WO2020207216A1 - 一种商品溯源码生成及查询方法和装置 - Google Patents

一种商品溯源码生成及查询方法和装置 Download PDF

Info

Publication number
WO2020207216A1
WO2020207216A1 PCT/CN2020/079837 CN2020079837W WO2020207216A1 WO 2020207216 A1 WO2020207216 A1 WO 2020207216A1 CN 2020079837 W CN2020079837 W CN 2020079837W WO 2020207216 A1 WO2020207216 A1 WO 2020207216A1
Authority
WO
WIPO (PCT)
Prior art keywords
traceability
product
source code
information
commodity
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/079837
Other languages
English (en)
French (fr)
Inventor
毛霖
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Newlixon Tech Co Ltd
Original Assignee
Newlixon Tech Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Newlixon Tech Co Ltd filed Critical Newlixon Tech Co Ltd
Priority to EP20788623.5A priority Critical patent/EP3955198A4/en
Priority to US17/599,554 priority patent/US20220198473A1/en
Publication of WO2020207216A1 publication Critical patent/WO2020207216A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION 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
    • G06Q30/00—Commerce
    • G06Q30/018—Certifying business or products
    • G06Q30/0185—Product, service or business identity fraud
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06N3/09—Supervised learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/95—Pattern authentication; Markers therefor; Forgery detection

Definitions

  • the invention relates to the technical field of information processing, in particular to a method and device for generating and querying product traceability source code.
  • the present invention aims to provide a method for product traceability.
  • a method for generating product traceability source code including:
  • acquiring the first image information of the commodity and performing fingerprint extraction on the first image information of the commodity includes:
  • the neural network model specifically uses the LeNet-5 convolutional neural network model based on the Tensorflow framework, where the LeNet-5 convolutional neural network model includes a first convolutional layer and a first pooling layer that are sequentially connected , The second convolutional layer, the second pooling layer, the first fully connected layer and the second fully connected layer;
  • the method further includes: acquiring the feature vector output by the second fully connected layer as the first fingerprint information of the commodity.
  • a method for querying product traceability source code including:
  • the method further includes:
  • the obtained second fingerprint information is compared with the first fingerprint information carried in the traceability source code, and the credibility information of the traceability source code is output and sent to the consumer terminal.
  • receiving the second image information of the commodity sent by the consumer terminal, and performing fingerprint extraction on the second image information includes:
  • Receive the product image sent by the consumer terminal perform image segmentation processing on the product image, input the segmented product image into the neural network model based on deep learning training, and obtain the differentiation degree of the product classification output by the neural network model
  • the feature vector of is used as the second fingerprint information of the product.
  • comparing the acquired second fingerprint information with the first fingerprint information carried in the traceability source code includes:
  • the obtained second fingerprint information is compared with the first fingerprint information pre-stored in the traceability source code, and the probability that the second fingerprint information is consistent with the first fingerprint information obtained through the chi-square check is used as the credibility information.
  • a device for generating product traceability source code including:
  • the first processing unit is used to obtain the traceability information of the commodity, and generate a traceability report based on the traceability information;
  • the second processing unit is configured to obtain the first image information of the commodity, perform fingerprint extraction on the first image information of the commodity, and obtain the first fingerprint information of the commodity;
  • the generating unit is used to generate the traceable source code of the commodity according to the traceability report and the first fingerprint information.
  • the second processing unit further includes:
  • the neural network model specifically adopts the LeNet-5 convolutional neural network model based on the Tensorflow framework, where the LeNet-5 convolutional neural network model includes a first convolutional layer connected in sequence , The first pooling layer, the second convolutional layer, the second pooling layer, the first fully connected layer and the second fully connected layer;
  • the feature vector output by the second fully connected layer is acquired as the first fingerprint information of the commodity.
  • the fourth aspect provides a product traceability source code query device, including:
  • the first receiving unit is configured to receive the traceability source code sent by the consumer terminal, wherein the traceability source code is the traceability source code generated by the commodity traceability source code generating device in claim 8;
  • the extraction unit is used to extract the traceability report corresponding to the source code
  • the sending unit is used to send the traceability report to the consumer terminal.
  • the device further includes:
  • the second receiving unit is configured to receive the second image information of the commodity sent by the consumer terminal;
  • the processing unit is configured to perform fingerprint extraction on the second image information to obtain the second fingerprint information of the commodity
  • the comparison unit is used to compare the acquired second fingerprint information with the first fingerprint information carried in the traceability source code, and output the credibility information of the traceability source code;
  • the sending unit is also used to send the credibility information to the consumer terminal.
  • the processing unit further includes: receiving a commodity image sent by a consumer terminal;
  • the processing unit further includes: performing image segmentation processing on the product image, inputting the segmented product image into a neural network model obtained based on deep learning training, and obtaining a feature vector output by the neural network model that distinguishes the product classification as the The second fingerprint information of the product.
  • the comparison unit further includes: comparing the acquired second fingerprint information with the first fingerprint information pre-stored in the traceability source code, and acquiring the second fingerprint information and the first fingerprint information through a chi-square check The probability of agreement is used as credibility information.
  • the beneficial effects of the present invention are: by adding the fingerprint information corresponding to the product image information in the product traceability source code, the corresponding binding of the product traceability source code and the real product can be realized, and the problem of the traceability label being misapplied is avoided, and at the same time, it improves The anti-counterfeiting performance of the traceable label.
  • the fingerprint information is extracted from the image information uploaded by the consumer, and compared with the fingerprint information carried in the traceability source code to verify the self-verification of the traceability report, which effectively improves the credibility of the traceability label.
  • the invention adopts software encryption, does not rely on the traditional anti-counterfeiting traceability label for hardware anti-counterfeiting technology, and has low production cost.
  • Figure 1 is a flow chart of a method for generating traceable source codes of commodities according to the present invention
  • Figure 2 is a diagram of a convolutional neural network architecture adopted by the present invention.
  • Figure 3 is a flow chart of a method for querying product traceability source code according to the present invention.
  • Figure 4 is a structural diagram of a device for generating traceable source codes for commodities of the present invention.
  • Fig. 5 is a structural diagram of a product traceability source code query device of the present invention.
  • S101 obtains the traceability information of the commodity, and generates a traceability report based on the traceability information
  • S102 acquire the first image information of the product, perform fingerprint extraction on the first image information of the product, and acquire the first fingerprint information of the product;
  • the step S102 includes:
  • the neural network model specifically uses the LeNet-5 convolutional neural network model based on the Tensorflow framework.
  • the architecture of the LeNet-5 convolutional neural network model includes the first convolutional layer, the first pooling layer, the second convolutional layer, the second pooling layer, and the first fully connected in sequence. Layer, second fully connected layer, softmax classification layer and classification output layer.
  • the trained LeNet-5 convolutional neural network model can classify the input product image, input the product image to the trained LeNet-5 convolutional neural network model, and its classification output layer can input the classification of the product , Its second fully connected layer can output a feature vector with distinguishing degree of product classification according to the input product image, classify the feature vector through the softmax classification layer, and output the classification of the product in the classification output layer;
  • the feature vector output by the second fully connected layer that is distinguishable for the product classification is used as the fingerprint information of the product image, which can effectively use the fingerprint information to describe the product characteristics, so that the fingerprint information and the product characteristics correspond one-to-one .
  • the output of the second fully connected layer of the LeNet-5 convolutional neural network model is a multi-dimensional array, which may specifically be an array with 84 elements.
  • the multi-dimensional feature vector output by the middle layer of the LeNet-5 convolutional neural network model is used as the fingerprint information of the product image, which can accurately describe the feature information of the product according to the classification feature of the product, so that the fingerprint information of the product is consistent with Commodities can be accurately corresponded, so that the traceable source code and the commodity correspond accurately, and the anti-counterfeiting effect of the traceable source code is improved.
  • S103 generates the traceable source code of the product according to the traceability report and the first fingerprint information.
  • the traceability information of the product is generated, a traceability report is generated, and the traceability report is encoded, and the traceability report code is used to generate the traceability source code.
  • the first fingerprint information obtained above is mixed into the traceability source code. , Become the traceable source code of the product.
  • the traceable source code is made into a traceable label, and the traceable label is added to the product packaging bag.
  • the traceability label can be in the form of a two-dimensional code, an RFID code, or the like.
  • FIG. 3 shows a method for querying product traceability source code, including:
  • S301 receives the traceability source code sent by the consumer terminal
  • the traceable source code is the traceable source code generated by any of the above-mentioned implementations of the commodity traceability source code generation method shown in FIG. 1.
  • the traceability source code carries product traceability report information and the first fingerprint information of the product.
  • consumers log in to a designated website through their consumer terminal, and enter the traceability code of the product in the website.
  • the consumer directly scans the traceable source code through APP, applet, etc., and sends the traceable source code directly to the server; and the server returns a traceability report corresponding to the traceable source code.
  • S302 extracts the traceability report corresponding to the source code, and sends the traceability report to the consumer terminal;
  • the server extracts the corresponding traceability report from the database according to the traceability source code sent by the consumer terminal, and sends the traceability report to the consumer terminal for display;
  • the report is a web interface, which is also provided with an interface for consumers to upload image information. Consumers can shoot or send the stored image information of the product to the server through the interface, and the server can further verify it.
  • S303 receives the second image information of the product sent by the consumer terminal, performs fingerprint extraction on the second image information, and obtains the second fingerprint information of the product;
  • this step S303 includes:
  • Receive the product image sent by the consumer terminal perform image segmentation processing on the product image, input the segmented product image into the neural network model based on deep learning training, and obtain the differentiation degree of the product classification output by the neural network model
  • the feature vector of is used as the second fingerprint information of the product.
  • the neural network model specifically uses the LeNet-5 convolutional neural network model based on the Tensorflow framework as shown in FIG. 2.
  • the consumer terminal after the consumer terminal receives the traceability report, it can further upload the image information of the product through the interface displaying the traceability report, and the server obtains the corresponding second fingerprint information according to the product image information uploaded by the consumer .
  • the image information of the product is a picture of the product taken by the consumer.
  • S304 compares the acquired second fingerprint information with the first fingerprint information carried in the traceability source code, and outputs the credibility information of the traceability source code and sends it to the consumer terminal.
  • the obtained second fingerprint information is compared with the first fingerprint information pre-stored in the traceability source code, and the probability that the second fingerprint information is consistent with the first fingerprint information is obtained through the chi-square check.
  • the probability that the second fingerprint information obtained through the chi-square check is consistent with the first fingerprint information is used as the credibility information.
  • the consumer finds the traceable source code on the product packaging, enters the traceable source code through the consumer terminal (mobile phone, computer, etc.) to the designated website, and queries the traceability information corresponding to the traceable source code; the server according to the received traceable source Query and return the corresponding traceability report to the consumer terminal; the consumer terminal displays the traceability report; further, the consumer uploads the image information of the product to the server through the traceability report display interface; the server performs the process based on the received image information Fingerprint information extraction, compare the extracted fingerprint information with the original fingerprint information carried by the traceability source code, determine the credibility of the traceability report by calculating its similarity, realize the self-verification of the traceability report, and integrate the credibility information Return to the consumer terminal for display.
  • the consumer terminal mobile phone, computer, etc.
  • the corresponding binding of the product traceability source code and the real product can be realized, which avoids the problem of the traceability label being misapplied and improves the traceability The anti-counterfeiting performance of the label.
  • the fingerprint information is extracted from the image information uploaded by the consumer, and compared with the fingerprint information carried in the traceability source code to verify the self-verification of the traceability report and effectively improve the credibility of the traceability label.
  • Tea leaves can be classified according to their leaf size and shape.
  • the leaf properties can be divided into one bud, one bud and one early leaf.
  • the length of the leaves can be divided into less than 1.5cm, 1.5-2cm, 2-3cm, 3cm and more.
  • the image of the tea leaves and the corresponding classification information are obtained, and the image of the tea leaves and the corresponding classification information are input into the LeNet-5 convolutional neural network model to complete the training of the model.
  • the traceability source code When generating the traceability source code, obtain the image information of the tea leaves for which the traceability source code is to be generated, collect the images of the batch of tea leaves, input the obtained tea images into the trained LeNet-5 convolutional neural network model, and output the model
  • the feature vector of is used as the first fingerprint information of the batch of tea, and the traceability source code and the traceability label of the batch of tea are generated according to the first fingerprint information and the traceability report of the batch of tea.
  • the consumer When querying the traceability information, the consumer enters the traceability source code on the tea packaging belt to the server, and views the traceability report returned by the server. In addition, the consumer can take a photo of the tea and pass the captured tea photo through the traceability report.
  • the network interface is uploaded to the server. After the server receives the tea photo uploaded by the consumer, it divides the tea photo, and enters the divided tea photo into the trained LeNet-5 convolutional neural network model to obtain The feature vector output by the trained LeNet-5 convolutional neural network model is used as the second fingerprint information.
  • the second fingerprint information obtained is compared with the first fingerprint information carried in the traceability source code, and the second fingerprint information is obtained through the chi-square test. The probability that the fingerprint information is consistent with the first fingerprint information is returned to the consumer as the credibility information of the traceability source code.
  • the above-mentioned embodiments of the present invention extract the image fingerprint information of the traceable product by using artificial intelligence neural network technology, bind the fingerprint information of the traceable product itself and its traceable source code, realize one-to-one correspondence of error codes, and improve the traceability source code. Credibility.
  • consumers can realize the self-verification of the traceable source code by uploading the image information of the traceable commodity after receiving the traceable commodity, which further improves the credibility of the traceable source code.
  • FIG. 4 shows a product traceability source code generation device, including:
  • the first processing unit 41 is configured to obtain traceability information of the commodity, and generate a traceability report based on the traceability information;
  • the second processing unit 42 is configured to obtain the first image information of the commodity, perform fingerprint extraction on the first image information of the commodity, and obtain the first fingerprint information of the commodity;
  • the generating unit 43 is configured to generate the traceable source code of the commodity according to the traceability report and the first fingerprint information.
  • the second processing unit 42 further includes:
  • the neural network model specifically uses the LeNet-5 convolutional neural network model based on the Tensorflow framework, where the LeNet-5 convolutional neural network model includes the first convolutions connected in sequence Layer, first pooling layer, second convolutional layer, second pooling layer, first fully connected layer, and second fully connected layer;
  • the feature vector output by the second fully connected layer is acquired as the first fingerprint information of the commodity.
  • FIG. 5 shows a product traceability source code query device, including:
  • the first receiving unit 51 is configured to receive the traceability source code sent by the consumer terminal, where the traceability source code is the traceability source code generated by the commodity traceability source code generating device in claim 8;
  • the extraction unit 52 is used to extract the traceability report corresponding to the traceability source code
  • the sending unit 53 is used to send the traceability report to the consumer terminal.
  • the device further includes:
  • the second receiving unit 54 is configured to receive the second image information of the commodity sent by the consumer terminal;
  • the processing unit 55 is configured to perform fingerprint extraction on the second image information to obtain the second fingerprint information of the commodity
  • the comparison unit 56 is configured to compare the obtained second fingerprint information with the first fingerprint information carried in the traceability source code, and output the credibility information of the traceability source code;
  • the sending unit 53 is also used to send the credibility information to the consumer terminal.
  • the processing unit further includes: receiving an image of the commodity sent by the consumer terminal;
  • the processing unit 55 further includes: performing image segmentation processing on the product image, inputting the segmented product image into a neural network model obtained based on deep learning training, and obtaining a feature vector output by the neural network model that distinguishes the product classification as The second fingerprint information of the product.
  • the comparing unit 56 further includes: comparing the obtained second fingerprint information with the first fingerprint information pre-stored in the traceability source code, and obtaining the second fingerprint information and the first fingerprint through the chi-square check The probability that the information is consistent is used as credibility information.
  • this generating and querying device is used to realize the functions of the above generating and querying methods.
  • Each module in the device corresponds to the steps of the above method and can implement different implementations in the above method. For details, please refer to the above about methods. Description, not detailed here.
  • each embodiment of the present invention can be integrated into one processing unit/module, or each unit/module can exist alone physically, or two or more units/modules.
  • the module is integrated in a unit/module.
  • the above-mentioned integrated unit/module can be implemented in the form of hardware or software functional unit/module.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Development Economics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • General Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Mathematical Physics (AREA)
  • Marketing (AREA)
  • Finance (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Accounting & Taxation (AREA)
  • Biophysics (AREA)
  • Economics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Image Analysis (AREA)
  • Collating Specific Patterns (AREA)

Abstract

本发明提供一种商品溯源码生成方法，包括：获取商品的溯源信息，基于溯源信息生成溯源报告；获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；根据溯源报告和第一指纹信息生成商品的溯源码。本发明通过在商品的溯源码中添加与商品图像信息对应的指纹信息，能够实现商品溯源码和真实商品的对应绑定，避免了溯源标签被乱套用等问题，同时提高了溯源标签的防伪性能。

Description

一种商品溯源码生成及查询方法和装置 技术领域
本发明涉及信息处理技术领域，特别是一种商品溯源码生成及查询方法和装置。
背景技术
食品安全对于人们日常生活意义重大，判断商品溯源信息的真伪一直以来都是食品溯源领域研究的热点。虽然，目前已经拥有各种各样防伪标签和防伪技术，例如：RFID标签、二维码内外码防伪等等，但这些防伪标签和防伪技术都是使用硬件(纸张、电子签)防伪，这就导致了一方面标签制作成本太高，另一方面实际只是提高了防伪的门槛，在理论上实际其依然可伪造。
发明内容
针对上述问题，本发明旨在提供一种商品溯源方法。
本发明的目的采用以下技术方案来实现：
第一方面，提供一种商品溯源码生成方法，包括：
获取商品的溯源信息，基于溯源信息生成溯源报告；
获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
根据溯源报告和第一指纹信息生成商品的溯源码。
在一种实施方式中，获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，包括：
采集商品图像，对该商品图像进行图像分割处理；将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第一指纹信息。
在一种实施方式中，神经网络模型具体采用基于Tensorflow框架搭建的LeNet-5卷积神经网络模型，其中LeNet-5卷积神经网络模型包括依次连接的第一卷积层、第一池化层、第二卷积层、第二池化层、第一全连接层和第二全连接层；
该方法还包括：获取第二全连接层输出的特征向量作为该商品的第一指纹信息。
第二方面，提供一种商品溯源码查询方法，包括：
接收由消费者终端发送的溯源码，其中溯源码为根据上述第一方面中任一种实施方式生成的溯源码；
提取与溯源码对应的溯源报告；
将溯源报告发送到消费者终端。
在一种实施方式中，该方法还包括：
接收消费者终端发送的商品的第二图像信息，对第二图像信息进行指纹提取，获取该商品的第二指纹信息；
将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，输出该溯源码的可信性信息并发送到消费者终端。
在一种实施方式中，接收消费者终端发送的商品的第二图像信息，对第二图像信息进行指纹提取，包括：
接收由消费者终端发送的商品图像，对该商品图像进行图像分割处理，将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第二指纹信息。
在一种实施方式中，将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，具体包括：
将获取的第二指纹信息与溯源码中预存的第一指纹信息进行比对，通过卡方检验获取第二指纹信息和第一指纹信息一致的概率作为可信性信息。
第三方面，提供一种商品溯源码生成装置，包括：
第一处理单元，用于获取商品的溯源信息，基于溯源信息生成溯源报告；
第二处理单元，用于获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
生成单元，用于根据溯源报告和第一指纹信息生成商品的溯源码。
在一种实施方式中，第二处理单元进一步包括：
用于采集商品图像，对该商品图像进行图像分割处理；将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第一指纹信息。
在一种实施方式中，第二处理单元中，神经网络模型具体采用基于Tensorflow框架搭建的LeNet-5卷积神经网络模型，其中LeNet-5卷积神经网络模型包括依次连接的第一卷积层、第一池化层、第二卷积层、第二池化层、第一全连接层和第二全连接层；
其中，获取第二全连接层输出的特征向量作为该商品的第一指纹信息。
第四方面，提供一种商品溯源码查询装置，包括：
第一接收单元，用于接收由消费者终端发送的溯源码，其中溯源码为由权利要求8中的商品溯源码生成装置生成的溯源码；
提取单元，用于提取与溯源码对应的溯源报告；
发送单元，用于将溯源报告发送到消费者终端。
在一种实施方式中，该装置还包括：
第二接收单元，用于接收消费者终端发送的商品的第二图像信息；
处理单元，用于对第二图像信息进行指纹提取，获取该商品的第二指纹信息；
比对单元，用于将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，输出该溯源码的可信性信息；
该发送单元，还用于将该可信性信息发送到消费者终端。
在一种实施方式中，该处理单元进一步包括：接收由消费者终端发送的商品图像；
处理单元进一步包括：对该商品图像进行图像分割处理，将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第二指纹信息。
在一种实施方式中，该比对单元，进一步包括：将获取的第二指纹信息与溯源码中预存的第一指纹信息进行比对，通过卡方检验获取第二指纹信息和第一指纹信息一致的概率作为可信性信息。
本发明的有益效果为：通过在商品的溯源码中添加与商品图像信息对应的指纹信息，能够实现商品溯源码和真实商品的对应绑定，避免了溯源标签被乱套用等问题，同时提高了溯源标签的防伪性能。
同时，根据消费者上传的图像信息提取指纹信息，与溯源码中携带的指纹信息进行比对验证，实现了溯源报告的自验证，有效提高了溯源标签的可信性。
本发明采用软件加密，不依赖传统防伪溯源标签对硬件防伪的技术，制作成本低。
附图说明
利用附图对本发明作进一步说明，但附图中的实施例不构成对本发明的任何限制，对于本领域的普通技术人员，在不付出创造性劳动的前提下，还可以根据以下附图获得其它的附图。
图1为本发明的一种商品溯源码生成方法流程图；
图2为本发明采用的一种卷积神经网络架构图；
图3为本发明的一种商品溯源码查询方法流程图；
图4为本发明的一种商品溯源码生成装置结构图；
图5为本发明的一种商品溯源码查询装置结构图。
具体实施方式
结合以下应用场景对本发明作进一步描述。
参见图1，其示出了一种商品溯源码生成方法，包括：
S101获取商品的溯源信息，基于溯源信息生成溯源报告；
S102获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
在一种实施方式中，该步骤S102包括：
采集商品图像，对该商品图像进行图像分割处理；将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第一指纹信息。
其中，该神经网络模型具体采用基于Tensorflow框架搭建的LeNet-5卷积神经网络模型。
如图2所示，该LeNet-5卷积神经网络模型的架构为包括依次连接的第一卷积层、第一池化层、第二卷积层、第二池化层、第一全连接层、第二全连接层、softmax分类层和分类输出层。
其中，该训练好的LeNet-5卷积神经网络模型能够对输入的商品图像进行分类处理，将商品图像输入到训练好的LeNet-5卷积神经网络模型，其分类输出层能够输入商品的分类，其第二全连接层能够根据输入的商品图像输出对商品分类具有区分度的特征向量，通过softmax分类层对该特征向量进行分类处理，在分类输出层输出商品的分类；
在一种实施方式中，采用第二全连接层输出的对商品分类具有区分度的特征向量作为商品图像的指纹信息，能够有效地采用指纹信息描述商品特征，使得指纹信息和商品特征一一对应。
在一种场景中，该LeNet-5卷积神经网络模型的第二全连接层的输出为一个多维数组，具体可以为一个具有84个元素的数组。
上述实施方式，采用LeNet-5卷积神经网络模型的中夹层输出的多维特征向量作为商品图像的指纹信息，能够根据商品的分类特征情况，准确地描述商品的特征信息，使得商品的指纹信息与商品能够精确地对应，从而使得溯源码和商品精确对应，提高溯源码的防伪效果。
S103根据溯源报告和第一指纹信息生成商品的溯源码。
在一种实施方式中，首先商品的溯源信息，生成溯源报告，并对溯源报告进行编码，使用溯源报告编码生成溯源码，在编码的过程中，将上述获取的第一指纹信息混入到溯源码中，成为该商品的溯源码。
在一种实施方式中，将溯源码制作成溯源标签，并将溯源标签加入到商品包装袋中。
其中，该溯源标签可以为二维码、RFID码等形式。
参见图3，其示出一种商品溯源码查询方法，包括：
S301接收由消费者终端发送的溯源码；
在一种实施方式中，该溯源码为由上述图1所示的任一种商品溯源码生成方法实施方式生成的溯源码。
进一步地，该溯源码携带有商品溯源报告信息以及该商品的第一指纹信息。
在一种实施方式中，消费者通过其消费者终端登录指定的网站，通过在网站中输入商品的溯源码。
或者，消费者通过APP、小程序、等直接扫描溯源码，将该溯源码直接发送到服务器；并由服务器返回与该溯源码对应的溯源报告。
S302提取与溯源码对应的溯源报告，并将该溯源报告发送到消费者终端；
在一种场景中，服务器根据消费者终端发送的溯源码从数据库中提取对应溯源报告，并将溯源报告发送到消费者终端进行显示；
进一步地，该报告为网页界面，其上还设有供消费者上传图像信息的接口，消费者能够通过该接口拍摄或者将存好的该商品的图像信息发送到服务器，由服务器进一步验证。
S303接收消费者终端发送的商品的第二图像信息，对第二图像信息进行指纹提取，获取该商品的第二指纹信息；
在一种实施方式中，该步骤S303包括：
接收由消费者终端发送的商品图像，对该商品图像进行图像分割处理，将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第二指纹信息。
其中，该神经网络模型具体采用如图2所示的基于Tensorflow框架搭建的LeNet-5卷积神经网络模型。
在一种实施方式中，消费者终端接收到溯源报告后，可以通过显示该溯源报告的界面进一步上传该商品的图像信息，由服务器根据该消费者上传的商品图像信息获取相应的第二指纹信息。
在一种场景中，该商品的图像信息为由消费者拍摄的该商品的图片。
S304将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，输出该溯源码的可信性信息并发送到消费者终端。
在一种实施方式中，将获取的第二指纹信息与溯源码中预存的第一指纹信息进行比对，通过卡方检验获取第二指纹信息和第一指纹信息一致的概率。
在一种实施方式中，以该通过卡方检验获取的第二指纹信息和第一指纹信息一致的概率作为可信性信息。
在一种场景中，消费者在商品包装上找到溯源码，通过其消费者终端(手机、电脑等)到指定的网站录入该溯源码，查询溯源码对应的溯源信息；服务器根据接收的溯源码查询并返回相应的溯源报告到消费者终端；消费者终端显示该溯源报告；进一步地，消费者通过该溯源报告显示界面上传该商品的图像信息到服务器；服务器根据接收到的图像信息对其进行指纹信息提取，将提取的指纹信息和该溯源码原本携带的指纹信息进行比对，通过计算其相似度判断该溯源报告的可信性，实现溯源报告的自验证，并将该可信性信息返回到消费者终端进行显示。
本发明上述实施方式，通过在商品的溯源码中添加与商品图像信息对应的指纹信息，能够实现商品溯源码和真实商品的对应绑定，避免了溯源标签被乱套用等问题，同时提高了溯源标签的防伪性能。
同时，根据消费者上传的图像信息提取指纹信息，与溯源码中携带的指纹信息进行比对验证，实现了溯源报告的自验证，有效提高了溯源标签的可信性。
上述实施方式全部采用软件加密，不依赖传统防伪溯源标签对硬件防伪的技术，制作成本低。
进一步以茶业为例，示出一种茶叶溯源码的生成和查询的实施方式，其中，茶叶可以因其叶片大小、叶片形状进行分类，其中叶片性状可以分为一芽、一芽一叶初展、一芽二叶初展、一芽三叶以上等，叶片长度可以分为小于1.5cm、1.5-2cm、2-3cm、3cm以上等。
在模型训练阶段，获取茶叶的图像及其对应的上述分类信息，将该茶叶的图像和对应的分类信息输入到LeNet-5卷积神经网络模型中完成对模型的训练。
在生成溯源码时，获取待生成溯源码的茶叶的图像信息，采集该批次茶叶的图像，将获取的茶叶图像输入到训练好的LeNet-5卷积神经网络模型中，并将该模型输出的特征向量作为该批次茶叶的第一指纹信息，并根据该第一指纹信息和该批次茶叶的溯源报告生成该批次茶叶的溯源码及溯源标签。
在查询溯源信息时，消费者向服务器录入该茶叶包装带上的溯源码，查看由服务器返回的溯源报告，并且，消费者可以拍摄该茶叶的照片，并将拍摄的茶叶照片通过溯源报告中的网络接口上传到服务器，服务器接收到由消费者上传的茶叶照片后，对该茶叶照片进行分割处理，并且将分割后的茶叶照片录入到训练好的LeNet-5卷积神经网络模型中，获取由该训练好的LeNet-5卷积神经网络模型输出的特征向量作为第二指纹信息，将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，通过卡方检验获取第二指纹信息与第一指纹信息一致的概率，作为该溯源码的可信性信息返回给消费者。
本发明上述实施方式，通过利用人工智能神经网络技术，提取被溯源商品的图像指纹信息，把被溯源商品本身的指纹信息和其溯源码进行绑定，实现误码一一对应，提高了溯源码的可信性。同时，消费者在拿到被溯源商品后能够通过上传被溯源商品的图像信息实现溯源码的自验证，进一步提高了溯源码的可信性。
参见图4，其示出一种商品溯源码生成装置，包括：
第一处理单元41，用于获取商品的溯源信息，基于溯源信息生成溯源报告；
第二处理单元42，用于获取商品的第一图像信息，对商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
生成单元43，用于根据溯源报告和第一指纹信息生成商品的溯源码。
在一种实施方式中，第二处理单元42进一步包括：
用于采集商品图像，对该商品图像进行图像分割处理；将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第一指纹信息。
在一种实施方式中，第二处理单元42中，神经网络模型具体采用基于Tensorflow框架搭建的LeNet-5卷积神经网络模型，其中LeNet-5卷积神经网络模型包括依次连接的第一卷积层、第一池化层、第二卷积层、第二池化层、第一全连接层和第二全连接层；
其中，获取第二全连接层输出的特征向量作为该商品的第一指纹信息。
参见图5，其示出一种商品溯源码查询装置，包括：
第一接收单元51，用于接收由消费者终端发送的溯源码，其中溯源码为由权利要求8中的商品溯源码生成装置生成的溯源码；
提取单元52，用于提取与溯源码对应的溯源报告；
发送单元53，用于将溯源报告发送到消费者终端。
在一种实施方式中，该装置还包括：
第二接收单元54，用于接收消费者终端发送的商品的第二图像信息；
处理单元55，用于对第二图像信息进行指纹提取，获取该商品的第二指纹信息；
比对单元56，用于将获取的第二指纹信息和溯源码中携带的第一指纹信息进行比对，输出该溯源码的可信性信息；
该发送单元53，还用于将该可信性信息发送到消费者终端。
在一种实施方式中，该处理单元进一步包括：接收消费者终端发送的商品的图像；
处理单元55进一步包括：对该商品图像进行图像分割处理，将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第二指纹信息。
在一种实施方式中，该比对单元56，进一步包括：将获取的第二指纹信息与溯源码中预存的第一指纹信息进行比对，通过卡方检验获取第二指纹信息和第一指纹信息一致的概率作为可信性信息。
需要说明的是，本生成和查询装置用于实现上述生成和查询方法的功能，装置中各模块与上述方法步骤相对应，并能够实施上述方法中的不同实施方式，具体可参见上述关于方法的描述，这里不再详细叙述。
需要说明的是，在本发明各个实施例中的各功能单元/模块可以集成在一个处理单元/模块中，也可以是各个单元/模块单独物理存在，也可以是两个或两个以上单元/模块集成在一个单元/模块中。上述集成的单元/模块既可以采用硬件的形式实现，也可以采用软件功能单元/模块的形式实现。
最后应当说明的是，以上实施例仅用以说明本发明的技术方案，而非对本发明保护范围的限制，尽管参照较佳实施例对本发明作了详细地说明，本领域的普通技术人员应当分析，可以对本发明的技术方案进行修改或者等同替换，而不脱离本发明技术方案的实质和范围。

Claims (9)

  1. 一种商品溯源码生成方法，其特征在于，包括：
    获取商品的溯源信息，基于所述溯源信息生成溯源报告；
    获取商品的第一图像信息，对所述商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
    根据所述溯源报告和第一指纹信息生成所述商品的溯源码。
  2. 根据权利要求1所述的一种商品溯源码生成方法，其特征在于，获取商品的第一图像信息，对所述商品的第一图像信息进行指纹提取，包括：
    采集商品图像，对该商品图像进行图像分割处理；将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取所述神经网络模型输出的对商品分类有区分度的特征向量作为该商品的第一指纹信息。
  3. 根据权利要求2所述一种商品溯源码生成方法，其特征在于，所述神经网络模型具体采用基于Tensorflow框架搭建的LeNet-5卷积神经网络模型，其中所述LeNet-5卷积神经网络模型包括依次连接的第一卷积层、第一池化层、第二卷积层、第二池化层、第一全连接层和第二全连接层；
    该方法还包括：获取所述第二全连接层输出的特征向量作为该商品的第一指纹信息。
  4. 一种商品溯源码查询方法，其特征在于，包括：
    接收由消费者终端发送的溯源码，其中所述溯源码为根据权利要求1-3中任一项所述商品溯源码生成方法生成的溯源码；
    提取与所述溯源码对应的溯源报告；
    将所述溯源报告发送到所述消费者终端。
  5. 根据权利要求4所述一种商品溯源码查询方法，其特征在于，该方法还包括：
    接收消费者终端发送的商品的第二图像信息，对所述第二图像信息进行指纹提取，获取该商品的第二指纹信息；
    将获取的所述第二指纹信息和所述溯源码中携带的第一指纹信息进行比对，输出该溯源码的可信性信息并发送到所述消费者终端。
  6. 根据权利要求5所述一种商品溯源码查询方法，其特征在于，所述接收消费者终端发送的商品的第二图像信息，对所述第二图像信息进行指纹提取，包括：
    接收由所述消费者终端发送的商品图像，对该商品图像进行图像分割处理，将分割处理后的商品图像输入到基于深度学习训练得到的神经网络模型，获取所述神经网络模型输出的 对商品分类有区分度的特征向量作为该商品的第二指纹信息。
  7. 根据权利要求5所述一种商品溯源码查询方法，其特征在于，所述将获取的所述第二指纹信息和所述溯源码中携带的第一指纹信息进行比对，具体包括：
    将获取的第二指纹信息与溯源码中预存的第一指纹信息进行比对，通过卡方检验获取第二指纹信息和第一指纹信息一致的概率作为所述可信性信息。
  8. 一种商品溯源码生成装置，其特征在于，包括：
    第一处理单元，用于获取商品的溯源信息，基于所述溯源信息生成溯源报告；
    第二处理单元，获取商品的第一图像信息，对所述商品的第一图像信息进行指纹提取，获取该商品的第一指纹信息；
    生成单元，用于根据所述溯源报告和第一指纹信息生成所述商品的溯源码。
  9. 一种商品溯源码查询装置，其特征在于，包括：
    第一接收单元，用于接收由消费者终端发送的溯源码，其中所述溯源码为由权利要求8中的所述商品溯源码生成装置生成的溯源码；
    提取单元，用于提取与所述溯源码对应的溯源报告；
    发送单元，用于将所述溯源报告发送到所述消费者终端。
PCT/CN2020/079837 2019-04-09 2020-03-18 一种商品溯源码生成及查询方法和装置 Ceased WO2020207216A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP20788623.5A EP3955198A4 (en) 2019-04-09 2020-03-18 METHOD AND DEVICE FOR GENERATION AND REQUEST OF THE PROBE CODE OF GOODS
US17/599,554 US20220198473A1 (en) 2019-04-09 2020-03-18 Method and apparatus for generating and querying traceability code of commodity

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910281876.0A CN110163629A (zh) 2019-04-09 2019-04-09 一种商品溯源码生成及查询方法和装置
CN201910281876.0 2019-04-09

Publications (1)

Publication Number Publication Date
WO2020207216A1 true WO2020207216A1 (zh) 2020-10-15

Family

ID=67639200

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/079837 Ceased WO2020207216A1 (zh) 2019-04-09 2020-03-18 一种商品溯源码生成及查询方法和装置

Country Status (4)

Country Link
US (1) US20220198473A1 (zh)
EP (1) EP3955198A4 (zh)
CN (1) CN110163629A (zh)
WO (1) WO2020207216A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112435042A (zh) * 2020-11-30 2021-03-02 重庆文理学院 基于区块链的供应链服务平台
CN113379720A (zh) * 2021-06-29 2021-09-10 云南昆船设计研究院有限公司 一种基于茶饼图像特征编码的茶饼防伪方法
CN114494890A (zh) * 2022-04-14 2022-05-13 广州市玄武无线科技股份有限公司 一种模型训练方法、商品图像管理方法及装置

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110163629A (zh) * 2019-04-09 2019-08-23 南京新立讯科技股份有限公司 一种商品溯源码生成及查询方法和装置
CN111159458A (zh) * 2019-12-30 2020-05-15 南京龟兔赛跑软件研究院有限公司 一种农产品图像处理与区块链交互识别方法及系统
CN111861158B (zh) * 2020-07-02 2025-02-14 广东菜丁控股集团有限公司 一种基于物联网的农产品质量溯源方法和系统
CN112069269B (zh) * 2020-08-27 2021-03-26 中润普达(深圳)大数据技术有限公司 基于大数据和多维特征的数据溯源方法及大数据云服务器
CN112488233B (zh) * 2020-12-09 2021-12-17 中国农业科学院农业资源与农业区划研究所 一种基于果纹图谱信息的编码和识别方法及装置
CN114386988A (zh) * 2021-12-28 2022-04-22 航天信息股份有限公司 一种防伪溯源方法、装置及电子设备
CN115035533B (zh) * 2022-08-10 2022-10-21 新立讯科技股份有限公司 一种数据鉴真处理方法、装置、计算机设备及存储介质
CN116957610B (zh) * 2023-08-08 2024-10-22 南京龟兔赛跑软件研究院有限公司 一种农产品全流程信息溯源录入管理方法、系统及介质
CN119130500B (zh) * 2024-11-12 2025-07-08 深圳国家金融科技测评中心有限公司 一种商品防伪方法及适用于其的射频识别标签介质
CN120338822B (zh) * 2025-04-07 2025-12-05 中科微智（北京）生物科技有限公司 Ai溯源码识别系统及转码封装方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100172538A1 (en) * 1993-11-18 2010-07-08 Rhoads Geoffrey B Hiding and Detecting Auxiliary Data in Media Materials and Signals
CN109034837A (zh) * 2018-06-29 2018-12-18 山东华夏维康农牧科技有限公司 多码溯源防伪的方法及系统
CN110163629A (zh) * 2019-04-09 2019-08-23 南京新立讯科技股份有限公司 一种商品溯源码生成及查询方法和装置

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012177845A2 (en) * 2011-06-23 2012-12-27 Pharmorx Security, Inc Systems and methods for tracking and authenticating goods
US8533075B1 (en) * 2013-02-13 2013-09-10 AgileQR, Inc. Systems and methods for biomedical track and trace

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100172538A1 (en) * 1993-11-18 2010-07-08 Rhoads Geoffrey B Hiding and Detecting Auxiliary Data in Media Materials and Signals
CN109034837A (zh) * 2018-06-29 2018-12-18 山东华夏维康农牧科技有限公司 多码溯源防伪的方法及系统
CN110163629A (zh) * 2019-04-09 2019-08-23 南京新立讯科技股份有限公司 一种商品溯源码生成及查询方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3955198A4 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112435042A (zh) * 2020-11-30 2021-03-02 重庆文理学院 基于区块链的供应链服务平台
CN113379720A (zh) * 2021-06-29 2021-09-10 云南昆船设计研究院有限公司 一种基于茶饼图像特征编码的茶饼防伪方法
CN113379720B (zh) * 2021-06-29 2022-08-09 云南昆船设计研究院有限公司 一种基于茶饼图像特征编码的茶饼防伪方法
CN114494890A (zh) * 2022-04-14 2022-05-13 广州市玄武无线科技股份有限公司 一种模型训练方法、商品图像管理方法及装置
CN114494890B (zh) * 2022-04-14 2022-08-23 广州市玄武无线科技股份有限公司 一种模型训练方法、商品图像管理方法及装置

Also Published As

Publication number Publication date
CN110163629A (zh) 2019-08-23
US20220198473A1 (en) 2022-06-23
EP3955198A1 (en) 2022-02-16
EP3955198A4 (en) 2022-12-21

Similar Documents

Publication Publication Date Title
WO2020207216A1 (zh) 一种商品溯源码生成及查询方法和装置
CN108734283B (zh) 神经网络系统
CN106203242B (zh) 一种相似图像识别方法及设备
US9830516B1 (en) Joint temporal segmentation and classification of user activities in egocentric videos
US10635946B2 (en) Eyeglass positioning method, apparatus and storage medium
US10031925B2 (en) Method and system of using image recognition and geolocation signal analysis in the construction of a social media user identity graph
CN105874474B (zh) 用于面部表示的系统和方法
CN102663448B (zh) 一种基于网络的增强现实物体识别分析方法
US20160358024A1 (en) Systems and methods for image processing
CN109639481A (zh) 一种基于深度学习的网络流量分类方法、系统及电子设备
CN110941978B (zh) 一种未识别身份人员的人脸聚类方法、装置及存储介质
CN103577421A (zh) 一种应用二维码的方法、设备和系统
CN107798243A (zh) 终端应用的检测方法和装置
CN112395979A (zh) 基于图像的健康状态识别方法、装置、设备及存储介质
TW202036367A (zh) 人臉識別方法及裝置
CN112149690A (zh) 一种基于生物图像特征识别的溯源方法和溯源系统
CN114359582B (zh) 一种基于神经网络的小样本特征提取方法及相关设备
CN110210513A (zh) 数据分类方法、装置及终端设备
CN106339719A (zh) 一种图像识别方法及装置
CN114998665B (zh) 一种图像类别识别方法、装置、电子设备及存储介质
CN108121943A (zh) 基于图片的判别方法及装置和计算设备
CN111126347B (zh) 人眼状态识别方法、装置、终端及可读存储介质
WO2025092452A1 (zh) 特征提取网络的训练方法、分类方法、装置及电子设备
CN112633200A (zh) 基于人工智能的人脸图像比对方法、装置、设备及介质
CN108389062A (zh) 一种基于图像处理的水果防伪溯源系统及方法

Legal Events

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

Ref document number: 20788623

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020788623

Country of ref document: EP

Effective date: 20211109