WO2020082732A1 - Automatic picture classification method, device, and computer readable storage medium - Google Patents

Automatic picture classification method, device, and computer readable storage medium Download PDF

Info

Publication number
WO2020082732A1
WO2020082732A1 PCT/CN2019/088636 CN2019088636W WO2020082732A1 WO 2020082732 A1 WO2020082732 A1 WO 2020082732A1 CN 2019088636 W CN2019088636 W CN 2019088636W WO 2020082732 A1 WO2020082732 A1 WO 2020082732A1
Authority
WO
WIPO (PCT)
Prior art keywords
picture
neural network
recognition model
training
image recognition
Prior art date
Application number
PCT/CN2019/088636
Other languages
French (fr)
Chinese (zh)
Inventor
肖玉宾
杨将
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 WO2020082732A1 publication Critical patent/WO2020082732A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present application relates to the field of image recognition, and in particular, to an automatic picture classification method, device, and computer-readable storage medium.
  • the present application provides an automatic picture classification method, device and computer-readable storage medium, and its main purpose is to improve the efficiency of picture annotation.
  • the present application provides an automatic picture classification method, which is applied to an electronic device.
  • the method includes:
  • the picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
  • the present application also provides an electronic device including a memory and a processor, and the memory stores a picture automatic classification program that can run on the processor, and the picture automatic classification program is used by the processor The following steps are implemented during execution:
  • the picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
  • the present application also provides a computer-readable storage medium on which an automatic picture classification program is stored, which can be executed by one or more processors, In order to realize the steps of the automatic picture classification method described above.
  • the automatic picture classification method, device and computer-readable storage medium provided by the present application recognize and annotate pictures by training a neural network picture recognition model and using the trained neural network picture recognition model, thereby improving the efficiency of picture annotation.
  • FIG. 1 is a schematic flowchart of an automatic picture classification method provided by an embodiment of the present application
  • FIG. 2 is a schematic diagram of an internal structure of an electronic device provided by an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of an Alexnet neural network used by an automatic picture classification method according to an embodiment of the present application
  • FIG. 4 is a schematic diagram of modules of an automatic picture classification program in an electronic device according to an embodiment of the application.
  • FIG. 1 it is a schematic flowchart of an automatic picture classification method provided by an embodiment of the present application.
  • the method may be executed by an apparatus, and the apparatus may be implemented by software and / or hardware.
  • the automatic picture classification method includes:
  • S101 Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID cards, residence permits, residence permits, driver ’s licenses, birth certificates, etc.
  • the preset number is 1000 sheets;
  • S102 Divide the image samples corresponding to each preset image category into a training subset with a first proportion and a verification subset with a second proportion, mix the image samples in each training subset to obtain a training set, and verify each validation The image samples in the subset are mixed to obtain the verification set; for example, the first ratio is 80% and the second ratio is 20%;
  • S105 Receive the current picture input by the user terminal; for example, the current picture is a scanned copy of the front and back of the ID card;
  • S107 Receive a query instruction for a picture category, search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for an ID card category, and find the data of the ID card category And display, to facilitate the quality control staff to review.
  • the method further includes: when the accuracy rate is less than the preset threshold, increase the number of ID picture samples corresponding to each preset picture category, and then re-divide the updated picture samples for each preset picture category into New training subset and validation subset.
  • step S103 includes:
  • the neural network image recognition model is the AlexNet neural network.
  • the neural network image recognition model includes 5 convolutional layers and 3 fully connected layers.
  • the output of the last fully connected layer is 1000.
  • the output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation.
  • the convolutional operation formula of the convolutional layer is:
  • l represents the lth layer
  • j represents the jth kernel of the convolution layer
  • M represents the region where the convolution kernel is located
  • k represents the convolution kernel
  • b represents the offset
  • x represents the value of the corresponding position of the feature map
  • f represents the activation function
  • I, j, l are natural numbers
  • the training set is substituted into the neural network image recognition model for training.
  • the architecture diagram of the AlexNet neural network is shown in FIG. 3, which includes 5 convolutional layers and 3 fully connected layers.
  • step S105 includes:
  • the user enters the ID card information: ID card number and name, and uploads a scanned copy of the front and back of the ID card.
  • y j is the jth neuron in an output vector
  • x i is the ith neuron in the input vector
  • w is a weight parameter in a fully connected layer.
  • the automatic picture classification method proposed in this embodiment recognizes and annotates pictures by training a neural network picture recognition model and uses the trained neural network picture recognition model, thereby improving the efficiency of picture annotation.
  • the present application also provides an electronic device 1.
  • FIG. 2 it is a schematic diagram of an internal structure of an electronic device provided by an embodiment of the present application.
  • the electronic device 1 may be a computer or a server.
  • the electronic device 1 includes at least a memory 11, a processor 13, a communication bus 15, and a network interface 17.
  • the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, and the like.
  • the memory 11 may be an internal storage unit of an electronic device in some embodiments, such as a hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in hard disk equipped on the electronic device, a smart memory card (Smart, Media, Card, SMC), and a secure digital (SD) card. Flash card (Flash Card), etc. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 can be used not only to store application software and various types of data installed in the electronic device 1, such as codes of the automatic picture classification program 111, but also to temporarily store data that has been or will be output.
  • the processor 13 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip for running the program code or processing stored in the memory 11 data.
  • CPU central processing unit
  • controller microcontroller
  • microprocessor or other data processing chip for running the program code or processing stored in the memory 11 data.
  • the communication bus 15 is used to realize connection and communication between these components.
  • the network interface 17 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the electronic device 1 and other electronic devices.
  • a standard wired interface and a wireless interface such as a WI-FI interface
  • the electronic device 1 may further include a user interface
  • the user interface may include a display (Display), an input unit such as a keyboard (Keyboard), and the optional user interface may further include a standard wired interface and a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light emitting diode) touch device, or the like.
  • the display may also be appropriately referred to as a display screen or a display unit, for displaying information processed in the electronic device and for displaying a visual user interface.
  • FIG. 2 only shows the electronic device 1 having the components 11-17. Those skilled in the art can understand that the structure shown in FIG. 2 does not constitute a limitation on the electronic device, and may include fewer or more than the illustration. Components, or a combination of certain components, or different component arrangements.
  • the automatic picture classification program is stored in the memory 11; the processor 13 implements the following steps when executing the electronic device stored in the memory 11:
  • the preset picture categories include ID cards, residence permits, household registration books, driver's licenses, birth certificates, etc.
  • the preset number is 1,000 ;
  • the picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is The image samples are mixed to obtain the verification set; for example, the first ratio is 80%, and the second ratio is 20%;
  • the current picture is a scanned copy of the front and back of the ID card
  • Receive a query instruction for a picture category search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for the ID card category, find the data of the ID card category and display , In order to facilitate the quality control staff to review.
  • the automatic picture classification program when executed by the processor, the following steps are also implemented: when the accuracy rate is less than the preset threshold, the number of certificate picture samples corresponding to each preset picture category is increased, and then each An updated picture sample of a preset picture category is divided into a new training subset and a verification subset.
  • the step of training a neural network image recognition model through the training set includes:
  • the neural network image recognition model is the AlexNet neural network.
  • the neural network image recognition model includes 5 convolutional layers and 3 fully connected layers.
  • the output of the last fully connected layer is 1000.
  • the output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation.
  • the convolutional operation formula of the convolutional layer is:
  • l represents the first layer
  • j represents the jth kernel of the convolution layer
  • M represents the region where the convolution kernel is located
  • k represents the convolution kernel
  • b represents the offset
  • x represents the value of the corresponding position of the feature map
  • f represents the activation function
  • the training set is substituted into the neural network image recognition model for training.
  • the architecture diagram of the AlexNet neural network is shown in FIG. 3, which includes 5 convolutional layers and 3 fully connected layers.
  • the step of receiving the current picture input by the user terminal includes:
  • the user enters the ID card information: ID card number and name, and uploads a scanned copy of the front and back of the ID card.
  • y j is the jth neuron in an output vector
  • x i is the ith neuron in the input vector
  • w is a weight parameter in a fully connected layer.
  • the electronic device proposed in this embodiment recognizes and marks pictures by training a neural network picture recognition model and uses the trained neural network picture recognition model, thereby improving the efficiency of picture marking.
  • embodiments of the present application also provide a computer-readable storage medium that stores an automatic picture classification program stored on the computer-readable storage medium, and the automatic picture classification program may be executed by one or more processors to implement Proceed as follows:
  • the preset picture categories include ID cards, residence permits, household registration books, driver's licenses, birth certificates, etc.
  • the preset number is 1,000 ;
  • the picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is The sample images are mixed to obtain the verification set; for example, the first ratio is 80%, and the second ratio is 20%;
  • the current picture is a scanned copy of the front and back of the ID card
  • Receive a query instruction for a picture category search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for the ID card category, find the data of the ID card category and display , In order to facilitate the quality control staff to review.
  • the automatic picture classification program may also be divided into one or more modules, and the one or more modules are stored in the memory 11 and are processed by one or more processors (this embodiment is The processor 13) is executed to complete this application.
  • the module referred to in this application refers to a series of computer program instruction segments capable of performing specific functions, and is used to describe the execution process of the automatic picture classification program in the electronic device.
  • FIG. 4 it is a schematic diagram of a program module of an automatic picture classification program in an embodiment of an electronic device of the present application.
  • the automatic picture classification program may be divided into a preparation module 10, a division module 20, and a training module 30.
  • the verification module 40, the receiving module 50, the identification module 60, and the search module 70 exemplarily:
  • the preparation module 10 is used to: prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID card, residence permit, household registration book, driver's license, and birth certificate Wait, the preset number is 1000 sheets;
  • the dividing module 20 is used to divide the picture samples corresponding to each preset picture category into a first proportion training subset and a second proportion verification subset, and mix the picture samples in each training subset to obtain training Set, and mix the image samples in each verification subset to obtain the verification set; for example, the first ratio is 80% and the second ratio is 20%;
  • the training module 30 is used to: train a neural network image recognition model through the training set;
  • the verification module 40 is used to verify the accuracy of the neural network image recognition model trained by the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
  • the receiving module 50 is used to: receive the current picture input by the user terminal; for example, the current picture is a scan of the front and back of the ID card;
  • the recognition module 60 is used to: use the trained neural network picture recognition model to perform category recognition and labeling on the current picture, and store the recognition result and the labeling result to the database;
  • the search module 70 is used to: receive a query instruction for a picture category, search for picture information in the database according to the query instruction and display the found picture information; for example, the query instruction is for the ID card category and find all Describe and display the information of the type of ID card to facilitate the quality control officer to review.
  • the above-mentioned preparation module 10, division module 20, training module 30, verification module 40, receiving module 50, identification module 60, and search module 70 and other program modules are implemented when the functions or operation steps are substantially the same as the above-mentioned embodiments, here No longer.
  • the methods in the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases the former is better Implementation.
  • the technical solution of the present application can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product is stored in a storage medium (such as ROM / RAM) , Magnetic disks, optical disks), including several instructions to enable a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to perform the method described in each embodiment of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Image Analysis (AREA)

Abstract

The present application relates to an image recognition technology, and discloses an automatic picture classification method, a device, and a computer readable storage medium. The method comprises: for each preset picture category, preparing a preset number of picture samples labeled with the corresponding picture category; grouping the picture samples corresponding to each preset picture category into a training subset of a first proportion and a verification subset of a second proportion, mixing the picture samples in the training subsets to obtain a training set, and mixing the picture samples in the verification subsets to obtain a verification set; training a neural network picture recognition model by means of the training set; verifying the accuracy of the trained neural network picture recognition model by means of the verification set; receiving the current picture input by a user terminal; performing category recognition and labeling on the current picture by using the trained neural network picture recognition model; and receiving a query instruction, searching a database for picture data according to the query instruction, and displaying the found picture data.

Description

图片自动分类方法、装置及计算机可读存储介质Picture automatic classification method, device and computer readable storage medium
本申请要求于2018年10月26日提交中国专利局，申请号为201811254121.3、发明名称为“图片自动分类方法、装置及计算机可读存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application requires the priority of the Chinese patent application submitted to the China Patent Office on October 26, 2018 with the application number 201811254121.3 and the invention titled "Automatic Picture Classification Method, Device, and Computer-readable Storage Medium" Incorporated in this application.
技术领域Technical field
本申请涉及图像识别领域，尤其涉及一种图片自动分类方法、装置及计算机可读存储介质。The present application relates to the field of image recognition, and in particular, to an automatic picture classification method, device, and computer-readable storage medium.
背景技术Background technique
目前很多事务办理(例如保险)都可以通过网络处理，客户直接使用输入终端打开操作界面将需要的信息及图片资料(通常的图片包括身份证、户口本、驾照、出生证等)上传，接收终端接收到信息及图片资料后，势必需要对所述图片进行标注分类。目前的操作方式是安排质控员在接收终端前对所述图片资料进行标注分类，这种方式势必影响工作效率。另外由于人工处理耗时，拉长了事务处理进度，这样势必影响客户的体验。At present, many transactions (such as insurance) can be processed through the network. The customer directly uses the input terminal to open the operation interface to upload the required information and picture materials (usual pictures include ID card, hukou book, driving license, birth certificate, etc.), and the receiving terminal After receiving the information and picture materials, the pictures must be marked and classified. The current operation method is to arrange the quality control personnel to label and classify the picture materials before receiving the terminal, which will inevitably affect work efficiency. In addition, due to the time-consuming manual processing, the transaction processing progress is lengthened, which will inevitably affect the customer experience.
发明内容Summary of the invention
本申请提供一种图片自动分类方法、装置及计算机可读存储介质，其主要目的在于提高图片标注的效率。The present application provides an automatic picture classification method, device and computer-readable storage medium, and its main purpose is to improve the efficiency of picture annotation.
为实现上述目的，本申请提供一种图片自动分类方法，应用于电子装置中，所述方法包括：To achieve the above purpose, the present application provides an automatic picture classification method, which is applied to an electronic device. The method includes:
为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category;
将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
接收由用户终端输入的当前图片；Receive the current picture input by the user terminal;
利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料。Receiving a query instruction for a picture category, searching for picture data in the database according to the query instruction and displaying the found picture data.
本申请还提供一种电子装置，所述电子装置包括存储器和处理器，所述存储器上存储有可在所述处理器上运行的图片自动分类程序，所述图片自动分类程序被所述处理器执行时实现如下步骤：The present application also provides an electronic device including a memory and a processor, and the memory stores a picture automatic classification program that can run on the processor, and the picture automatic classification program is used by the processor The following steps are implemented during execution:
为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category;
将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
接收由用户终端输入的当前图片；Receive the current picture input by the user terminal;
利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料。Receiving a query instruction for a picture category, searching for picture data in the database according to the query instruction and displaying the found picture data.
此外，为实现上述目的，本申请还提供一种计算机可读存储介质，所述计算机可读存储介质上存储有图片自动分类程序，所述图片自动分类程序可被一个或者多个处理器执行，以实现上述的图片自动分类方法的步骤。In addition, in order to achieve the above object, the present application also provides a computer-readable storage medium on which an automatic picture classification program is stored, which can be executed by one or more processors, In order to realize the steps of the automatic picture classification method described above.
本申请提供的图片自动分类方法、装置及计算机可读存储介质通过训练神经网络图片识别模型并利用训练好的神经网络图片识别模型来识别及标注图片，从而提升图片标注的效率。The automatic picture classification method, device and computer-readable storage medium provided by the present application recognize and annotate pictures by training a neural network picture recognition model and using the trained neural network picture recognition model, thereby improving the efficiency of picture annotation.
附图说明BRIEF DESCRIPTION
图1为本申请一实施例提供的图片自动分类方法的流程示意图；FIG. 1 is a schematic flowchart of an automatic picture classification method provided by an embodiment of the present application;
图2为本申请一实施例提供的电子装置的内部结构示意图；2 is a schematic diagram of an internal structure of an electronic device provided by an embodiment of the present application;
图3为本申请一实施例提供的图片自动分类方法利用到的Alexnet神经网络的结构示意图；FIG. 3 is a schematic structural diagram of an Alexnet neural network used by an automatic picture classification method according to an embodiment of the present application;
图4为本申请一实施例提供的电子装置中图片自动分类程序的模块示意图。4 is a schematic diagram of modules of an automatic picture classification program in an electronic device according to an embodiment of the application.
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。The implementation, functional characteristics and advantages of the present application will be further described in conjunction with the embodiments and with reference to the drawings.
具体实施方式detailed description
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.
本申请提供一种图片自动分类方法。参照图1所示，为本申请一实施例提供的图片自动分类方法的流程示意图。该方法可以由一个装置执行，该装置可以由软件和/或硬件实现。This application provides an automatic picture classification method. Referring to FIG. 1, it is a schematic flowchart of an automatic picture classification method provided by an embodiment of the present application. The method may be executed by an apparatus, and the apparatus may be implemented by software and / or hardware.
在本实施例中，图片自动分类方法包括：In this embodiment, the automatic picture classification method includes:
S101，为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；例如，预设图片类别包括身份证、居住证、户口本、驾照、出生证等，预设数量为1000张；S101. Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID cards, residence permits, residence permits, driver ’s licenses, birth certificates, etc. The preset number is 1000 sheets;
S102，将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；例如第一比例为80％，第二比例为20％；S102: Divide the image samples corresponding to each preset image category into a training subset with a first proportion and a verification subset with a second proportion, mix the image samples in each training subset to obtain a training set, and verify each validation The image samples in the subset are mixed to obtain the verification set; for example, the first ratio is 80% and the second ratio is 20%;
S103，通过所述训练集训练神经网络图片识别模型；S103, training a neural network image recognition model through the training set;
S104，通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；S104. Verify the accuracy of the trained neural network image recognition model through the verification set. When the accuracy is greater than or equal to a preset threshold, the training ends;
S105，接收由用户终端输入的当前图片；例如当前图片为身份证的正反面扫描件；S105. Receive the current picture input by the user terminal; for example, the current picture is a scanned copy of the front and back of the ID card;
S106，利用训练好的神经网络图片识别模型对所述当前图片进行类别识 别及标注，并存储识别结果及标注结果至数据库；S106, using the trained neural network image recognition model to identify and label the current picture, and store the recognition results and the labeling results to the database;
S107，接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料；例如查询指令针对的是身份证类别的，找到所述身份证类别的资料并显示，以方便质控员审查。S107: Receive a query instruction for a picture category, search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for an ID card category, and find the data of the ID card category And display, to facilitate the quality control staff to review.
进一步地，所述方法还包括：当准确率小于所述预设阈值，增加每一个预设图片类别对应的证件图片样本的数量，再重新将每一预设图片类别更新后的图片样本分为新的训练子集及验证子集。Further, the method further includes: when the accuracy rate is less than the preset threshold, increase the number of ID picture samples corresponding to each preset picture category, and then re-divide the updated picture samples for each preset picture category into New training subset and validation subset.
进一步地，所述步骤S103包括：Further, the step S103 includes:
构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
Figure PCTCN2019088636-appb-000001
Figure PCTCN2019088636-appb-000001
其中l表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Where l represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function , I, j, l are natural numbers;
将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
所述AlexNet神经网络的架构示意图如图3所示，包含5个卷积层及3个全连接层。The architecture diagram of the AlexNet neural network is shown in FIG. 3, which includes 5 convolutional layers and 3 fully connected layers.
进一步地，所述步骤S105包括：Further, the step S105 includes:
显示用户输入界面；Display user input interface;
接收由用户输入的所述图片信息；及Receiving the picture information input by the user; and
接收由用户上传的所述图片资料。Receive the picture data uploaded by the user.
例如在用户输入界面上，用户输入身份证的信息：身份证号及名字，并上传身份证的正反面扫描件。For example, on the user input interface, the user enters the ID card information: ID card number and name, and uploads a scanned copy of the front and back of the ID card.
进一步地，所述全连接层表达式为：Further, the fully connected layer expression is:
y j＝f(x i)＝b j+∑ iw ij.x j y j = f (x i ) = b j + ∑ i w ij .x j
其中y j是一个输出向量中的第j个神经元，x i是输入向量中的第i个神经元，w是一个全连层中的权重参数。 Where y j is the jth neuron in an output vector, x i is the ith neuron in the input vector, and w is a weight parameter in a fully connected layer.
本实施例提出的图片自动分类方法通过训练神经网络图片识别模型并利用训练好的神经网络图片识别模型来识别及标注图片，从而提升图片标注的效率。The automatic picture classification method proposed in this embodiment recognizes and annotates pictures by training a neural network picture recognition model and uses the trained neural network picture recognition model, thereby improving the efficiency of picture annotation.
本申请还提供一种电子装置1。参照图2所示，为本申请一实施例提供的电子装置的内部结构示意图。The present application also provides an electronic device 1. Referring to FIG. 2, it is a schematic diagram of an internal structure of an electronic device provided by an embodiment of the present application.
在本实施例中，电子装置1可以是电脑或服务器。所述电子装置1至少包括存储器11、处理器13，通信总线15，以及网络接口17。In this embodiment, the electronic device 1 may be a computer or a server. The electronic device 1 includes at least a memory 11, a processor 13, a communication bus 15, and a network interface 17.
其中，存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如，SD或DX存储器等)、磁性存储器、磁盘、光盘等。存储器11在一些实施例中可以是电子装置的内部存储单元，例如所述电子装置的硬盘。存储器11在另一些实施例中也可以是电子装置的外部存储设备，例如电子装置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的内部存储单元也包括外部存储设备。存储器11不仅可以用于存储安装于电子装置1的应用软件及各类数据，例如图片自动分类程序111的代码等，还可以用于暂时地存储已经输出或者将要输出的数据。Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, and the like. The memory 11 may be an internal storage unit of an electronic device in some embodiments, such as a hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in hard disk equipped on the electronic device, a smart memory card (Smart, Media, Card, SMC), and a secure digital (SD) card. Flash card (Flash Card), etc. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 can be used not only to store application software and various types of data installed in the electronic device 1, such as codes of the automatic picture classification program 111, but also to temporarily store data that has been or will be output.
处理器13在一些实施例中可以是一中央处理器(Central Processing Unit,CPU)、控制器、微控制器、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。In some embodiments, the processor 13 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip for running the program code or processing stored in the memory 11 data.
通信总线15用于实现这些组件之间的连接通信。The communication bus 15 is used to realize connection and communication between these components.
网络接口17可选的可以包括标准的有线接口、无线接口(如WI-FI接口)，通常用于在电子装置1与其他电子设备之间建立通信连接。The network interface 17 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the electronic device 1 and other electronic devices.
可选地，电子装置1还可以包括用户接口，用户接口可以包括显示器(Display)、输入单元比如键盘(Keyboard)，可选的用户接口还可以包括标准的有线接口、无线接口。可选地，在一些实施例中，显示器可以是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。其中，显示器也可以适当的称为显示屏或显示单元，用于显示在电子装置中处理的信息以及用于显示可视化的用户界面。Optionally, the electronic device 1 may further include a user interface, and the user interface may include a display (Display), an input unit such as a keyboard (Keyboard), and the optional user interface may further include a standard wired interface and a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light emitting diode) touch device, or the like. Among them, the display may also be appropriately referred to as a display screen or a display unit, for displaying information processed in the electronic device and for displaying a visual user interface.
图2仅示出了具有组件11～17的电子装置1，本领域技术人员可以理解的是，图2示出的结构并不构成对电子装置的限定，可以包括比图示更少或者更多的部件，或者组合某些部件，或者不同的部件布置。FIG. 2 only shows the electronic device 1 having the components 11-17. Those skilled in the art can understand that the structure shown in FIG. 2 does not constitute a limitation on the electronic device, and may include fewer or more than the illustration. Components, or a combination of certain components, or different component arrangements.
在图2所示的电子装置1的实施例中，存储器11中存储有图片自动分类程序；处理器13执行存储器11中存储的电子装置时实现如下步骤：In the embodiment of the electronic device 1 shown in FIG. 2, the automatic picture classification program is stored in the memory 11; the processor 13 implements the following steps when executing the electronic device stored in the memory 11:
为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；例如，预设图片类别包括身份证、居住证、户口本、驾照、出生证等，预设数量为1000张；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID cards, residence permits, household registration books, driver's licenses, birth certificates, etc. The preset number is 1,000 ;
将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；例如第一比例为80％，第二比例为20％；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is The image samples are mixed to obtain the verification set; for example, the first ratio is 80%, and the second ratio is 20%;
通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
接收由用户终端输入的当前图片；例如当前图片为身份证的正反面扫描件；Receive the current picture input by the user terminal; for example, the current picture is a scanned copy of the front and back of the ID card;
利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料；例如查询指令针对的是身份证类别的，找到所述身份证类别的资料并显示，以方便质控员审查。Receive a query instruction for a picture category, search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for the ID card category, find the data of the ID card category and display , In order to facilitate the quality control staff to review.
进一步地，所述图片自动分类程序被所述处理器执行时还实现如下步骤：当准确率小于所述预设阈值，增加每一个预设图片类别对应的证件图片样本的数量，再重新将每一预设图片类别更新后的图片样本分为新的训练子集及验证子集。Further, when the automatic picture classification program is executed by the processor, the following steps are also implemented: when the accuracy rate is less than the preset threshold, the number of certificate picture samples corresponding to each preset picture category is increased, and then each An updated picture sample of a preset picture category is divided into a new training subset and a verification subset.
进一步地，所述通过所述训练集训练神经网络图片识别模型的步骤包括：Further, the step of training a neural network image recognition model through the training set includes:
构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
Figure PCTCN2019088636-appb-000002
Figure PCTCN2019088636-appb-000002
其中l表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Where l represents the first layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function , I, j, l are natural numbers;
将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
所述AlexNet神经网络的架构示意图如图3所示，包含5个卷积层及3个全连接层。The architecture diagram of the AlexNet neural network is shown in FIG. 3, which includes 5 convolutional layers and 3 fully connected layers.
进一步地，所述接收由用户终端输入的当前图片的步骤包括：Further, the step of receiving the current picture input by the user terminal includes:
显示用户输入界面；Display user input interface;
接收由用户输入的所述图片信息；及Receiving the picture information input by the user; and
接收由用户上传的所述图片资料。Receive the picture data uploaded by the user.
例如在用户输入界面上，用户输入身份证的信息：身份证号及名字，并上传身份证的正反面扫描件。For example, on the user input interface, the user enters the ID card information: ID card number and name, and uploads a scanned copy of the front and back of the ID card.
进一步地，所述图片自动分类程序被所述处理器执行时还实现如下步骤：所述全连接层表达式为：Further, when the automatic picture classification program is executed by the processor, the following steps are also implemented: the fully connected layer expression is:
y j＝f(x i)＝b j+∑ iw ij.x j y j = f (x i ) = b j + ∑ i w ij .x j
其中y j是一个输出向量中的第j个神经元，x i是输入向量中的第i个神经元，w是一个全连层中的权重参数。 Where y j is the jth neuron in an output vector, x i is the ith neuron in the input vector, and w is a weight parameter in a fully connected layer.
本实施例提出的电子装置通过训练神经网络图片识别模型并利用训练好的神经网络图片识别模型来识别及标注图片，从而提升图片标注的效率。The electronic device proposed in this embodiment recognizes and marks pictures by training a neural network picture recognition model and uses the trained neural network picture recognition model, thereby improving the efficiency of picture marking.
此外，本申请实施例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上存储有的图片自动分类程序，所述图片自动分类程序可被一个或多个处理器执行，以实现如下操作：In addition, the embodiments of the present application also provide a computer-readable storage medium that stores an automatic picture classification program stored on the computer-readable storage medium, and the automatic picture classification program may be executed by one or more processors to implement Proceed as follows:
为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；例如，预设图片类别包括身份证、居住证、户口本、驾照、出生证等，预设数量为1000张；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID cards, residence permits, household registration books, driver's licenses, birth certificates, etc. The preset number is 1,000 ;
将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；例如第一比例为80％， 第二比例为20％；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is The sample images are mixed to obtain the verification set; for example, the first ratio is 80%, and the second ratio is 20%;
通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
接收由用户终端输入的当前图片；例如当前图片为身份证的正反面扫描件；Receive the current picture input by the user terminal; for example, the current picture is a scanned copy of the front and back of the ID card;
利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料；例如查询指令针对的是身份证类别的，找到所述身份证类别的资料并显示，以方便质控员审查。Receive a query instruction for a picture category, search for picture data in the database according to the query instruction and display the found picture data; for example, the query instruction is for the ID card category, find the data of the ID card category and display , In order to facilitate the quality control staff to review.
本申请计算机可读存储介质具体实施方式与上述电子装置和方法各实施例基本相同，在此不作累述。The specific implementation of the computer-readable storage medium of the present application is basically the same as the foregoing embodiments of the electronic device and method, and will not be repeated here.
可选地，在其他实施例中，图片自动分类程序还可以被分割为一个或者多个模块，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13)所执行以完成本申请，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，用于描述图片自动分类程序在电子装置中的执行过程。Optionally, in other embodiments, the automatic picture classification program may also be divided into one or more modules, and the one or more modules are stored in the memory 11 and are processed by one or more processors (this embodiment is The processor 13) is executed to complete this application. The module referred to in this application refers to a series of computer program instruction segments capable of performing specific functions, and is used to describe the execution process of the automatic picture classification program in the electronic device.
例如，参照图4所示，为本申请电子装置一实施例中的图片自动分类程序的程序模块示意图，该实施例中，图片自动分类程序可以被分割为准备模块10、划分模块20、训练模块30、验证模块40、接收模块50、识别模块60及查找模块70，示例性地：For example, referring to FIG. 4, it is a schematic diagram of a program module of an automatic picture classification program in an embodiment of an electronic device of the present application. In this embodiment, the automatic picture classification program may be divided into a preparation module 10, a division module 20, and a training module 30. The verification module 40, the receiving module 50, the identification module 60, and the search module 70, exemplarily:
所述准备模块10用于：为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；例如，预设图片类别包括身份证、居住证、户口本、驾照、出生证等，预设数量为1000张；The preparation module 10 is used to: prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category; for example, the preset picture categories include ID card, residence permit, household registration book, driver's license, and birth certificate Wait, the preset number is 1000 sheets;
所述划分模块20用于：将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；例如第一比例为80％，第二比例为20％；The dividing module 20 is used to divide the picture samples corresponding to each preset picture category into a first proportion training subset and a second proportion verification subset, and mix the picture samples in each training subset to obtain training Set, and mix the image samples in each verification subset to obtain the verification set; for example, the first ratio is 80% and the second ratio is 20%;
所述训练模块30用于：通过所述训练集训练神经网络图片识别模型；The training module 30 is used to: train a neural network image recognition model through the training set;
所述验证模块40用于：通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；The verification module 40 is used to verify the accuracy of the neural network image recognition model trained by the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
所述接收模块50用于：接收由用户终端输入的当前图片；例如当前图片为身份证的正反面扫描件；The receiving module 50 is used to: receive the current picture input by the user terminal; for example, the current picture is a scan of the front and back of the ID card;
所述识别模块60用于：利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；The recognition module 60 is used to: use the trained neural network picture recognition model to perform category recognition and labeling on the current picture, and store the recognition result and the labeling result to the database;
所述查找模块70用于：接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料；例如查询指令针对的是身份证类别的，找到所述身份证类别的资料并显示，以方便质控员审查。The search module 70 is used to: receive a query instruction for a picture category, search for picture information in the database according to the query instruction and display the found picture information; for example, the query instruction is for the ID card category and find all Describe and display the information of the type of ID card to facilitate the quality control officer to review.
上述准备模块10、划分模块20、训练模块30、验证模块40、接收模块50、识别模块60及查找模块70等程序模块被执行时所实现的功能或操作步骤与上述实施例大体相同，在此不再赘述。The above-mentioned preparation module 10, division module 20, training module 30, verification module 40, receiving module 50, identification module 60, and search module 70 and other program modules are implemented when the functions or operation steps are substantially the same as the above-mentioned embodiments, here No longer.
需要说明的是，上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。并且本文中的术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、装置、物品或者方法不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、装置、物品或者方法所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、装置、物品或者方法中还存在另外的相同要素。It should be noted that the sequence numbers of the above embodiments of the present application are for description only, and do not represent the advantages and disadvantages of the embodiments. And the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method that includes a series of elements includes not only those elements, but also includes no explicit The other elements listed may also include elements inherent to this process, device, article, or method. Without more restrictions, the element defined by the sentence "include one ..." does not exclude that there are other identical elements in the process, device, article or method that includes the element.
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在如上所述的一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，或者网络设备等)执行本申请各个实施例所述的方法。Through the description of the above embodiments, those skilled in the art can clearly understand that the methods in the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases the former is better Implementation. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product is stored in a storage medium (such as ROM / RAM) , Magnetic disks, optical disks), including several instructions to enable a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to perform the method described in each embodiment of the present application.
以上仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the description and drawings of this application, or directly or indirectly used in other related technical fields , The same reason is included in the scope of patent protection of this application.

Claims (20)

  1. 一种图片自动分类方法，应用于电子装置中，其特征在于，所述方法包括：An automatic picture classification method, applied to an electronic device, characterized in that the method includes:
    为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category;
    将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
    通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
    通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
    接收由用户终端输入的当前图片；Receive the current picture input by the user terminal;
    利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
    接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料。Receiving a query instruction for a picture category, searching for picture data in the database according to the query instruction and displaying the found picture data.
  2. 如权利要求1所述的图片自动分类方法，其特征在于，所述方法还包括：当准确率小于所述预设阈值，增加每一个预设图片类别对应的证件图片样本的数量，再重新将每一预设图片类别更新后的图片样本分为新的训练子集及验证子集。The automatic picture classification method according to claim 1, wherein the method further comprises: when the accuracy rate is less than the preset threshold, increasing the number of certificate picture samples corresponding to each preset picture category, and then re-classifying The updated picture samples of each preset picture category are divided into new training subsets and verification subsets.
  3. 如权利要求1所述的图片自动分类方法，其特征在于，所述接收由用户终端输入的当前图片的步骤包括：The automatic picture classification method according to claim 1, wherein the step of receiving the current picture input by the user terminal comprises:
    显示用户输入界面；Display user input interface;
    接收由用户输入的所述图片信息；及Receiving the picture information input by the user; and
    接收由用户上传的所述图片资料。Receive the picture data uploaded by the user.
  4. 如权利要求1所述的图片自动分类方法，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The automatic picture classification method according to claim 1, wherein the step of training a neural network picture recognition model through the training set comprises:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个 全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100001
    Figure PCTCN2019088636-appb-100001
    其中l表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Where l represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function , I, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  5. 如权利要求2所述的图片自动分类方法，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The automatic picture classification method according to claim 2, wherein the step of training a neural network picture recognition model through the training set comprises:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100002
    Figure PCTCN2019088636-appb-100002
    其中l表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Where l represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function , I, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  6. 如权利要求3所述的图片自动分类方法，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The automatic picture classification method according to claim 3, wherein the step of training a neural network picture recognition model through the training set comprises:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100003
    Figure PCTCN2019088636-appb-100003
    其中l表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积 核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Where l represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function , I, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  7. 如权利要求4-6任一项所述的图片自动分类方法，其特征在于，所述全连接层表达式为：The automatic picture classification method according to any one of claims 4-6, wherein the fully connected layer expression is:
    y j＝f(x i)＝b j+∑ iw ij.x j y j = f (x i ) = b j + ∑ i w ij .x j
    其中y j是一个输出向量中的第j个神经元，x i是输入向量中的第i个神经元，w是一个全连层中的权重参数。 Where y j is the jth neuron in an output vector, x i is the ith neuron in the input vector, and w is a weight parameter in a fully connected layer.
  8. 一种电子装置，其特征在于，所述电子装置包括存储器和处理器，所述存储器上存储有可在所述处理器上运行的图片自动分类程序，所述图片自动分类程序被所述处理器执行时实现如下步骤：An electronic device, characterized in that the electronic device includes a memory and a processor, and the memory stores a picture automatic classification program that can run on the processor, and the picture automatic classification program is used by the processor The following steps are implemented during execution:
    为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category;
    将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
    通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
    通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
    接收由用户终端输入的当前图片；Receive the current picture input by the user terminal;
    利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
    接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料。Receiving a query instruction for a picture category, searching for picture data in the database according to the query instruction and displaying the found picture data.
  9. 如权利要求8所述的电子装置，其特征在于，所述图片自动分类程序被所述处理器执行时还实现如下步骤：当准确率小于所述预设阈值，增加每一个预设图片类别对应的证件图片样本的数量，再重新将每一预设图片类别更新后的图片样本分为新的训练子集及验证子集。The electronic device according to claim 8, wherein when the automatic picture classification program is executed by the processor, the following steps are further implemented: when the accuracy rate is less than the preset threshold, increase the number corresponding to each preset picture category The number of ID picture samples, and then re-divide the updated picture samples of each preset picture category into new training subsets and verification subsets.
  10. 如权利要求8所述的电子装置，其特征在于，所述接收由用户终端输入的当前图片的步骤包括：The electronic device according to claim 8, wherein the step of receiving the current picture input by the user terminal comprises:
    显示用户输入界面；Display user input interface;
    接收由用户输入的所述图片信息；及Receiving the picture information input by the user; and
    接收由用户上传的所述图片资料。Receive the picture data uploaded by the user.
  11. 如权利要求8所述的电子装置，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The electronic device according to claim 8, wherein the step of training a neural network image recognition model through the training set comprises:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100004
    Figure PCTCN2019088636-appb-100004
    其中表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Which represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function, i, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  12. 如权利要求9所述的电子装置，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The electronic device according to claim 9, wherein the step of training a neural network image recognition model through the training set comprises:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100005
    Figure PCTCN2019088636-appb-100005
    其中表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Which represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function, i, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  13. 如权利要求10所述的电子装置，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The electronic device of claim 10, wherein the step of training a neural network image recognition model through the training set includes:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个 全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100006
    Figure PCTCN2019088636-appb-100006
    其中表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Which represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function, i, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  14. 如权利要求11-13任一项所述的电子装置，其特征在于，所述全连接层表达式为：The electronic device according to any one of claims 11-13, wherein the expression of the fully connected layer is:
    y j＝f(x i)＝b j+∑ iw ij.x j y j = f (x i ) = b j + ∑ i w ij .x j
    其中y j是一个输出向量中的第j个神经元，x i是输入向量中的第i个神经元，w是一个全连层中的权重参数。 Where y j is the jth neuron in an output vector, x i is the ith neuron in the input vector, and w is a weight parameter in a fully connected layer.
  15. 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上存储有图片自动分类程序，所述图片自动分类程序可被一个或者多个处理器执行，以实现如下步骤：A computer-readable storage medium, characterized in that an automatic picture classification program is stored on the computer-readable storage medium, and the automatic picture classification program may be executed by one or more processors to implement the following steps:
    为每一个预设图片类别准备预设数量的标注有对应的图片类别的图片样本；Prepare a preset number of picture samples marked with corresponding picture categories for each preset picture category;
    将每一个预设图片类别对应的图片样本分为第一比例的训练子集和第二比例的验证子集，将各个训练子集中的图片样本进行混合以得到训练集，并将各个验证子集中的图片样本进行混合以得到验证集；The picture samples corresponding to each preset picture category are divided into a first proportion training subset and a second proportion verification subset, the picture samples in each training subset are mixed to obtain a training set, and each validation subset is Samples of the photos are mixed to get the verification set;
    通过所述训练集训练神经网络图片识别模型；Training a neural network image recognition model through the training set;
    通过所述验证集验证训练的所述神经网络图片识别模型的准确率，当准确率大于或者等于预设阈值时，训练结束；Verifying the accuracy of the trained neural network image recognition model through the verification set, and when the accuracy is greater than or equal to a preset threshold, the training ends;
    接收由用户终端输入的当前图片；Receive the current picture input by the user terminal;
    利用训练好的神经网络图片识别模型对所述当前图片进行类别识别及标注，并存储识别结果及标注结果至数据库；Use the trained neural network image recognition model to classify and label the current picture, and store the recognition results and labeling results to the database;
    接收针对一图片类别的查询指令，根据所述查询指令在所述数据库中查找图片资料并显示找到的图片资料。Receiving a query instruction for a picture category, searching for picture data in the database according to the query instruction and displaying the found picture data.
  16. 如权利要求15所述的计算机可读存储介质，其特征在于，所述图片自动分类程序被所述处理器执行时还实现如下步骤：当准确率小于所述预设 阈值，增加每一个预设图片类别对应的证件图片样本的数量，再重新将每一预设图片类别更新后的图片样本分为新的训练子集及验证子集。The computer-readable storage medium of claim 15, wherein when the automatic picture classification program is executed by the processor, the following steps are further implemented: when the accuracy rate is less than the preset threshold, each preset is added The number of ID picture samples corresponding to the picture categories, and then re-divide the updated picture samples of each preset picture category into new training subsets and verification subsets.
  17. 如权利要求15所述的计算机可读存储介质，其特征在于，所述接收由用户终端输入的当前图片的步骤包括：The computer-readable storage medium of claim 15, wherein the step of receiving the current picture input by the user terminal includes:
    显示用户输入界面；Display user input interface;
    接收由用户输入的所述图片信息；及Receiving the picture information input by the user; and
    接收由用户上传的所述图片资料。Receive the picture data uploaded by the user.
  18. 如权利要求15所述的计算机可读存储介质，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The computer-readable storage medium of claim 15, wherein the step of training a neural network image recognition model through the training set includes:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100007
    Figure PCTCN2019088636-appb-100007
    其中表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Which represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function, i, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  19. 如权利要求16或17所述的计算机可读存储介质，其特征在于，所述通过所述训练集训练神经网络图片识别模型的步骤包括：The computer-readable storage medium according to claim 16 or 17, wherein the step of training a neural network image recognition model through the training set includes:
    构建神经网络图片识别模型，所述神经网络图片识别模型为AlexNet神经网络，所述神经网络图片识别模型包括5个卷积层及3个全连接层，最后一个全连接层的输出是具有1000个输出的输出层，在每一个卷积层以及全连接层后紧跟的操作是激活操作，卷积层卷积操作公式为：Construct a neural network image recognition model. The neural network image recognition model is the AlexNet neural network. The neural network image recognition model includes 5 convolutional layers and 3 fully connected layers. The output of the last fully connected layer is 1000. The output layer of the output, the operation immediately following each convolutional layer and the fully connected layer is the activation operation. The convolutional operation formula of the convolutional layer is:
    Figure PCTCN2019088636-appb-100008
    Figure PCTCN2019088636-appb-100008
    其中表示第l层，j表示卷积层的第j个核，M表示卷积核所在区域，k表示卷积核，b表示偏置，x表示特征图对应位置的值，f表示激活函数，i、j、l为自然数；Which represents the lth layer, j represents the jth kernel of the convolution layer, M represents the region where the convolution kernel is located, k represents the convolution kernel, b represents the offset, x represents the value of the corresponding position of the feature map, and f represents the activation function, i, j, l are natural numbers;
    将所述训练集代入所述神经网络图片识别模型进行训练。The training set is substituted into the neural network image recognition model for training.
  20. 如权利要求19所述的计算机可读存储介质，其特征在于，所述全连接层表达式为：The computer-readable storage medium of claim 19, wherein the fully connected layer expression is:
    y j＝f(x i)＝b j+∑ iw ij.x j y j = f (x i ) = b j + ∑ i w ij .x j
    其中y j是一个输出向量中的第j个神经元，x i是输入向量中的第i个神经元，w是一个全连层中的权重参数。 Where y j is the jth neuron in an output vector, x i is the ith neuron in the input vector, and w is a weight parameter in a fully connected layer.
PCT/CN2019/088636 2018-10-26 2019-05-27 Automatic picture classification method, device, and computer readable storage medium WO2020082732A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811254121.3 2018-10-26
CN201811254121.3A CN109657694A (en) 2018-10-26 2018-10-26 Picture automatic classification method, device and computer readable storage medium

Publications (1)

Publication Number Publication Date
WO2020082732A1 true WO2020082732A1 (en) 2020-04-30

Family

ID=66110009

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/088636 WO2020082732A1 (en) 2018-10-26 2019-05-27 Automatic picture classification method, device, and computer readable storage medium

Country Status (2)

Country Link
CN (1) CN109657694A (en)
WO (1) WO2020082732A1 (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111597373A (en) * 2020-05-19 2020-08-28 清华大学 Image classification method based on convolutional neural network and connected graph and related equipment
CN111680642A (en) * 2020-06-11 2020-09-18 石家庄铁道大学 Terrain classification method and device
CN111739617A (en) * 2020-06-15 2020-10-02 中国医学科学院医学信息研究所 Medical image artificial intelligence quality control marking method, device, equipment and storage medium
CN111783968A (en) * 2020-06-30 2020-10-16 山东信通电子股份有限公司 Power transmission line monitoring method and system based on cloud edge cooperation
CN112731160A (en) * 2020-12-25 2021-04-30 东莞新能安科技有限公司 Battery hysteresis model training method, and method and device for estimating battery SOC
CN113269229A (en) * 2021-04-22 2021-08-17 中国科学院信息工程研究所 Training method for enhancing generalization ability of deep learning classification model
CN114677563A (en) * 2022-04-08 2022-06-28 李燕秋 Neural network online learning method and system based on block chain
CN115114467A (en) * 2021-03-17 2022-09-27 腾讯科技（深圳）有限公司 Training method and device of picture neural network model
CN116738297A (en) * 2023-08-15 2023-09-12 北京快舒尔医疗技术有限公司 Diabetes typing method and system based on depth self-coding

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109657694A (en) * 2018-10-26 2019-04-19 平安科技（深圳）有限公司 Picture automatic classification method, device and computer readable storage medium
CN110222736A (en) * 2019-05-20 2019-09-10 北京字节跳动网络技术有限公司 Method, apparatus, electronic equipment and the computer readable storage medium of training classifier
CN110275974A (en) * 2019-06-28 2019-09-24 武汉轻工大学 Data adding method, device, terminal device and the storage medium of sample data set
CN110490221B (en) * 2019-07-05 2023-06-23 平安科技（深圳）有限公司 Multi-label classification method, electronic device and computer readable storage medium
CN113051969A (en) * 2019-12-26 2021-06-29 深圳市超捷通讯有限公司 Object recognition model training method and vehicle-mounted device
CN111241235A (en) * 2019-12-28 2020-06-05 深圳市九洲电器有限公司 Network disk searching method based on intelligent voice and related products
CN111680740A (en) * 2020-06-04 2020-09-18 京东方科技集团股份有限公司 Neural network training method and device and electrical load distinguishing method and device
CN111510752B (en) * 2020-06-18 2021-04-23 平安国际智慧城市科技股份有限公司 Data transmission method, device, server and storage medium
CN112053343A (en) * 2020-09-02 2020-12-08 平安科技（深圳）有限公司 User picture data processing method and device, computer equipment and storage medium
CN112132230B (en) * 2020-10-09 2024-02-20 腾讯科技（深圳）有限公司 Image classification method and device
CN112348089B (en) * 2020-11-10 2024-01-16 中南民族大学 Working state identification method, server, storage medium and device
CN113449814B (en) * 2021-07-20 2022-10-04 曲阜师范大学 Picture level classification method and system
CN114943875A (en) * 2022-06-09 2022-08-26 苏州大学 Visual analysis method for cable element identification
US11727673B1 (en) 2022-06-09 2023-08-15 Soochow University Visual analysis method for cable element identification

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102567742A (en) * 2010-12-15 2012-07-11 中国科学院电子学研究所 Automatic classification method of support vector machine based on selection of self-adapting kernel function
CN104517122A (en) * 2014-12-12 2015-04-15 浙江大学 Image target recognition method based on optimized convolution architecture
CN104992177A (en) * 2015-06-12 2015-10-21 安徽大学 Network pornographic image detection method based on deep convolutional neural network
CN105243398A (en) * 2015-09-08 2016-01-13 西安交通大学 Method of improving performance of convolutional neural network based on linear discriminant analysis criterion
CN109657694A (en) * 2018-10-26 2019-04-19 平安科技（深圳）有限公司 Picture automatic classification method, device and computer readable storage medium

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107527060B (en) * 2016-06-20 2022-01-25 青岛海尔智能技术研发有限公司 Refrigerating device storage management system and refrigerating device
CN106934364A (en) * 2017-03-09 2017-07-07 腾讯科技（上海）有限公司 The recognition methods of face picture and device
CN106951848B (en) * 2017-03-13 2018-12-28 平安科技（深圳）有限公司 The method and system of picture recognition
CN107292823A (en) * 2017-08-20 2017-10-24 平安科技（深圳）有限公司 Electronic installation, the method for invoice classification and computer-readable recording medium
CN107704514A (en) * 2017-08-31 2018-02-16 努比亚技术有限公司 A kind of photo management method, device and computer-readable recording medium

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102567742A (en) * 2010-12-15 2012-07-11 中国科学院电子学研究所 Automatic classification method of support vector machine based on selection of self-adapting kernel function
CN104517122A (en) * 2014-12-12 2015-04-15 浙江大学 Image target recognition method based on optimized convolution architecture
CN104992177A (en) * 2015-06-12 2015-10-21 安徽大学 Network pornographic image detection method based on deep convolutional neural network
CN105243398A (en) * 2015-09-08 2016-01-13 西安交通大学 Method of improving performance of convolutional neural network based on linear discriminant analysis criterion
CN109657694A (en) * 2018-10-26 2019-04-19 平安科技（深圳）有限公司 Picture automatic classification method, device and computer readable storage medium

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111597373B (en) * 2020-05-19 2023-06-20 清华大学 Picture classifying method and related equipment based on convolutional neural network and connected graph
CN111597373A (en) * 2020-05-19 2020-08-28 清华大学 Image classification method based on convolutional neural network and connected graph and related equipment
CN111680642A (en) * 2020-06-11 2020-09-18 石家庄铁道大学 Terrain classification method and device
CN111680642B (en) * 2020-06-11 2023-06-23 石家庄铁道大学 Terrain classification method and device
CN111739617A (en) * 2020-06-15 2020-10-02 中国医学科学院医学信息研究所 Medical image artificial intelligence quality control marking method, device, equipment and storage medium
CN111783968A (en) * 2020-06-30 2020-10-16 山东信通电子股份有限公司 Power transmission line monitoring method and system based on cloud edge cooperation
CN111783968B (en) * 2020-06-30 2024-05-31 山东信通电子股份有限公司 Power transmission line monitoring method and system based on cloud edge cooperation
CN112731160A (en) * 2020-12-25 2021-04-30 东莞新能安科技有限公司 Battery hysteresis model training method, and method and device for estimating battery SOC
CN115114467A (en) * 2021-03-17 2022-09-27 腾讯科技（深圳）有限公司 Training method and device of picture neural network model
CN115114467B (en) * 2021-03-17 2024-05-14 腾讯科技（深圳）有限公司 Training method and device for picture neural network model
CN113269229A (en) * 2021-04-22 2021-08-17 中国科学院信息工程研究所 Training method for enhancing generalization ability of deep learning classification model
CN114677563A (en) * 2022-04-08 2022-06-28 李燕秋 Neural network online learning method and system based on block chain
CN116738297A (en) * 2023-08-15 2023-09-12 北京快舒尔医疗技术有限公司 Diabetes typing method and system based on depth self-coding
CN116738297B (en) * 2023-08-15 2023-11-21 北京快舒尔医疗技术有限公司 Diabetes typing method and system based on depth self-coding

Also Published As

Publication number Publication date
CN109657694A (en) 2019-04-19

Similar Documents

Publication Publication Date Title
WO2020082732A1 (en) Automatic picture classification method, device, and computer readable storage medium
CN111476227B (en) Target field identification method and device based on OCR and storage medium
WO2019024234A1 (en) Vehicle loss-related identification photo classification method and system, electronic device, and readable storage medium
US9400806B2 (en) Image triggered transactions
US20190392200A1 (en) Facial Recognition Pet Identifying System
WO2019174130A1 (en) Bill recognition method, server, and computer readable storage medium
US10572725B1 (en) Form image field extraction
CN111695439A (en) Image structured data extraction method, electronic device and storage medium
WO2019071662A1 (en) Electronic device, bill information identification method, and computer readable storage medium
WO2019071660A1 (en) Bill information identification method, electronic device, and readable storage medium
WO2019218514A1 (en) Method for extracting webpage target information, device, and storage medium
WO2018188199A1 (en) Method and device for identifying characters of claim settlement bill, server and storage medium
CN110097329B (en) Information auditing method, device, equipment and computer readable storage medium
WO2021012494A1 (en) Deep learning-based face recognition method and apparatus, and computer-readable storage medium
WO2019242144A1 (en) Electronic device, preference tendency prediction method and computer readable storage medium
CN112380343B (en) Problem analysis method, device, electronic equipment and storage medium
WO2019037259A1 (en) Electronic device, method and system for categorizing invoices, and computer-readable storage medium
US20180181915A1 (en) Systems and Methods to Improve Job Posting Structure and Presentation
WO2019184219A1 (en) Method, server, and storage medium for smart search
WO2014133571A1 (en) Presentation of image of source of tax data through tax preparation application
CN112016273A (en) Document directory generation method and device, electronic equipment and readable storage medium
WO2014100172A1 (en) Mobile-enabled systems and processes for intelligent research platform
WO2019223146A1 (en) Stomach cancer identification method, device, and storage medium
CN109190062B (en) Crawling method and device for target corpus data and storage medium
CN109784339A (en) Picture recognition test method, device, computer equipment and storage medium

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: 19877120

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: 19877120

Country of ref document: EP

Kind code of ref document: A1