Intelligent claim settlement method and device, computer device, and storage medium
By obtaining road surface texture fingerprints and physical tilt angles through semantic segmentation and slope calculation, and combining them with a map database to verify the authenticity of claims photos, the problem of identifying fake crime scenes has been solved, thus improving the security and accuracy of insurance claims.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- PING AN TECH (SHENZHEN) CO LTD
- Filing Date
- 2026-04-16
- Publication Date
- 2026-06-05
AI Technical Summary
Current technology cannot effectively identify whether a vehicle accident scene in a claim photo is a fake collision scene, resulting in low security for insurance claims.
The target road surface texture fingerprint is obtained by semantic segmentation, and the physical inclination angle of the target road surface is obtained by slope calculation. The reference road surface information is queried from the map database to identify risks and verify the authenticity of the road surface in the claim photo.
It improves the security of self-service insurance claims, prevents fraudsters from simulating a consistent road surface at a fake scene, provides multi-dimensional environmental evidence, and reduces the manual input of claims investigation.
Smart Images
Figure CN122155874A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of artificial intelligence technology and image processing technology, and is applicable to the fields of financial technology and healthcare. In particular, it relates to an intelligent claims processing method, device, equipment, and storage medium. Background Technology
[0002] In the field of artificial intelligence technology, self-service insurance claims have been implemented on the server side. In the fields of fintech and healthcare, users upload photos of vehicle accident scenes to the server. After the server approves the photos, it can process the corresponding insurance claim for the user (such as vehicle damage insurance, third-party liability insurance, medical insurance, etc.).
[0003] Currently, existing deep learning models typically identify "road background" from claim photos during the review process, but they cannot verify whether the vehicle accident scene in the claim photo is a fake collision scene, resulting in low security for insurance claims. Summary of the Invention
[0004] This invention provides an intelligent claims method, apparatus, device, and storage medium to solve the technical problem of low security in self-service insurance claims due to the difficulty in identifying fake accident scenes.
[0005] Firstly, an intelligent claims processing method is provided, applied on the server side, the method comprising: In response to detecting that a user has triggered a self-service insurance claim on the client, the system takes photos of the road surface at the accident scene through the client to obtain claim photos; Semantic segmentation is performed on the claim photos to obtain the target road surface texture fingerprint; Based on the claimed photos, the slope is calculated to obtain the physical inclination angle of the target road surface; The client obtains the target location information of the user and queries a preset map database based on the target location information to obtain reference road surface information. Based on the reference road surface information, risk identification is performed on the target road surface texture fingerprint and the target road surface physical inclination angle to obtain claim risk information; Based on the claim risk information and the claim photos, the insurance claim will be processed for the user.
[0006] Secondly, a smart claims processing device is provided for use on a server side, the device comprising: The claims trigger module is used to respond to the detection that a user has triggered the self-service insurance claims on the client, and to take pictures of the road surface at the scene of the vehicle accident through the client to obtain claims photos; The semantic segmentation module is used to perform semantic segmentation on the claim photo to obtain the target road surface texture fingerprint; The slope calculation module is used to calculate the slope based on the claim photos to obtain the physical inclination angle of the target road surface; The information query module is used to obtain the target location information of the user through the client, and query the preset map database based on the target location information to obtain reference road surface information; The risk identification module is used to identify risks based on the reference road surface information, the texture fingerprint of the target road surface, and the physical inclination angle of the target road surface, to obtain claims risk information. The insurance claims module is used to process insurance claims for the user based on the claim risk information and the claim photos.
[0007] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned intelligent claims settlement method.
[0008] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the aforementioned intelligent claims settlement method.
[0009] In the aforementioned intelligent claims method, intelligent claims device, computer equipment, and storage medium, the server responds to the detection that a user has triggered a self-service insurance claims service on the client side. The server takes a picture of the road surface at the accident scene through the client to obtain a claims photo. Next, the server performs semantic segmentation on the claims photo to obtain a target road surface texture fingerprint. Then, the server calculates the slope based on the claims photo to obtain the target road surface physical inclination angle. Next, the server obtains the user's target location information through the client and queries a preset map database based on the target location information to obtain reference road surface information. Then, the server performs risk identification based on the reference road surface information, the target road surface texture fingerprint, and the target road surface physical inclination angle to obtain claims risk information. Finally, the server processes the insurance claim for the user based on the claims risk information and the claims photo. In this invention, for users who wish to apply for insurance claims through the self-service insurance claims service, the server no longer simply verifies whether the background in the claims photo is a road, but obtains the target road surface texture fingerprint through semantic segmentation and also obtains the target road surface physical inclination angle through slope calculation. Because the physical slope angle and texture fingerprint of a road surface are physical facts tied to its geographical location, it is extremely difficult for fraudsters to simulate a completely identical road surface at a fake scene. By then using reference road information to identify the target road surface texture fingerprint and physical slope angle, it is possible to determine whether the road surface in the claim photo is genuine, thereby improving the security of self-service insurance claims. Attached Figure Description
[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 This is a schematic diagram of an application environment for an intelligent claims settlement method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an intelligent claims settlement method according to an embodiment of the present invention; Figure 3 yes Figure 2 A schematic diagram of a specific implementation method for step S20; Figure 4 yes Figure 2 A schematic diagram of a specific implementation method for step S30; Figure 5 yes Figure 2 A schematic diagram of a specific implementation method for step S50; Figure 6This is a schematic diagram of the structure of an intelligent claims processing device according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 8 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation
[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0013] First, let's clarify some of the terms used in this invention: Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.
[0014] Natural Language Processing (NLP): NLP uses computers to process, understand, and utilize human language (such as Chinese and English). It is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, often referred to as computational linguistics. NLP includes syntactic analysis, semantic analysis, and discourse understanding. It is commonly used in machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information and image processing, information extraction and filtering, text classification and clustering, sentiment analysis, and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computation.
[0015] Large Language Model (LLM): An LLM is a model built using deep learning methods and trained on large-scale text data, capable of understanding and generating natural language. It possesses the ability to perform reasoning, answering questions, writing text, translating, summarizing, and conversing on various language tasks, and typically exhibits strong versatility and transfer learning capabilities.
[0016] In the field of artificial intelligence technology, self-service insurance claims have been implemented, allowing users to report incidents simply by taking photos of the road surface and damaged vehicles. However, existing technologies face the following serious challenges: (1) Accident scenes are easy to fake: Fraudsters can use damaged parts to create fake collision scenes in repair shops or remote road sections. Existing GPS verification can be modified by forgery software, and existing image AI can only identify "the background is a road" and cannot determine the real physical properties of the road section from a single photo.
[0017] (2) Lack of correlation with road surface physical characteristics: Each road section (such as asphalt road, cement road, anti-skid road) has its own unique physical "fingerprint". Currently, the claims app collects photos, but completely ignores the ground slope (tilt angle) data collected by the mobile phone sensor at the moment of shooting.
[0018] (3) Lack of physical-geographic logical alignment: The claims system cannot automatically verify whether the road surface roughness and drainage slope shown in the photo match the actual road administration data at that latitude and longitude in the cloud map database.
[0019] Based on this, embodiments of the present invention provide intelligent claims processing methods, devices, computer equipment, and storage media to solve the problems existing in the prior art.
[0020] The intelligent claims processing method provided in this invention can be applied to, for example... Figure 1In this application environment, the client communicates with the server via a network. The server can respond to a user triggering a self-service insurance claim on the client by taking a photo of the road surface at the accident scene through the client, obtaining a claim photo; the server performs semantic segmentation on the claim photo to obtain a target road surface texture fingerprint; the server calculates the slope based on the claim photo to obtain the target road surface physical inclination angle; the server obtains the user's target location information through the client and queries a preset map database based on the target location information to obtain reference road surface information; the server performs risk identification based on the target road surface texture fingerprint and the target road surface physical inclination angle based on the reference road surface information to obtain claim risk information; and the server processes the insurance claim for the user based on the claim risk information and the claim photo. In this invention, for users who wish to apply for insurance claims through the self-service insurance claim service, the server no longer simply verifies whether the background in the claim photo is a road, but instead obtains the target road surface texture fingerprint through semantic segmentation and also obtains the target road surface physical inclination angle through slope calculation. Because the physical slope angle and texture fingerprint of a road surface are physical facts bound to geographical location, it is extremely difficult for fraudsters to simulate a completely identical road surface at a fake scene. Then, by using reference road surface information to identify the risk of the target road surface texture fingerprint and physical slope angle, it can be determined whether the road surface in the claim photo is real, thereby improving the security of self-service insurance claims. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.
[0021] Please see Figure 2 As shown, Figure 2 A flowchart of an intelligent claims settlement method provided in an embodiment of the present invention includes the following steps S10-S60.
[0022] S10: In response to detecting that a user has triggered a self-service insurance claim on the client, the system takes photos of the road surface at the accident scene through the client to obtain claim photos.
[0023] The intelligent claims processing method provided by this invention can be applied to insurance claims engines in various application scenarios. The insurance claims engine can be implemented through a server-side platform, which can provide users with self-service insurance claims. In the fields of fintech and healthcare, users upload photos of vehicle accident scenes (also known as claims photos) and related claims materials, which are first received and accepted by the server. Submitted materials include, but are not limited to, vehicle accident scene photos, vehicle registration certificate / license, driver's license, insurance policy, claims application form, on-site inspection notes, and necessary timestamps and location information. The server performs format verification, integrity checks, and basic information extraction (such as license plate, vehicle model, accident time, location, and other metadata) on these uploaded claims materials. Subsequently, the process enters the intelligent review stage, utilizing multimodal analysis and a rule engine to comprehensively evaluate image quality, scene elements (such as collision location, degree of damage, presence of obstruction or forgery), material consistency, historical claims records, and risk signals. If the review is successful, the server automatically triggers the insurance claims process, completing the determination of claims and disbursement arrangements for different types of insurance (such as vehicle damage insurance, third-party liability insurance, and medical insurance).
[0024] The aforementioned road surface at the scene of a vehicle accident refers to the observable road surface area and its immediate surrounding environment at the time of the accident, serving as the physical space for recording the conditions of the accident, the extent of damage, and evidence. Claim photos include both the road surface area and the damaged vehicle area.
[0025] In one example, a user opens an insurance app on a client (such as a mobile phone) and clicks on the self-service insurance claims entry point. The client then generates a claims request and sends it to the server. Based on the received claims request, the server determines that the user has triggered the self-service insurance claims on the client and can then send a photo capture command to the client. The client, upon receiving the photo capture command, activates its camera and takes photos of the accident scene to obtain the claims photos.
[0026] S20: Perform semantic segmentation on the claim photos to obtain the target road surface texture fingerprint.
[0027] In one embodiment, a deep learning model is used to perform semantic segmentation on the claim photo to extract the microscopic features of the road surface area in the claim photo and obtain the target road surface texture fingerprint.
[0028] Target road surface texture fingerprint is a unique set of observable and quantifiable identification information of the surface texture features of the road surface at the accident scene. It is used to distinguish road surface material, texture density, wear level, wet / dry state, etc., thereby helping to determine the potential impact of road conditions on the occurrence of accidents and improving the credibility of claims evidence.
[0029] The target road surface texture fingerprint includes road material fingerprint and road wear fingerprint. Road material fingerprint: distinguishes the porosity and particle distribution of road materials such as asphalt, cement, and paving stones. Road wear fingerprint: calculates the fractal dimension of the road surface texture to characterize the road's age and frictional properties. Different road sections have unique fractal dimensions due to varying traffic volumes.
[0030] In some embodiments of the present invention, such as Figure 3 As shown, step S20, which involves semantic segmentation of the claim photo to obtain the target road surface texture fingerprint, includes the following steps: S21: Perform road surface inspection on the claim photos to obtain the target road surface area; S22: Use the first deep learning model to analyze the material of the target road surface area and obtain the road surface material fingerprint; S23: Use the second deep learning model to identify wear in the target road surface area and obtain the road surface wear fingerprint; S24: The target road surface texture fingerprint is obtained by fusing the road surface material fingerprint and the road surface wear fingerprint.
[0031] In step S21, road surface detection can be performed on the claim photo using an object detection model or an object segmentation model to obtain the target road surface area. The object detection model or object segmentation model can be an instance-oriented segmentation model (such as Mask R-CNN, DeepLabv3, Segment Anything, etc.) or a lightweight segmentation model to ensure inference performance. The target road surface area does not contain vehicles.
[0032] In one embodiment, step S21, which is to perform road surface detection on the claim photo to obtain the target road surface area, may include the following steps: performing road surface detection on the claim photo using multiple target segmentation models to obtain candidate road surface areas; performing vehicle debris detection on each candidate road surface area to obtain the proportion of vehicle debris in the candidate road surface area; sorting the candidate road surface areas based on the proportion of debris, and taking the candidate road surface area with the smallest proportion of debris as the target road surface area.
[0033] As can be seen from the above scheme, considering that there will be damaged vehicle bodies and scattered vehicle fragments at the scene of a vehicle accident, and that vehicle fragments will affect the recognition of road surface texture fingerprints, a multi-object segmentation model is introduced to detect road surface and vehicle fragments during the detection of the road surface area. This reduces the adverse effects of vehicle fragments on subsequent texture fingerprint recognition and helps to improve the accuracy of texture fingerprint recognition.
[0034] In steps S22-S23, the first and second deep learning models can be selected from existing models according to actual needs. For example, the first deep learning model can be a Transformer-based model, and the second deep learning model can also be a Transformer-based model. When training the first deep learning model, a first dataset is used, including: on-site photographs of the road surface, manually labeled road surface areas (segmentation masks), and material labels. When training the second deep learning model, a second dataset is used, including: on-site photographs of the road surface, manually labeled road surface areas (segmentation masks), and wear labels. The specific process of training the model using the dataset is similar to the conventional training process and will not be described in detail here.
[0035] Finally, in step S24, the server fuses the road surface material fingerprint and the road surface wear fingerprint to obtain the target road surface texture fingerprint.
[0036] As can be seen from the above steps S21-S24, the target road surface area is first segmented based on the claim photo, and then the road surface material fingerprint and road surface wear fingerprint are extracted separately, which improves the accuracy of road surface texture fingerprint extraction.
[0037] S30: Calculate the slope based on the claim photos to obtain the physical inclination angle of the target road surface.
[0038] Physical slope of road surface: The degree of inclination of the road surface relative to the horizontal plane in a geographic or local coordinate system. The physical slope of the road surface can also be expressed as the road inclination angle: the angle at which the road surface rises or falls forward, reflecting the longitudinal upward and downward slope trend of the road surface.
[0039] In some embodiments of the present invention, such as Figure 4 As shown, step S30, which involves calculating the slope based on the claim photos to obtain the physical inclination angle of the target road surface, includes the following steps: S31: Perform target detection on the claim photos to obtain the vehicle outline and the target road surface area; S32: Based on the vehicle profile and road surface area, perform boundary analysis to obtain the vanishing point of the contact line between the vehicle profile and the road surface. S33: Generate a road surface parallel line that is parallel to the target road surface area; S34: Based on the vanishing point and the road surface parallel line, the inclination angle is calculated to obtain the physical inclination angle of the target road surface.
[0040] In step S31, the claim photo can be used to detect objects using an object detection model to obtain the vehicle outline and the target road surface area. Object detection models include Mask R-CNN, DeepLabv3, Segment Anything, etc.
[0041] In step S32, the vehicle outline and road surface area can be mapped to a predetermined space to obtain the vehicle outline equation and the road surface equation; then, based on the vehicle outline equation and the road surface equation, numerical calculations are performed to obtain the contact line and the vanishing point of the contact line in the claim photo.
[0042] In step S33, parallel lines parallel to the road surface can be generated based on the road surface equation.
[0043] In step S34, for example, the camera center height is h, the pitch angle is 0 (horizontal); the road surface has a physical tilt angle α along the x-direction, and the ground normal vector is n = ( sin α, 0, cos α); the camera intrinsic parameters are known, and the vanishing point P = (u_v, v_v); by the relationship between projection and vanishing point, the mapping between the direction of the symmetric vector and the direction of the real world can be obtained, thus solving for α: the projection direction of the road surface direction on the image plane is consistent with the direction pointed to by the vanishing point P; by relating the unit direction vector v_dir in the image plane to the projection of the ground normal vector in the camera coordinate system, the physical tilt angle α of the target road surface can be solved.
[0044] As can be seen from the above scheme, by combining the vanishing point of the contact line between the vehicle outline extracted from the claim photos and the road surface area, the physical inclination angle of the road surface can be deduced. The calculation logic is simple and easy to implement.
[0045] S40: Obtain the user's target location information through the client, and query the preset map database based on the target location information to obtain reference road surface information.
[0046] The client collects and submits the user's target location information, which may include the user's current latitude and longitude, as well as necessary contextual information (such as user authorization, accuracy requirements, timestamps, etc.). The target location information may also be accompanied by environmental information (altitude, speed, direction sensor data, etc.) to improve query accuracy.
[0047] The server receives the target location information from the client and performs parameter verification and authentication. The server then queries the map database service, aggregates and transforms the data into client-friendly road surface information. The server returns reference road surface information (such as road segment name, road width, road condition labels, speed limits, reference lines, etc.) and metadata (accuracy, source, timestamp).
[0048] The map database stores road network geometry information, attribute information (road segment name, road grade, speed limit, road direction, administrative boundary, etc.), and additional data related to traffic events.
[0049] It should be noted that location information is collected only after obtaining user authorization; and a data minimization principle is set for sensitive areas or uses.
[0050] The reference road surface information includes a reference road surface texture fingerprint and a reference road surface physical dip angle. The reference road surface texture fingerprint is obtained based on real road surface photographs. The reference road surface physical dip angle is obtained either based on real road surface photographs or based on measurements of the real road surface using physical measurement tools.
[0051] S50: Based on reference road surface information, risk identification is performed on the target road surface texture fingerprint and physical inclination angle to obtain claims risk information. Specifically, a large language model can be used to assess the risks of reference road surface information (such as road segment attributes, texture fingerprint, physical inclination angle, etc.), target road surface texture fingerprint, and target road surface physical inclination angle, outputting claims risk information that can be used for claims decision-making. Claims risk information may include: claims risk level or claims risk score (such as low / medium / high, score range 0-100), a list of influencing factors (texture fingerprint anomalies, inclination angle anomalies, signs of road surface deterioration, etc.), a measure of the difference between the reference road surface and the target road surface (distance, angle, texture similarity, etc.), and suggestions for the claims process (whether on-site inspection is required, whether there is a risk of false alarm, and a list of evidence to be collected).
[0052] In some embodiments of the present invention, reference is made to... Figure 5 Step S50, which involves risk identification based on the target road surface texture fingerprint and physical inclination angle using reference road surface information to obtain claim risk information, includes the following steps: S51: Compare the target road surface texture fingerprint with the reference road surface texture fingerprint to obtain texture difference information; S52: Compare the slope angles of the target road surface and the reference road surface to obtain slope angle difference information; S53: Based on the fusion of texture difference information and tilt angle difference information, claim risk information is obtained.
[0053] In step S51, the differences between the target road surface texture fingerprint and the reference road surface texture fingerprint are quantified by comparing them, and texture difference information characterizing the texture differences and their uncertainties is output. Texture difference information may include: texture difference vectors or scores (such as similarity scores, distance metrics, local difference maps); difference distribution statistics (mean, variance, confidence interval); regional level difference labels (such as obvious abnormality / consistency of a certain section of road surface texture); explanatory descriptions (which texture features contribute the most, inference of potential causes); and suggestions for the claims process (whether on-site inspection is required, suggestions for evidence lists, etc.).
[0054] In step S52, the difference between the actual physical inclination angle of the target road surface and the known inclination angle of the reference road surface is quantified by comparing the two, and inclination angle difference information characterizing the inclination angle difference and its uncertainty is output. The inclination angle difference information may include: inclination angle difference value; a list of influencing factors (sensor error, environmental factors, changes in road surface condition, etc.); and suggestions for the claims process (such as the need for on-site inspection, whether there is a risk of misjudgment, and evidence list guidance).
[0055] In step S53, texture difference information and tilt angle difference information can be fused using a large language model to obtain claims risk information.
[0056] As can be seen from the above scheme, combining texture fingerprints and physical tilt angles to achieve risk identification improves the richness of claims risk information and reduces the manual input of claims investigation.
[0057] In one example, claims risk information includes a claims risk score. : .in, As the first weight, As the second weight, For the target road surface texture fingerprint, To reference the fingerprint pattern on the road surface, The target road surface physical inclination angle, This is for reference to the physical slope of the road surface. If the claim risk score is... If the temperature is too high, the GPS may show "asphalt expressway" but the fingerprint image shows "cement rural road," immediately triggering a risk warning.
[0058] In some embodiments of the present invention, after step S50 or before step S60, the intelligent claims method may further include: obtaining a gravity acceleration vector through a client during shooting, and generating a device physical attitude angle based on the gravity acceleration vector; obtaining the relative angle of the user holding the client through the client during shooting; adding the relative angle and the target road surface physical tilt angle to obtain the road surface physical attitude angle; matching the device physical attitude angle and the road surface physical attitude angle to obtain an attitude matching residual; and updating the claims risk information based on the attitude matching residual.
[0059] As can be seen from the above scheme, using the gravitational acceleration vector collected by the client (such as a mobile phone sensor) at the moment of shooting enriches the information on claims risk and further improves the security of insurance claims.
[0060] In some embodiments of the present invention, the step of matching the physical attitude angle of the equipment and the physical attitude angle of the road surface to obtain the attitude matching residual may include the following steps: performing a sine operation on the physical attitude angle of the equipment to obtain a first attitude mapping value; performing a sine operation on the physical attitude angle of the road surface to obtain a second attitude mapping value; and calculating the difference between the first attitude mapping value and the second attitude mapping value to obtain the attitude matching residual. For example, the attitude matching residual... : .in, The physical attitude angle of the device. The target road surface physical inclination angle, The relative angle between the user holding the device and the surrounding area. If the attitude matching residual... If the value is far above the threshold, it means that the user is "horizontally reproducing" an old photo with a sense of slope, or that they are in a repair shop on flat ground but have faked an accident scene on a slope.
[0061] S60: Processes insurance claims for users based on claim risk information and claim photos. It utilizes generated claim risk information and claim photos to assist in the approval and payment decisions for insurance claims.
[0062] In some embodiments of the present invention, the claims risk information includes a claims risk score and suggestions for the claims process; the steps for processing insurance claims for the user based on the claims risk information and the claims photo include the following steps: if the claims risk score is less than a preset risk threshold, the claims photo is determined to have passed security verification, and an insurance claims plan is generated for the user based on the claims photo and suggestions; if the claims risk score is greater than or equal to the preset risk threshold, the claims photo is determined to have failed security verification, and the self-service insurance claims service is interrupted.
[0063] The insurance claim process here can include: whether to proceed with standard claims processing, and whether a second review is required. Evidence chain index: the set of evidence corresponding to the risk information (photos, text descriptions, test reports, sensor data). Risk classification and action recommendations: for example, high risk requires on-site inspection, medium risk requires supplementary materials, and low risk allows direct release. Compensation recommendations and scope: preliminary compensation scope, deductible, list of pending claim materials, and timeline reminders.
[0064] As can be seen from the above scheme, the insurance risk score can determine whether to interrupt the self-service insurance claims service. When the self-service insurance claims service is not interrupted, an insurance claims plan is generated based on the suggested information and claims photos, which expands the input data and improves the accuracy of insurance claims plan generation, thereby improving the security of insurance claims.
[0065] As can be seen, in the above steps S10 to S60, for users who wish to apply for insurance claims through the self-service insurance claims service, the server no longer simply verifies whether the background in the claim photo is a road. Instead, it obtains the target road surface texture fingerprint through semantic segmentation and also obtains the target road surface physical inclination angle through slope calculation. Since the road surface physical inclination angle and road surface texture fingerprint are physical facts bound to geographical location, it is extremely difficult for fraudsters to simulate a completely identical road surface in a fake scene. Then, by using reference road information to perform risk identification on the target road surface texture fingerprint and target road surface physical inclination angle, it can be determined whether the road surface in the claim photo is real, thereby improving the security of self-service insurance claims.
[0066] This invention also has the following beneficial effects: It elevates the anti-counterfeiting measures for claims from the "pixel level" to the "geophysical level." It precisely combats staged photo fraud and insurance fraud by reproducing photos: Because slope α and road surface texture T are physical facts bound to geographical location, it is extremely difficult for fraudsters to simulate perfectly matching gravity components and road fractal features at a fake scene. It enhances the authority of automatic damage assessment: The generated claims risk information provides a second independent chain of evidence besides visual evidence, reducing the manual input of claims investigations. Finally, it provides multi-dimensional environmental evidence: Road surface fingerprints can serve as "fingerprint evidence" of the accident scene, possessing significant physical evidence value in legal proceedings or insurance fraud investigations.
[0067] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0068] In one embodiment, an intelligent claims processing device is provided, applied to a server. This intelligent claims processing device corresponds one-to-one with the intelligent claims processing methods described in the above embodiments. For example... Figure 6 As shown, the intelligent claims processing device includes: a claims triggering module 101, a semantic segmentation module 102, a slope calculation module 103, an information query module 104, a risk identification module 105, and an insurance claims module 106. Detailed descriptions of each functional module are as follows: The claims triggering module 101 is used to respond to the detection that a user has triggered the self-service insurance claims on the client, and to take pictures of the road surface at the scene of the vehicle accident through the client to obtain claims photos; Semantic segmentation module 102 is used to perform semantic segmentation on the claim photo to obtain the target road surface texture fingerprint; The slope calculation module 103 is used to calculate the slope based on the claim photos to obtain the physical inclination angle of the target road surface; The information query module 104 is used to obtain the target location information of the user through the client, and query the preset map database based on the target location information to obtain reference road surface information; Risk identification module 105 is used to identify risks based on the reference road surface information, the texture fingerprint of the target road surface, and the physical inclination angle of the target road surface to obtain claims risk information; The insurance claims module 106 is used to process insurance claims for the user based on the claim risk information and the claim photos.
[0069] In one embodiment, the reference road surface information includes a reference road surface texture fingerprint and a reference road surface physical tilt angle, and the risk identification module 105 is specifically used for: Texture comparison is performed between the target road surface texture fingerprint and the reference road surface texture fingerprint to obtain texture difference information; The inclination angle is compared with the physical inclination angle of the target road surface and the physical inclination angle of the reference road surface to obtain inclination angle difference information; The claim risk information is obtained by fusing the texture difference information and the tilt angle difference information.
[0070] In one embodiment, before processing the insurance claim for the user based on the claim risk information and the claim photo, the intelligent claims device further includes an information update module, used for: During shooting, the gravity acceleration vector is obtained through the client, and the physical attitude angle of the device is generated based on the gravity acceleration vector; During shooting, the relative angle at which the user holds the client is obtained through the client. The physical attitude angle of the road surface is obtained by adding the relative angle and the physical inclination angle of the target road surface. The attitude matching residual is obtained by matching the physical attitude angle of the equipment and the physical attitude angle of the road surface. The claims risk information is updated based on the attitude matching residual.
[0071] In one embodiment, the information update module is specifically used for: Perform a sine calculation on the physical attitude angle of the device to obtain the first attitude mapping value; The second attitude mapping value is obtained by performing a sine calculation on the physical attitude angle of the road surface. The attitude matching residual is obtained by calculating the difference between the first attitude mapping value and the second attitude mapping value.
[0072] In one embodiment, the semantic segmentation module 102 is specifically used for: The road surface is inspected using the claim photos to obtain the target road surface area; wherein, the target road surface area does not contain any vehicles; The target road surface region is analyzed using a first deep learning model to obtain the road surface material fingerprint. A second deep learning model is used to identify wear in the target road surface area to obtain a road surface wear fingerprint. The target road surface texture fingerprint is obtained by fusing the road surface material fingerprint and the road surface wear fingerprint.
[0073] In one embodiment, the slope calculation module 103 is specifically used for: Target detection is performed on the claimed photos to obtain the vehicle outline and the target road surface area; Based on the vehicle profile and the road surface area, an intersection analysis is performed to obtain the vanishing point of the contact line between the vehicle profile and the road surface. Generate a road surface parallel line that is parallel to the target road surface area; The physical inclination angle of the target road surface is obtained by inverse calculation based on the vanishing point and the road surface parallel line.
[0074] In one embodiment, the claims risk information includes a claims risk score and suggestions for the claims process; the insurance claims module 106 is specifically used for: If the claim risk score is less than the preset risk threshold, the claim photo is determined to have passed the security verification, and an insurance claim plan is generated for the user based on the claim photo and the suggestion information. If the claim risk score is greater than or equal to the preset risk threshold, the claim photo is determined to have failed security verification, and the self-service insurance claim service is interrupted.
[0075] This invention provides an intelligent claims processing device. For users seeking insurance claims through self-service, the server no longer simply verifies whether the background in the claim photo is a road. Instead, it obtains the target road surface texture fingerprint through semantic segmentation and calculates the target road surface's physical inclination angle through slope calculation. Since the road surface's physical inclination angle and texture fingerprint are physical facts bound to geographical location, it is extremely difficult for fraudsters to simulate a completely identical road surface in a fake scenario. Then, by using reference road information to perform risk identification on the target road surface texture fingerprint and physical inclination angle, the device can determine whether the road surface in the claim photo is real, thereby improving the security of self-service insurance claims.
[0076] For specific limitations regarding the intelligent claims processing device, please refer to the limitations of the intelligent claims processing method above, which will not be repeated here. Each module in the aforementioned intelligent claims processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.
[0077] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a smart claims processing server-side method.
[0078] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a smart claims processing method on the client side.
[0079] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: In response to detecting that a user has triggered a self-service insurance claim on the client, the system takes photos of the road surface at the accident scene through the client to obtain claim photos; Semantic segmentation is performed on the claim photos to obtain the target road surface texture fingerprint; Based on the claimed photos, the slope is calculated to obtain the physical inclination angle of the target road surface; The client obtains the target location information of the user and queries a preset map database based on the target location information to obtain reference road surface information. Based on the reference road surface information, risk identification is performed on the target road surface texture fingerprint and the target road surface physical inclination angle to obtain claim risk information; Based on the claim risk information and the claim photos, the insurance claim will be processed for the user.
[0080] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: In response to detecting that a user has triggered a self-service insurance claim on the client, the system takes photos of the road surface at the accident scene through the client to obtain claim photos; Semantic segmentation is performed on the claim photos to obtain the target road surface texture fingerprint; Based on the claimed photos, the slope is calculated to obtain the physical inclination angle of the target road surface; The client obtains the target location information of the user and queries a preset map database based on the target location information to obtain reference road surface information. Based on the reference road surface information, risk identification is performed on the target road surface texture fingerprint and the target road surface physical inclination angle to obtain claim risk information; Based on the claim risk information and the claim photos, the insurance claim will be processed for the user.
[0081] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.
[0082] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0083] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.
[0084] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.
[0085] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. An intelligent claims settlement method, characterized in that, Applied to the server side, the method includes: In response to detecting that a user has triggered a self-service insurance claim on the client, the system takes photos of the road surface at the accident scene through the client to obtain claim photos; Semantic segmentation is performed on the claim photos to obtain the target road surface texture fingerprint; Based on the claimed photos, the slope is calculated to obtain the physical inclination angle of the target road surface; The client obtains the target location information of the user and queries a preset map database based on the target location information to obtain reference road surface information. Based on the reference road surface information, risk identification is performed on the target road surface texture fingerprint and the target road surface physical inclination angle to obtain claim risk information; Based on the claim risk information and the claim photos, the insurance claim will be processed for the user.
2. The method according to claim 1, characterized in that, The reference road surface information includes a reference road surface texture fingerprint and a reference road surface physical inclination angle. The step of identifying risks based on the reference road surface information, specifically the target road surface texture fingerprint and the target road surface physical inclination angle, to obtain claims risk information includes: Texture comparison is performed between the target road surface texture fingerprint and the reference road surface texture fingerprint to obtain texture difference information; The inclination angle is compared with the physical inclination angle of the target road surface and the physical inclination angle of the reference road surface to obtain inclination angle difference information; The claim risk information is obtained by fusing the texture difference information and the tilt angle difference information.
3. The method according to claim 1, characterized in that, Before processing the insurance claim for the user based on the claim risk information and the claim photo, the method further includes: During shooting, the gravity acceleration vector is obtained through the client, and the physical attitude angle of the device is generated based on the gravity acceleration vector; During shooting, the relative angle at which the user holds the client is obtained through the client. The physical attitude angle of the road surface is obtained by adding the relative angle and the physical inclination angle of the target road surface. The attitude matching residual is obtained by matching the physical attitude angle of the equipment and the physical attitude angle of the road surface. The claims risk information is updated based on the attitude matching residual.
4. The method according to claim 3, characterized in that, The step of matching the physical attitude angle of the device and the physical attitude angle of the road surface to obtain the attitude matching residual includes: Perform a sine calculation on the physical attitude angle of the device to obtain the first attitude mapping value; The second attitude mapping value is obtained by performing a sine calculation on the physical attitude angle of the road surface. The attitude matching residual is obtained by calculating the difference between the first attitude mapping value and the second attitude mapping value.
5. The method according to any one of claims 1 to 4, characterized in that, The semantic segmentation of the claim photo to obtain the target road surface texture fingerprint includes: The road surface is inspected using the claim photos to obtain the target road surface area; wherein, the target road surface area does not contain any vehicles; The target road surface region is analyzed using a first deep learning model to obtain the road surface material fingerprint. A second deep learning model is used to identify wear in the target road surface area to obtain a road surface wear fingerprint. The target road surface texture fingerprint is obtained by fusing the road surface material fingerprint and the road surface wear fingerprint.
6. The method according to any one of claims 1 to 4, characterized in that, The step of calculating the slope based on the claim photos to obtain the physical inclination angle of the target road surface includes: Target detection is performed on the claimed photos to obtain the vehicle outline and the target road surface area; Based on the vehicle profile and the road surface area, an intersection analysis is performed to obtain the vanishing point of the contact line between the vehicle profile and the road surface. Generate a road surface parallel line that is parallel to the target road surface area; The physical inclination angle of the target road surface is obtained by inverse calculation based on the vanishing point and the road surface parallel line.
7. The method according to any one of claims 1 to 4, characterized in that, The claims risk information includes a claims risk score and suggestions for the claims process; the process of processing insurance claims for the user based on the claims risk information and the claims photos includes: If the claim risk score is less than the preset risk threshold, the claim photo is determined to have passed the security verification, and an insurance claim plan is generated for the user based on the claim photo and the suggestion information. If the claim risk score is greater than or equal to the preset risk threshold, the claim photo is determined to have failed security verification, and the self-service insurance claim service is interrupted.
8. An intelligent claims processing device, characterized in that, Applied to the server side, the device includes: The claims trigger module is used to respond to the detection that a user has triggered the self-service insurance claims on the client, and to take pictures of the road surface at the scene of the vehicle accident through the client to obtain claims photos; The semantic segmentation module is used to perform semantic segmentation on the claim photo to obtain the target road surface texture fingerprint; The slope calculation module is used to calculate the slope based on the claim photos to obtain the physical inclination angle of the target road surface; The information query module is used to obtain the target location information of the user through the client, and query the preset map database based on the target location information to obtain reference road surface information; The risk identification module is used to identify risks based on the reference road surface information, the texture fingerprint of the target road surface, and the physical inclination angle of the target road surface, to obtain claims risk information. The insurance claims module is used to process insurance claims for the user based on the claim risk information and the claim photos.
9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.