimating device 200 that detects an accident through the above-described process may acquire an image of the accident vehicle through the user terminal.

1, the user terminals 100 and 400 are respectively shown in singular, but a plurality of user terminals may be connected to the AI automatic estimator 200 and the management server 300.

The user terminals 100 and 400 refer to a communication terminal capable of transmitting and receiving data in a wired/wireless communication environment. Here, the user terminals 100 and 400 may be a user's personal computer or a user's portable terminal. Although the portable terminal is shown in FIG. 1 as a smart phone, the spirit of the present invention is not limited to this, and as described above, a program capable of connecting to a communication network is mounted or a terminal connected to a communication module can be borrowed without limitation.

To explain this in more detail, user terminals 100 and 400 include computers (eg, desktops, laptops, tablets, etc.), media computing platforms (eg, cables, satellite set-top boxes, digital video recorders), and handheld computing. Devices (eg, PDAs, email clients, etc.), may include any form of mobile phone, or any form of other type of computing or communication platform, but the invention is not so limited.

The AI automatic estimator 200 detects the vehicle part of the vehicle of each accident from the images of the accident vehicle, infers the damage level for each vehicle part, and calculates an automatic quote by applying the inferred contents to the standard hydraulic acceptance criteria. Can.

The management server 300 performs a function of updating an algorithm for inferring more accurate vehicle parts and damage levels for each vehicle part. The updated algorithm may be periodically provided to the AI automatic estimation device 200. The management server 300 receives an image of an actual accident vehicle, an actual estimate for the accident, and the like, and corrects modeling data for automatic calculation by a person in charge of the accident vehicle (insurer employee, industrial employee, etc.). The modeling data refers to commands constituting a partial detection algorithm, a failure determination algorithm, and the like, and may be used to re-adjust the partial algorithms applied to the published algorithm to determine the vehicle portion detection and damage degree.

The communication network 500 connects the plurality of user terminals 100 and 400, the AI automatic estimation device 200, and the management server 300. That is, the communication network 500 refers to a communication network that provides a connection path for the user terminals 100 and 400 to transmit and receive data after accessing the AI automatic estimator 200 and the management server 300. The communication network 400 includes, for example, wired networks such as Local Area Networks (LANs), Wide Area Networks (WANs), Metropolitan Area Networks (MANs), and Integrated Service Digital Networks (ISDNs), wireless LANs, CDMA, Bluetooth, satellite communication, etc. It may cover a wireless network, but the scope of the present invention is not limited thereto.

2 is a block diagram showing the structure of an AI automatic estimation device 200 according to embodiments of the present invention.

As shown in FIG. 2, the AI automatic estimation device 200 may include a processor 210, a memory 220, and a communication unit 230 to calculate an automatic quotation.

The processor 210 typically controls the overall operation of the device 200. For example, the processor 210 may control the communication unit 230 and the like by executing programs stored in the memory 220.

The communication unit 230 may include one or more components that enable communication between the user terminals 100 and 400, the management servers 300, and the AI automatic estimating device 200. The communication unit 230 is a Bluetooth communication unit, a Bluetooth Low Energy (BLE) communication unit, a near field communication unit (Near Field Communication unit), a WLAN (Wi-Fi) communication unit, a Zigbee communication unit, an infrared (IrDA, infrared data association) communication unit, a WFD ( Wi-Fi Direct) communication unit, UWB (ultra wideband) communication unit, Ant+ communication unit, and the like, may include a short-range communication unit, a mobile communication network.

The memory 220 includes a reception unit 221 that receives data received through the communication unit 230, a damage level determination unit 240 that analyzes the received data to determine a damage level, a damage level, and pre-stored standard repair recognition standards Includes a job information generating unit 222 for generating job information in consideration of, and an automatic quote generating unit 223 for generating an automatic quote considering the job information and the current wage, and receiving feedback in addition to generating the automatic quote It may include a feedback receiving unit 224, a logic optimization unit 225 for updating the damage level algorithm.

When the vehicle accident is received from the user terminal 100, the receiver 221 receives images of the vehicle. The reception unit 221 may further receive specific information on the damaged portion through a third party to prevent further repair of the previous damage caused by the owner's negligence.

In order to prevent a false report of an insurance accident, the reception unit 221 may identify whether the accident vehicle matches the user's vehicle, thereby preventing damage to another accident vehicle. When the receiver 221 compares the registered vehicle (vehicle, year, etc.) information using the identification information of the user terminal with the vehicle model, year, etc., obtained through the image, determines whether the accident vehicle is matched, and if it is determined that it is a match As long as it starts, automatic quotation creation will begin. In this case, information such as a vehicle model and year may be obtained through a vehicle information request signal including a vehicle number obtained through an image.

The damage level determination unit 240 may detect a vehicle part from images taken of an accident vehicle, and determine a damage level for each detected vehicle part. The damage level determination unit 240 may detect images taken by the accident vehicle according to an irregular boundary line of a vehicle part using a partial detection algorithm. As the partial detection algorithm, one of various methods such as YOLO, SSD, R CNN, MASK R CNN, and TEST R CNN, which are standardized algorithms, are used, but are not limited thereto, and are partially or totally free can be changed. The partial detection algorithm is modeled by the management server 300 and follows the selected algorithm based on the ratio of accurately detecting a vehicle part in the image of the accident vehicle.

Vehicle parts can be recognized by being classified by vehicle type. For example, the appearance of the SUV and sedan models and the parts of each vehicle may be recognized differently. In this embodiment, the vehicle portion is a front bumper, left fender, head lamp left, right fender, head lamp right, left door, right door, left rear door, right rear door, left rear fender, rear combi lamp, right rear fender, rear combi, It may include at least one of a trunk lid, a rear bumper, a hood, and the like.

In order to accurately determine the damaged portion, a standard vehicle image may be managed for each vehicle portion for each vehicle type. The vehicle part according to the vehicle model may be determined by comparing the vehicle parts included in the captured image with a standard vehicle image for each vehicle part of the corresponding vehicle type.

The damage level determination unit 240 may analyze each of the object areas detected for each vehicle part, and determine the damage level. The damage level determination unit 240 may use a failure determination algorithm modeled by the management server 300. The damage determination algorithm includes a correspondence relationship between the properties of the images obtained by repeatedly learning the data set of the image and the damage level of the accident vehicle and the damage level. The damage determination algorithm is an image recognition algorithm, and various methods such as standardized models ResNet, Inception, and Inception-ReNnet can be used, but are not limited thereto.

The work information generation unit 222 may generate work information of the accident vehicle by applying the damage levels for each part of the vehicle and the vehicle parts obtained from the image of the accident vehicle to the standard repair acceptance criteria prescribed according to the vehicle model and the year.

The work information of the accident vehicle includes ancillary work, and among the ancillary work, the scope of accreditation work is determined by the standard acceptance criteria. Examples of work information provided by this embodiment include front bumper exchange, front bumper repair, left shoulder exchange, left shoulder sheet metal, headlamp left exchange, right shoulder exchange, right shoulder sheet metal, head lamp right exchange, left door exchange, Left door sheet metal, right door exchange, right door sheet metal, left and rear door sheet metal, left and rear door sheet metal, upper and rear door sheet metal, upper and rear door sheet metal, left and right shoulder sheet replacement, left and rear sheet metal sheeting, rear combi lamp (left), right and rear sheet metal replacement, right and rear sheet metal replacement , Rear combi lamp (right), trunk lead exchange, trunk lead sheet metal, rear bumper exchange, rear bumper sheet metal, hood exchange, hood sheet metal. At this time, when it is determined to be sheet metal, the time of the sheet metal may be further displayed as 0.5, 1, 1.5, 2, or the like.

Here, the standard repair acceptance criteria may include distinct repair costs according to vehicle model, year, paint coat, factory M/H, repair method, and the like. Vehicles registered to the standard acceptance standards include nf Sonata, Sonata Transform, yf Sonata, if Sonata, New Morning, All New Morning, All New Morning, Santa Fe, Sportage, SM5, Starex, Carnival, Sorento, Tucson, K5, etc. It can be.

The automatic quote generating unit 223 generates an automatic quote based on the work information of the accident vehicle. In consideration of the vehicle type and year of the accident vehicle, the repair cost of each recognized work included in the work information can be extracted, and an automatic quotation including the approved work details and the repair cost can be generated. The automatic estimator generating unit 223 generates work information for the accident vehicle by considering the damage level and the damage level for each damaged part, the vehicle name, the vehicle model, the model year, and the painting coat, etc. using a pre-stored table. And acquire detailed tasks related to the work information, generate an automatic quotation by reflecting the standardized working time of the detailed tasks and the number of maintenance contracts made by each company, and provide the automatic quotation to the user terminal Can.

The feedback receiving unit 224 receives feedback on the generated automatic quotation. At this time, the feedback receiving unit 224 receives, from the terminal of an industrial company that actually repairs the suitability of the recognition work, the suitability of the damaged part, the suitability of the damage level, and the suitability of each repair cost in the automatic quote. The feedback received from the industry can be used to update the scope of the accreditation work of the standard acceptance criteria for pre-stored and the repair cost of each accreditation work. In addition, the feedback may be transmitted to the management server 300 and used to update the partial detection algorithm and the damage determination algorithm.

The logic optimization unit 225 performs a function of optimizing the standard hydraulic acceptance criteria used to generate an automatic quote. The logic optimizer 225 updates the standard repair acceptance criteria based on the price of each vehicle part of the vehicle manufacturer when a new model of the vehicle is introduced.

When the discontinued vehicle is generated, the logic optimization unit 225 may process an accident estimate of the discontinued vehicle so that it cannot be automatically estimated and request an actual estimate from the terminal of the industrial company.

The logic optimizer 225 may stop generating an automatic quote and request an actual treatment quote through the terminal of an insurance company or an industrial company when the entire damaged vehicle image is received.

Through this, the AI automatic estimation device 200 according to the embodiments of the present invention can automatically infer the damage level of the accident vehicle and generate an automatic quotation without receiving an industrial company. The AI automatic estimator 200 can suppress the occurrence of a repair process that exceeds the degree of damage and reduce errors in damage estimation. The AI automatic estimator 200 can automatically generate an estimate with only the input of an image that minimizes the user's input. The AI automatic estimating apparatus 200 may simply provide an incident and brief damage calculation through a terminal carried by the user. The AI automatic estimating apparatus 200 may increase the accuracy of the degree of damage by updating the algorithm for calculating the damage level for each vehicle part and vehicle part from the image.

3 is a block diagram showing the structure of the breakage level determining unit 240.

As illustrated in FIG. 3, the breakage level determining unit 240 includes a breakage site inference unit 241 that detects a damaged vehicle part, and a breakage level inference part 242 that stochastically infers a breakage level for each breakage part. It can contain.

The damaged portion inference unit 241 may detect object regions corresponding to the damaged portion of the accident vehicle from one or more images of the accident vehicle using an algorithm learned through a set of one or more images that are labeled and input.

The damaged part inference part 241 may be expressed by including a label in each of the object areas corresponding to the damaged part.

The damaged part inference unit 241 detects object areas including areas recognized as bumpers, fenders, and doors of the accident vehicle in the received one or more image sets, but is detected to match the irregular borders of the bumpers, fenders, and doors. Can.

The breakage level inference unit 242 acquires breakage information for an object region corresponding to a breakage region using a breakage determination algorithm, and is arranged in the order of one or more breakage levels inferred from the breakage information and a matching probability for each breakage level You can create a list of breakage levels. The breakage level inference unit 242 filters each specifically labeled object region in a method adopted through a CNN convolutional neural network layer, and one or more breakage levels and breakage levels inferred from breakage information that is result data based on the filtered images It is possible to generate a list of damage levels arranged in the order of magnitude of star matching probability. The CNN layer may perform a function of selecting a filtering method for filtering attributes associated with a breakage level among attributes included in an image. The labeled object area may be filtered by at least one of a Sobel mask or a canny edge detector. The object region is characterized in that after filtering, the contrast is uniformized by a uniformity algorithm. The damage level of each object area can be set to one of normal, sheet metal 0.5, sheet metal 1, sheet metal 1.5, sheet metal 2, and exchange. The breakage level list is arranged in the order of the matching probability for normal, sheet metal, 0.5, sheet metal 1, sheet metal 1.5, sheet metal 2, exchange, and can provide the highest level of failure level with the highest matching probability.

The breakage level inference unit 242 determines the breakage level in the highest rank as the breakage level for the corresponding breakage site.

The damaged portion inference unit 241 calculates whether an error of the damaged portion of the input one or more images occurs, and when the error level exceeds a preset first value, labels the image of the accident vehicle and transmits it to the management server to learn again Can perform the function. When the accuracy of the damage site value is less than 50%, the detected damage site can be determined as an error. The accuracy of the damage site value can be obtained through information provided by the partial detection algorithm.

The damage level inference unit 242 receives the error level of the damage level, and when the error level exceeds a preset second value, transmits the object area and the damage level corresponding to the damaged part to the management server to be learned again. Can perform a function. When the matching probability of the breakage level is less than 50%, the inferred breakage level may be determined as an error. The accuracy of the breakage level value can be obtained by the breakdown level list provided by the breakage determination algorithm.

4 is a diagram showing the structure of a management server according to embodiments of the present invention.

The management server 300 may include a processor 310, a memory 320, and a communication unit 330.

The processor 310 typically controls the overall operation of the management server 300. For example, the processor 310 may control the communication unit 330 and the like by executing programs stored in the memory 320.

The communication unit 330 may include one or more components that enable communication between the user terminals 100 and 400, the AI automatic estimation device 200, and the management server 300. The communication unit 330 is a Bluetooth communication unit, a Bluetooth Low Energy (BLE) communication unit, a near field communication unit (Near Field Communication unit), a WLAN (Wi-Fi) communication unit, a Zigbee communication unit, an infrared (IrDA) infrared data association (WFD) communication unit, WFD ( Wi-Fi Direct) communication unit, UWB (ultra wideband) communication unit, Ant+ communication unit, and the like, may include a short-range communication unit, a mobile communication network.

The memory 320 may include a data input unit 321, an attribute extraction unit 322, a function application unit 323, a collection unit 324, a removal unit 325, and a model building unit 326.

The data input unit 321 performs a function of inputting data for a model to be built.

The attribute extraction unit 322 extracts various attributes (color values for each pixel, brightness values, transparency, and sharpness) included in the input image. The attribute extraction unit 322 extracts as many attributes as possible from the image without a reference. The attribute extraction unit 322 may further perform a sampling process for reducing the amount of attributes.

The function application unit 323 hierarchically generates a correspondence relationship between the attributes as images and damage levels are input. Modeling between the properties of the images and the damage level can be constructed by inferring the correspondence relationship so that the properties of the image correspond to the damage level and repeating the process of calculating the accuracy of the correspondence through the input data.

The model building unit 324 repeatedly performs the attribute extraction unit and the function application unit, thereby constructing modeling having the highest accuracy in the input data.

5 is a flowchart of an AI automatic estimation method according to embodiments of the present invention.

In S100, the AI automatic estimator receives a set of images of the accident vehicle. The set of images taken by the accident vehicle can be controlled and photographed by the AI automatic estimator. For example, when an accident is detected through the vehicle's black box, the camera of the owner's terminal may be automatically executed and controlled to be captured at the moment the accident vehicle is recognized.

In S200, the AI automatic estimator can determine the vehicle type and year of the accident vehicle by considering the image set. The AI automatic estimating device may determine the vehicle type and year of the accident vehicle based on the image, but may be determined based on the subscription information entered when registering for auto insurance.

In S300, the AI automatic estimator can detect the object regions corresponding to the vehicle part in the image set by instant segmentation.

In S400, the AI automatic estimator calculates the degree of damage using the features extracted from the object regions by using the failure level algorithm, and the one or more failure levels inferred from the degree of damage are arranged in the order of matching probability. You can create a level list.

In S500, the AI automatic estimating apparatus determines the damage level having the largest matching probability as the damage level of the object area.

In S600, the AI automatic estimator converts the work information according to the breakage levels of the object areas, and the AI automatic estimator can calculate costs incurred from detailed tasks included in the work information using standard acceptance criteria.

In S700, the AI automatic quotation device can generate the automatic quotation by synthesizing the calculated costs.

The AI automatic estimating device may perform a third-party verification procedure for the damaged vehicle part in order to prevent the damage not caused by the vehicle accident from being repaired.

The AI automatic estimator is designed to prevent illegal gains by registering photos of other accident vehicles.

6 is an exemplary diagram representing damaged parts recognized by the AI automatic estimator.

The damage level determination unit of the AI automatic estimation device detects object areas corresponding to the vehicle part included in the image of the accident vehicle. The damage level determination unit detects an object area corresponding to the front bumper, front shoulder L, front shoulder R, front door L, and front door R of the accident vehicle. At this time, the detected object regions may have an irregular boundary directly surrounding the corresponding portion in order to obtain the exact name of the damaged portion without overlapping each other. As illustrated in FIG. 6, the damage level determining unit is a front bumper area FBP, a front fender R area FFR, a front door R area FDR, and a rear door R area RDR regardless of the shooting angle, image resolution, and the like. , It is possible to detect and label the rear R area (RFR) from the accident vehicle image. The damage level determination unit recognizes the damaged portion regardless of an image capturing method by recognizing a vehicle portion through images taken according to various shooting angles and image resolutions by modeling the management server.

7 is an exemplary view of a damage level list provided by the AI automatic estimation device according to embodiments of the present invention.

The AI automatic estimator can set the damage level according to the degree of damage to the recognized vehicle part as one of a bad level, a middle level, a burn level, and an exchange level. At this time, the AI automatic estimating device may infer the damage level and the matching probability for each damaged portion, and determine the damage level having the largest matching probability as the failure level of the corresponding damaged portion. As illustrated in FIG. 7, a list in which the FFR region has the highest probability of the middle hand level, the FDR region has the highest probability of the bad debt level, and the FBP region may be provided with a list having the highest probability of the middle hand level. Accordingly, the AI automatic estimating apparatus determines the middle level for the FFR region, the bad level for the FDR region, and the middle level for the FBP region.

The determined damage level may be provided to the owner of the vehicle, the responsible person in charge of the accident vehicle, the industrial company server of the accident vehicle, and the like.

8 and 9 are diagrams for explaining an image pre-processing process performed by the breakage level determination unit of the AI automatic estimator.

8 is an example of images before pre-processing, and FIG. 9 is an example of images processed through a pre-processing process.

The damage determination algorithm that determines the damage level for each vehicle part used by the damage level determination unit is an unspecified attribute included in the image after receiving the image of the vehicle part in the accident vehicle and the damage level (lvl) for the vehicle part. Through the classification process according to the correspondence between (features), the relationship with the damage level lvl can be randomly determined. The damage determination algorithm intuitively learns and learns the association between the properties of the unprogrammed image and the damage level by learning through quality data, that is, the image and the exact damage level (lvl) according to the image. It is possible to infer the damage level (lvl) through the intuition. In addition, the damage determination algorithm itself performs a process of selecting a pre-processing method of an image (image) that increases the accuracy of inferring the damage level (lvl) from among the previously disclosed pre-processing methods. As illustrated in FIGS. 8 and 9, the damage determination algorithm may specifically preprocess an image using a Sobel mask, a canny edge detector filter, and a contrast equalization algorithm.

Preprocessing Img11, img 21, img31, img41, img51, img61, img71, img81, and img91 in a method selected by the damage determination algorithm, img 12, img22, img32, img42, img52, img62 , img72, img82, and img92 are created.

FIG. 10 is an exemplary view of a screen comparing an estimate automatically generated by the AI automatic quotation device and an actual quotation from an industrial company.

The AI automatic quotation device can calculate the damage of the accident vehicle at the user's request and provide the automatic quotation accordingly, and further perform the process of comparing the provided automatic quotation with the actual quotation generated by the industrial company.

As shown in FIG. 10, when the AI auto-estimate device determines the detected vehicle part and the damage level for each vehicle part, it automatically generates work details in consideration of the vehicle model, vehicle name, year, painting coat, and damage level for each vehicle part. can do. For example, when a bumper sheet metal and painting are required, the time required for the bumper sheet metal is calculated according to the damage level, the labor cost is calculated by applying the hourly wage of the current technicians to the required time, and the painting work included in the painting work , Drying operations, etc. are created in detail.

At this time, the work details are generated in accordance with the preset standard repair acceptance criteria according to the damage level for each vehicle part, and detailed work details included in the work details are extracted.

The automatic quotation can be generated considering the incidental work such as the waiting time for vehicle repair, the drying time after painting, and the detachment time of other parts when the bumper is replaced, as well as the cost of returning the damaged part.

11 is an exemplary diagram of result data obtained through an image. A management number is assigned to each vehicle type, model year, painting, and repair part, and can be stored by setting the scope of accreditation work during ancillary work in correspondence with the management number. The AI automatic estimation device according to the embodiments of the present invention may search for a management number corresponding to the detected vehicle part, and extract a recognition work range for each damage level included in the management number. You can generate an automatic quote by listing the scope of accredited work and the repair cost for each work.

The scope of the accreditation work can be revised through the actual estimate of the feedbacked industry. For example, in the case of the approved work, the painting time was 3 hours, but according to the actual estimate of the industrial company, if the painting time is increased to 4 hours, 30 minutes, which is 50 percent of the increased time, can be added as the recognized work range. In addition, the database can be standardized and stored for repair cost and damages recognition standards.

12A, 12B, and 12C are exemplary views of standard hydraulic acceptance criteria used in the AI automatic estimator. The AI automatic estimator can search for the code corresponding to the breakage level for each damaged part obtained through the image, and obtain the work history and the amount of labor for each work history.

The standard repair base (role base) has a breakdown of work according to the degree of damage by vehicle type.

As shown in FIGS. 12A and 12B, a code list for classifying and identifying the vehicle model and a code list for identifying each vehicle part of the vehicle model may be provided.

The code to classify and identify the vehicle model can be registered and managed in detail, such as Avante HD, Avante MD, Avante Hybrid LPi, Avante AD.

The codes to identify each vehicle part include front bumper replacement, front bumper repair, left shoulder replacement, left shoulder sheet metal, headlamp left exchange, right shoulder exchange, right shoulder sheet, headlamp right exchange, left door exchange, left door sheet metal, Right door exchange, right door sheet metal, left and back door exchange, left and rear door sheet metal, right and rear door exchange, right and rear door sheet metal, left and right rear plate exchange, left and rear rear plate, rear combi lamp, right and rear rear plate exchange, right and rear rear plate, etc. .

As illustrated in FIG. 12C, an automatic estimate of the accident vehicle may be generated in a table including the damaged object name, vehicle type, description, work type, part amount, labor amount, main work code, and auxiliary work code.

The work details that should be executed on the accident vehicle by the identification code (rcp_nv_seqno) given for each registered vehicle accident are the front bumper side bracket (left), the front bumper, the heating and drying ratio, the front bumper side bracket (right), the radiator grill, and the front bumper. Exchange, pain-time work may be involved.

Each work is classified as a main work or an auxiliary work, and the work time is determined according to the damage level of the 00001 event and the standard work time obtained by the standard repair acceptance criteria. For example, considering the degree of breakage of the incident, the front bumper replacement takes 1.91 hours. The painting of the front bumper exchange is set to take 1.98 hours and an automatic quote can be generated based on it.

Through this, if the damaged part and the damaged level are determined through the image recognition technology, an estimate for repairing the vehicle in question can be automatically generated by applying the vehicle model, year, damaged part, and damaged level to the standard repair acceptance criteria.

The device described above may be implemented with hardware components, software components, and/or combinations of hardware components and software components. For example, the devices and components described in the embodiments may include, for example, processors, controllers, arithmetic logic units (ALUs), digital signal processors (micro signal processors), microcomputers, field programmable gate arrays (FPGAs). , A programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose computers or special purpose computers. The processing device may perform an operating system (OS) and one or more software applications running on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For convenience of understanding, a processing device may be described as one being used, but a person having ordinary skill in the art, the processing device may include a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include. For example, the processing device may include a plurality of processors or a processor and a controller. In addition, other processing configurations, such as parallel processors, are possible.

The software may include a computer program, code, instruction, or a combination of one or more of these, and configure the processing device to operate as desired, or process independently or collectively You can command the device. Software and/or data may be interpreted by a processing device or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodied in the transmitted signal wave. The software may be distributed on networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and usable by those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic media such as floptical disks. -Hardware devices specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language code that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described by a limited embodiment and drawings, those skilled in the art can make various modifications and variations from the above description. For example, the described techniques are performed in a different order than the described method, and/or the components of the described system, structure, device, circuit, etc. are combined or combined in a different form from the described method, or other components Alternatively, even if replaced or substituted by equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

10: AI automatic quotation system
100, 400: user terminal
101: accident vehicle
200: AI automatic estimator
300: management server

Claims (7)

The AI automatic estimating apparatus receives the entire image of the accident vehicle from the user terminal, and receives a partial image of the accident site from the user terminal;
Determining, by the AI automatic estimating apparatus, the vehicle model and the year of the accident vehicle by using the entire image and pre-stored standard vehicle images;
Among the bumper, fender, and door objects areas corresponding to the damaged part of the accident vehicle from one or more images photographed the accident vehicle using an algorithm learned through one or more image sets in which the AI automatic estimator is labeled and input. Detecting to match one irregular boundary line, and expressing by including a label in each of the object areas corresponding to the damaged area;
The AI automatic estimator uses a damage determination algorithm to filter the object regions corresponding to the labeled damage region through a CNN layer with a filter of either a Sobel mask or a canny edge detector, and the object regions are converted to a contrast level by a uniformity algorithm. Uniformizing and generating a list of damage levels arranged in the order of one or more damage levels and a matching probability for each damage level, based on the object regions; And
The AI automatic estimating device generates work information for the accident vehicle in consideration of the breakage level and the breakage level for each breakage part, the vehicle model, and the model year, obtains detailed tasks related to the work information, and details AI automatic quotation method comprising; applying a standardized working time of the work and generating an automatic quotation by reflecting the number of maintenance made by individual contracts by company, and providing the automatic quotation to the user terminal. According to claim 1,
And the AI automatic estimating device delivering the automatic quotation to the industrial terminal where the accident vehicle has been moved.
According to claim 2,
Receiving the AI automatic estimator from a terminal of an industrial company that actually repairs the suitability for the automatic quote, the suitability for the detailed work included in the automatic quote, and the suitability for the cost of the detailed work; And
AI correction method, including; taking into account the received suitability, modifying the logic to generate an automatic quote.
A receiving unit for receiving the entire image of the accident vehicle and the partial image of the accident site from the user terminal through the communication unit;
Using the entire image and pre-stored standard vehicle images, the vehicle type and year of the accident vehicle are determined,
The object regions corresponding to the damaged parts of the accident vehicle are matched to irregular boundaries of one of the bumpers, fenders, and doors from one or more images photographed of the accident vehicle using an algorithm learned through a set of one or more images that are labeled and input. To detect and include a label in each of the object areas corresponding to the damaged area
Using the damage determination algorithm, the labeled object regions and the corresponding object regions are filtered through a CNN layer with a filter of either a Sobel mask or a canny edge detector, and the object regions are equalized by a uniformity algorithm, and the object regions are equalized. A breakage level determination unit that generates a list of breakage levels arranged in order of one or more breakage levels and a matching probability for each breakage level based on the breakage site; And
Create work information for the accident vehicle in consideration of the breakage level and the breakage level for each breakage site, the vehicle type, and the model year, obtain detailed work related to the work information, and apply standardized work time of the work And an automatic quote generating unit for generating an automatic quote by reflecting the number of maintenance made of individual contracts for each company and providing the automatic quote to the user terminal. According to claim 4,
The communication unit
AI automatic estimating device, characterized in that for delivering the automatic quote to the industrial terminal of the accident vehicle is moved. The method of claim 5,
A feedback receiving unit received from an industrial terminal that actually repairs the suitability for the automatic quotation, the suitability for the detailed work included in the automatic estimate, and the cost for the detailed work; And
AI automatic estimating device further comprising a logic optimization unit for modifying the logic for generating an automatic quotation, in consideration of the received suitability.
A computer program stored in a computer-readable storage medium for executing the method of claim 1 using a computer.

Images