Freight claims processing method, device, system and storage medium

Abstract

The present invention relates to the field of logistics management technology, and discloses a freight claim processing method, device, system, and storage medium for reducing labor costs, ensuring the authenticity of claim certification materials, and automatically determining damages through a cargo damage assessment model, thereby avoiding unreasonable damage assessments caused by subjective personnel, shortening freight claim processing time, and improving freight claim processing efficiency. The freight claim processing method includes: obtaining a claim request, and retrieving insurance policy information and freight node information based on the claim request; when the claim type is cargo loss, verifying the claim request based on the insurance policy information and freight node information to obtain a verification result; when the verification result is passed, inputting real-time video data, cargo video data, and target freight order information into a preset cargo damage assessment model to obtain a cargo loss quantity and cargo loss degree information layer; generating a claim result based on the cargo loss quantity, cargo loss degree information, target freight order information, and insurance policy information.

Description

Technical Field

[0001] The present invention relates to the field of logistics management technology, and in particular to a freight claims processing method, device, system and storage medium. Background Art

[0002] In the logistics and freight industry, logistics companies can take out insurance from insurance companies. During the storage, loading and unloading, processing, distribution, transshipment, cross-border, air and sea transportation and other operations, logistics companies can ask insurance companies to make claims for damage to goods, loss of goods, delayed delivery time, etc. due to accidents, so as to transfer the risks of logistics companies and protect their rights and interests. Compared with other types of insurance, the subject matter of freight insurance is liquid and the risks it faces are relatively complex.

[0003] In the traditional freight claims process, logistics companies need to sign a cooperation agreement with insurance companies for each logistics order. When goods are damaged or lost, logistics company employees report to the insurance company. The insurance company uses a combination of online and offline methods to handle claims. First, a claims specialist is dispatched to conduct on-site inspections of the damage and loss of goods, and to collect claim proof materials, such as on-site photos, responsibility and damage assessment materials from relevant departments, proof of repair costs, relevant documents of freight orders, etc., and upload them to internal auditors for review, arbitration, damage assessment, and claims.

[0004] Most cargo transportation accidents occur in other places or even overseas. It is difficult for insurance companies to grasp the specific dynamic process, resulting in frequent fraud. As the business volume increases, the method of sending claims specialists to conduct offline surveys and collect claims proof materials increases the personnel costs of insurance companies and logistics companies. Auditors may cause unreasonable arbitration and loss assessment due to company interests, personal subjectivity, insufficient training, etc. At the same time, due to the complicated claims process and long processing time, customer complaints occur. Traditional freight claims methods cannot meet the growing demand for fast and accurate claims for freight orders. Summary of the Invention

[0005] The present invention provides a freight claims processing method, device, system and storage medium, which are used to solve the problems of high labor costs, high fraud risks, subjective and unreasonable loss assessment process of insurance companies, and complicated and time-consuming claims process in traditional freight claims processing.

[0006] The first aspect of the present invention provides a freight claim processing method, including: obtaining a claim request, and retrieving policy information and freight node information according to the claim request, the claim request including target freight order information and claim information, and the claim information including at least the claim type and claim proof information; when the claim type is cargo loss, verifying the claim request according to the policy information and freight node information to obtain a verification result, the claim proof information including at least real-time video data, and the freight node information including cargo video data of the corresponding freight node; when the verification result is passed, inputting the real-time video data, cargo video data and target freight order information into a preset cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information, the cargo damage assessment model including an input layer, a feature extraction layer, a feature pyramid layer and an output layer; generating a claim result according to the cargo loss quantity, cargo loss degree information, the target freight order information and the policy information.

[0007] In a feasible implementation, when the verification result is passed, the real-time video data, cargo video data and target freight order information are input into a preset cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information, including: when the verification result is passed, the real-time video data and cargo video data are preprocessed by the preset cargo damage assessment model to obtain multiple claim key frame images and multiple standard key frame images; the multiple claim key frame images and multiple standard key frame images are feature extracted by the cargo damage assessment model to obtain a target feature map, and the target feature map marks the corresponding cargo loss area; the target feature map and the target freight order information are processed by the cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information.

[0008] In a feasible embodiment, when the verification result is passed, the real-time video data and the cargo video data are preprocessed by a preset cargo damage assessment model to obtain multiple claim key frame images and multiple standard key frame images, including: extracting frames from the cargo video data at multiple angles preset by the cargo damage assessment model to obtain multiple standard key frame images, and the multiple standard key frame images are used to indicate the cargo conditions at different angles in the corresponding freight node; extracting features from the multiple standard key frame images to obtain cargo feature information of each standard key frame image; extracting frames from the real-time video data according to the cargo feature information and multiple angles to obtain multiple claim key frame images.

[0009] In a feasible implementation, feature extraction is performed on multiple claim key frame images and multiple standard key frame images through a cargo damage assessment model to obtain a target feature map, including: performing key point matching on each claim key frame image and the corresponding standard key frame image, and calculating the corresponding homography matrix; mapping each standard key frame image to the corresponding claim key frame image according to the corresponding homography matrix; outputting multiple difference points according to a preset difference threshold and clustering them to obtain a damage feature map; processing the damage feature map of each claim key frame image according to a preset scale gradient to obtain damage feature maps of multiple scales corresponding to each claim key frame image; fusing the damage feature maps of multiple scales corresponding to each claim key frame image to output a target feature map.

[0010] In a feasible implementation, key point matching is performed on each claim key frame image and the corresponding standard key frame image, and the corresponding homography matrix is ​​calculated, including: extracting the key points of each claim key frame image and the corresponding standard key frame image, and calculating the descriptor corresponding to each key point through a description algorithm; performing key point matching on each claim key frame image and the corresponding standard key frame image according to the matching algorithm to obtain multiple paired key points; and calculating the multiple paired key points to obtain the corresponding homography matrix.

[0011] In a feasible implementation, when the claim type is cargo loss, the claim request is verified according to the insurance policy information and the freight node information to obtain a verification result, and the claim proof materials include at least real-time video data, and the freight node information includes cargo video data of the corresponding freight node, including: when the claim type is cargo loss, the target freight order information and the claim type are verified according to the insurance policy information to obtain a first verification result; the shooting timestamp and shooting location information corresponding to the real-time video data are parsed, and verified according to the freight node information to obtain a second verification result; the cargo feature information of the real-time video data and the cargo video data is extracted, and the cargo feature information is compared to obtain a third verification result; the first verification result, the second verification result and the third verification result are combined into a verification result.

[0012] In a feasible implementation, before obtaining a claim request, it also includes: retrieving target freight order information and setting the claim type corresponding to the target freight order information; when the claim type is cargo loss, determining the claim freight node and obtaining the corresponding cargo video data; inputting the cargo video data into a preset shooting guidance model to generate shooting guidance information; generating real-time video data according to the shooting guidance information, shooting time and shooting location, the real-time video data including the shooting timestamp and shooting location information; generating a claim request according to the real-time video data, the target freight order information and the claim type.

[0013] The second aspect of the present invention provides a freight claim processing device, including: an acquisition module, used to obtain a claim request, and retrieve policy information and freight node information according to the claim request, the claim request includes target freight order information and claim information, and the claim information includes at least claim type and claim proof information; a verification module, used to verify the claim request according to the policy information and freight node information when the claim type is cargo loss, and obtain a verification result, the claim proof information includes at least real-time video data, and the freight node information includes cargo video data of the corresponding freight node; a damage assessment module, used to input the real-time video data, cargo video data and target freight order information into a preset cargo damage assessment model when the verification result is passed, to obtain the cargo loss quantity and cargo loss degree information, the cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer and an output layer; a claim module, used to generate a claim result based on the cargo loss quantity, cargo loss degree information, target freight order information and policy information.

[0014] In a feasible embodiment, the damage assessment module includes: a preprocessing unit, which is used to preprocess the real-time video data and cargo video data through a preset cargo damage assessment model when the verification result is passed, to obtain multiple claim key frame images and multiple standard key frame images; a feature extraction unit, which is used to extract features from multiple claim key frame images and multiple standard key frame images through the cargo damage assessment model to obtain a target feature map, and the target feature map marks the corresponding cargo loss area; an estimation unit is used to process the target feature map and the target freight order information through the cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information.

[0015] In a feasible embodiment, the preprocessing unit is specifically used to: extract frames from the cargo video data through multiple angles preset by the cargo damage assessment model to obtain multiple standard key frame images, and the multiple standard key frame images are used to indicate the cargo conditions at different angles in the corresponding freight node; extract features from the multiple standard key frame images to obtain cargo feature information of each standard key frame image; extract frames from the real-time video data according to the cargo feature information and multiple angles to obtain multiple claims key frame images.

[0016] In a feasible embodiment, the feature extraction unit includes: a processing subunit, which is used to perform key point matching on each claim key frame image and the corresponding standard key frame image respectively, and calculate the corresponding homography matrix; a mapping subunit, which is used to map each standard key frame image to the corresponding claim key frame image according to the corresponding homography matrix; a feature extraction subunit, which is used to output multiple difference points according to a preset difference threshold and cluster them to obtain a damage feature map; a pyramid subunit, which is used to process the damage feature map of each claim key frame image according to a preset scale gradient to obtain damage feature maps of multiple scales corresponding to each claim key frame image; and a fusion subunit, which is used to fuse the damage feature maps of multiple scales corresponding to each claim key frame image and output a target feature map.

[0017] In a feasible implementation, the processing subunit is specifically used to: extract the key points of each claim key frame image and the corresponding standard key frame image, and calculate the descriptor corresponding to each key point through a description algorithm; perform key point matching on each claim key frame image and the corresponding standard key frame image according to a matching algorithm to obtain multiple paired key points; calculate the multiple paired key points to obtain the corresponding homography matrix.

[0018] In a feasible implementation, the verification module is specifically used to: when the claim type is cargo loss, verify the target freight order information and the claim type according to the insurance policy information to obtain a first verification result; parse the shooting timestamp and shooting location information corresponding to the real-time video data, and verify according to the freight node information to obtain a second verification result; extract cargo feature information from the real-time video data and the cargo video data, and compare the cargo feature information to obtain a third verification result; combine the first verification result, the second verification result, and the third verification result into a verification result.

[0019] In a feasible embodiment, a freight claim processing device also includes: a generation module, used to retrieve target freight order information and set the claim type corresponding to the target freight order information; when the claim type is cargo loss, determine the claim freight node and obtain the corresponding cargo video data; input the cargo video data into a preset shooting guidance model to generate shooting guidance information; generate real-time video data according to the shooting guidance information, shooting time and shooting location, the real-time video data including the shooting timestamp and shooting location information; generate a claim request according to the real-time video data, the target freight order information and the claim type.

[0020] The third aspect of the present invention provides a freight claims processing system, which is used to execute the above-mentioned freight claims processing method. The freight claims processing system includes: a client platform, which is used to generate an insurance request based on a freight order and send it to the insurance platform, and generate a claim request and send it to the arbitration platform; the insurance platform is used to review the insurance request and generate policy information based on the insurance request, logistics company information, historical claims information and preset insurance rules; the arbitration platform is used to review, determine the damage and make claims based on the claim request and policy information.

[0021] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein, which, when executed on a computer, causes the computer to execute the above-mentioned freight claims processing method.

[0022] In the technical solution provided by the present invention, a claim request is obtained, and the insurance policy information and freight node information are retrieved according to the claim request, the claim request includes the target freight order information and the claim information, and the claim information includes at least the claim type and claim proof information; when the claim type is cargo loss, the claim request is verified according to the insurance policy information and the freight node information to obtain a verification result, the claim proof information includes at least real-time video data, and the freight node information includes the cargo video data of the corresponding freight node; when the verification result is passed, the real-time video data, the cargo video data and the target freight order information are input into a preset cargo damage assessment model to obtain the cargo loss quantity and cargo loss degree information, the cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer and an output layer; a claim result is generated according to the cargo loss quantity, cargo loss degree information, the target freight order information and the insurance policy information. In the embodiments of the present invention, the labor costs of insurance companies in the freight claims process are reduced, the authenticity of claim certification materials is guaranteed, and damage is automatically determined through a cargo damage assessment model, thereby avoiding unreasonable damage assessments caused by subjective personnel, simplifying the freight claims processing process, shortening the freight claims processing time, and improving the efficiency of freight claims processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of an embodiment of a method for handling freight claims in an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of another embodiment of a method for handling freight claims in an embodiment of the present invention;

[0025] Figure 3 Schematic diagram of an embodiment of a freight claims processing device in an embodiment of the present invention;

[0026] Figure 4 Schematic diagram of another embodiment of a freight claims processing device according to an embodiment of the present invention;

[0027] Figure 5 This is a schematic diagram of an embodiment of a freight claims processing system according to an embodiment of the present invention;

[0028] Figure 6 FIG. 1 is a schematic diagram of an embodiment of an arbitration platform in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The present invention provides a freight claims processing method, device, system and storage medium for reducing the labor costs of insurance companies in the freight claims process, ensuring the authenticity of claim certification materials, automatically determining damages through a cargo damage assessment model, avoiding unreasonable damage assessments caused by subjective damage assessments by personnel, simplifying the freight claims processing process, and shortening the freight claims processing time.

[0030] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or system that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or systems.

[0031] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of a method for handling freight claims in an embodiment of the present invention includes:

[0032] 101. Obtain a claim request and retrieve insurance policy information and freight node information based on the claim request. The claim request includes target freight order information and claim information. The claim information includes at least the claim type and claim certification information.

[0033] It is understandable that the execution subject of the present invention may be a freight claim processing device or a freight claim processing system, which is not limited herein. The embodiment of the present invention is described by taking the freight claim processing system as the execution subject.

[0034] The freight claims processing system includes a client platform, which is used to generate insurance requests based on freight orders and send them to the insurance platform, as well as submit claim requests to the arbitration platform; the insurance platform is used to obtain the insurance requests submitted by the client platform and the historical claim information of the arbitration platform, and review and generate policy information based on the insurance requests, historical claim information, logistics company information and preset insurance rules; the arbitration platform is used to obtain the claim requests submitted by the client platform, and retrieve the policy information generated by the insurance platform, as well as the freight order information uploaded at the time of insurance, and the freight node information during the freight process, to conduct freight claim review, damage assessment and claim settlement. By using the arbitration platform to perform the review, damage assessment and claim settlement work on behalf of the insurance company, it can avoid unreasonable damage assessment caused by subjective damage assessment by personnel and reduce the labor cost of the insurance company in the freight claims process.

[0035] In this embodiment, a claim request includes target freight order information, claim information, and user information. The claim information includes the claim type and claim verification information, which is used by the arbitration platform to enter the claim process. Claim types include cargo loss (including cargo damage and cargo loss), delivery delay, and return. The arbitration platform can perform corresponding claim procedures based on different claim types to meet the needs of various freight claims. Claim verification information can include real-time video data, liability and damage assessment information from relevant departments, repair cost verification information, and relevant documents of the freight order. Claim verification information can be uploaded by the client platform without waiting for the claims specialist to conduct an on-site inspection. It can be uploaded independently, reducing the insurance company's labor costs and improving the efficiency of freight claims processing. Among them, claims types such as delivery delay and return can complete the entire freight process traceability through blockchain, and their claims process is relatively simple. Cargo damage and cargo loss are difficult in actual freight claims processing. This embodiment mainly explains cargo loss.

[0036] The data exchange process between the above platforms can be implemented through a data interface between the platforms or based on blockchain technology. Furthermore, the claim request can also include a data interface or access key. After the arbitration platform obtains the claim request, it extracts the insurance policy information and freight node information corresponding to the claim request based on the data interface or access key to implement data access control between different platforms. Among them, the insurance policy information includes the corresponding insured freight order information, insurance coverage, and corresponding claims rules; the freight node information can include cargo video data of each node process, time information of each freight node, document information of each freight node, freight order number, cargo order information, transportation tool information, etc. Among them, the cargo video data refers to the process video recording of the operation of the cargo order at each freight node, such as cargo outbound video, cargo inbound video, cargo loading and unloading video, and cargo processing video; the document information of each freight node can include transportation contracts, delivery notes, bills of lading, various transaction vouchers, etc.; the cargo order information can include information such as cargo type, quantity, and amount, so as to facilitate the traceability of each freight node, provide standard information for the loss assessment process, and arbitration and liability determination in the freight claims process.

[0037] 102. When the claim type is cargo loss, the claim request is verified based on the policy information and freight node information to obtain a verification result. The claim proof materials include at least real-time video data, and the freight node information includes the cargo video data of the corresponding freight node.

[0038] The freight claims processing system verifies the claim request and first verifies the user information of the claim request to verify whether the user is a whitelist user. The whitelist user can be set by the insurance platform based on historical claims information and logistics company information to indicate that the user is considered a trustworthy user by the insurance company and can execute freight claims processing through the arbitration platform; for claim requests that pass the user verification, the target freight order information and claim type are verified according to the policy information to determine whether the corresponding claim fields such as the freight order number and the selected claim type are within the claim scope of the policy. The arbitration platform rejects claim requests that are not within the claim scope and verifies the authenticity of the claim proof materials for claim requests that pass the claim scope verification.

[0039] When the claim type is cargo loss, the authenticity verification of the claim proof materials may include: parsing the real-time video data of the claim proof materials to obtain the corresponding shooting timestamp and shooting location information; verifying the shooting timestamp and shooting location information of the real-time video data according to the freight node information to determine that the shooting time of the real-time video data is within the preset time range and the shooting location is within the preset location range; obtaining the claim key frame image of the real-time video data and the standard key frame image of the cargo video data, and comparing the cargo feature information in the claim key frame image and the standard key frame image to determine whether they are the same batch of cargo.

[0040] The above-mentioned cargo characteristic information includes special markers on the cargo, the container number on the side wall of the container, cargo identity information, cargo packaging information, corner points of each cargo and cargo accessories, among which special markers include anti-pressure, moisture-proof and fragile signs. Cargo information may include steel stamps, nameplates, trademarks, corresponding text descriptions (such as freight order details), digital descriptions (such as freight bill numbers) and other cargo information. Cargo packaging information may include packaging material, size and shape, etc. Cargo accessories may include binding straps, padding and wedges, etc. Cargo characteristic information may also be other characteristic information that can identify the cargo.

[0041] In this embodiment, the real-time video data and cargo video data are composed of multiple frames of image data containing time series information, which are more difficult to forge than a single image. The video data can also further reflect the correlation between the cargo and the freight node; and by marking the timestamp and location information and comparing the cargo feature information, it can further avoid the use of historical claims proof materials and other means to commit insurance fraud. The real-time video data and cargo video data can be framed using preset parameters to extract key frame images with better shooting quality, which reduces the requirements for the photographer, facilitates the selection of claim key frame images and corresponding standard key frame images, reduces the computer resources for pre-processing, and improves the accuracy of damage assessment.

[0042] It should be further explained that the claim proof materials may include the relevant departments' liability and damage assessment materials, maintenance cost proof materials, and relevant documents of freight orders. They can be uploaded by the customer through the client platform and verified by checking the freight order number, signature information of the relevant departments / institutions, etc. on the claim proof materials. Blockchain nodes or data interfaces of the relevant departments and maintenance institutions can also be established separately to ensure the authenticity and reliability of the claim proof materials through direct acquisition. The details will not be repeated here.

[0043] 103. When the verification result is passed, the real-time video data, the cargo video data and the target freight order information are input into the preset cargo damage assessment model to obtain the cargo loss quantity and cargo loss degree information.

[0044] When the user information, claim scope, and claim supporting materials of the claim request are verified, the freight claims processing system inputs the real-time video data, cargo video data, and target freight order information into the preset cargo damage assessment model, and automatically performs a damage assessment through the cargo damage assessment model to obtain information on the quantity and degree of cargo loss.

[0045] The cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer, and an output layer, wherein the output layer includes a fully connected layer, a regression layer, and a classification layer.

[0046] The freight claims processing system inputs real-time video data and cargo video data into the input layer of the cargo damage assessment model. This preprocesses the real-time video data and cargo video data through the input layer, generating multiple claim keyframe images and multiple standard keyframe images. The feature extraction layer uses these multiple standard keyframe images as a comparison benchmark to extract a damage feature map for each claim keyframe image. The feature pyramid layer performs residual connections and feature fusion on the damage feature maps of each claim keyframe image based on a preset scale gradient, generating multi-scale damage feature maps. A fully connected layer fuses the multi-scale damage feature maps of each claim keyframe image to output a target feature map, which labels the corresponding cargo loss area. The target freight order information and target feature map are then input into the regression and classification layers for processing, generating information on the cargo loss quantity and extent. The cargo damage assessment model automatically and rapidly generates high-quality cargo images (keyframe images) that meet the requirements of damage assessment processing. These keyframe images include both claim keyframe images and standard keyframe images. The model automatically extracts cargo damage areas and estimates the cargo loss quantity and extent, improving the efficiency and accuracy of damage assessment.

[0047] The above-mentioned claim key frame image refers to the cargo image extracted from real-time video data for use in damage assessment when cargo loss occurs, and the standard key frame image refers to the cargo image extracted from cargo video data for use as a comparison standard; the cargo video data can be the cargo video data of the previous freight node, or it can be cargo video data corresponding to different types of cargo.

[0048] It should be further explained that the cargo damage assessment model training process of this embodiment can be carried out according to actual conditions. The output layer of the initial damage assessment model can be trained with video data before and after different types of cargo losses to improve the accuracy and efficiency of the output layer in matching, extracting, and processing key frame images. Then, the key frame images before and after different types of cargo losses are used for training. The output target feature map, cargo loss quantity, and cargo loss degree information are compared with the target feature map, actual cargo loss quantity, and actual cargo loss degree information of the corresponding manually corrected cargo loss area. The loss function composed of the above three parts is calculated. When the loss function converges, the parameters of the initial damage assessment model are updated and optimized by the back propagation method to obtain the cargo damage assessment model. The specific process is not limited in this embodiment.

[0049] 104. Generate a claim settlement result based on the cargo loss quantity, cargo loss extent information, target freight order information and insurance policy information.

[0050] The freight claims processing system parses the insurance policy information and obtains multiple loss score ranges; calculates the cargo loss score based on the cargo loss quantity, cargo loss degree information and the preset cargo loss score function; determines the target loss score range based on the cargo loss score, and determines the claim amount based on the claim amount base, claim coefficient and claim amount upper limit of the target loss score range; makes claim allocations based on the claim amount, generates claim results, writes them into the blockchain, and broadcasts them.

[0051] The formula for the cargo loss score function is as follows:

[0052]

[0053] Where S represents the cargo damage assessment score; N represents the number of cargo types in the target freight order information; i represents the i-th cargo; w i S represents the impact weight of the i-th type of goods on the damage assessment score; i The damage assessment score of the i-th type of goods can be calculated using the following formula:

[0054] S i =λ×Q i +(1-λ)×D i

[0055] Where λ is a specified constant used to balance Q i With D i The relative size of Q i It represents the loss quantity score of the i-th type of goods. It can calculate the loss ratio between the quantity of the i-th type of goods and the loss quantity of the i-th type of goods, and determine Q according to the loss ratio and the preset loss ratio interval. i , the greater the loss ratio Qi The higher the score, the better; i It represents the loss degree score of the i-th type of goods. Multiple loss degree levels can be set according to the i-th type of goods. D is determined based on the loss degree information of the i-th type of goods and multiple loss degree levels. i , the higher the loss level, the i The higher the score.

[0056] In the embodiment of the present invention, the labor cost of the insurance company in the freight claims process is reduced, the authenticity of the claim certification materials is guaranteed, and the damage is automatically determined through the cargo damage assessment model, thereby avoiding unreasonable damage assessment caused by subjective damage assessment by personnel, simplifying the freight claims processing process, shortening the freight claims processing time, and improving the freight claims processing efficiency.

[0057] See also Figure 2 Another embodiment of the freight claim processing method in the embodiment of the present invention includes:

[0058] 201. Obtain a claim request, and retrieve insurance policy information and freight node information according to the claim request. The claim request includes target freight order information and claim information. The claim information includes at least claim type and claim certification information.

[0059] Before obtaining a claim request, the freight claims processing system also includes: retrieving target freight order information and setting the claim type corresponding to the target freight order information; when the claim type is cargo loss, determining the claim freight node and obtaining the corresponding cargo video data; inputting the cargo video data into a preset shooting guidance model to generate shooting guidance information; generating real-time video data based on the shooting guidance information, shooting time, and shooting location, the real-time video data including the shooting timestamp and shooting location information; and generating a claim request based on the real-time video data, the target freight order information, and the claim type. Generating shooting guidance information from cargo video data allows ordinary users to complete shooting that meets the requirements of damage assessment processing, improving the user's interactive experience in damage assessment. Moreover, since the shooting guidance information is generated based on cargo video data, the resulting real-time video data has a higher degree of match with the cargo video data used for comparison compared to directly setting the shooting angle.

[0060] The freight claims processing system obtains a claim request and retrieves the insurance policy information and freight node information based on the claim request. The claim request includes the target freight order information and claim information. The claim information includes at least the claim type and claim proof materials. The claim proof materials include at least real-time video data. The freight node information includes the cargo video data of the corresponding freight node.

[0061] 202. When the claim type is cargo loss, the claim request is verified according to the insurance policy information and the freight node information to obtain the verification result.

[0062] When the claim type is cargo loss, the freight claims processing system verifies the target freight order information and claim type according to the insurance policy information to obtain a first verification result; parses the shooting timestamp and shooting location information corresponding to the real-time video data, and verifies it according to the freight node information to obtain a second verification result; extracts cargo feature information from the real-time video data and the cargo video data, and compares the cargo feature information to obtain a third verification result; and combines the first verification result, the second verification result, and the third verification result into a verification result.

[0063] 203. When the verification result is passed, the real-time video data and the cargo video data are preprocessed by a preset cargo damage assessment model to obtain multiple claim key frame images and multiple standard key frame images.

[0064] The cargo damage assessment model consists of an input layer, a feature extraction layer, a feature pyramid layer, and an output layer. The output layer includes a fully connected layer, a regression layer, and a classification layer. The freight claims processing system inputs real-time video data and cargo video data into the input layer of the cargo damage assessment model. This layer pre-processes the real-time video data and cargo video data to generate multiple claim keyframe images and multiple standard keyframe images.

[0065] Specifically, the freight claims processing system extracts frames from the cargo video data through multiple angles preset by the input layer to obtain multiple standard key frame images. The multiple standard key frame images are used to indicate the cargo conditions at different angles in the corresponding freight node. The multiple standard key frame images may include the front view, left view, and right view of the cargo, and may also include cargo images at other angles to improve the accuracy of damage assessment; feature extraction is performed on the multiple standard key frame images to obtain cargo feature information of each standard key frame image; the real-time video data is extracted according to the cargo feature information and multiple angles to obtain multiple claim key frame images, and each claim key frame image corresponds to a standard key frame image.

[0066] Furthermore, the display effect of the standard key frame image corresponding to each claim key frame image is improved through one or more methods of grayscale, geometric transformation and image enhancement, which solves the problem of low accuracy of feature extraction of claim key frame images through standard key frame images in the later stage due to differences in shooting angle, shooting distance, shooting light and shooting equipment, and improves the matching degree between claim key frame images and standard key frame images.

[0067] 204. Feature extraction is performed on multiple claim key frame images and multiple standard key frame images through the cargo damage assessment model to obtain a target feature map.

[0068] The freight claims processing system uses a feature extraction layer to match keypoints between each claim keyframe image and its corresponding standard keyframe image, and calculates the corresponding homography matrix. Each standard keyframe image is mapped to its corresponding claim keyframe image based on the corresponding homography matrix. Keypoints with a difference threshold are marked as difference points, resulting in a damage feature map that annotates multiple difference points with damage labels. The damage feature map for each claim keyframe image is then residually connected and fused using the scale gradients preset in the feature pyramid layer to generate damage feature maps at multiple scales corresponding to each claim keyframe image. The fully connected layer fuses the damage feature maps at multiple scales corresponding to each claim keyframe image, producing a target feature map that annotates the corresponding cargo loss area. By fusing multi-scale damage feature maps, the accuracy of damage assessment can be further improved.

[0069] In a feasible implementation, key point matching is performed on each claim key frame image and the corresponding standard key frame image respectively through the feature extraction layer, and the corresponding homography matrix is ​​calculated, including: extracting key points of each claim key frame image and the corresponding standard key frame image respectively through the feature extraction layer, and calculating the descriptor corresponding to each key point through the description algorithm; key point matching is performed on each claim key frame image and the corresponding standard key frame image through the matching algorithm to obtain multiple paired key points, and the homography matrix is ​​calculated through the paired key points to determine the transformation relationship between the images.

[0070] In a feasible implementation, multiple difference points are output according to a preset difference threshold and clustered to obtain a corresponding cargo loss area, including: calculating the difference between the key points before and after mapping in each claim key frame image according to the descriptor; marking the difference points in each claim key frame image according to the difference and the preset difference threshold to obtain multiple difference points; clustering the multiple difference points according to the descriptor of each difference point to obtain the cargo loss area.

[0071] In this embodiment, key points are used to indicate key information of the cargo image, which may be corner points, edge points, and intersection points of the cargo in the cargo image. Descriptors are used to indicate the characteristic representation of each key point, such as the position, shape, color, size, depth, and semantic label of the key point. The characteristic information of the cargo can be obtained by clustering the key points and their descriptors. The difference threshold can be set according to actual conditions or by using a statistical distribution graph of similarity to determine whether the key points before and after mapping are similar within an error range, that is, the similarity of the key points before and after mapping. Points that are not within the difference threshold range can be marked as difference points. The difference between the key points before and after mapping can be comprehensively considered by comparing factors such as color difference, shape difference, and position difference of the key points and their descriptors. For color difference, the color difference can be measured by calculating the color space distance of the key points before and after mapping, while the shape difference can be evaluated by calculating the geometric features of the key points before and after mapping, such as contour lines and curvature.

[0072] The description algorithm described above can be the Scale-Invariant Feature Transform (SIFT) algorithm, the Speeded-Up Robust Features (SURF) description algorithm, or the Oriented Fast and Rotated BRIEF (ORB) algorithm, used to generate the descriptor corresponding to each key point. Matching algorithms include the Brute-Force algorithm, the Fast Library for Approximate Nearest Neighbors (FLANN) matching algorithm, the Random Sample Consensus (RANSAC) matching algorithm, the Hough transform algorithm, etc., used to match the key points corresponding to the same descriptor in each claim keyframe image and the corresponding standard keyframe image. The selection of the description algorithm, the matching algorithm, the training process of the cargo loss assessment model, and the process of backpropagation optimization of the model parameters based on the loss function can be configured according to actual conditions and are not specifically limited in this embodiment.

[0073] 205. The target feature map and target freight order information are processed by the cargo loss assessment model to obtain cargo loss quantity and cargo loss degree information.

[0074] The freight claims processing system regresses the target feature map and target freight order information through the regression layer to obtain the quantity of cargo loss; and classifies the target feature map and target freight order information through the classification layer to obtain the degree of cargo loss information.

[0075] For the above regression, regression algorithms such as linear regression, polynomial regression, and random forest regression may be used; for the above classification, classification algorithms such as logistic regression, support vector machine classification (SVC), decision tree classification, and neural network classification may be used.

[0076] 206. Generate a claim settlement result based on the cargo loss quantity, cargo loss extent information, target freight order information, and insurance policy information.

[0077] Step 206 is similar to the above-mentioned step 104 and can be executed with reference to step 104, which will not be repeated here.

[0078] In the embodiment of the present invention, the labor cost of the insurance company in the freight claims process is reduced, the authenticity of the claim certification materials is guaranteed, high-quality cargo images that meet the damage assessment processing requirements are extracted through a preset cargo damage assessment model, and a target feature map that marks the damaged area of ​​the cargo is automatically extracted. By estimating the target feature map, information on the quantity and degree of cargo loss is obtained, thereby improving the efficiency and accuracy of damage assessment, avoiding unreasonable damage assessment caused by subjective personnel, simplifying the freight claims processing process, shortening the freight claims processing time, and improving the efficiency of freight claims processing.

[0079] The above describes the freight claim processing method in the embodiment of the present invention. The following describes the freight claim processing device in the embodiment of the present invention. Figure 3 In one embodiment of the present invention, a freight claim processing device includes:

[0080] The acquisition module 301 is used to obtain a claim request and retrieve the insurance policy information and freight node information according to the claim request. The claim request includes the target freight order information and the claim information. The claim information includes at least the claim type and claim certification information.

[0081] Verification module 302 is used to verify the claim request based on the policy information and freight node information when the claim type is cargo loss, and obtain a verification result. The claim verification information includes at least real-time video data, and the freight node information includes cargo video data corresponding to the freight node;

[0082] The damage assessment module 303 is used to input the real-time video data, the cargo video data, and the target freight order information into a preset cargo damage assessment model when the verification result is passed, to obtain the cargo loss quantity and cargo loss extent information. The cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer, and an output layer;

[0083] The claim settlement module 304 is used to generate a claim settlement result based on the cargo loss quantity, cargo loss extent information, target freight order information and insurance policy information.

[0084] In the embodiment of the present invention, the labor cost of the insurance company in the freight claims process is reduced, the authenticity of the claim certification materials is guaranteed, and the damage is automatically determined through the cargo damage assessment model, thereby avoiding unreasonable damage assessment caused by subjective damage assessment by personnel, simplifying the freight claims processing process, shortening the freight claims processing time, and improving the freight claims processing efficiency.

[0085] See also Figure 4 Another embodiment of the freight claims processing device in the embodiment of the present invention includes:

[0086] The acquisition module 301 is used to obtain a claim request and retrieve the insurance policy information and freight node information according to the claim request. The claim request includes the target freight order information and the claim information. The claim information includes at least the claim type and claim certification information.

[0087] Verification module 302 is used to verify the claim request based on the policy information and freight node information when the claim type is cargo loss, and obtain a verification result. The claim verification information includes at least real-time video data, and the freight node information includes cargo video data corresponding to the freight node;

[0088] The damage assessment module 303 is used to input the real-time video data, the cargo video data, and the target freight order information into a preset cargo damage assessment model when the verification result is passed, to obtain the cargo loss quantity and cargo loss extent information. The cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer, and an output layer;

[0089] The claim settlement module 304 is used to generate a claim settlement result based on the cargo loss quantity, cargo loss extent information, target freight order information and insurance policy information.

[0090] Optionally, the damage assessment module 303 includes:

[0091] The pre-processing unit 3031 is configured to pre-process the real-time video data and the cargo video data using a preset cargo damage assessment model when the verification result is passed, to obtain a plurality of claim key frame images and a plurality of standard key frame images;

[0092] A feature extraction unit 3032 is configured to extract features from a plurality of claim key frame images and a plurality of standard key frame images using a cargo loss assessment model to obtain a target feature map, wherein the target feature map marks corresponding cargo loss areas;

[0093] The estimation unit 3033 is used to process the target feature map and the target freight order information through the cargo loss assessment model to obtain the cargo loss quantity and cargo loss degree information.

[0094] Optionally, the preprocessing unit 3031 is specifically used to: extract frames from the cargo video data through multiple angles preset by the cargo damage assessment model to obtain multiple standard key frame images, and the multiple standard key frame images are used to indicate the cargo conditions at different angles in the corresponding freight node; extract features from the multiple standard key frame images to obtain cargo feature information of each standard key frame image; extract frames from the real-time video data according to the cargo feature information and multiple angles to obtain multiple claims key frame images.

[0095] Optionally, the feature extraction unit 3032 includes:

[0096] The processing subunit 30321 is used to perform key point matching on each claim key frame image and the corresponding standard key frame image, and calculate the corresponding homography matrix;

[0097] A mapping subunit 30322 is configured to map each standard key frame image to a corresponding claim key frame image according to a corresponding homography matrix;

[0098] The feature extraction subunit 30323 is used to output multiple difference points according to a preset difference threshold and perform clustering to obtain a damage feature map;

[0099] The pyramid subunit 30324 is used to process the damage feature map of each claim key frame image according to a preset scale gradient to obtain damage feature maps of multiple scales corresponding to each claim key frame image;

[0100] The fusion subunit 30325 is used to fuse the damage feature maps of multiple scales corresponding to each claim key frame image and output a target feature map.

[0101] Optionally, the processing sub-unit 30321 is specifically used to: extract the key points of each claim key frame image and the corresponding standard key frame image, and calculate the descriptor corresponding to each key point through the description algorithm; perform key point matching on each claim key frame image and the corresponding standard key frame image according to the matching algorithm to obtain multiple pairs of key points; calculate the multiple pairs of key points to obtain the corresponding homography matrix.

[0102] Optionally, the verification module 302 is specifically used to: when the claim type is cargo loss, verify the target freight order information and the claim type according to the insurance policy information to obtain a first verification result; parse the shooting timestamp and shooting location information corresponding to the real-time video data, and verify according to the freight node information to obtain a second verification result; extract cargo feature information from the real-time video data and the cargo video data, and compare the cargo feature information to obtain a third verification result; combine the first verification result, the second verification result and the third verification result into a verification result.

[0103] Optionally, a freight claim processing device also includes a generation module 305, which is used to retrieve target freight order information and set the claim type corresponding to the target freight order information; when the claim type is cargo loss, determine the claim freight node and obtain the corresponding cargo video data; input the cargo video data into a preset shooting guidance model to generate shooting guidance information; generate real-time video data according to the shooting guidance information, shooting time and shooting location, and the real-time video data includes a shooting timestamp and shooting location information; generate a claim request according to the real-time video data, the target freight order information and the claim type.

[0104] In the embodiment of the present invention, real-time video data can be shot and uploaded according to the shooting guidance information, thereby reducing the labor cost of the insurance company in the freight claims process, ensuring the authenticity of the claim certification materials through the verification module, extracting high-quality cargo images that meet the damage processing requirements through the preset cargo damage assessment model, and automatically extracting the target feature map that marks the damaged area of ​​the cargo. By estimating the target feature map, the amount of cargo loss and the degree of cargo loss information are obtained, thereby improving the efficiency and accuracy of damage assessment, avoiding unreasonable damage assessment caused by subjective damage assessment by personnel, simplifying the freight claims processing process, shortening the freight claims processing time, and improving the efficiency of freight claims processing.

[0105] above Figure 3 and Figure 4 The freight claims processing apparatus in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The freight claims processing system in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0106] Figure 5 is a schematic diagram of the structure of a freight claims processing system provided by an embodiment of the present invention. The freight claims processing system 500 may vary significantly due to different configurations or performances, and may include: a client platform 510 for generating an insurance request based on a freight order and sending it to an insurance platform 520, and submitting a claim request to an arbitration platform 530;

[0107] The insurance platform 520 is used to obtain the insurance application submitted by the client platform 510 and the historical claims information of the arbitration platform 530, and to review and generate the insurance policy information based on the insurance application, historical claims information, logistics enterprise information and pre-set insurance rules;

[0108] The arbitration platform 530 is used to obtain the claim request submitted by the client platform 510, and retrieve the insurance policy information generated by the insurance platform 520, as well as the freight order information uploaded when the insurance is taken out, and the freight node information during the freight process, to conduct freight claim review, damage assessment and claim settlement. The arbitration platform 530 performs the freight claim review, damage assessment and claim settlement work on behalf of the insurance company to obtain the claim settlement result; this can avoid unreasonable damage assessment caused by subjective damage assessment by personnel, and reduce the labor cost of the insurance company in the freight claim settlement process.

[0109] Specifically, before making a claim, the client platform 510 generates an insurance request based on the freight order; the insurance platform 520 can access the blockchain based on the insurance request, and dynamically configure it according to the corresponding cargo order, logistics company information and corresponding historical claims information to obtain the policy information; the arbitration platform 530 establishes a blockchain-based smart contract based on the freight order and policy information; the client platform 510 writes the freight node information of each freight node during the freight order operation into the block.

[0110] Specifically, policies have two scenarios: claim and non-claim. If the user does not file a claim within the preset claim time, the insurance platform 520 marks the freight order as non-claim, updates the logistics company's corresponding historical claim information, and writes it to the blockchain. When a claim is filed, the user can use the client platform 510 to select the corresponding freight order and generate a claim request for broadcast. Based on the claim request, the arbitration platform 530 can access the corresponding policy information and freight node information on the blockchain, perform verification, determine damage, and process the claim, writing the claim result as historical claim information to the blockchain. The arbitration platform 530, based on blockchain technology, can monitor the dynamic changes in cargo transportation, prevent information tampering, and reduce the risk of fraud in the claims process. The insurance platform 520 can also intuitively view the historical claims of logistics companies, enabling more comprehensive risk assessment and policy configuration. The arbitration platform 530, based on blockchain smart contracts, can largely avoid risks. As long as the claim conditions are met, the claim can be completed, or the policy can be automatically terminated if the goods arrive safely without any accidents. This not only avoids false claims but also improves claims efficiency.

[0111] Furthermore, the client platform 510 can also shoot real-time video data according to the shooting guidance information, combine the claim type and claim proof information into a claim request, write the claim request into the blockchain, and broadcast it on the blockchain. Users do not need to wait for the claims specialist on site, but can shoot real-time video data that meets the requirements according to the shooting guidance information, and can upload the damage assessment in time, simplifying the freight claims process, shortening the freight claims processing time, and reducing the insurance company's labor costs.

[0112] Specifically, the insurance platform 520 is used to review insurance requests and generate policy information based on approved insurance requests, historical claims information, logistics company information and preset insurance rules; the historical claims information comes from the arbitration platform 530, and may also include historical claims information entered through other channels.

[0113] Further, Figure 6 FIG2 is a schematic diagram of the structure of an arbitration platform 530 provided in an embodiment of the present invention, comprising one or more central processing units (CPUs) 5310 (e.g., one or more processors), a memory 5320, and one or more storage media 5330 (e.g., one or more mass storage devices) storing applications 5333 or data 5332. The memory 5320 and storage medium 5330 may be either transient or persistent storage. The program stored in the storage medium 5330 may include one or more modules (not shown), each of which may include a series of instruction operations on the arbitration platform 530. Furthermore, the processor 5310 may be configured to communicate with the storage medium 5330 to execute the series of instruction operations stored in the storage medium 5330 on the arbitration platform 530.

[0114] The arbitration platform 5330 may also include one or more power supplies 5340, one or more wired or wireless network interfaces 5350, one or more input and output interfaces 5360, and / or one or more operating systems 5331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 6 The illustrated arbitration platform structure does not constitute a limitation on the arbitration platform, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0115] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the freight claims processing method.

[0116] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0118] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for handling freight claims, characterized in that: The freight claim processing method is applied to a freight claim processing system, including: Obtaining a claim request, and retrieving policy information and freight node information according to the claim request, wherein the claim request includes target freight order information and claim information, and the claim information includes at least claim type and claim certification information; When the claim type is cargo loss, the claim request is verified according to the policy information and the freight node information to obtain a verification result, wherein the claim certification materials include at least real-time video data, and the freight node information includes cargo video data corresponding to the freight node; When the verification result is passed, the real-time video data, the cargo video data, and the target freight order information are input into a preset cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information, wherein the cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer, and an output layer; generating a claim settlement result based on the cargo loss quantity, the cargo loss extent information, the target freight order information, and the insurance policy information; When the verification result is passed, the real-time video data, the cargo video data, and the target freight order information are input into a preset cargo damage assessment model to obtain cargo loss quantity and cargo loss degree information, including: When the verification result is passed, the real-time video data and the cargo video data are preprocessed using a preset cargo damage assessment model to obtain a plurality of claim key frame images and a plurality of standard key frame images; Performing feature extraction on the multiple claim key frame images and the multiple standard key frame images using a cargo loss assessment model to obtain a target feature map, wherein the target feature map marks a corresponding cargo loss area; Processing the target feature map and the target freight order information through a cargo loss assessment model to obtain cargo loss quantity and cargo loss degree information; The cargo damage assessment model is used to extract features from the multiple claim key frame images and the multiple standard key frame images to obtain a target feature map, including: Perform key point matching on each claim key frame image and the corresponding standard key frame image, and calculate the corresponding homography matrix; Mapping each standard key frame image to a corresponding claim key frame image according to the corresponding homography matrix; Output multiple difference points according to the preset difference threshold and cluster them to obtain the damage feature map; Processing the damage feature map of each claim key frame image according to a preset scale gradient to obtain damage feature maps of multiple scales corresponding to each claim key frame image; The damage feature maps of multiple scales corresponding to each claim key frame image are fused to output a target feature map.

2. The freight claim processing method according to claim 1, characterized in that: When the verification result is passed, the real-time video data and the cargo video data are preprocessed using a preset cargo damage assessment model to obtain a plurality of claim key frame images and a plurality of standard key frame images, including: Extract frames from the cargo video data at multiple angles preset by the cargo damage assessment model to obtain multiple standard key frame images, wherein the multiple standard key frame images are used to indicate the cargo conditions at different angles in the corresponding freight node; Perform feature extraction on multiple standard key frame images to obtain cargo feature information of each standard key frame image; The real-time video data is frame-extracted according to the cargo characteristic information and the multiple angles to obtain multiple claim key frame images.

3. The freight claim processing method according to claim 1, characterized in that: The key point matching of each claim key frame image and the corresponding standard key frame image is performed, and the corresponding homography matrix is ​​calculated, including: Extract the key points of each claim key frame image and the corresponding standard key frame image, and calculate the descriptor corresponding to each key point through the description algorithm; According to the matching algorithm, key points of each claim key frame image and the corresponding standard key frame image are matched to obtain multiple pairs of key points; Calculate the multiple pairs of key points to obtain corresponding homography matrices.

4. The freight claim processing method according to claim 1, characterized in that: When the claim type is cargo loss, the claim request is verified according to the policy information and the freight node information to obtain a verification result, wherein the claim certification materials include at least real-time video data, and the freight node information includes cargo video data corresponding to the freight node, including: When the claim type is cargo loss, verifying the target freight order information and the claim type according to the insurance policy information to obtain a first verification result; parsing the shooting timestamp and shooting location information corresponding to the real-time video data, and verifying them according to the freight node information to obtain a second verification result; extracting cargo characteristic information from the real-time video data and the cargo video data, and comparing the cargo characteristic information to obtain a third verification result; The first verification result, the second verification result, and the third verification result are combined into a verification result.

5. The freight claim processing method according to any one of claims 1 to 4, characterized in that: Before obtaining the claim request, it also includes: Retrieve target freight order information and set the claim type corresponding to the target freight order information; When the claim type is cargo loss, determining the claim freight node and obtaining corresponding cargo video data; Inputting the cargo video data into a preset shooting guidance model to generate shooting guidance information; generating real-time video data according to the shooting guide information, the shooting time and the shooting location, wherein the real-time video data includes a shooting timestamp and shooting location information; A claim request is generated according to the real-time video data, the target freight order information, and the claim type.

6. A freight claims processing device, characterized in that: The freight claim processing device includes: An acquisition module, configured to acquire a claim request and retrieve policy information and freight node information based on the claim request, wherein the claim request includes target freight order information and claim information, and the claim information includes at least claim type and claim certification information; a verification module, configured to verify the claim request based on the policy information and the freight node information to obtain a verification result when the claim type is cargo loss, wherein the claim verification information includes at least real-time video data, and the freight node information includes cargo video data corresponding to the freight node; a damage assessment module, configured to input the real-time video data, the cargo video data, and the target freight order information into a preset cargo damage assessment model when the verification result is passed, to obtain information on the quantity and extent of cargo loss, wherein the cargo damage assessment model includes an input layer, a feature extraction layer, a feature pyramid layer, and an output layer; A claims settlement module, configured to generate a claims settlement result based on the cargo loss quantity, the cargo loss extent information, the target freight order information, and the insurance policy information; The damage assessment module includes: a preprocessing unit for preprocessing the real-time video data and the cargo video data using a preset cargo damage assessment model when the verification result is passed, thereby obtaining a plurality of claim key frame images and a plurality of standard key frame images; a feature extraction unit for extracting features from the plurality of claim key frame images and the plurality of standard key frame images using the cargo damage assessment model, thereby obtaining a target feature map, wherein the target feature map marks the corresponding cargo loss area; and an estimation unit for processing the target feature map and the target freight order information using the cargo damage assessment model to obtain information on the cargo loss quantity and cargo loss extent. The feature extraction unit includes: a processing subunit, which is used to match key points of each claim key frame image and the corresponding standard key frame image respectively, and calculate the corresponding homography matrix; a mapping subunit, which is used to map each standard key frame image to the corresponding claim key frame image according to the corresponding homography matrix; a feature extraction subunit, which is used to output multiple difference points according to a preset difference threshold and cluster them to obtain a damage feature map; a pyramid subunit, which is used to process the damage feature map of each claim key frame image according to a preset scale gradient to obtain damage feature maps of multiple scales corresponding to each claim key frame image; and a fusion subunit, which is used to fuse the damage feature maps of multiple scales corresponding to each claim key frame image and output a target feature map.

7. A freight claims processing system, characterized in that: The freight claim processing system implements the freight claim processing method according to any one of claims 1 to 5, and the freight claim processing system includes: The client platform is used to generate insurance requests based on freight orders and send them to the insurance platform, and to generate claims requests and send them to the arbitration platform; The insurance platform is used to review the insurance application and generate policy information based on the insurance application, logistics company information, historical claims information and pre-set insurance rules; The arbitration platform is used to review, determine damages and settle claims based on the claim request and policy information.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is read and executed, the method for handling freight claims as described in any one of claims 1 to 5 is executed.

Images