WO2020155790A1 - 理赔信息提取方法和装置、电子设备 - Google Patents

理赔信息提取方法和装置、电子设备 Download PDF

Info

Publication number
WO2020155790A1
WO2020155790A1 PCT/CN2019/119352 CN2019119352W WO2020155790A1 WO 2020155790 A1 WO2020155790 A1 WO 2020155790A1 CN 2019119352 W CN2019119352 W CN 2019119352W WO 2020155790 A1 WO2020155790 A1 WO 2020155790A1
Authority
WO
WIPO (PCT)
Prior art keywords
image data
image
classification
settlement
key information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/119352
Other languages
English (en)
French (fr)
Inventor
吴博坤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Alibaba Group Holding Ltd
Original Assignee
Alibaba Group Holding Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Alibaba Group Holding Ltd filed Critical Alibaba Group Holding Ltd
Publication of WO2020155790A1 publication Critical patent/WO2020155790A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/08Insurance
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/10Image acquisition

Definitions

  • One or more embodiments of this specification relate to the field of computer application technology, and in particular to a method and device for extracting claims information, and electronic equipment.
  • This specification proposes a method for extracting claims information, which includes:
  • the key information used for claims settlement is extracted from the image data of each image category obtained by classification.
  • the method further includes:
  • the first classification model is a convolutional neural network CNN model.
  • the image category obtained by classifying the image data in the image data set includes one or more of the following image categories:
  • Document image document image; scene image; damage image; other images.
  • the extraction of key information for claim settlement from the image data of each category obtained by classification includes:
  • the relevant personnel information and relevant vehicle information of the claim settlement case are extracted from the image data of the certificate image as the key information for claim settlement.
  • the extraction of key information for claim settlement from the image data of each category obtained by classification includes:
  • the liability ratio of the relevant person in the claim settlement case is extracted from the image data of the document image as the key information for claim settlement.
  • the extraction of key information for claim settlement from the image data of each category obtained by classification includes:
  • the image data of the scene image is obtained by classification
  • the image data of the scene image is input into the second classification model for classification calculation
  • the accident type corresponding to the image data of the scene image is determined based on the classification result, so as to The accident type is used as key information for claim settlement;
  • the second classification model is a machine learning model trained based on a number of scene image samples labeled with the accident type.
  • the second classification model is a CNN model.
  • This specification also proposes a device for extracting claims information, which includes:
  • the first obtaining module is used to obtain a collection of image data related to a claim settlement case
  • the classification module is configured to input the image data in the image data set into a first classification model for classification calculation, and classify the image data in the image data set based on the classification calculation result; wherein, the first The classification model is a machine learning model trained based on a number of image data samples labeled with image categories;
  • the extraction module is used to extract key information used for claim settlement from the image data of each image category obtained by classification.
  • the device further includes:
  • the second acquisition module is used to acquire the extracted key information for claim settlement
  • the claim settlement module is used for claim settlement processing based on the key information used for claim settlement.
  • the first classification model is a convolutional neural network CNN model.
  • the image category obtained by classifying the image data in the image data set includes one or more of the following image categories:
  • Document image document image; scene image; damage image; other images.
  • the extraction module is specifically configured to:
  • the relevant personnel information and relevant vehicle information of the claim settlement case are extracted from the image data of the certificate image as key information for claim settlement.
  • the extraction module is specifically configured to:
  • the liability ratio of the relevant person in the claim settlement case is extracted from the image data of the document image as the key information for claim settlement.
  • the extraction module is specifically configured to:
  • the image data of the scene image is obtained by classification
  • the image data of the scene image is input into the second classification model for classification calculation
  • the accident type corresponding to the image data of the scene image is determined based on the classification result, so as to The accident type is used as key information for claim settlement;
  • the second classification model is a machine learning model trained based on a number of scene image samples labeled with the accident type.
  • the second classification model is a CNN model.
  • This specification also proposes an electronic device, which includes:
  • Memory for storing machine executable instructions
  • the key information used for claims settlement is extracted from the image data of each image category obtained by classification.
  • the image data set related to the claim settlement case may be input to the classification model, so that the image data in the image data set can be classified by the classification model.
  • the key information for claim settlement can be extracted automatically from the image data of each image category obtained by classification.
  • Fig. 1 is a flowchart of a method for extracting claims information according to an exemplary embodiment of this specification
  • Figure 2 is a hardware structure diagram of an electronic device where a claim information extraction device is shown in an exemplary embodiment of this specification;
  • Fig. 3 is a block diagram of a device for extracting claims information according to an exemplary embodiment of this specification.
  • first, second, third, etc. may be used in this specification to describe various information, the information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other.
  • first information may also be referred to as second information, and similarly, the second information may also be referred to as first information.
  • word “if” as used herein can be interpreted as "when” or “when” or "in response to determination”.
  • This manual aims to provide a technical solution for a claim settlement case, to classify the image data in the image data set related to the claim settlement case, and to extract the key information for the claim settlement from the image data of each category obtained by the classification. .
  • a collection of image data related to the claim settlement case can be obtained first.
  • the image data set may include image data of at least one image category related to the claim case, such as: image data of a certificate image; image data of a document image; and image data of a scene image of the car accident.
  • the image data in the image data set can be classified based on the classification model.
  • the classification model may be a machine learning model trained based on several image data samples labeled with classification labels.
  • the key information for claim settlement can be extracted from the image data of each image category obtained by classification, for example: the relevant personnel information of the claim case can be extracted from the image data of the document image; from the image data of the document image Extract the responsibility ratio of the relevant personnel of the claim settlement case; and determine the type of accident corresponding to the claim settlement case based on the image data of the scene image.
  • the image data set related to the claim settlement case may be input to the classification model, so that the image data in the image data set can be classified by the classification model.
  • the key information for claim settlement can be extracted automatically from the image data of each image category obtained by classification.
  • FIG. 1 is a flowchart of a method for extracting claims information according to an exemplary embodiment of this description.
  • This method can be applied to electronic devices such as servers, mobile phones, tablet devices, notebook computers, and PDAs (Personal Digital Assistants), and includes the following steps:
  • Step 102 Acquire a collection of image data related to a claim settlement case
  • Step 104 Input the image data in the image data set into a first classification model for classification calculation, and classify the image data in the image data set based on the classification calculation result; wherein, the first classification model A machine learning model trained based on several image data samples labeled with image categories;
  • Step 106 Extract key information used for claim settlement from the image data of each image category obtained by classification.
  • a collection of image data related to the claim settlement case may be obtained first.
  • the image data set may contain image data of at least one image category related to the claim settlement case.
  • the image categories of the image data contained in the image data set may include one or more of the following image categories: document images; document images; live images; damage images; other images.
  • the image data set may contain the image data of the certificate image (for example: the driver’s license image of the owner of the damaged vehicle, the Image data of the damaged vehicle’s license, etc.); image data of the document image (for example: image data of an accident liability certificate, etc.); image data of the scene image of the car accident; image data of the damaged image of the damaged vehicle; and Image data belonging to the first 4 image categories (called image data of other images).
  • the certificate image for example: the driver’s license image of the owner of the damaged vehicle, the Image data of the damaged vehicle’s license, etc.
  • image data of the document image for example: image data of an accident liability certificate, etc.
  • image data of the scene image of the car accident image data of the damaged image of the damaged vehicle
  • Image data belonging to the first 4 image categories called image data of other images.
  • cameras deployed near the accident site corresponding to the claim settlement case may be used to obtain at least one image captured by these cameras, and the captured image may be used as the image data in the image data set related to the claim settlement case.
  • cameras deployed near the accident site corresponding to the claim settlement case can be used to obtain videos captured by these cameras, and image frames in these videos can be extracted to use these image frames as a collection of image data related to the claim settlement case Image data.
  • the image data in the image data set can be input into a preset classification model (referred to as the first classification model) for classification calculation.
  • the first classification model a preset classification model for classification calculation.
  • the first classification model may be a machine learning model such as a commonly used Convolutional Neural Networks (CNN) model.
  • CNN Convolutional Neural Networks
  • a preset number of image data can be obtained from the image data set corresponding to the historical claim case (that is, the claim case in which the claims processing has been executed before), and the image data can be labeled with classification labels.
  • the classification label can be used to characterize the image category to which the image data belongs, for example: the image data related to the document in the image data can be labeled as the classification label; the image data related to the document in the image data can be labeled as the document
  • the images are used as classification labels; the image data related to the accident scene in these images are labeled as the classification labels; the images related to the damaged parts of the accident vehicles in these images are labeled as the classification labels; Images that are not related to documents, documents, accident scenes, and damaged parts of the accident vehicle are labeled with other images as classification labels.
  • the preset number of image data samples is 100
  • 100 images can be obtained from the image data collection corresponding to historical claims, and image categories can be marked for these images. Later, you can use these 100 images with image categories as training samples, and use backpropagation to train the 100 images with image categories based on the CNN algorithm to obtain the first classification model. .
  • the image data in the image data set related to the claim case can be classified and calculated based on the trained first classification model, so that the image data in the image data set can be classified based on the result of the classification calculation, that is, Determine the image category to which the image data in the image data set belongs.
  • the key information for claim settlement can be extracted from the image data of each category.
  • the image data of the certificate image obtained from the classification may be based on the OCR (Optical Character Recognition) algorithm.
  • OCR Optical Character Recognition
  • the relevant personnel information and relevant vehicle information of the claim settlement case are extracted from the data, and the extracted relevant personnel information and relevant vehicle information of the claim settlement case are used as the key information for the settlement of claims.
  • the image data of the certificate image obtained by the classification can be input to the character string detection model, so that the character string detection model obtains the image area containing the target character string based on the image data.
  • the target character string can be a character string used to characterize personnel information such as a name or ID number, or a character string used to characterize vehicle information such as a license plate number.
  • the string detection model may be a machine learning model such as a commonly used CNN model.
  • the image data marked with the image area containing the target string can be used as training samples, and the back propagation method can be used to train on these image data samples based on a preset machine learning algorithm (for example: CNN algorithm) , In order to obtain a character string detection model for detecting the image area containing the target character string from the image data of the above-mentioned document image.
  • a preset machine learning algorithm for example: CNN algorithm
  • the name of the relevant person in the claim case can be obtained based on the target character string used to characterize the name; the ID number of the relevant person in the claim case can be obtained based on the target character string used to characterize the ID number;
  • the target string used to characterize the license plate number is identified to obtain the license plate number of the relevant vehicle in the claim case.
  • the relevant personnel information such as the name and ID number, as well as the relevant vehicle information such as the license plate number obtained for the claim settlement case can be used as the key information for the claim settlement.
  • the character string recognition model may be a Recurrent Neural Network (RNN) model based on a CTC (Connectionist Temporal Classification) loss function.
  • RNN Recurrent Neural Network
  • CTC Connectionist Temporal Classification
  • the image data containing the character string labeled with the text content corresponding to the character string can be used as a training sample, and the method of back propagation can be used based on a preset machine learning algorithm (for example: RNN algorithm based on CTC loss function ), training these image data samples to obtain a character string recognition model for recognizing the target character string in the aforementioned image area.
  • a preset machine learning algorithm for example: RNN algorithm based on CTC loss function
  • the image data of the document image in the image data set related to the above-mentioned claim settlement can be obtained from the classification based on the OCR algorithm and the NLP (Natural Language Processing) algorithm. From the image data of the document image, extract the liability ratio of the relevant person in the claim settlement case, and use the extracted liability ratio of the relevant person in the claim settlement case as the key information for claim settlement.
  • NLP Natural Language Processing
  • the image data of the document image obtained by the classification may be input to the character string detection model, so that the character string detection model obtains the image area containing the target character string based on the image data.
  • the target character string may be a character string used to characterize the responsibility information of the relevant person in the claim settlement case.
  • the target string in the image data of the accident responsibility certificate can be the text describing the responsibility of the person involved in the accident in the accident responsibility certificate (for example, the relevant person A assumes the main responsibility , Relevant Person B assumes secondary responsibility, etc.) The corresponding character string.
  • the string detection model may be a machine learning model such as a commonly used CNN model.
  • the image data marked with the image area containing the target string can be used as training samples, and the back propagation method can be used to train on these image data samples based on a preset machine learning algorithm (for example: CNN algorithm) , In order to obtain a character string detection model for detecting the image area containing the target character string from the image data of the above-mentioned receipt image.
  • a preset machine learning algorithm for example: CNN algorithm
  • the character string recognition model may be an RNN model based on the CTC loss function.
  • the image data containing the character string labeled with the text content corresponding to the character string can be used as a training sample, and the method of back propagation can be used based on a preset machine learning algorithm (for example: RNN algorithm based on CTC loss function ), training these image data samples to obtain a character string recognition model for recognizing the target character string in the aforementioned image area.
  • a preset machine learning algorithm for example: RNN algorithm based on CTC loss function
  • the liability information of the relevant persons in the claim settlement case can be analyzed based on the NPL algorithm to obtain the proportion of the liability of the relevant persons in the claim settlement case, and the obtained claim settlement
  • the liability ratio of the relevant persons in the case is used as the key information for claims settlement.
  • the responsibility information of the relevant person in the claim settlement case includes "relevant person A assumes primary responsibility” and "relevant person B assumes secondary responsibility”
  • the responsibility information can be analyzed based on the NPL algorithm, thereby It can be determined that the responsibility ratio of related personnel A is greater than 50%, and the responsibility proportion of related personnel B is less than 50%.
  • the character string detection model used to detect the image area containing the target character string from the image data of the aforementioned document image is the same as the character string detection model used to detect the image area containing the target character string from the image data of the aforementioned document image
  • the string detection model of can be the same string detection model, or two different string detection models, which are not limited in this specification.
  • the character string recognition model used to recognize the target character string in the image area of the document image may be the same character as the character string recognition model used to recognize the target character string in the image area of the document image
  • the string recognition model can also be two different string recognition models, which are not limited in this specification.
  • the image data of the live image in the image data set related to the above-mentioned claim settlement case can be input into a preset classification model (referred to as the second Classification calculation in the classification model).
  • the second classification model may be a machine learning model such as a commonly used CNN model.
  • the image data of the scene image marked with the accident type can be used as training samples, and the back propagation method can be used to train on these scene image samples based on a preset machine learning algorithm (for example: CNN algorithm).
  • the second classification model used to determine the accident type corresponding to the image data of the scene image is obtained.
  • the types of accidents can include single-vehicle accidents, double-vehicle accidents, and multiple-vehicle accidents.
  • the car accidents shown in the scene images used as training samples can be marked as “single vehicle accidents”, and the car accidents shown in these scene images containing two vehicles can be marked as “double Car accidents”, the car accidents shown in these scene images containing three or more vehicles are marked as "multi-vehicle accidents.”
  • the image data of the scene image can be classified and calculated based on the trained second classification model, so that the image data of the scene image can be classified based on the result of the classification calculation, that is, the accident corresponding to the scene image is determined Type of accident.
  • the key information extracted from each category of image data for claim settlement.
  • the extracted key information used for claim settlement can be obtained, and the claim settlement process can be performed based on the key information used for claim settlement.
  • the key information for claims can be input to the claims system loaded on the electronic device, so that the claims system can automatically enter the key information for claims.
  • the claims settlement system can then perform subsequent claims settlement processing based on the information of the claim settlement case.
  • the claims system can also classify and store the image data set related to the claim case according to the foregoing classification results, that is, classify and store the image data in the image data set according to the image category to which it belongs.
  • the image data set related to the claim settlement case may be input to the classification model, so that the image data in the image data set can be classified by the classification model.
  • the key information for claim settlement can be extracted automatically from the image data of each image category obtained by classification.
  • this specification also provides an embodiment of the claim information extraction device.
  • the embodiments of the claim settlement information extraction device in this specification can be applied to electronic equipment.
  • the device embodiments can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a logical device, it is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory through the processor of the electronic device where it is located. From a hardware perspective, as shown in Figure 2, it is a hardware structure diagram of the electronic equipment where the claims information extraction device is located in this specification, except for the processor, memory, network interface, and non-volatile memory shown in Figure 2. In addition, the electronic device where the device is located in the embodiment usually extracts actual functions based on the claims information, and may also include other hardware, which will not be repeated here.
  • FIG. 3 is a block diagram of an apparatus for extracting claims information according to an exemplary embodiment of this specification.
  • the device 30 can be applied to the electronic equipment shown in FIG. 2, including:
  • the first obtaining module 301 is configured to obtain a collection of image data related to a claim settlement case
  • the classification module 302 is configured to input the image data in the image data set into the first classification model for classification calculation, and classify the image data in the image data set based on the classification calculation result; wherein, the A classification model is a machine learning model trained based on a number of image data samples labeled with image categories;
  • the extraction module 303 is configured to extract key information for claim settlement from the image data of each image category obtained by classification.
  • the device 30 may further include:
  • the second obtaining module 304 is configured to obtain the extracted key information used for claim settlement
  • the claim settlement module 305 is configured to perform claim settlement processing based on the key information used for claim settlement.
  • the first classification model may be a convolutional neural network CNN model.
  • the image category obtained by classifying the image data in the image data set may include one or more of the following image categories:
  • Document image document image; scene image; damage image; other images.
  • the extraction module 303 may be specifically used for:
  • the relevant personnel information and relevant vehicle information of the claim settlement case are extracted from the image data of the certificate image as key information for claim settlement.
  • the extraction module 303 may be specifically used for:
  • the liability ratio of the relevant person in the claim settlement case is extracted from the image data of the document image as the key information for claim settlement.
  • the extraction module 303 may be specifically used for:
  • the image data of the scene image is obtained by classification
  • the image data of the scene image is input into the second classification model for classification calculation
  • the accident type corresponding to the image data of the scene image is determined based on the classification result, so as to The accident type is used as key information for claim settlement;
  • the second classification model is a machine learning model trained based on a number of scene image samples labeled with the accident type.
  • the second classification model may be a CNN model.
  • the relevant part can refer to the part of the description of the method embodiment.
  • the device embodiments described above are merely illustrative.
  • the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in One place, or it can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution in this specification. Those of ordinary skill in the art can understand and implement it without creative work.
  • a typical implementation device is a computer.
  • the specific form of the computer can be a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email receiving and sending device, and a game control A console, a tablet computer, a wearable device, or a combination of any of these devices.
  • the electronic device includes a processor and a memory for storing machine executable instructions; wherein the processor and the memory are usually connected to each other through an internal bus.
  • the device may also include an external interface to be able to communicate with other devices or components.
  • the processor is prompted to:
  • the key information used for claims settlement is extracted from the image data of each image category obtained by classification.
  • the processor is also prompted to:
  • the first classification model is a convolutional neural network CNN model.
  • the image categories obtained by classifying the image data in the image data set include one or more of the following image categories:
  • Document image document image; scene image; damage image; other images.
  • the processor is prompted to:
  • the relevant personnel information and relevant vehicle information of the claim settlement case are extracted from the image data of the certificate image as key information for claim settlement.
  • the processor is prompted to:
  • the liability ratio of the relevant person in the claim settlement case is extracted from the image data of the document image as the key information for claim settlement.
  • the processor is prompted to:
  • the image data of the scene image is obtained by classification
  • the image data of the scene image is input into the second classification model for classification calculation
  • the accident type corresponding to the image data of the scene image is determined based on the classification result, so as to The accident type is used as key information for claim settlement;
  • the second classification model is a machine learning model trained based on a number of scene image samples labeled with the accident type.
  • the second classification model is a CNN model.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Multimedia (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Technology Law (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)

Abstract

一种理赔信息提取方法和装置、电子设备,所述方法包括:获取与理赔案件相关的图像数据集合(102);将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型(104);从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息(106)。

Description

理赔信息提取方法和装置、电子设备 技术领域
本说明书一个或多个实施例涉及计算机应用技术领域,尤其涉及一种理赔信息提取方法和装置、电子设备。
背景技术
现如今,在发生了车祸事故,接到车主报案之后,通常需要收集大量照片用于理赔。后续在处理相应的理赔案件时,通常需要由该理赔案件的相关责任人对这些照片进行人工归档,并对这些照片进行分析,得到与该理赔案件相关的理赔信息。然而,这样不仅需要消耗大量的人力资源,理赔信息的提取效率也较低。
发明内容
本说明书提出一种理赔信息提取方法,所述方法包括:
获取与理赔案件相关的图像数据集合;
将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
可选地,所述方法还包括:
获取提取出的所述用于理赔的关键信息;
基于所述用于理赔的关键信息进行理赔处理。
可选地,所述第一分类模型为卷积神经网络CNN模型。
可选地,对所述图像数据集合中的图像数据进行分类得到的图像类别包括以下图像类别中的一个或多个:
证件图像;单据图像;现场图像;损伤图像;其他图像。
可选地,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
可选地,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于理赔的关键信息。
可选地,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
可选地,所述第二分类模型为CNN模型。
本说明书还提出一种理赔信息提取装置,所述装置包括:
第一获取模块,用于获取与理赔案件相关的图像数据集合;
分类模块,用于将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
提取模块,用于从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
可选地,所述装置还包括:
第二获取模块,用于获取提取出的所述用于理赔的关键信息;
理赔模块,用于基于所述用于理赔的关键信息进行理赔处理。
可选地,所述第一分类模型为卷积神经网络CNN模型。
可选地,对所述图像数据集合中的图像数据进行分类得到的图像类别包括以下图像 类别中的一个或多个:
证件图像;单据图像;现场图像;损伤图像;其他图像。
可选地,所述提取模块具体用于:
如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
可选地,所述提取模块具体用于:
如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于理赔的关键信息。
可选地,所述提取模块具体用于:
如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
可选地,所述第二分类模型为CNN模型。
本说明书还提出一种电子设备,所述电子设备包括:
处理器;
用于存储机器可执行指令的存储器;
其中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
获取与理赔案件相关的图像数据集合;
将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
在上述技术方案中,对于某个理赔案件而言,可以将与该理赔案件相关的图像数据 集合输入至分类模型,以由该分类模型对该图像数据集合中的图像数据进行分类。后续,可以自动从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。采用这样的方式,与常用的针对理赔案件的图像数据进行人工分类和分析的方式相比,可以提高理赔信息的提取效率,减少人力资源的消耗。
附图说明
图1是本说明书一示例性实施例示出的一种理赔信息提取方法的流程图;
图2是本说明书一示例性实施例示出的一种理赔信息提取装置所在电子设备的硬件结构图;
图3是本说明书一示例性实施例示出的一种理赔信息提取装置的框图。
具体实施方式
这里将详细地对示例性实施例进行说明,其示例表示在附图中。下面的描述涉及附图时,除非另有表示,不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本说明书一个或多个实施例相一致的所有实施方式。相反,它们仅是与如所附权利要求书中所详述的、本说明书一个或多个实施例的一些方面相一致的装置和方法的例子。
在本说明书使用的术语是仅仅出于描述特定实施例的目的,而非旨在限制本说明书。在本说明书和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式,除非上下文清楚地表示其他含义。还应当理解,本文中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。
应当理解,尽管在本说明书可能采用术语第一、第二、第三等来描述各种信息,但这些信息不应限于这些术语。这些术语仅用来将同一类型的信息彼此区分开。例如,在不脱离本说明书范围的情况下,第一信息也可以被称为第二信息,类似地,第二信息也可以被称为第一信息。取决于语境,如在此所使用的词语“如果”可以被解释成为“在……时”或“当……时”或“响应于确定”。
本说明书旨在提供一种针对理赔案件,对与该理赔案件相关的图像数据集合中的图像数据进行分类,并从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息的技术方案。
在具体实现时,对于某个理赔案件而言,可以先获取与该理赔案件相关的图像数据集合。
其中,图像数据集合中可以包含与理赔案件相关的至少一个图像类别的图像数据,例如:证件图像的图像数据;单据图像的图像数据;以及该车祸事故的现场图像的图像数据等。
在获取到该图像数据集合后,可以基于分类模型对该图像数据集合中的图像数据进行分类。
其中,该分类模型可以是基于若干被标注了分类标签的图像数据样本训练出的机器学习模型。
后续,即可从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息,例如:可以从证件图像的图像数据中提取该理赔案件的相关人员信息;从单据图像的图像数据中提取该理赔案件的相关人员的责任比例;以及基于现场图像的图像数据确定该理赔案件对应的事故类型等。
在上述技术方案中,对于某个理赔案件而言,可以将与该理赔案件相关的图像数据集合输入至分类模型,以由该分类模型对该图像数据集合中的图像数据进行分类。后续,可以自动从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。采用这样的方式,与常用的针对理赔案件的图像数据进行人工分类和分析的方式相比,可以提高理赔信息的提取效率,减少人力资源的消耗。
下面通过具体实施例对本说明书进行描述。
请参考图1,图1是本说明一示例性实施例示出的一种理赔信息提取方法的流程图。该方法可以应用于服务器、手机、平板设备、笔记本电脑、掌上电脑(Personal Digital Assistants,PDAs)等电子设备,包括如下步骤:
步骤102,获取与理赔案件相关的图像数据集合;
步骤104,将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
步骤106,从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
在本实施例中,对于某个理赔案件而言,可以先获取与该理赔案件相关的图像数据 集合。
其中,图像数据集合中可以包含与理赔案件相关的至少一个图像类别的图像数据。
在示出的一种实施方式中,该图像数据集合中包含的图像数据的图像类别可以包括以下图像类别中的一个或多个:证件图像;单据图像;现场图像;损伤图像;其他图像。
举例来说,假设该理赔案件为针对车祸事故中的受损车辆的理赔案件,则该图像数据集合中可以包含证件图像的图像数据(例如:该受损车辆的车主的驾驶证图像、该受损车辆的行车证图像等);单据图像的图像数据(例如:事故责任认定书的图像数据等);该车祸事故的现场图像的图像数据;该受损车辆的损伤图像的图像数据;以及不属于前4个图像类别的图像数据(称为其他图像的图像数据)。
在实际应用中,可以获取由用户上传的对驾驶证和行车证等有效证件、事故责任认定书等单据以及事故车辆的受损部位等进行拍摄得到的至少一张图像,并将拍摄得到的图像作为与该理赔案件相关的图像数据集合中的图像数据。
或者,可以利用部署在该理赔案件对应的事故现场附近的摄像头,获取这些摄像头拍摄得到的至少一张图像,并将拍摄得到的图像作为与该理赔案件相关的图像数据集合中的图像数据。
或者,可以利用部署在该理赔案件对应的事故现场附近的摄像头,获取这些摄像头拍摄得到的视频,并提取这些视频中的图像帧,以将这些图像帧作为与该理赔案件相关的图像数据集合中的图像数据。
在获取到与该理赔案件相关的图像数据集合后,可以将该图像数据集合中的图像数据输入至预设的分类模型(称为第一分类模型)中进行分类计算。
其中,第一分类模型可以是常用的卷积神经网络(Convolutional Neural Networks,CNN)模型等机器学习模型。
需要说明的是,可以先从历史理赔案件(即以前执行完理赔处理的理赔案件)对应的图像数据集合中获取预设数量的图像数据,并为这些图像数据标注分类标签。
其中,分类标签可以是用于表征图像数据所属的图像类别,例如:可以为这些图像数据中与证件相关的图像数据标注证件图像作为分类标签;为这些图像数据中与单据相关的图像数据标注单据图像作为分类标签;为这些图像中与事故现场相关的图像数据标注现场图像作为分类标签;为这些图像中与事故车辆的受损部位相关的图像标注损伤图 像作为分类标签;并为这些图像中与证件、单据、事故现场以及事故车辆的受损部位均不相关的图像标注其他图像作为分类标签。
后续,可以将这些被标注了图像类别的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:CNN算法),针对这些图像数据样本进行训练,以得到用于对与上述理赔案件相关的图像数据集合中的图像数据进行分类的上述第一分类模型。
举例来说,假设预设的图像数据样本的数量为100张,则可以从历史理赔案件对应的图像数据集合中获取100张图像,并为这些图像标注图像类别。后续,可以将这100张被标注了图像类别的图像作为训练样本,采用反向传播的方式,基于CNN算法,针对这100张被标注了图像类别的图像进行训练,以得到该第一分类模型。
这样,可以基于已训练好的上述第一分类模型对与上述理赔案件相关的图像数据集合中的图像数据进行分类计算,从而可以基于分类计算结果对该图像数据集合中的图像数据进行分类,即确定该图像数据集合中的图像数据所属的图像类别。
在确定了该图像数据集合中的图像数据所属的图像类别后,可以从各个类别的图像数据中分别提取用于理赔的关键信息。
在示出的一种实施方式中,针对与上述理赔案件相关的图像数据集合中的证件图像的图像数据,可以基于OCR(Optical Character Recognition,光学字符识别)算法,从分类得到的证件图像的图像数据中提取该理赔案件的相关人员信息和相关车辆信息,并将提取出的该理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
具体地,可以将分类得到的证件图像的图像数据输入至字符串检测模型,以由该字符串检测模型基于该图像数据获取其中包含目标字符串的图像区域。
对于证件图像的图像数据来说,目标字符串可以是用于表征姓名或身份证号码等人员信息的字符串,也可以是用于表征车牌号等车辆信息的字符串。
其中,字符串检测模型可以是常用的CNN模型等机器学习模型。
类似地,可以将被标注了包含目标字符串的图像区域的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:CNN算法),针对这些图像数据样本进行训练,以得到用于从上述证件图像的图像数据中检测出包含目标字符串的图像区域的字符串检测模型。
而在获取到包含目标字符串的图像区域后,则可以继续将该图像区域输入至字符串识别模型,以由该字符串识别模型对该图像区域中的目标字符串进行识别,得到该理赔案件的相关人员信息和相关车辆信息,并将得到的该理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
举例来说,可以基于用于表征姓名的目标字符串识别得到该理赔案件的相关人员的姓名;基于用于表征身份证号码的目标字符串识别得到该理赔案件的相关人员的身份证号码;基于用于表征车牌号的目标字符串识别得到该理赔案件的相关车辆的车牌号等。后续,可以将针对该理赔案件得到的姓名和身份证号码等相关人员信息,以及车牌号等相关车辆信息作为用于理赔的关键信息。
其中,字符串识别模型可以是基于CTC(Connectionist Temporal Classification)损失函数的循环神经网络(Recurrent Neural Network,RNN)模型。
类似地,可以将被标注了字符串对应的文字内容的包含该字符串的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:基于CTC损失函数的RNN算法),针对这些图像数据样本进行训练,以得到用于对上述图像区域中的目标字符串进行识别的字符串识别模型。
在示出的一种实施方式中,针对与上述理赔案件相关的图像数据集合中的单据图像的图像数据,可以基于OCR算法,以及NLP(Natural Language Processing,自然语言处理)算法,从分类得到的单据图像的图像数据中提取该理赔案件的相关人员的责任比例,并将提取出的该理赔案件的相关人员的责任比例作为用于理赔的关键信息。
具体地,可以将分类得到的单据图像的图像数据输入至字符串检测模型,以由该字符串检测模型基于该图像数据获取其中包含目标字符串的图像区域。
对于单据图像的图像数据而言,目标字符串可以是用于表征该理赔案件的相关人员的责任信息的字符串。以事故责任认定书为例,事故责任认定书的图像数据中的目标字符串可以是该事故责任认定书中对事故的相关人员所承担的责任进行描述的文字(例如:相关人员A承担主要责任、相关人员B承担次要责任等)对应的字符串。
其中,字符串检测模型可以是常用的CNN模型等机器学习模型。
类似地,可以将被标注了包含目标字符串的图像区域的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:CNN算法),针对这些图像数据样本进行训练,以得到用于从上述单据图像的图像数据中检测出包含目标字符串的图 像区域的字符串检测模型。
而在获取到包含目标字符串的图像区域后,则可以继续将该图像区域输入至字符串识别模型,以由该字符串识别模型对该图像区域中的目标字符串进行识别,得到该理赔案件的相关人员的责任信息。
其中,字符串识别模型可以是基于CTC损失函数的RNN模型。
类似地,可以将被标注了字符串对应的文字内容的包含该字符串的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:基于CTC损失函数的RNN算法),针对这些图像数据样本进行训练,以得到用于对上述图像区域中的目标字符串进行识别的字符串识别模型。
进一步地,在得到该理赔案件的相关人员的责任信息后,可以基于NPL算法对该理赔案件的相关人员的责任信息进行分析,得到该理赔案件的相关人员的责任比例,并将得到的该理赔案件的相关人员的责任比例作为用于理赔的关键信息。
举例来说,在识别得到的该理赔案件的相关人员的责任信息包括“相关人员A承担主要责任”和“相关人员B承担次要责任”时,可以基于NPL算法对该责任信息进行分析,从而可以确定相关人员A的责任比例大于50%,相关人员B的责任比例小于50%。
需要说明的是,用于从上述证件图像的图像数据中检测出包含目标字符串的图像区域的字符串检测模型,与用于从上述单据图像的图像数据中检测出包含目标字符串的图像区域的字符串检测模型可以是同一个字符串检测模型,也可以不同的两个字符串检测模型,本说明书对此不作限定。同样地,用于对证件图像的图像区域中的目标字符串进行识别的字符串识别模型,与用于对单据图像的图像区域中的目标字符串进行识别的字符串识别模型可以是同一个字符串识别模型,也可以不同的两个字符串识别模型,本说明书对此不作限定。
在示出的一种实施方式中,针对与上述理赔案件相关的图像数据集合中的现场图像的图像数据,可以将分类得到的现场图像的图像数据输入至预设的分类模型(称为第二分类模型)中进行分类计算。
其中,第二分类模型可以是常用的CNN模型等机器学习模型。
类似地,可以将被标注了事故类型的现场图像的图像数据作为训练样本,采用反向传播的方式,基于预设的机器学习算法(例如:CNN算法),针对这些现场图像样本进行训练,以得到用于确定现场图像的图像数据对应的事故类型的上述第二分类模 型。
其中,事故类型可以包括单车事故,双车事故,多车事故等。
举例来说,可以将作为训练样本的现场图像中显示的车祸事故仅包含一辆车辆的图像标注为“单车事故”,将这些现场图像中显示的车祸事故包含两辆车辆的图像标注为“双车事故”,将这些现场图像中显示的车祸事故包含三辆及以上车辆的图像标注为“多车事故”。
这样,可以基于已训练好的上述第二分类模型对上述现场图像的图像数据进行分类计算,从而可以基于分类计算结果对该现场图像的图像数据进行分类,即确定该现场图像对应的事故所属的事故类型。
在实际应用中,还可以对从各个类别的图像数据中分别提取出的用于理赔的关键信息加以利用。具体地,可以获取提取出的用于理赔的关键信息,并基于该用于理赔的关键信息进行理赔处理。
举例来说,在获取到该用于理赔的关键信息后,可以将该用于理赔的关键信息输入至装载在电子设备上的理赔系统,以由该理赔系统自行录入该用于理赔的关键信息作为上述理赔案件的信息,从而可以由该理赔系统基于该理赔案件的信息进行后续的理赔处理。
在实际应用中,该理赔系统还可以按照上述分类结果对与该理赔案件相关的图像数据集合进行分类存储,即将该图像数据集合中的图像数据按照其所属的图像类别进行分类存储。
在上述技术方案中,对于某个理赔案件而言,可以将与该理赔案件相关的图像数据集合输入至分类模型,以由该分类模型对该图像数据集合中的图像数据进行分类。后续,可以自动从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。采用这样的方式,与常用的针对理赔案件的图像数据进行人工分类和分析的方式相比,可以提高理赔信息的提取效率,减少人力资源的消耗。
与前述理赔信息提取方法的实施例相对应,本说明书还提供了理赔信息提取装置的实施例。
本说明书理赔信息提取装置的实施例可以应用在电子设备上。装置实施例可以通过软件实现,也可以通过硬件或者软硬件结合的方式实现。以软件实现为例,作为一个逻辑意义上的装置,是通过其所在电子设备的处理器将非易失性存储器中对应的计算 机程序指令读取到内存中运行形成的。从硬件层面而言,如图2所示,为本说明书理赔信息提取装置所在电子设备的一种硬件结构图,除了图2所示的处理器、内存、网络接口、以及非易失性存储器之外,实施例中装置所在的电子设备通常根据该理赔信息提取的实际功能,还可以包括其他硬件,对此不再赘述。
请参考图3,图3是本说明书一示例性实施例示出的一种理赔信息提取装置的框图。该装置30可以应用于图2所示的电子设备,包括:
第一获取模块301,用于获取与理赔案件相关的图像数据集合;
分类模块302,用于将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
提取模块303,用于从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
在本实施例中,所述装置30还可以包括:
第二获取模块304,用于获取提取出的所述用于理赔的关键信息;
理赔模块305,用于基于所述用于理赔的关键信息进行理赔处理。
在本实施例中,所述第一分类模型可以为卷积神经网络CNN模型。
在本实施例中,对所述图像数据集合中的图像数据进行分类得到的图像类别可以包括以下图像类别中的一个或多个:
证件图像;单据图像;现场图像;损伤图像;其他图像。
在本实施例中,所述提取模块303具体可以用于:
如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
在本实施例中,所述提取模块303具体可以用于:
如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于 理赔的关键信息。
在本实施例中,所述提取模块303具体可以用于:
如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
在本实施例中,所述第二分类模型可以为CNN模型。
上述装置中各个模块的功能和作用的实现过程具体详见上述方法中对应步骤的实现过程,在此不再赘述。
对于装置实施例而言,由于其基本对应于方法实施例,所以相关之处参见方法实施例的部分说明即可。以上所描述的装置实施例仅仅是示意性的,其中所述作为分离部件说明的模块可以是或者也可以不是物理上分开的,作为模块显示的部件可以是或者也可以不是物理模块,即可以位于一个地方,或者也可以分布到多个网络模块上。可以根据实际的需要选择其中的部分或者全部模块来实现本说明书方案的目的。本领域普通技术人员在不付出创造性劳动的情况下,即可以理解并实施。
上述实施例阐明的系统、装置、模块或模块,具体可以由计算机芯片或实体实现,或者由具有某种功能的产品来实现。一种典型的实现设备为计算机,计算机的具体形式可以是个人计算机、膝上型计算机、蜂窝电话、相机电话、智能电话、个人数字助理、媒体播放器、导航设备、电子邮件收发设备、游戏控制台、平板计算机、可穿戴设备或者这些设备中的任意几种设备的组合。
与上述理赔信息提取方法实施例相对应,本说明书还提供了一种电子设备的实施例。该电子设备包括:处理器以及用于存储机器可执行指令的存储器;其中,处理器和存储器通常通过内部总线相互连接。在其他可能的实现方式中,所述设备还可能包括外部接口,以能够与其他设备或者部件进行通信。
在本实施例中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
获取与理赔案件相关的图像数据集合;
将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基 于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
在本实施例中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器还被促使:
获取提取出的所述用于理赔的关键信息;
基于所述用于理赔的关键信息进行理赔处理。
在本实施例中,所述第一分类模型为卷积神经网络CNN模型。
在本实施例中,对所述图像数据集合中的图像数据进行分类得到的图像类别包括以下图像类别中的一个或多个:
证件图像;单据图像;现场图像;损伤图像;其他图像。
在本实施例中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
在本实施例中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于理赔的关键信息。
在本实施例中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
在本实施例中,所述第二分类模型为CNN模型。
本领域技术人员在考虑说明书及实践这里公开的发明后,将容易想到本说明书的其它实施方案。本说明书旨在涵盖本说明书的任何变型、用途或者适应性变化,这些变型、用途或者适应性变化遵循本说明书的一般性原理并包括本说明书未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的,本说明书的真正范围和精神由下面的权利要求指出。
应当理解的是,本说明书并不局限于上面已经描述并在附图中示出的精确结构,并且可以在不脱离其范围进行各种修改和改变。本说明书的范围仅由所附的权利要求来限制。
以上所述仅为本说明书一个或多个实施例的较佳实施例而已,并不用以限制本说明书一个或多个实施例,凡在本说明书一个或多个实施例的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本说明书一个或多个实施例保护的范围之内。

Claims (17)

  1. 一种理赔信息提取方法,所述方法包括:
    获取与理赔案件相关的图像数据集合;
    将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
    从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
  2. 根据权利要求1所述的方法,所述方法还包括:
    获取提取出的所述用于理赔的关键信息;
    基于所述用于理赔的关键信息进行理赔处理。
  3. 根据权利要求1所述的方法,所述第一分类模型为卷积神经网络CNN模型。
  4. 根据权利要求1所述的方法,对所述图像数据集合中的图像数据进行分类得到的图像类别包括以下图像类别中的一个或多个:
    证件图像;单据图像;现场图像;损伤图像;其他图像。
  5. 根据权利要求4所述的方法,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
    如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
  6. 根据权利要求4所述的方法,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
    如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于理赔的关键信息。
  7. 根据权利要求4所述的方法,所述从分类得到的各个类别的图像数据中分别提取用于理赔的关键信息,包括:
    如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
  8. 根据权利要求7所述的方法,所述第二分类模型为CNN模型。
  9. 一种理赔信息提取装置,所述装置包括:
    第一获取模块,用于获取与理赔案件相关的图像数据集合;
    分类模块,用于将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
    提取模块,用于从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
  10. 根据权利要求9所述的装置,所述装置还包括:
    第二获取模块,用于获取提取出的所述用于理赔的关键信息;
    理赔模块,用于基于所述用于理赔的关键信息进行理赔处理。
  11. 根据权利要求9所述的装置,所述第一分类模型为卷积神经网络CNN模型。
  12. 根据权利要求9所述的装置,对所述图像数据集合中的图像数据进行分类得到的图像类别包括以下图像类别中的一个或多个:
    证件图像;单据图像;现场图像;损伤图像;其他图像。
  13. 根据权利要求12所述的装置,所述提取模块具体用于:
    如果分类得到所述证件图像的图像数据,则基于光学字符识别OCR算法,从所述证件图像的图像数据中提取所述理赔案件的相关人员信息和相关车辆信息作为用于理赔的关键信息。
  14. 根据权利要求12所述的装置,所述提取模块具体用于:
    如果分类得到所述单据图像的图像数据,则基于OCR算法和自然语言处理NLP算法,从所述单据图像的图像数据中提取所述理赔案件的相关人员的责任比例作为用于理赔的关键信息。
  15. 根据权利要求12所述的装置,所述提取模块具体用于:
    如果分类得到所述现场图像的图像数据,则将所述现场图像的图像数据输入至第二分类模型中进行分类计算,并基于分类结果确定与所述现场图像的图像数据对应的事故类型,以将所述事故类型作为用于理赔的关键信息;其中,所述第二分类模型为基于若干被标注了事故类型的现场图像样本训练出的机器学习模型。
  16. 根据权利要求15所述的装置,所述第二分类模型为CNN模型。
  17. 一种电子设备,所述电子设备包括:
    处理器;
    用于存储机器可执行指令的存储器;
    其中,通过读取并执行所述存储器存储的与理赔信息提取的控制逻辑对应的机器可执行指令,所述处理器被促使:
    获取与理赔案件相关的图像数据集合;
    将所述图像数据集合中的图像数据输入至第一分类模型中进行分类计算,并基于分类计算结果对所述图像数据集合中的图像数据进行分类;其中,所述第一分类模型为基于若干被标注了图像类别的图像数据样本训练出的机器学习模型;
    从分类得到的各个图像类别的图像数据中分别提取用于理赔的关键信息。
PCT/CN2019/119352 2019-01-31 2019-11-19 理赔信息提取方法和装置、电子设备 Ceased WO2020155790A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910097463.7 2019-01-31
CN201910097463.7A CN109903172A (zh) 2019-01-31 2019-01-31 理赔信息提取方法和装置、电子设备

Publications (1)

Publication Number Publication Date
WO2020155790A1 true WO2020155790A1 (zh) 2020-08-06

Family

ID=66944493

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/119352 Ceased WO2020155790A1 (zh) 2019-01-31 2019-11-19 理赔信息提取方法和装置、电子设备

Country Status (3)

Country Link
CN (1) CN109903172A (zh)
TW (1) TWI712980B (zh)
WO (1) WO2020155790A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112132527A (zh) * 2020-08-07 2020-12-25 精英数智科技股份有限公司 基于企业监测数据的理赔查勘方法及装置
CN114298845A (zh) * 2021-11-29 2022-04-08 泰康保险集团股份有限公司 一种理赔票据处理方法和装置
US20230142507A1 (en) * 2021-11-08 2023-05-11 Ping An Technology (Shenzhen) Co., Ltd. Systems and methods for estimating monetary loss to an accident damaged vehicle
CN116342300A (zh) * 2023-05-26 2023-06-27 凯泰铭科技(北京)有限公司 一种保险理赔人员特征分析方法、装置和设备
CN118505400A (zh) * 2024-05-06 2024-08-16 国网福建省电力有限公司 基于深度学习的电力杆塔保险理赔报告生成方法、系统、设备及介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109903172A (zh) * 2019-01-31 2019-06-18 阿里巴巴集团控股有限公司 理赔信息提取方法和装置、电子设备
CN111401438B (zh) * 2020-03-13 2023-08-25 德联易控科技(北京)有限公司 图像分拣方法、装置及系统
CN111681116A (zh) * 2020-04-27 2020-09-18 中国平安财产保险股份有限公司 理赔资料核实方法、装置、设备及存储介质
CN111680693A (zh) * 2020-05-28 2020-09-18 泰康保险集团股份有限公司 理赔业务批量处理方法及装置
CN114881671B (zh) * 2022-05-17 2024-10-22 飞鸟鱼信息科技有限公司 理赔数据的处理方法及装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780048A (zh) * 2016-11-28 2017-05-31 中国平安财产保险股份有限公司 一种智能车险的自助理赔方法、自助理赔装置及系统
CN107220648A (zh) * 2017-04-11 2017-09-29 平安科技(深圳)有限公司 理赔单据的字符识别方法及服务器
CN107292749A (zh) * 2017-08-04 2017-10-24 平安科技(深圳)有限公司 车损证件照片的分类方法、系统及可读存储介质
US20170351934A1 (en) * 2015-03-05 2017-12-07 Omron Corporation Object recognition device, object recognition method, and program
CN107610091A (zh) * 2017-07-31 2018-01-19 阿里巴巴集团控股有限公司 车险图像处理方法、装置、服务器及系统
CN109903172A (zh) * 2019-01-31 2019-06-18 阿里巴巴集团控股有限公司 理赔信息提取方法和装置、电子设备

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20160011916A (ko) * 2014-07-23 2016-02-02 삼성전자주식회사 얼굴 인식을 통한 사용자 식별 방법 및 장치
CN108986468A (zh) * 2018-08-01 2018-12-11 平安科技(深圳)有限公司 交通事故的处理方法、装置、计算机设备及计算机存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170351934A1 (en) * 2015-03-05 2017-12-07 Omron Corporation Object recognition device, object recognition method, and program
CN106780048A (zh) * 2016-11-28 2017-05-31 中国平安财产保险股份有限公司 一种智能车险的自助理赔方法、自助理赔装置及系统
CN107220648A (zh) * 2017-04-11 2017-09-29 平安科技(深圳)有限公司 理赔单据的字符识别方法及服务器
CN107610091A (zh) * 2017-07-31 2018-01-19 阿里巴巴集团控股有限公司 车险图像处理方法、装置、服务器及系统
CN107292749A (zh) * 2017-08-04 2017-10-24 平安科技(深圳)有限公司 车损证件照片的分类方法、系统及可读存储介质
CN109903172A (zh) * 2019-01-31 2019-06-18 阿里巴巴集团控股有限公司 理赔信息提取方法和装置、电子设备

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112132527A (zh) * 2020-08-07 2020-12-25 精英数智科技股份有限公司 基于企业监测数据的理赔查勘方法及装置
US20230142507A1 (en) * 2021-11-08 2023-05-11 Ping An Technology (Shenzhen) Co., Ltd. Systems and methods for estimating monetary loss to an accident damaged vehicle
US12148043B2 (en) * 2021-11-08 2024-11-19 Ping An Technology (Shenzhen) Co., Ltd Systems and methods for estimating monetary loss to an accident damaged vehicle
CN114298845A (zh) * 2021-11-29 2022-04-08 泰康保险集团股份有限公司 一种理赔票据处理方法和装置
CN116342300A (zh) * 2023-05-26 2023-06-27 凯泰铭科技(北京)有限公司 一种保险理赔人员特征分析方法、装置和设备
CN118505400A (zh) * 2024-05-06 2024-08-16 国网福建省电力有限公司 基于深度学习的电力杆塔保险理赔报告生成方法、系统、设备及介质

Also Published As

Publication number Publication date
TWI712980B (zh) 2020-12-11
CN109903172A (zh) 2019-06-18
TW202030683A (zh) 2020-08-16

Similar Documents

Publication Publication Date Title
TWI712980B (zh) 理賠資訊提取方法和裝置、電子設備
CN109858555B (zh) 基于图像的数据处理方法、装置、设备及可读存储介质
CN111695392B (zh) 基于级联的深层卷积神经网络的人脸识别方法及系统
CN109409241B (zh) 视频核验方法、装置、设备及可读存储介质
CN107977633A (zh) 人脸图像的年龄识别方法、装置及存储介质
CN114093022A (zh) 活动检测装置、活动检测系统及活动检测方法
WO2019127273A1 (zh) 一种多人脸检测方法、装置、服务器、系统及存储介质
WO2020164278A1 (zh) 一种图像处理方法、装置、电子设备和可读存储介质
CN113221918B (zh) 目标检测方法、目标检测模型的训练方法及装置
WO2018166116A1 (zh) 车损识别方法、电子装置及计算机可读存储介质
WO2017220032A1 (zh) 基于深度学习的车牌分类方法、系统、电子装置及存储介质
TW202125332A (zh) 一種目標運動軌跡構建方法、設備以及計算機存儲介質
CN108241836A (zh) 用于安检的方法及装置
CN110033011A (zh) 车祸事故处理方法和装置、电子设备
WO2019056503A1 (zh) 门店监控评价方法、装置及存储介质
CN110717407A (zh) 基于唇语密码的人脸识别方法、装置及存储介质
CN105760488A (zh) 基于多层次特征融合的图像表达方法和装置
CN114663871A (zh) 图像识别方法、训练方法、装置、系统及存储介质
CN110222582B (zh) 一种图像处理方法和相机
US10423817B2 (en) Latent fingerprint ridge flow map improvement
EP4388507A1 (en) Training models for object detection
CN115359575A (zh) 身份识别方法、装置和计算机设备
CN115953744A (zh) 一种基于深度学习的车辆识别追踪方法
CN112613471A (zh) 人脸活体检测方法、装置及计算机可读存储介质
WO2025020700A1 (zh) 故障确定方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19912485

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19912485

Country of ref document: EP

Kind code of ref document: A1