WO2019061662A1 - 电子装置、投保牲畜识别方法和计算机可读存储介质 - Google Patents

电子装置、投保牲畜识别方法和计算机可读存储介质 Download PDF

Info

Publication number
WO2019061662A1
WO2019061662A1 PCT/CN2017/108769 CN2017108769W WO2019061662A1 WO 2019061662 A1 WO2019061662 A1 WO 2019061662A1 CN 2017108769 W CN2017108769 W CN 2017108769W WO 2019061662 A1 WO2019061662 A1 WO 2019061662A1
Authority
WO
WIPO (PCT)
Prior art keywords
photo
face
animal
training
preset
Prior art date
Application number
PCT/CN2017/108769
Other languages
English (en)
French (fr)
Inventor
王健宗
王晨羽
马进
肖京
Original Assignee
平安科技(深圳)有限公司
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 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2019061662A1 publication Critical patent/WO2019061662A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR 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

Definitions

  • the present application relates to the field of insurance, and in particular to an electronic device, an insured livestock identification method, and a computer readable storage medium.
  • animal husbandry is one of the important sources of human access to food
  • the pig industry is an important component of the livestock industry.
  • livestock death is a frequent occurrence.
  • livestock death it will usually cause huge economic losses for these farmers. It may less inhibit the enthusiasm of potential farmers to engage in animal husbandry, and cause potential obstacles to the development of animal husbandry; on the other hand, increase the possibility of farmers reducing the probability of death of animals through abnormal routes (for example, drug control pathways). This poses a great real threat to the food safety.
  • the main object of the present application is to provide an electronic device, an insured livestock identification method, and a computer readable storage medium, aimed at realizing a low cost, high efficiency, and remote batch identification livestock identification scheme.
  • a first aspect of the present application provides an electronic device comprising a memory, a processor having stored thereon an insured livestock identification system operable on the processor, the insured livestock identification system being executed by the processor Implement the following steps:
  • the determined similarity is greater than the preset threshold, it is determined that the face recognition of the animal passes, or if the determined similarity is less than or equal to the preset threshold, determining that the face recognition of the animal fails.
  • a second aspect of the present application provides a method for identifying an insured animal, the method comprising the steps of:
  • the determined similarity is greater than the preset threshold, it is determined that the face recognition of the animal passes, or if the determined similarity is less than or equal to the preset threshold, determining that the face recognition of the animal fails.
  • a third aspect of the present application provides a computer readable storage medium storing an insured animal identification system, the insured livestock identification system being executable by at least one processor to cause the at least one processor Perform the following steps:
  • the determined similarity is greater than the preset threshold, it is determined that the face recognition of the animal passes, or if the determined similarity is less than or equal to the preset threshold, determining that the face recognition of the animal fails.
  • the technical solution of the present application determines the standard corresponding to the identity identifier from the associated data of the system through the identification of the insured animal in the claim application and the current facial photo of the animal after receiving the claim for the claim for the injured animal.
  • a photo of the animal's face and inputting the current facial photo of the animal and the standard animal face photo into a pre-trained preset type recognition model to derive a current facial photo of the animal and a determined standard animal face
  • the similarity of the partial photos and then comparing the similarity with a preset threshold, and confirming whether the facial recognition passes according to the comparison result, thereby determining whether the current facial photograph of the animal and the determined standard livestock facial photograph belong to the same animal.
  • the user when the livestock is out of danger, the user only needs to send the claim application with the current face photo and identity of the dangerous animal to the system, and the insurance company's system is based on the claim sent by the user.
  • the accompanying current facial photos and identifications identify and verify the claims of the animals, without on-site verification and identification, and can quickly and quickly process claims, reducing costs and improving efficiency. Rate, and can be remotely identified in batches, which is more convenient.
  • FIG. 1 is a schematic flow chart of an embodiment of a method for identifying an insured livestock according to the present application
  • FIG. 2 is a training flowchart of a preset type recognition model in an embodiment of the method for identifying an insured livestock according to the present application;
  • FIG. 3 is a schematic diagram of an operating environment of an embodiment of an insured livestock identification system of the present application.
  • FIG. 4 is a block diagram of a program of an embodiment of the insured livestock identification system of the present application.
  • FIG. 1 is a schematic flow chart of an embodiment of an insured livestock identification method according to the present application.
  • the method for identifying the insured livestock includes:
  • Step S10 after receiving the claim with the identity of the insured animal and the current facial photo of the animal, determining the corresponding identifier according to the pre-stored identity and the associated data of the standard livestock face photo.
  • Standard livestock face photo
  • an insurance company insures its livestock (for example, pigs, raises, cattle, etc.), it needs to provide a standard face photo for each animal to be insured to the insurance company, and the insurance company provides for each insured animal separately.
  • the unique identity ID i.e., the identity of the insured animal, such as a numerical number
  • the identity of each insured animal is associated with its standard livestock face photo within the system to store the associated data.
  • the user has a sick animal dying, the user sends a claim to the insurance company with the current facial photo and identity of the dead animal, and the insurance company's system receives the claim sent by the user after receiving the claim. Extracting the identity mark and the current face photo in the claim application, and determining the standard livestock face photo corresponding to the extracted identity mark according to the association data of the pre-stored identity mark and the standard livestock face photo in the system. .
  • Step S20 inputting the current face photo of the animal and the determined standard livestock face photo into the pre-trained preset type recognition model, and determining that the current face photo of the animal is similar to the determined standard livestock face photo. degree;
  • the system has a pre-trained preset type recognition model for identifying the similarity of the matching facial photos; after determining the standard livestock face photo corresponding to the identification in the claim application, the system will The current facial photograph of the animal and the determined standard livestock facial photograph are input into the pre-trained preset type recognition model, thereby obtaining the current facial photograph of the animal and the determined standard livestock face according to the preset type identification model output. The similarity of the photos.
  • step S30 if the determined similarity is greater than the preset threshold, it is determined that the face recognition of the animal passes, or if the determined similarity is less than or equal to the preset threshold, it is determined that the face recognition of the animal fails.
  • a judgment threshold having a preset similarity result in the system ie, a preset threshold, for example, 90%
  • a preset threshold for example 90%
  • determining the current face photo of the animal The determined standard livestock face photograph is a photograph of the face of the same animal, and the face recognition of the animal passes, and the insurance company can determine that the animal is an insured animal.
  • the similarity result output by the preset type recognition model is less than or equal to the preset threshold, it is determined that the face recognition of the animal has failed.
  • the identity of the insured animal in the claim application and the current facial photo of the animal are determined, and the corresponding identifier of the identity is determined from the associated data of the system.
  • a standard animal face photograph and inputting the current facial photograph of the animal and the standard livestock facial photograph into a pre-trained preset type recognition model to obtain a current facial photograph of the animal and the determined standard livestock
  • the similarity of the face photo then comparing the similarity with a preset threshold, and confirming whether the face recognition passes according to the comparison result, thereby determining whether the current face photo of the animal and the determined standard livestock face photo belong to the same animal .
  • the user when the livestock is out of danger, the user only needs to send the claim application with the current face photo and identity of the dangerous animal to the system, and the insurance company's system is based on the claim sent by the user.
  • the attached current facial photos and identifications identify and verify the claims of the animals, without on-site verification and identification, can quickly and quickly process claims, reduce costs, improve efficiency, and remote batch identification, more convenient.
  • the preset type recognition model adopted by the embodiment is a twin neural network model (Siamese network model), and the preset type recognition model includes a first sub-network model, a second sub-network model, and a result calculation module. among them:
  • the first sub-network model (for example, a convolutional neural network model) is configured to perform feature extraction on the current facial photo of the animal, and output a first feature vector;
  • the second sub-network model (for example, a convolutional neural network model) is configured to perform feature extraction on the determined standard livestock face photo, and output a second feature vector;
  • the result calculation module is configured to calculate a vector distance between the first feature vector and the second feature vector according to a predetermined feature vector distance calculation function, where the vector distance is the current face photo of the animal and the determined standard livestock The similarity of face photos.
  • step S20 includes:
  • the first sub-network model performs feature extraction on the current facial photo of the animal, and outputs a first feature vector
  • the second sub-network model performs feature extraction on the determined standard livestock face photo, and outputs a second feature vector
  • the result calculation module calculates a vector distance between the first feature vector and the second feature vector according to a predetermined feature vector distance calculation function (ie, the similarity between the current face photo of the animal and the determined standard livestock two photos) ).
  • E W (X1, X2) ⁇ G W (X1) -G W (X2) ⁇ ;
  • G W (X) represents a network mapping function of the preset type identification model, and the parameter is W, and the parameter W is trained by the preset category identification model; the network mapping function G W (X) is directed to Two different input feature vectors X1 and X2, respectively output low-dimensional space results are G W (X1) and G W (X2), G W (X1) is obtained by X1 through network mapping, and G W (X2) is Obtained by X2 through network mapping.
  • the model Given a network mapping function G W (X), the model is identified by training the preset type to find a set of parameters W that satisfy: when X1 and X2 belong to the same animal face, features
  • FIG. 2 is a training flowchart of a preset type recognition model in an embodiment of the method for identifying an insured livestock.
  • the training process of the preset type recognition model is as follows:
  • Step E1 obtaining a preset number of face photos of the insured livestock and a preset number of facial photos of the claiming livestock;
  • the preset number is 100,000, that is, a photograph of the face of 100,000 insured animals and a photograph of the face of 100,000 claims of the insured animal (ie, a photograph of the face of the animal in which the insured event occurred).
  • step E2 all the acquired facial photos are randomly paired to obtain a preset number of facial photo pairs, and the first photo label is attached to the facial photo pair belonging to the same animal, and the facial photo pair not belonging to the same livestock is marked.
  • Second label
  • a preset number of face photos of the insured livestock and a preset number of face photos of the claiming livestock are randomly paired to obtain a preset number of face photo pairs (each of the face photo pairs) Both include a photo of the face of two animals).
  • the identity of the face photos of the two animals in the face photo it can be confirmed whether the two face photos of each face photo pair are the face photos of the same animal, and the face photos belonging to the same animal are
  • Step E3 dividing the face photo into a first percentage training set and a second hundred a verification set of the ratio, the sum of the first percentage and the second percentage being less than or equal to 100%;
  • Step E4 training the preset type recognition model by using a photo in the training set, and verifying the accuracy of the preset type recognition model of the training by using the verification set after the training is completed;
  • the preset type recognition model is trained by using the face photo of the training set, and after the preset type recognition model is trained, the preset type recognition model is performed by using the face photo of the verification set.
  • the accuracy rate verification obtains the accuracy of the preset type recognition model after the training set is completed.
  • step E5 if the accuracy rate is greater than the preset threshold, the model training ends;
  • the verification threshold of the accuracy rate (that is, the preset threshold, for example, 98.5%) is preset in the system, and is used to check the training effect of the preset type identification model; if the preset is set by the verification set If the accuracy of the type identification model verification is greater than the preset threshold, then the training of the preset type recognition model reaches the expected standard, and the model training is ended.
  • step E6 if the accuracy rate is less than or equal to the preset threshold, the number of samples of the face photo is increased, and the above steps E2, E3, and E4 are re-executed based on the samples of the increased face photo.
  • the accuracy of the verification of the preset type identification model by the verification set is less than or equal to the preset threshold, it indicates that the training of the preset type recognition model has not reached the expected standard, and may be the number of training sets. Not enough or the number of verification sets is not enough, so in this case, increase the number of samples of the face photo (for example, increase the fixed number each time or increase the random number each time), and then re-execute the above based on this Steps E2, E3 and E4 are cyclically executed until the requirement of step E5 is reached, and the model training is ended.
  • the present application also proposes an insured livestock identification system.
  • FIG. 3 is a schematic diagram of the operating environment of the preferred embodiment of the insured livestock identification system 10 of the present application.
  • the insured livestock identification system 10 is installed and operated in the electronic device 1.
  • the electronic device 1 can be a desktop computer, a notebook, a palmtop computer, a server, etc. Computing device.
  • the electronic device 1 may include, but is not limited to, a memory 11, a processor 12, and a display 13.
  • Figure 3 shows only the electronic device 1 with components 11-13, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a hard disk or memory of the electronic device 1.
  • the memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (SMC), and a secure digital (SD). Card, flash card, etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 is used to store application software and various types of data installed in the electronic device 1, such as program codes of the insured livestock identification system 10.
  • the memory 11 can also be used to temporarily store data that has been output or is about to be output.
  • the processor 12 in some embodiments, may be a Central Processing Unit (CPU), microprocessor or other data processing chip for running program code or processing data stored in the memory 11, such as performing insured livestock identification. System 10 and so on.
  • CPU Central Processing Unit
  • microprocessor or other data processing chip for running program code or processing data stored in the memory 11, such as performing insured livestock identification. System 10 and so on.
  • the display 13 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch sensor, or the like in some embodiments.
  • the display 13 is for displaying information processed in the electronic device 1 and a user interface for displaying visualization, such as a business customization interface or the like.
  • the components 11-13 of the electronic device 1 communicate with one another via a system bus.
  • FIG. 4 is a block diagram of a program of an embodiment of the insured livestock identification system 10 of the present application.
  • the insured livestock identification system 10 can be partitioned into one or more modules, one or more modules being stored in the memory 11 and being processed by one or more processors (the processor 12 in this embodiment) Executed to complete the application.
  • the insured livestock identification system 10 can be segmented into a first determination module 101, a second determination module 102, and a determination module 103.
  • a module referred to in this application refers to a series of computer program instructions that are capable of performing a particular function, and are more suitable than the program for describing the execution of the insured livestock identification system 10 in the electronic device 1, wherein:
  • the first determining module 101 is configured to determine, according to the association data of the pre-stored identity identifier and the standard livestock face photo, after receiving the claim with the identity of the insured animal and the current facial photo of the animal a standard livestock face photo corresponding to the identity;
  • an insurance company insures its livestock (for example, pigs, raises, cattle, etc.), it needs to provide a standard face photo for each animal to be insured to the insurance company, and the insurance company provides for each insured animal separately.
  • the unique identity ID ie the identity of the insured animal, such as a numeric number
  • the identity of the livestock is associated with its standard livestock face photo, and the associated data is stored.
  • the user has a sick animal dying, the user sends a claim to the insurance company with the current facial photo and identity of the dead animal, and the insurance company's system receives the claim sent by the user after receiving the claim. Extracting the identity mark and the current face photo in the claim application, and determining the standard livestock face photo corresponding to the extracted identity mark according to the association data of the pre-stored identity mark and the standard livestock face photo in the system. .
  • the second determining module 102 is configured to input the current facial photo of the animal and the determined standard animal facial photo into the pre-trained preset type recognition model, determine the current facial photo of the animal and the determined standard livestock The similarity of facial photos;
  • the system has a pre-trained preset type recognition model for identifying the similarity of the matching facial photos; after determining the standard livestock face photo corresponding to the identification in the claim application, the system will The current facial photograph of the animal and the determined standard livestock facial photograph are input into the pre-trained preset type recognition model, thereby obtaining the current facial photograph of the animal and the determined standard livestock face according to the preset type identification model output. The similarity of the photos.
  • the determining module 103 is configured to determine that the face recognition of the animal passes when the determined similarity is greater than the preset threshold, or determine that the face recognition of the animal fails when the determined similarity is less than or equal to the preset threshold.
  • a judgment threshold having a preset similarity result in the system ie, a preset threshold, for example, 90%
  • a preset threshold for example 90%
  • determining the current face photo of the animal The determined standard livestock face photograph is a photograph of the face of the same animal, and the face recognition of the animal passes, and the insurance company can determine that the animal is an insured animal.
  • the similarity result output by the preset type recognition model is less than or equal to the preset threshold, it is determined that the face recognition of the animal has failed.
  • the identity of the insured animal in the claim application and the current facial photo of the animal are determined, and the corresponding identifier of the identity is determined from the associated data of the system.
  • a standard animal face photograph and inputting the current facial photograph of the animal and the standard livestock facial photograph into a pre-trained preset type recognition model to obtain a current facial photograph of the animal and the determined standard livestock
  • the similarity of the face photo then comparing the similarity with a preset threshold, and confirming whether the face recognition passes according to the comparison result, thereby determining whether the current face photo of the animal and the determined standard livestock face photo belong to the same animal .
  • the user when the livestock is out of danger, the user only needs to send the claim application with the current face photo and identity of the dangerous animal to the system, and the insurance company's system is based on the claim sent by the user.
  • the attached current facial photos and identifications identify and verify the claims of the animals, without on-site verification and identification, can quickly and quickly process claims, reduce costs, improve efficiency, and remote batch identification, more convenient.
  • the preset type recognition model adopted by the embodiment is a twin neural network model (Siamese network model), and the preset type recognition model includes a first sub-network model, a second sub-network model, and a result calculation module. among them:
  • the first sub-network model (for example, a convolutional neural network model) is configured to perform feature extraction on the current facial photo of the animal, and output a first feature vector;
  • the second sub-network model (for example, a convolutional neural network model) is configured to perform feature extraction on the determined standard livestock face photo, and output a second feature vector;
  • the result calculation module is configured to calculate a vector distance between the first feature vector and the second feature vector according to a predetermined feature vector distance calculation function, where the vector distance is the current face photo of the animal and the determined standard livestock The similarity of face photos.
  • the second determining module 102 includes:
  • a first extraction sub-module configured to control the first sub-network model to perform feature extraction on a current facial photo of the animal, and output a first feature vector
  • a first extraction sub-module configured to control the second sub-network model to perform feature extraction on the determined standard livestock facial photo, and output a second feature vector
  • a predetermined feature vector distance calculation function ie, a current face photo of the animal and a determined standard The similarity between the two photos of the animals.
  • G W (X) represents a network mapping function of the preset type identification model, and the parameter is W, and the parameter W is trained by the preset category identification model; the network mapping function G W (X) is directed to Two different input feature vectors X1 and X2, respectively output low-dimensional space results are G W (X1) and G W (X2), G W (X1) is obtained by X1 through network mapping, and G W (X2) is Obtained by X2 through network mapping.
  • the model Given a network mapping function G W (X), the model is identified by training the preset type to find a set of parameters W that satisfy: when X1 and X2 belong to the same animal face, features
  • the training process of the preset type recognition model is as follows:
  • Step E1 obtaining a preset number of face photos of the insured livestock and a preset number of facial photos of the claiming livestock;
  • the preset number is 100,000, that is, a photograph of the face of 100,000 insured animals and a photograph of the face of 100,000 claims of the insured animal (ie, a photograph of the face of the animal in which the insured event occurred).
  • step E2 all the acquired facial photos are randomly paired to obtain a preset number of facial photo pairs, and the first photo label is attached to the facial photo pair belonging to the same animal, and the facial photo pair not belonging to the same livestock is marked.
  • Second label
  • a preset number of face photos of the insured livestock and a preset number of face photos of the claiming livestock are randomly paired to obtain a preset number of face photo pairs (each of the face photo pairs) Both include a photo of the face of two animals).
  • the identity of the face photos of the two animals in the face photo it can be confirmed whether the two face photos of each face photo pair are the face photos of the same animal, and the face photos belonging to the same animal are
  • Step E3 dividing the facial photo into a first percentage training set and a second percentage verification set, wherein the sum of the first percentage and the second percentage is less than or equal to 100% ;
  • Step E4 training the preset type recognition model by using a photo in the training set, and verifying the accuracy of the preset type recognition model of the training by using the verification set after the training is completed;
  • the preset type recognition model is trained by using the face photo of the training set, and after the preset type recognition model is trained, the preset type recognition model is performed by using the face photo of the verification set.
  • the accuracy rate verification obtains the accuracy of the preset type recognition model after the training set is completed.
  • step E5 if the accuracy rate is greater than the preset threshold, the model training ends;
  • the verification threshold of the accuracy rate (that is, the preset threshold, for example, 98.5%) is preset in the system, and is used to check the training effect of the preset type identification model; if the preset is set by the verification set If the accuracy of the type identification model verification is greater than the preset threshold, then the training of the preset type recognition model reaches the expected standard, and the model training is ended.
  • step E6 if the accuracy rate is less than or equal to the preset threshold, the facial photo is increased.
  • the number of samples of the slice, and the above steps E2, E3, E4 are re-executed based on the sample of the increased face photo.
  • the accuracy of the verification of the preset type identification model by the verification set is less than or equal to the preset threshold, it indicates that the training of the preset type recognition model has not reached the expected standard, and may be the number of training sets. Not enough or the number of verification sets is not enough, so in this case, increase the number of samples of the face photo (for example, increase the fixed number each time or increase the random number each time), and then re-execute the above based on this Steps E2, E3 and E4 are cyclically executed until the requirement of step E5 is reached, and the model training is ended.
  • the present application further provides a computer readable storage medium storing an insured livestock identification system, the insured livestock identification system being executable by at least one processor to cause the at least one process The insured livestock identification method of any of the above embodiments is performed.

Landscapes

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

Abstract

本申请公开一种投保牲畜识别方法、电子装置及存储介质,所述方法包括:在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。本申请技术方案实现了低成本、高效率且可远程批量的对出险牲畜识别。

Description

电子装置、投保牲畜识别方法和计算机可读存储介质
本申请基于巴黎公约申明享有2017年9月30日递交的申请号为CN 201710914928.4、名称为“电子装置、投保牲畜识别方法和计算机可读存储介质”中国专利申请的优先权,该中国专利申请的整体内容以参考的方式结合在本申请中。
技术领域
本申请涉及保险领域,特别涉及一种电子装置、投保牲畜识别方法和计算机可读存储介质。
背景技术
目前,畜牧业是人类获取食物的重要来源之一,而养猪业是畜牧业中的重要组成部门。在牲畜养殖过程中,牲畜生病死亡是一个经常发生的事件,对大多数养殖户而言,若发生牲畜生病死亡事件,则通常会造成这些养殖户巨大的经济损失,这种风险一方面或多或少抑制了潜在养殖户投身畜牧业的积极性,给畜牧业的发展造成潜在的阻碍;另一方面增加了养殖户通过非正常途径(例如,药物控制途径)降低牲畜生病死亡概率的可能性,从而对该食品安全构成极大的现实威胁。
为了最大程度降低这种风险带来的影响,很多保险公司推出了牲畜险,以保险的方式为养殖户规避这种风险。为了配合畜牧险的开展,目前出现了许多识别被保牲畜身份的识别方案,例如,为被投保的猪植入芯片、DNA识别、打耳标等方式对被保的猪进行身份识别,但这类现有识别方案成本较高、效率低下、无法远程批量识别。
发明内容
本申请的主要目的是提供一种电子装置、投保牲畜识别方法和计算机可读存储介质,旨在实现低成本、高效率且可远程批量识别的牲畜识别方案。
本申请第一方面提供一种电子装置,包括存储器、处理器,所述存储器上存储有可在所述处理器上运行的投保牲畜识别系统,所述投保牲畜识别系统被所述处理器执行时实现如下步骤:
在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
本申请第二方面提供一种投保牲畜识别方法,该方法包括步骤:
在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
本申请第三方面提供一种计算机可读存储介质,所述计算机可读存储介质存储有投保牲畜识别系统,所述投保牲畜识别系统可被至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
本申请技术方案,在收到出险牲畜的理赔申请后,通过理赔申请中的被保险的牲畜的身份标识和该牲畜的当前脸部照片,从系统的关联数据中确定出该身份标识对应的标准牲畜脸部照片,并将该牲畜当前的脸部照片与所述标准牲畜脸部照片输入预先训练好的预设类型识别模型中,以得出该牲畜的当前脸部照片与确定的标准牲畜脸部照片的相似度,然后,将该相似度与预设阈值比较,根据比较结果确认脸部识别是否通过,从而确定该牲畜的当前脸部照片与确定的标准牲畜脸部照片是否属于同一牲畜。与现有技术相比,本方案在牲畜出险时,用户只需向系统发送带有出险牲畜的当前脸部照片和身份标识的理赔申请即可,保险公司的系统则根据用户发送的理赔申请中附带的当前脸部照片和身份标识对理赔申请的牲畜的进行识别验证,无需现场验证识别,能够及时快速的处理理赔申请,降低了成本,提升了效 率,并且可以远程批量识别,更加的方便。
附图说明
图1为本申请投保牲畜识别方法一实施例的流程示意图;
图2为本申请投保牲畜识别方法一实施例中预设类型识别模型的训练流程图;
图3为本申请投保牲畜识别系统一实施例的运行环境示意图;
图4为本申请投保牲畜识别系统一实施例的程序模块图。
具体实施方式
以下结合附图对本申请的原理和特征进行描述,所举实例只用于解释本申请,并非用于限定本申请的范围。
如图1所示,图1为本申请投保牲畜识别方法一实施例的流程示意图。
本实施例中,该投保牲畜识别方法包括:
步骤S10,在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
用户在保险公司为其牲畜(例如,猪、养、牛等)投保时,需要针对每一个要投保的牲畜分别提供标准脸部照片给保险公司,保险公司则针对每一个被保险的牲畜分别提供唯一的身份ID(即被保险的牲畜的身份标识,例如数字编号)给用户,并在系统内将每一个投保的牲畜的身份标识与其标准牲畜脸部照片关联,将关联数据进行存储。当用户有被保险的牲畜发生生病死亡情况时,用户向保险公司发送带有该死亡牲畜的当前脸部照片和身份标识的理赔申请,保险公司的系统在接收到用户发送的该理赔申请后,提取出该理赔申请中的身份标识和当前脸部照片,并根据系统中预先存储的身份标识与标准牲畜脸部照片的关联数据,确定出该提取出的身份标识所对应的标准牲畜脸部照片。
步骤S20,将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
系统中具有预先训练好的预设类型识别模型,该模型用于识别比对脸部照片的相似度;系统在确定了该理赔申请中的身份标识所对应的标准牲畜脸部照片后,将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,从而根据预设类型识别模型输出得到该牲畜的当前脸部照片与确定的标准牲畜脸部照片的相似度。
步骤S30,若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
系统中具有预先设置的相似度结果的判断阈值(即预设阈值,例如90%),当预设类型识别模型输出的相似度结果大于该预设阈值,那么判定该牲畜的当前脸部照片与确定的标准牲畜脸部照片为同一个牲畜的脸部照片,该牲畜的脸部识别通过,保险公司则可确定该牲畜为被保险的牲畜。反之,当预设类型识别模型输出的相似度结果小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
本实施例技术方案,在收到出险牲畜的理赔申请后,通过理赔申请中的被保险的牲畜的身份标识和该牲畜的当前脸部照片,从系统的关联数据中确定出该身份标识对应的标准牲畜脸部照片,并将该牲畜当前的脸部照片与所述标准牲畜脸部照片输入预先训练好的预设类型识别模型中,以得出该牲畜的当前脸部照片与确定的标准牲畜脸部照片的相似度,然后,将该相似度与预设阈值比较,根据比较结果确认脸部识别是否通过,从而确定该牲畜的当前脸部照片与确定的标准牲畜脸部照片是否属于同一牲畜。与现有技术相比,本方案在牲畜出险时,用户只需向系统发送带有出险牲畜的当前脸部照片和身份标识的理赔申请即可,保险公司的系统则根据用户发送的理赔申请中附带的当前脸部照片和身份标识对理赔申请的牲畜的进行识别验证,无需现场验证识别,能够及时快速的处理理赔申请,降低了成本,提升了效率,并且可以远程批量识别,更加的方便。
优选地,本实施例采用的所述预设类型识别模型为孪生神经网络模型(Siamese网络模型),所述预设类型识别模型包括第一子网络模型、第二子网络模型及结果计算模块,其中:
该第一子网络模型(例如,卷积神经网络模型),用于对该牲畜的当前脸部照片进行特征提取,输出第一特征向量;
该第二子网络模型(例如,卷积神经网络模型),用于对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
该结果计算模块,用于根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离,该向量距离即为该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度。
也就是,步骤S20包括:
所述第一子网络模型对该牲畜当前脸部照片进行特征提取,输出第一特征向量;
所述第二子网络模型对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
所述结果计算模块根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离(即该牲畜的当前脸部照片与确定的标准牲畜两部照片的相似度)。
本实施例优选所述预先确定的特征向量距离计算函数为:EW(X1,X2)=‖GW(X1)-GW(X2)‖;
其中,GW(X)代表所述预设类型识别模型的网络映射函数,其参数为W,该参数W由所述预设类别识别模型训练得出;该网络映射函数GW(X)针对两个不同的输入特征向量X1和X2,分别输出低维空间结果为GW(X1)和GW(X2),GW(X1)是由X1经过网络映射得到的,GW(X2)是由X2经过网络映射得到的。
给定网络映射函数GW(X),通过训练所述预设类型识别模型,以找出一组参数W,该组参数W满足:使当X1和X2属于同一个牲畜脸部的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖小于第一阈值,以及当X1和X2属于不同的类别的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖大于第二阈值;所述第一阈值小于或者等于所述第二阈值。
如图2所示,图2为本申请投保牲畜识别方法一实施例中预设类型识别模型的训练流程图。
在本实施例中,所述预设类型识别模型的训练过程如下:
步骤E1,获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
例如,预设数量为10万个,即获取10万个投保牲畜的脸部照片以及10万个出险理赔牲畜的脸部照片(即发生保险事故的牲畜的脸部照片)。
步骤E2,将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
将获取的预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片进行两两随机配对,从而得到预设数量的脸部照片对(每个所述脸部照片对均包括两张牲畜的脸部照片)。根据脸部照片对中两张牲畜的脸部照片分别对应的身份标识,可确认各个脸部照片对的两张脸部照片是否为同一牲畜的脸部照片,给属于同一牲畜的脸部照片对标注第一标签(例如,label=1),以及给不属于同一牲畜的脸部照片对标注第二标签(例如,label=-1),以供制作训练集或验证集。
步骤E3,将所述脸部照片对分为第一百分比的训练集和第二百 分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
从所述脸部照片对中分出一个训练集和一个验证集,所述训练集和验证集分别占所述脸部照片对的第一百分比和第二百分比,所述第一百分比和第二百分比之和小于或者等于100%,即可以是将整个所述脸部照片对刚好分成所述训练集和验证集(例如,所述第一百分比为70%,所述第二百分比为30%),也可以是将所述脸部照片对的一部分分成所述训练集和验证集(例如,所述第一百分比为65%,所述第二百分比为25%)。
步骤E4,利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
用所述训练集中的脸部照片对训练所述预设类型识别模型,训练完所述预设类型识别模型后,再用所述验证集中的脸部照片对对所述预设类型识别模型进行准确率验证,得到经所述训练集训练完成后的所述预设类型识别模型的准确率。
步骤E5,若准确率大于预设阈值,则模型训练结束;
系统中预先设置了准确率的验证阈值(即所述预设阈值,例如98.5%),用于对所述预设类型识别模型的训练效果进行检验;若通过所述验证集对所述预设类型识别模型验证得到的准确率大于所述预设阈值,那么说明该预设类型识别模型的训练达到了预期标准,此时则结束模型训练。
步骤E6,若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
若是通过所述验证集对所述预设类型识别模型验证得到的准确率小于或等于所述预设阈值,那么说明该预设类型识别模型的训练还没有达到了预期标准,可能是训练集数量不够或验证集数量不够,所以,在这种情况时,则增加所述脸部照片的样本数量(例如,每次增加固定数量或每次增加随机数量),然后在这基础上,重新执行上述步骤E2、E3和E4,如此循环执行,直至达到了步骤E5的要求,则结束模型训练。
此外,本申请还提出一种投保牲畜识别系统。
请参阅图3,是本申请投保牲畜识别系统10较佳实施例的运行环境示意图。
在本实施例中,投保牲畜识别系统10安装并运行于电子装置1中。电子装置1可以是桌上型计算机、笔记本、掌上电脑及服务器等 计算设备。该电子装置1可包括,但不仅限于,存储器11、处理器12及显示器13。图3仅示出了具有组件11-13的电子装置1,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
存储器11在一些实施例中可以是电子装置1的内部存储单元,例如该电子装置1的硬盘或内存。存储器11在另一些实施例中也可以是电子装置1的外部存储设备,例如电子装置1上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。进一步地,存储器11还可以既包括电子装置1的内部存储单元也包括外部存储设备。存储器11用于存储安装于电子装置1的应用软件及各类数据,例如投保牲畜识别系统10的程序代码等。存储器11还可以用于暂时地存储已经输出或者将要输出的数据。
处理器12在一些实施例中可以是一中央处理器(Central Processing Unit,CPU),微处理器或其他数据处理芯片,用于运行存储器11中存储的程序代码或处理数据,例如执行投保牲畜识别系统10等。
显示器13在一些实施例中可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode,有机发光二极管)触摸器等。显示器13用于显示在电子装置1中处理的信息以及用于显示可视化的用户界面,例如业务定制界面等。电子装置1的部件11-13通过系统总线相互通信。
请参阅图4,是本申请投保牲畜识别系统10一实施例的程序模块图。在本实施例中,投保牲畜识别系统10可以被分割成一个或多个模块,一个或者多个模块被存储于存储器11中,并由一个或多个处理器(本实施例为处理器12)所执行,以完成本申请。例如,在图4中,投保牲畜识别系统10可以被分割成第一确定模块101、第二确定模块102及判断模块103。本申请所称的模块是指能够完成特定功能的一系列计算机程序指令段,比程序更适合于描述投保牲畜识别系统10在电子装置1中的执行过程,其中:
第一确定模块101,用于在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
用户在保险公司为其牲畜(例如,猪、养、牛等)投保时,需要针对每一个要投保的牲畜分别提供标准脸部照片给保险公司,保险公司则针对每一个被保险的牲畜分别提供唯一的身份ID(即被保险的牲畜的身份标识,例如数字编号)给用户,并在系统内将每一个投保 的牲畜的身份标识与其标准牲畜脸部照片关联,将关联数据进行存储。当用户有被保险的牲畜发生生病死亡情况时,用户向保险公司发送带有该死亡牲畜的当前脸部照片和身份标识的理赔申请,保险公司的系统在接收到用户发送的该理赔申请后,提取出该理赔申请中的身份标识和当前脸部照片,并根据系统中预先存储的身份标识与标准牲畜脸部照片的关联数据,确定出该提取出的身份标识所对应的标准牲畜脸部照片。
第二确定模块102,用于将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
系统中具有预先训练好的预设类型识别模型,该模型用于识别比对脸部照片的相似度;系统在确定了该理赔申请中的身份标识所对应的标准牲畜脸部照片后,将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,从而根据预设类型识别模型输出得到该牲畜的当前脸部照片与确定的标准牲畜脸部照片的相似度。
判断模块103,用于在确定的相似度大于预设阈值时,确定该牲畜的脸部识别通过,或者,在确定的相似度小于或者等于预设阈值时,确定该牲畜的脸部识别失败。
系统中具有预先设置的相似度结果的判断阈值(即预设阈值,例如90%),当预设类型识别模型输出的相似度结果大于该预设阈值,那么判定该牲畜的当前脸部照片与确定的标准牲畜脸部照片为同一个牲畜的脸部照片,该牲畜的脸部识别通过,保险公司则可确定该牲畜为被保险的牲畜。反之,当预设类型识别模型输出的相似度结果小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
本实施例技术方案,在收到出险牲畜的理赔申请后,通过理赔申请中的被保险的牲畜的身份标识和该牲畜的当前脸部照片,从系统的关联数据中确定出该身份标识对应的标准牲畜脸部照片,并将该牲畜当前的脸部照片与所述标准牲畜脸部照片输入预先训练好的预设类型识别模型中,以得出该牲畜的当前脸部照片与确定的标准牲畜脸部照片的相似度,然后,将该相似度与预设阈值比较,根据比较结果确认脸部识别是否通过,从而确定该牲畜的当前脸部照片与确定的标准牲畜脸部照片是否属于同一牲畜。与现有技术相比,本方案在牲畜出险时,用户只需向系统发送带有出险牲畜的当前脸部照片和身份标识的理赔申请即可,保险公司的系统则根据用户发送的理赔申请中附带的当前脸部照片和身份标识对理赔申请的牲畜的进行识别验证,无需现场验证识别,能够及时快速的处理理赔申请,降低了成本,提升了效率,并且可以远程批量识别,更加的方便。
优选地,本实施例采用的所述预设类型识别模型为孪生神经网络模型(Siamese网络模型),所述预设类型识别模型包括第一子网络模型、第二子网络模型及结果计算模块,其中:
该第一子网络模型(例如,卷积神经网络模型),用于对该牲畜的当前脸部照片进行特征提取,输出第一特征向量;
该第二子网络模型(例如,卷积神经网络模型),用于对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
该结果计算模块,用于根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离,该向量距离即为该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度。
也就是,第二确定模块102包括:
第一提取子模块,用于控制所述第一子网络模型对该牲畜当前脸部照片进行特征提取,输出第一特征向量;
第一提取子模块,用于控制所述第二子网络模型对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
确定子模块,用于控制所述结果计算模块根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离(即该牲畜的当前脸部照片与确定的标准牲畜两部照片的相似度)。
本实施例优选所述预先确定的特征向量距离计算函数为:EW(X1,X2)=‖GW(X1)-GW(X2)‖;
其中,GW(X)代表所述预设类型识别模型的网络映射函数,其参数为W,该参数W由所述预设类别识别模型训练得出;该网络映射函数GW(X)针对两个不同的输入特征向量X1和X2,分别输出低维空间结果为GW(X1)和GW(X2),GW(X1)是由X1经过网络映射得到的,GW(X2)是由X2经过网络映射得到的。
给定网络映射函数GW(X),通过训练所述预设类型识别模型,以找出一组参数W,该组参数W满足:使当X1和X2属于同一个牲畜脸部的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖小于第一阈值,以及当X1和X2属于不同的类别的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖大于第二阈值;所述第一阈值小于或者等于所述第二阈值。
进一步地,在本实施例中,所述预设类型识别模型的训练过程如下:
步骤E1,获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
例如,预设数量为10万个,即获取10万个投保牲畜的脸部照片以及10万个出险理赔牲畜的脸部照片(即发生保险事故的牲畜的脸部照片)。
步骤E2,将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
将获取的预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片进行两两随机配对,从而得到预设数量的脸部照片对(每个所述脸部照片对均包括两张牲畜的脸部照片)。根据脸部照片对中两张牲畜的脸部照片分别对应的身份标识,可确认各个脸部照片对的两张脸部照片是否为同一牲畜的脸部照片,给属于同一牲畜的脸部照片对标注第一标签(例如,label=1),以及给不属于同一牲畜的脸部照片对标注第二标签(例如,label=-1),以供制作训练集或验证集。
步骤E3,将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
从所述脸部照片对中分出一个训练集和一个验证集,所述训练集和验证集分别占所述脸部照片对的第一百分比和第二百分比,所述第一百分比和第二百分比之和小于或者等于100%,即可以是将整个所述脸部照片对刚好分成所述训练集和验证集(例如,所述第一百分比为70%,所述第二百分比为30%),也可以是将所述脸部照片对的一部分分成所述训练集和验证集(例如,所述第一百分比为65%,所述第二百分比为25%)。
步骤E4,利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
用所述训练集中的脸部照片对训练所述预设类型识别模型,训练完所述预设类型识别模型后,再用所述验证集中的脸部照片对对所述预设类型识别模型进行准确率验证,得到经所述训练集训练完成后的所述预设类型识别模型的准确率。
步骤E5,若准确率大于预设阈值,则模型训练结束;
系统中预先设置了准确率的验证阈值(即所述预设阈值,例如98.5%),用于对所述预设类型识别模型的训练效果进行检验;若通过所述验证集对所述预设类型识别模型验证得到的准确率大于所述预设阈值,那么说明该预设类型识别模型的训练达到了预期标准,此时则结束模型训练。
步骤E6,若准确率小于或者等于预设阈值,则增加所述脸部照 片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
若是通过所述验证集对所述预设类型识别模型验证得到的准确率小于或等于所述预设阈值,那么说明该预设类型识别模型的训练还没有达到了预期标准,可能是训练集数量不够或验证集数量不够,所以,在这种情况时,则增加所述脸部照片的样本数量(例如,每次增加固定数量或每次增加随机数量),然后在这基础上,重新执行上述步骤E2、E3和E4,如此循环执行,直至达到了步骤E5的要求,则结束模型训练。
进一步地,本申请还提出一种计算机可读存储介质,所述计算机可读存储介质存储有投保牲畜识别系统,所述投保牲畜识别系统可被至少一个处理器执行,以使所述至少一个处理器执行上述任一实施例中的投保牲畜识别方法。
以上所述仅为本发明的优选实施例,并非因此限制本发明的专利范围,凡是在本发明的发明构思下,利用本发明说明书及附图内容所作的等效结构变换,或直接/间接运用在其他相关的技术领域均包括在本发明的专利保护范围内。

Claims (20)

  1. 一种电子装置,其特征在于,所述电子装置包括存储器、处理器,所述存储器上存储有可在所述处理器上运行的投保牲畜识别系统,所述投保牲畜识别系统被所述处理器执行时实现如下步骤:
    在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
    将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
    若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
  2. 如权利要求1所述的电子装置,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  3. 如权利要求1所述的电子装置,其特征在于,所述预设类型识别模型为孪生神经网络模型,所述预设类型识别模型包括第一子网络模型、第二子网络模型及结果计算模块,其中:
    该第一子网络模型,用于对该牲畜的当前脸部照片进行特征提取,输出第一特征向量;
    该第二子网络模型,用于对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
    该结果计算模块,用于根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离,该向量距离即为该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度。
  4. 如权利要求3所述的电子装置,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  5. 如权利要求3所述的电子装置,其特征在于,所述预先确定的特征向量距离计算函数为:EW(X1,X2)=‖GW(X1)-GW(X2)‖;
    其中,GW(X)代表所述预设类型识别模型的网络映射函数,其参数为W,该参数W由所述预设类别识别模型训练得出;该网络映射函数GW(X)针对两个不同的输入特征向量X1和X2,分别输出低维空间结果为GW(X1)和GW(X2),GW(X1)是由X1经过网络映射得到的,GW(X2)是由X2经过网络映射得到的。
  6. 如权利要求5所述的电子装置,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的 验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  7. 如权利要求5所述的电子装置,所述网络映射函数GW(X)的参数W满足:使得当X1和X2属于同一个牲畜脸部的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖小于第一阈值;当X1和X2属于不同的类别的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖大于第二阈值;所述第一阈值小于或者等于第二阈值。
  8. 如权利要求7所述的电子装置,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  9. 一种投保牲畜识别方法,其特征在于,该方法包括步骤:
    在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
    将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预 先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
    若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
  10. 如权利要求9所述的投保牲畜识别方法,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  11. 如权利要求9所述的投保牲畜识别方法,其特征在于,所述预设类型识别模型为孪生神经网络模型,所述预设类型识别模型包括第一子网络模型、第二子网络模型及结果计算模块,其中:
    该第一子网络模型,用于对该牲畜的当前脸部照片进行特征提取,输出第一特征向量;
    该第二子网络模型,用于对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
    该结果计算模块,用于根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离,该向量距离即为该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度。
  12. 如权利要求11所述的投保牲畜识别方法,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  13. 如权利要求11所述的投保牲畜识别方法,其特征在于,所述预先确定的特征向量距离计算函数为:EW(X1,X2)=‖GW(X1)-GW(X2)‖;
    其中,GW(X)代表所述预设类型识别模型的网络映射函数,其参数为W,该参数W由所述预设类别识别模型训练得出;该网络映射函数GW(X)针对两个不同的输入特征向量X1和X2,分别输出低维空间结果为GW(X1)和GW(X2),GW(X1)是由X1经过网络映射得到的,GW(X2)是由X2经过网络映射得到的。
  14. 如权利要求13所述的投保牲畜识别方法,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、 E4。
  15. 如权利要求13所述的投保牲畜识别方法,所述网络映射函数GW(X)的参数W满足:使得当X1和X2属于同一个牲畜脸部的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖小于第一阈值;当X1和X2属于不同的类别的时候,特征向量距离度量EW(X1,X2)=‖GW(X1)-GW(X2)‖大于第二阈值;所述第一阈值小于或者等于第二阈值。
  16. 如权利要求15所述的投保牲畜识别方法,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
  17. 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质存储有投保牲畜识别系统,所述投保牲畜识别系统可被至少一个处理器执行,以使所述至少一个处理器执行如下步骤:
    在收到带有被保险的牲畜的身份标识和该牲畜的当前脸部照片的理赔申请后,根据预先存储的身份标识和标准牲畜脸部照片的关联数据,确定该身份标识对应的标准牲畜脸部照片;
    将该牲畜的当前脸部照片和确定的标准牲畜脸部照片输入到预先训练好的预设类型识别模型中,确定该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度;
    若确定的相似度大于预设阈值,则确定该牲畜的脸部识别通过,或者,若确定的相似度小于或者等于预设阈值,则确定该牲畜的脸部识别失败。
  18. 如权利要求17所述的计算机可读存储介质,其特征在于,所述预设类型识别模型为孪生神经网络模型,所述预设类型识别模型包括第一子网络模型、第二子网络模型及结果计算模块,其中:
    该第一子网络模型,用于对该牲畜的当前脸部照片进行特征提取,输出第一特征向量;
    该第二子网络模型,用于对确定的标准牲畜脸部照片进行特征提取,输出第二特征向量;
    该结果计算模块,用于根据预先确定的特征向量距离计算函数计算出所述第一特征向量与第二特征向量的向量距离,该向量距离即为该牲畜的当前脸部照片和确定的标准牲畜脸部照片的相似度。
  19. 如权利要求18所述的计算机可读存储介质,其特征在于,所述预先确定的特征向量距离计算函数为:EW(X1,X2)=‖GW(X1)-GW(X2)‖;
    其中,GW(X)代表所述预设类型识别模型的网络映射函数,其参数为W,该参数W由所述预设类别识别模型训练得出;该网络映射函数GW(X)针对两个不同的输入特征向量X1和X2,分别输出低维空间结果为GW(X1)和GW(X2),GW(X1)是由X1经过网络映射得到的,GW(X2)是由X2经过网络映射得到的。
  20. 如权利要求19所述的计算机可读存储介质,其特征在于,所述预设类型识别模型的训练过程如下:
    E1、获取预设数量的投保牲畜的脸部照片和预设数量的出险理赔牲畜的脸部照片;
    E2、将获取的所有脸部照片进行两两随机配对获得预设数量的脸部照片对,对属于同一牲畜的脸部照片对标注第一标签,对不属于同一牲畜的脸部照片对标注第二标签;
    E3、将所述脸部照片对分为第一百分比的训练集和第二百分比的验证集,所述第一百分比和第二百分比之和小于或者等于100%;
    E4、利用训练集中的照片对对所述预设类型识别模型进行训练,并在训练完成后利用验证集对训练的所述预设类型识别模型的准确率进行验证;
    E5、若准确率大于预设阈值,则模型训练结束;
    E6、若准确率小于或者等于预设阈值,则增加所述脸部照片的样本数量,并基于增加后的脸部照片的样本重新执行上述步骤E2、E3、E4。
PCT/CN2017/108769 2017-09-30 2017-10-31 电子装置、投保牲畜识别方法和计算机可读存储介质 WO2019061662A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710914928.4A CN107766807A (zh) 2017-09-30 2017-09-30 电子装置、投保牲畜识别方法和计算机可读存储介质
CN201710914928.4 2017-09-30

Publications (1)

Publication Number Publication Date
WO2019061662A1 true WO2019061662A1 (zh) 2019-04-04

Family

ID=61267818

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/108769 WO2019061662A1 (zh) 2017-09-30 2017-10-31 电子装置、投保牲畜识别方法和计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN107766807A (zh)
WO (1) WO2019061662A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111144378A (zh) * 2019-12-30 2020-05-12 众安在线财产保险股份有限公司 一种目标对象的识别方法及装置
CN112308722A (zh) * 2019-07-28 2021-02-02 四川谦泰仁投资管理有限公司 一种基于红外摄像的养殖业保险申报请求验证系统

Families Citing this family (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108664878A (zh) * 2018-03-14 2018-10-16 广州影子控股股份有限公司 基于卷积神经网络的猪只身份识别方法
CN108389061A (zh) * 2018-03-19 2018-08-10 平安科技(深圳)有限公司 电子装置、基于区块链的牲畜保险理赔方法及存储介质
CN108492198A (zh) * 2018-03-27 2018-09-04 平安科技(深圳)有限公司 牲畜保险理赔方法、装置、计算机设备和存储介质
CN108549860A (zh) * 2018-04-09 2018-09-18 深源恒际科技有限公司 一种基于深度神经网络的牛脸识别方法
CN108830138B (zh) * 2018-04-26 2021-05-07 平安科技(深圳)有限公司 牲畜识别方法、装置及存储介质
CN108764045B (zh) * 2018-04-26 2019-11-26 平安科技(深圳)有限公司 牲畜识别方法、装置及存储介质
CN108921026A (zh) * 2018-06-01 2018-11-30 平安科技(深圳)有限公司 动物身份的识别方法、装置、计算机设备和存储介质
CN108876297A (zh) * 2018-06-14 2018-11-23 平安健康保险股份有限公司 用于商业医保的自动审核方法、装置、设备及存储介质
CN108960258A (zh) * 2018-07-06 2018-12-07 江苏迪伦智能科技有限公司 一种基于自学习深度特征的模板匹配方法
CN111210227A (zh) * 2018-11-22 2020-05-29 重庆小雨点小额贷款有限公司 一种数据处理方法、装置、服务器及计算机可读存储介质
CN109886826B (zh) * 2019-01-04 2023-07-28 平安科技(深圳)有限公司 基于区块链的牲畜养殖流通追踪方法、电子装置及存储介质
CN110610125A (zh) * 2019-07-31 2019-12-24 平安科技(深圳)有限公司 基于神经网络的牛脸识别方法、装置、设备及存储介质
CN110737885A (zh) * 2019-10-16 2020-01-31 支付宝(杭州)信息技术有限公司 豢养物的身份认证方法及装置
CN110705512A (zh) * 2019-10-16 2020-01-17 支付宝(杭州)信息技术有限公司 豢养物身份特征检测方法以及装置
CN110728244B (zh) * 2019-10-16 2022-06-14 蚂蚁胜信(上海)信息技术有限公司 豢养物身份信息采集的引导方法以及装置
CN110909683B (zh) * 2019-11-25 2022-05-03 蚂蚁胜信(上海)信息技术有限公司 基于保障项目的保障核验方法以及装置
CN110956149A (zh) * 2019-12-06 2020-04-03 中国平安财产保险股份有限公司 宠物身份核验方法、装置、设备及计算机可读存储介质
CN114097642A (zh) * 2020-08-27 2022-03-01 卡尤迪智农科技(北京)有限公司 电子耳标、远程监测其拆卸的方法和设备、远程核保方法
CN112541432A (zh) * 2020-12-11 2021-03-23 上海品览数据科技有限公司 一种基于深度学习的视频牲畜身份认证系统及方法
CN115641458B (zh) * 2022-10-14 2023-06-20 吉林鑫兰软件科技有限公司 用于待统计目标养殖的ai识别系统及银行风控应用方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102654903A (zh) * 2011-03-04 2012-09-05 井维兰 一种人脸比对方法
CN103593598A (zh) * 2013-11-25 2014-02-19 上海骏聿数码科技有限公司 基于活体检测和人脸识别的用户在线认证方法及系统
CN103902961A (zh) * 2012-12-28 2014-07-02 汉王科技股份有限公司 一种人脸识别方法及装置
CN104899579A (zh) * 2015-06-29 2015-09-09 小米科技有限责任公司 人脸识别方法和装置
CN107229947A (zh) * 2017-05-15 2017-10-03 邓昌顺 一种基于动物识别的金融保险方法及系统

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105678622A (zh) * 2016-01-07 2016-06-15 平安科技(深圳)有限公司 车险理赔照片的分析方法及系统
CN106778607A (zh) * 2016-12-15 2017-05-31 国政通科技股份有限公司 一种基于人脸识别的人与身份证同一性认证装置及方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102654903A (zh) * 2011-03-04 2012-09-05 井维兰 一种人脸比对方法
CN103902961A (zh) * 2012-12-28 2014-07-02 汉王科技股份有限公司 一种人脸识别方法及装置
CN103593598A (zh) * 2013-11-25 2014-02-19 上海骏聿数码科技有限公司 基于活体检测和人脸识别的用户在线认证方法及系统
CN104899579A (zh) * 2015-06-29 2015-09-09 小米科技有限责任公司 人脸识别方法和装置
CN107229947A (zh) * 2017-05-15 2017-10-03 邓昌顺 一种基于动物识别的金融保险方法及系统

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112308722A (zh) * 2019-07-28 2021-02-02 四川谦泰仁投资管理有限公司 一种基于红外摄像的养殖业保险申报请求验证系统
CN111144378A (zh) * 2019-12-30 2020-05-12 众安在线财产保险股份有限公司 一种目标对象的识别方法及装置
CN111144378B (zh) * 2019-12-30 2023-10-31 众安在线财产保险股份有限公司 一种目标对象的识别方法及装置

Also Published As

Publication number Publication date
CN107766807A (zh) 2018-03-06

Similar Documents

Publication Publication Date Title
WO2019061662A1 (zh) 电子装置、投保牲畜识别方法和计算机可读存储介质
WO2019205375A1 (zh) 牲畜识别方法、装置及存储介质
WO2019109526A1 (zh) 人脸图像的年龄识别方法、装置及存储介质
WO2015165365A1 (zh) 一种人脸识别方法及系统
WO2019179035A1 (zh) 电子装置、基于区块链的牲畜保险理赔方法及存储介质
WO2019105163A1 (zh) 目标人物的搜索方法和装置、设备、程序产品和介质
WO2019196303A1 (zh) 用户身份验证方法、服务器及存储介质
WO2022105179A1 (zh) 生物特征图像识别方法、装置、电子设备及可读存储介质
CN110163096B (zh) 人物识别方法、装置、电子设备和计算机可读介质
WO2018090641A1 (zh) 识别保险单号码的方法、装置、设备及计算机可读存储介质
CN110866443B (zh) 人像存储方法、人脸识别方法、设备及存储介质
WO2019041524A1 (zh) 聚类标签生成方法、电子设备及计算机可读存储介质
CN113707300A (zh) 基于人工智能的搜索意图识别方法、装置、设备及介质
WO2019085338A1 (zh) 电子装置、基于图像的年龄分类方法、系统及存储介质
CN113887438A (zh) 人脸图像的水印检测方法、装置、设备及介质
CN110956149A (zh) 宠物身份核验方法、装置、设备及计算机可读存储介质
CN112070506A (zh) 风险用户识别方法、装置、服务器及存储介质
CN116311370A (zh) 一种基于多角度特征的牛脸识别方法及其相关设备
CN112668575A (zh) 关键信息提取方法、装置、电子设备及存储介质
CN113706249B (zh) 数据推荐方法、装置、电子设备及存储介质
US20130236065A1 (en) Image semantic clothing attribute
CN113268328A (zh) 批处理方法、装置、计算机设备和存储介质
CN116563040A (zh) 基于牲畜识别的农险查勘方法、装置、设备及存储介质
KR20190022430A (ko) 소셜 정보 기반의 리스크 이벤트의 식별 시스템, 방법, 전자장치 및 저장매체
CN114818685B (zh) 关键词提取方法、装置、电子设备及存储介质

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: 17926357

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC , EPO FORM 1205A DATED 28.09.2020.

122 Ep: pct application non-entry in european phase

Ref document number: 17926357

Country of ref document: EP

Kind code of ref document: A1