KR102426803B1 - Method, device and system for automatically uploading overseas sales page of product based on artificial intelligence - Google Patents

Abstract

According to an embodiment, a method for automatically uploading an overseas sales page for a product based on artificial intelligence comprises the steps of: receiving a request for generating an overseas sales page for a first product from a first seller terminal; generating a first input signal by encoding the image of the first product; inputting the first input signal to a first artificial neural network and obtaining a first output signal based on a result of the input for the first artificial neural network; identifying a first category, which is a type of the first product, based on the first output signal; identifying a first site being operated using the first category as a main category, among overseas sales sites; generating a first page, which is a sales page for the first product, based on a first template used in the first site; and uploading the first page to the first site and registering the same. Therefore, the method can automatically generate and upload a sales page suitable for a template for each overseas sales site.

Description

Automatic upload method, device and system for overseas sales page of artificial intelligence-based products {METHOD, DEVICE AND SYSTEM FOR AUTOMATICALLY UPLOADING OVERSEAS SALES PAGE OF PRODUCT BASED ON ARTIFICIAL INTELLIGENCE}

The embodiments below relate to a technology for automatically uploading an overseas sales page of a product based on artificial intelligence.

In order to sell a product abroad, a product sales page is uploaded to an overseas sales site, and the product is sold overseas.

These overseas sales sites require the use of their own templates to upload sales pages.

Accordingly, sellers who sell products have the inconvenience of creating and uploading sales pages one by one according to the template for each overseas sales site, and the process of creating and uploading sales pages is not computerized. There are difficult and difficult problems in selling products through

Therefore, in selling products through overseas sales sites, there is an increasing demand for automatically generating and uploading sales pages, and research on related technologies is required.

Korean Patent Publication No. 10-2021-0133722 Korean Patent Publication No. 10-2020-0097112 Korean Patent Registration No. 10-1298785 Korean Patent No. 10-2365608

According to an embodiment, receiving a request for generating an overseas sales page for a first product from a first seller terminal, encoding an image of the first product to generate a first input signal, and applying the first input signal to a first artificial neural network input to , obtain a first output signal based on a result of the input of the first artificial neural network, and identify a first category that is a type of a first product based on the first output signal, among overseas sales sites A first site that operates the first category as a main category is checked, and based on a first template that is a template used in the first site, a first page that is a sales page of a first product is generated, and the first page is displayed. An object of the present invention is to provide a method, apparatus, and system for automatically uploading and registering an overseas sales page of an AI-based product by uploading it to the first site.

Objects of the present invention are not limited to the objects mentioned above, and other objects not mentioned will be clearly understood from the description below.

According to an embodiment, in a method for automatically uploading an overseas sales page of a product based on artificial intelligence, performed by a device, the method comprising: receiving a request for generating an overseas sales page for a first product from a first seller terminal; ; generating a first input signal by encoding the image of the first product; inputting the first input signal to a first artificial neural network, and obtaining a first output signal based on a result of the input of the first artificial neural network; identifying a first category that is a type of the first product based on the first output signal; checking a first site operating the first category as a main category among overseas sales sites; generating a first page that is a sales page of the first product based on a first template that is a template used in the first site; and uploading the first page to the first site and registering the first page.

The generating of the first page may include: checking whether the first product is being sold on the first site; when it is confirmed that the first product is being sold at the first site, checking a first price that is the lowest price of the first product being sold at the first site; checking whether a second price set by the first seller is higher than the first price; when it is confirmed that the second price is higher than the first price, transmitting a notification message informing that the first product is being sold abroad at the first price to the first seller terminal; generating the first page by setting the second price as the selling price of the first product when it is confirmed that the second price is not higher than the first price; and generating the first page by setting the second price as the selling price of the first product when it is confirmed that the first product is not being sold on the first site.

The method for automatically uploading the overseas sales page of the artificial intelligence-based product includes, after the step of uploading and registering the first page to the first site, the number of views of customers who have viewed the first page, then checking the number of interests of customers who put the first product in their shopping cart and the number of purchases of customers who purchased the first product through the first page; calculating a purchase ratio based on a value obtained by dividing the number of purchases by the number of inquiries, and calculating an interest ratio based on a value obtained by dividing the number of interests by the number of inquiries; checking whether the purchase ratio is higher than a preset first reference value; if it is confirmed that the purchase ratio is higher than the first reference value, providing a sales strategy for recommending maintaining the current sales policy of the first product to the first seller terminal; when it is determined that the purchase ratio is not higher than the first reference value, checking whether the interest ratio is higher than a preset second reference value; providing a sales strategy for recommending a discount event for the first product to the first seller terminal when it is confirmed that the interest rate is higher than the second reference value; and if it is confirmed that the interest rate is not higher than the second reference value, providing a sales strategy for recommending a change to the detailed information of the first product registered on the first page to the first seller terminal. may include more.

According to an embodiment, receiving a request for generating an overseas sales page for a first product from a first seller terminal, encoding an image of the first product to generate a first input signal, and applying the first input signal to a first artificial neural network input to , obtain a first output signal based on a result of the input of the first artificial neural network, and identify a first category that is a type of a first product based on the first output signal, among overseas sales sites A first site that operates the first category as a main category is checked, and based on a first template that is a template used in the first site, a first page that is a sales page of a first product is generated, and the first page is displayed. By uploading and registering on the first site, there is an effect that a sales page can be automatically created and uploaded according to a template for each overseas sales site.

On the other hand, the effects according to the embodiments are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those of ordinary skill in the art from the following description.

1 is a diagram schematically showing the configuration of a system according to an embodiment.
2 is a flowchart illustrating a process of automatically uploading an overseas sales page based on artificial intelligence according to an embodiment.
3 is a flowchart illustrating a process of generating a sales page by setting a sales price of a product according to an exemplary embodiment.
4 is a flowchart illustrating a process of recommending a product sales strategy through a purchase ratio and an interest ratio according to an exemplary embodiment.
5 is a flowchart illustrating a process of registering a thumbnail of a sales page on a sales site according to a representative color of a product according to an exemplary embodiment.
6 is a diagram for explaining a first artificial neural network according to an embodiment.
7 is a diagram for explaining a method of learning a first artificial neural network according to an embodiment.
8 is a flowchart for explaining a process of managing the amount of products taken out according to an exemplary embodiment.
9 is a diagram for explaining a second artificial neural network according to an embodiment.
10 is a flowchart for explaining a process of confirming the amount of products taken out through product sales details according to an exemplary embodiment.
11 is an exemplary diagram of a configuration of an apparatus according to an embodiment.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, since various changes may be made to the embodiments, the scope of the patent application is not limited or limited by these embodiments. It should be understood that all modifications, equivalents and substitutes for the embodiments are included in the scope of the rights.

Specific structural or functional descriptions of the embodiments are disclosed for purposes of illustration only, and may be changed and implemented in various forms. Accordingly, the embodiments are not limited to a specific disclosure form, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical spirit.

Although terms such as first or second may be used to describe various elements, these terms should be interpreted only for the purpose of distinguishing one element from another. For example, a first component may be termed a second component, and similarly, a second component may also be termed a first component.

When a component is referred to as being “connected” to another component, it may be directly connected or connected to the other component, but it should be understood that another component may exist in between.

Terms used in the examples are used for the purpose of description only, and should not be construed as limiting. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present specification, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It is to be understood that it does not preclude the possibility of the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiment belongs. Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

In addition, in the description with reference to the accompanying drawings, the same components are assigned the same reference numerals regardless of the reference numerals, and the overlapping description thereof will be omitted. In the description of the embodiment, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the embodiment, the detailed description thereof will be omitted.

The embodiments may be implemented in various types of products, such as personal computers, laptop computers, tablet computers, smart phones, televisions, smart home appliances, intelligent cars, kiosks, wearable devices, and the like.

In an embodiment, the artificial intelligence (AI) system is a computer system that implements human-level intelligence, and is a system in which a machine learns and makes decisions on its own, unlike an existing rule-based smart system. The more the AI system is used, the better the recognition rate and the more accurate understanding of the seller's tastes, and the existing rule-based smart systems are gradually being replaced by deep learning-based AI systems.

Artificial intelligence technology consists of machine learning and element technologies using machine learning. Machine learning is an algorithm technology that categorizes/learns characteristics of input data by itself, and element technology uses machine learning algorithms such as deep learning to simulate functions such as cognition and judgment of the human brain. It consists of technical fields such as understanding, reasoning/prediction, knowledge expression, and motion control.

The various fields where artificial intelligence technology is applied are as follows. Linguistic understanding is a technology for recognizing and applying/processing human language/text, and includes natural language processing, machine translation, dialogue system, question and answer, and speech recognition/synthesis. Visual understanding is a technology for recognizing and processing objects like human vision, and includes object recognition, object tracking, image search, human recognition, scene understanding, spatial understanding, image improvement, and the like. Inferential prediction is a technology for logically reasoning and predicting by judging information, and includes knowledge/probability-based reasoning, optimization prediction, preference-based planning, and recommendation. Knowledge expression is a technology that automatically processes human experience information into knowledge data, and includes knowledge construction (data generation/classification) and knowledge management (data utilization). Motion control is a technology for controlling autonomous driving of a vehicle and movement of a robot, and includes motion control (navigation, collision, driving), manipulation control (action control), and the like.

In general, in order to apply the machine learning algorithm to real life, learning is performed in the Trial and Error method due to the characteristics of the basic methodology of machine learning. In particular, deep learning requires hundreds of thousands of iterations. It is impossible to execute this in the actual physical external environment, so instead, the actual physical external environment is implemented on a computer and learning is performed through simulation.

1 is a diagram schematically showing the configuration of a system according to an embodiment.

Referring to FIG. 1 , a system according to an embodiment may include a plurality of seller terminals 100 and devices 200 capable of communicating with each other through a communication network.

First, the communication network may be configured regardless of its communication mode, such as wired and wireless, and may be implemented in various forms so that communication between a server and a server and communication between a server and a terminal are performed.

Each of the plurality of seller terminals 100 may be a desktop computer, a notebook computer, a tablet, a smart phone, or the like. Preferably, as shown in FIG. 1 , the first seller terminal 110 , the second seller terminal 120 , the third seller terminal 130 , etc. may be a smartphone, and may be employed differently depending on the embodiment. .

Each of the plurality of seller terminals 100 may be configured to perform all or part of an arithmetic function, a storage/referencing function, an input/output function, and a control function of a typical computer. A plurality of seller terminals 100 may be configured to communicate with the device 200 in a wired or wireless manner.

Each of the plurality of seller terminals 100 is connected to a web page operated by a person or group that provides a service using the device 200 , or is developed and distributed by a person or group that provides a service using the device 200 . One application can be installed. Each of the plurality of seller terminals 100 may be linked with the device 200 through a web page or an application.

Each of the plurality of seller terminals 100 may access the device 200 through a web page or an application provided by the device 200 .

Hereinafter, for convenience of explanation, the operation of the first seller terminal 110 will be mainly described, but the operation of the first seller terminal 110 may be performed instead of the second seller terminal 120 in other seller terminals, such as to be.

The device 200 may be a self-server owned by a person or organization that provides services using the device 200, a cloud server, or a peer-to-peer (p2p) set of distributed nodes. may be The device 200 may be configured to perform all or part of an arithmetic function, a storage/referencing function, an input/output function, and a control function of a typical computer. The device 200 may include at least one artificial neural network that performs an inference function.

The device 200 may be configured to communicate with a plurality of seller terminals 100 by wire and wirelessly, and controls the operation of each of the plurality of seller terminals 100 , and which information is displayed on each screen of the plurality of seller terminals 100 . You can control what is displayed.

On the other hand, for convenience of explanation, in FIG. 1 , only the first seller terminal 110 , the second seller terminal 120 and the third seller terminal 130 are shown among the plurality of seller terminals 100 , but the number of terminals is implemented It may vary depending on the example. As long as the processing capacity of the apparatus 200 allows, the number of terminals is not particularly limited.

According to an embodiment, the device 200 classifies a product category based on artificial intelligence, classifies and selects a market for uploading to an overseas market according to the characteristics of the product, and a sales page according to the overseas market and product properties You can create a template of , and automatically create and upload an overseas sales page through this.

In the present invention, artificial intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, perceptual ability, etc., and implements it with a computer, and the concepts of machine learning, symbolic logic, etc. may include Machine Learning (ML) is an algorithm technology that classifies or learns characteristics of input data by itself. Artificial intelligence technology is an algorithm of machine learning that can analyze input data, learn the results of the analysis, and make judgments or predictions based on the results of the learning. In addition, technologies that use machine learning algorithms to simulate functions such as cognition and judgment of the human brain can also be understood as a category of artificial intelligence. For example, technical fields such as verbal comprehension, visual comprehension, reasoning/prediction, knowledge expression, and motion control may be included.

Machine learning may refer to the process of training a neural network model using experience of processing data. With machine learning, computer software could mean improving its own data processing capabilities. The neural network model is constructed by modeling the correlation between data, and the correlation may be expressed by a plurality of parameters. A neural network model extracts and analyzes features from given data to derive correlations between data, and repeating this process to optimize parameters of a neural network model can be called machine learning. For example, a neural network model may learn a mapping (correlation) between an input and an output with respect to data given as an input/output pair. Alternatively, the neural network model may learn the relationship by deriving regularity between the given data even when only input data is given.

The artificial intelligence learning model or neural network model may be designed to implement a human brain structure on a computer, and may include a plurality of network nodes that simulate neurons of a human neural network and have weights. A plurality of network nodes may have a connection relationship with each other by simulating a synaptic activity of a neuron through which a neuron sends and receives a signal through a synapse. In the AI learning model, a plurality of network nodes can exchange data according to a convolutional connection relationship while being located in layers of different depths. The AI learning model may be, for example, an artificial neural network model, a convolutional neural network model, or the like. As an embodiment, the AI learning model may be machine-learned according to a method such as supervised learning, unsupervised learning, reinforcement learning, or the like. Machine learning algorithms for performing machine learning include Decision Tree, Bayesian Network, Support Vector Machine, Artificial Neural Network, Ada-boost. , Perceptron, Genetic Programming, Clustering, etc. may be used.

Among them, CNNs are a type of multilayer perceptrons designed to use minimal preprocessing. CNN consists of one or several convolutional layers and general artificial neural network layers on top of it, and additionally utilizes weights and pooling layers. Thanks to this structure, CNN can fully utilize the input data of the two-dimensional structure. Compared with other deep learning structures, CNN shows good performance in both video and audio fields. CNNs can also be trained through standard back-passing. CNNs are easier to train than other feed-forward neural network techniques and have the advantage of using fewer parameters.

Convolutional networks are neural networks that contain sets of nodes with bound parameters. The increased size of available training data and the availability of computational power, combined with advances in algorithms such as piecewise linear units and dropout training, have greatly improved many computer vision tasks. For huge data sets, such as those available for many tasks today, overfitting is not important, and increasing the size of the network improves test accuracy. Optimal use of computing resources becomes a limiting factor. To this end, a distributed, scalable implementation of deep neural networks may be used.

2 is a flowchart illustrating a process of automatically uploading an overseas sales page based on artificial intelligence according to an embodiment.

Referring to FIG. 2 , first, in step S201 , the device 200 may receive a request for generating an overseas sales page for a first product from the first seller terminal 110 . Here, the request to create an overseas sales page for the first product is a request to create and upload a sales page of the first product on an overseas sales site in order to sell the first product in an overseas market. 1 It may include detailed information of the product, first seller information, and the like.

In step S202 , the device 200 may generate a first input signal by encoding the image of the first product.

Specifically, the device 200 may generate the first input signal by encoding pixels of the image of the first product into color information. The color information may include, but is not limited to, RGB color information, brightness information, and saturation information. The apparatus 200 may convert the color information into a numerical value, and may encode the image of the first product in the form of a data sheet including the value.

In step S203 , the device 200 may input the first input signal to the pre-trained first artificial neural network in the device 200 .

The first artificial neural network according to an embodiment is composed of a feature extraction neural network and a classification neural network, and the feature extraction neural network sequentially stacks a convolutional layer and a pooling layer on an input signal. The convolution layer includes a convolution operation, a convolution filter, and an activation function. The calculation of the convolution filter is adjusted according to the matrix size of the target input, but a 9X9 matrix is generally used. The activation function generally uses, but is not limited to, a ReLU function, a sigmoid function, and a tanh function. The pooling layer is a layer that reduces the size of the input matrix, and uses a method of extracting representative values by tying pixels in a specific area. In general, the average value or the maximum value is often used for the calculation of the pooling layer, but is not limited thereto. The operation is performed using a square matrix, usually a 9x9 matrix. The convolutional layer and the pooling layer are repeated alternately until the corresponding input becomes small enough while maintaining the difference.

According to an embodiment, the classification neural network has a hidden layer and an output layer. Classification of the first artificial neural network for product type detection The neural network consists of 5 or less hidden layers, and may include a total of 50 or less hidden layer nodes. The activation function of the hidden layer uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. There is a total of one output layer node in the classification neural network, and the output value for the type of product can be output to the output layer node. A detailed description of the first artificial neural network will be described later with reference to FIG. 6 .

In step S204 , the device 200 may obtain a first output signal based on a result of the input of the first artificial neural network.

In step S205 , the device 200 may identify the first category, which is the type of the first product, based on the first output signal.

For example, as a result of checking the output value of the first output signal, when the output value is 1, the device 200 identifies the type of the first product as a cosmetic category, and when the output value is 2, sets the type of the first product to clothing When the category is identified and the output value is 3, the type of the first product may be identified as the home appliance category.

In step S206 , the device 200 may identify a first site operating a first category as a main category among overseas sales sites. To this end, main category information divided by overseas sales sites is stored in the database of the device 200 , and the device 200 searches the database to check the first site operating the first category as the main category. can

In operation S207 , the device 200 may generate a first page that is a sales page of the first product based on the first template that is a template used in the first site.

Specifically, the device 200 may obtain the first template information from the server operating the first site, and may generate a sales page frame based on the first template information, The first page may be created by inserting an image of the first product, detailed information of the first product, and first seller information.

When generating the first page, the device 200 may additionally provide an automatic translation function. For example, in the case where, in the detailed information of the first product, contents describing specifications, functions, etc. of the first product are written in the first language and the first site is being serviced based on the second language, the device 200 ) may generate the first page by inserting the translated content into the second language after processing the content of the detailed information of the first product to be automatically translated from the first language to the second language.

In step S208 , the device 200 may upload the first page to the first site and register it. In this case, the device 200 may transmit an upload request for the first page to a server operating the first site, and process the first page to be registered in the first site.

3 is a flowchart illustrating a process of generating a sales page by setting a sales price of a product according to an exemplary embodiment.

Referring to FIG. 3 , first, in step S301 , the device 200 may identify a first site operating a first category among overseas sales sites as a main category.

In step S302 , the device 200 may check whether the first product is being sold on the first site.

Specifically, the device 200 may search sales pages of products sold on the first site through crawling, and check whether the first product is being sold on the first site. In this case, the device 200 checks the product name of the first product in the detailed information of the first product, and checks whether the first product is being sold on the first site through a keyword search based on the product name of the first product. Also, it is possible to check whether the first product is being sold on the first site through an image search based on the image of the first product.

If it is determined in step S302 that the first product is being sold on the first site, in step S303 , the device 200 may check a first price that is the lowest price of the first product being sold on the first site.

Specifically, when the first product is sold through one sales page on the first site, the device 200 may check the price sold on the corresponding sales page as the first price, and the first product on the first site When the sale is performed through the plurality of sale pages, a price sold at the lowest price among the plurality of sale pages may be identified as the first price.

In step S304 , the device 200 may check whether the second price set by the first seller is higher than the first price. Here, the second price may be confirmed through detailed information of the first product.

If it is confirmed in step S304 that the second price is higher than the first price, in step S305, the device 200 sends a notification message to the first seller terminal 110 informing that the first product is being sold abroad at the first price. ) can be transmitted.

If it is determined in step S304 that the second price is not higher than the first price, in step S306 , the device 200 may set the second price as the selling price of the first product to generate the first page. Here, the second price may be displayed as the selling price of the first product in the first page.

On the other hand, if it is determined in step S302 that the first product is not being sold on the first site, in step S306 , the device 200 sets the second price as the selling price of the first product to generate the first page. can

4 is a flowchart illustrating a process of recommending a product sales strategy through a purchase ratio and an interest ratio according to an exemplary embodiment.

Referring to FIG. 4 , first, in step S401 , the device 200 uploads and registers the first page to the first site, the number of views of customers who have viewed the first page, and the first page after inquiring the first page The number of interests of customers who put the page in their shopping cart and the number of purchases of customers who purchased the first product through the first page may be checked, respectively.

Specifically, the device 200 clicks a thumbnail of the first page to check detailed information about the first product on the first site, and based on the number of views of the first page, the first page You can check the number of views of customers who have viewed . Here, the thumbnail of the first page is miniaturized by reducing the first page so that it can be recognized at a glance, and may consist of a product name of the first product, an image of the first product, etc., and when the thumbnail of the first page is clicked, the 1 Pages can be linked and displayed on the screen.

Also, the device 200 checks the number of interests of customers who have placed the first page in their shopping cart based on the number of customers who have placed the first product in their shopping cart because they are interested in the first product among customers who have viewed the first page. can

Also, the device 200 may check the number of purchases by customers who purchased the first product through the first page, based on the number of customers who actually purchased the first product among customers who put the first product in their shopping cart.

In step S402 , the device 200 may calculate a purchase ratio based on a value obtained by dividing the number of purchases by the number of inquiries, and may calculate an interest ratio based on a value obtained by dividing the number of interests by the number of inquiries.

For example, when the number of views is 100, the number of interests is 30, and the number of purchases is 10, the device 200 may calculate the purchase rate as 10% and the interest rate as 30%.

In step S403 , the device 200 may determine whether the purchase ratio is higher than the first reference value. Here, the first reference value may be set differently according to embodiments.

If it is confirmed in step S403 that the purchase ratio is higher than the first reference value, in step S405 , the device 200 provides the first seller terminal 110 with a sales strategy that recommends maintaining the current sales policy of the first product. can

That is, when it is confirmed that the purchase ratio is higher than the first reference value, the device 200 analyzes that the number of purchases is higher than the number of inquiries, and sets a sales strategy that maintains the current without changing the sales policy of the first product. Can recommend to seller.

If it is determined in step S403 that the purchase ratio is not higher than the first reference value, in step S404 , the device 200 may determine whether the interest ratio is higher than the second reference value. Here, the second reference value may be set differently according to embodiments.

If it is determined in step S404 that the interest rate is higher than the second reference value, in step S406 , the device 200 may provide a sales strategy for recommending a discount event for the first product to the first seller terminal 110 .

That is, if the purchase ratio is lower than the first reference value but the interest ratio is higher than the second reference value, the device 200 analyzes that the number of purchases is not high compared to the number of views, but the number of interests is higher than the number of inquiries, A sales strategy for increasing the number of purchases of the first product through the discount event for the first product may be recommended to the first seller.

If it is determined in step S404 that the interest rate is not higher than the second reference value, in step S407, the device 200 applies a sales strategy for recommending a change to the detailed information of the first product registered on the first page to the first seller It may be provided to the terminal 110 .

That is, when it is confirmed that the purchase rate is lower than the first reference value and the interest rate is lower than the second reference value, the device 200 analyzes that both the number of purchases and the number of interest are not high compared to the number of views, A sales strategy of changing an image, a description of the first product, and the like may be recommended to the first seller.

5 is a flowchart illustrating a process of registering a thumbnail of a sales page on a sales site according to a representative color of a product according to an exemplary embodiment.

Referring to FIG. 5 , first, in step S501 , the device 200 analyzes the color of the first product for each pixel based on the image of the first product, and classifies all pixels of the first product according to color series. can do. Here, the color series may be divided into color series such as red, orange, yellow, green, blue, indigo, and purple.

For example, when the RGB value of the first pixel among pixels included in the first product is (250, 10, 0) and the RGB value of the second pixel is (250, 0, 10), the device 200 is The first pixel and the second pixel may be classified into a red color series. That is, the device 200 may classify pixels exhibiting a color somewhat similar to the red color into a red color series. In this case, when the image of the first product consists of 10,000 pixels, the device 200 may analyze the color for each 10,000 pixels and classify the 10,000 pixels according to a color series, respectively.

In operation S502 , as a result of classifying all pixels of the first product according to color series, the device 200 may identify the first color series having the largest number of pixels.

For example, as a result of classifying all pixels of the first product according to color series, the number of pixels classified as red color series is 5,000, the number of pixels classified as yellow color series is 3,000, and pixels classified as blue color series When the number of is 1,000, the device 200 may identify the first color series as the red color series.

In operation S503 , the device 200 may calculate the first ratio through a value obtained by dividing the number of pixels of the first color series by the total number of pixels of the first product.

For example, when the number of pixels of the first color series is 5,000 and the total number of pixels of the first product is 10,000, the device 200 may calculate the first ratio as 50%.

In step S504 , the device 200 may determine whether the first ratio is greater than the reference ratio. Here, the reference ratio may be set differently according to embodiments.

If it is determined in step S504 that the first ratio is greater than the reference ratio, in step S505 , the device 200 may set the representative color of the first product as the first color series.

In step S506 , the device 200 may register the thumbnail of the first page in the posting space of the first color series among the posting spaces for each color series provided on the first site. To this end, a posting space is provided for each color series in the first site.

That is, when the type of the first product is identified as the first category and the representative color of the first product is set as the first color series, the device 200 is provided in the first site operating the first category as the main category. The thumbnail of the first page may be registered in the posting space of the first color series.

Meanwhile, if it is determined in step S504 that the first ratio is smaller than the reference ratio, in step S507 , the device 200 classifies all pixels of the first product according to color series. The second color series can be confirmed.

For example, as a result of classifying all pixels of the first product according to color series, the number of pixels classified as red color series is 3,000, the number of pixels classified as yellow color series is 2,000, and pixels classified as blue color series When the number of is 1,000, the device 200 may identify the first color series as the red color series and the second color series as the yellow color series.

In step S508 , the device 200 sums the number of pixels of the first color series and the number of pixels of the second color series, and calculates the second ratio through a value obtained by dividing the sum of the number of pixels by the total number of pixels of the first product can do.

For example, if the number of pixels in the first color series is 3,000, the number of pixels in the second color series is 2,000, and the total number of pixels in the first product is 10,000, the device 200 may calculate the second ratio as 50% have.

In step S509 , the device 200 may check whether the second ratio is greater than the reference ratio.

If it is determined in step S509 that the second ratio is greater than the reference ratio, in step S510 , the device 200 may derive the third color by mixing the first color and the second color.

For example, when the first color series is a red color series and the second color series is a yellow color series, the device 200 may derive an orange color by mixing a red color and a yellow color.

In step S511 , the device 200 may set the representative color of the first product as the third color series.

In operation S512 , the device 200 may register the thumbnail of the first page in the posting space of the third color series among the posting spaces for each color series provided on the first site.

Meanwhile, if it is determined in step S509 that the second ratio is smaller than the reference ratio, in step S513 , the device 200 may set the representative color of the first product as unclassified.

In operation S514 , the device 200 may register the thumbnail of the first page in the unclassified posting space among the posting spaces for each color series provided on the first site.

6 is a diagram for explaining a first artificial neural network according to an embodiment.

The first artificial neural network 600 according to an embodiment may take a first input signal generated by encoding a product image as an input, and may output information indicating the type of product.

Encoding according to an embodiment may be performed by storing color information for each pixel of an image in the form of a digitized data sheet, and the color information may include RGB color, brightness information, and saturation information of one pixel. , but not limited thereto.

According to an embodiment, the first artificial neural network 600 is composed of a feature extraction neural network 610 and a classification neural network 620, and the feature extraction neural network 610 performs a task of separating a product and a background from an image. In addition, the classification neural network 620 may perform a task of identifying which category the product in the image belongs to.

The method in which the feature extraction neural network 610 distinguishes the product from the background is that the change in each value of color information from the data sheet of the first input signal encoding the image is 30% in 6 or more of 8 pixels including one pixel A group of pixels detected as having an abnormal change may be used as the boundary between the product and the background, but is not limited thereto.

The feature extraction neural network 610 stacks the input signal by sequentially stacking a convolutional layer and a pooling layer. The convolution layer includes a convolution operation, a convolution filter, and an activation function. The calculation of the convolution filter is adjusted according to the matrix size of the target input, but generally a 9X9 matrix is used. The activation function generally uses, but is not limited to, a ReLU function, a sigmoid function, and a tanh function. The pooling layer is a layer that reduces the size of the input matrix, and uses a method of extracting representative values by tying pixels in a specific area. In general, the average value or the maximum value is often used for the calculation of the pooling layer, but is not limited thereto. The operation is performed using a square matrix, usually a 9x9 matrix. The convolutional layer and the pooling layer are iterated alternately until the corresponding input becomes small enough while maintaining the difference.

The classification neural network 620 confirms the shape of the product separated from the background through the feature extraction neural network 610, and checks whether the product in the image belongs to which category by checking whether it is similar to the representative product for each predefined category can figure out Information stored in the database of the device 200 may be used for product comparison.

The classification neural network 620 prioritizes the task of identifying the category to which the product belongs, and may also grasp the shape and size of the product.

The classification neural network 620 has a hidden layer and an output layer, and is composed of five or less hidden layers, including a total of 50 or less hidden layer nodes, and the activation function of the hidden layer is a ReLU function and a sigmoid function. and tanh functions, but is not limited thereto.

The classification neural network 620 may include only one output layer node in total.

The output of the classification neural network 620 is an output value for the type of product, and may indicate to which category the product type belongs. For example, when the output value is 1, it indicates that the type of product is a cosmetic category, when the output value is 2, it indicates that the type of product is a clothing category, and when the output value is 3, the type of product is a household appliance category can indicate that

According to one embodiment, the first artificial neural network 600 receives the first learning signal generated by the corrected correct answer input by the user when the user discovers a problem with the output according to the first artificial neural network 600 to learn. can The problem of output according to the first artificial neural network 600 may mean a case in which output values indicating different categories of products are output.

The first learning signal according to an embodiment is created based on the error between the correct answer and the output value, and in some cases, SGD using delta, a batch method, or a method following a backpropagation algorithm may be used. The first artificial neural network 600 performs learning by modifying the existing weights according to the first learning signal, and may use momentum in some cases. A cost function can be used to calculate the error, and a cross entropy function can be used as the cost function. Hereinafter, the learning contents of the first artificial neural network 600 will be described with reference to FIG. 7 .

7 is a diagram for explaining a method of learning a first artificial neural network according to an embodiment.

According to an embodiment, the learning apparatus may train the first artificial neural network 600 . The learning device may be a separate entity different from the device 200 , but is not limited thereto.

According to an embodiment, the first artificial neural network 600 includes an input layer to which training samples are input and an output layer for outputting training outputs, and can be learned based on a difference between the training outputs and the first labels. have. Here, the first labels may be defined based on a representative image for each type of product. The first artificial neural network 600 is connected to a group of a plurality of nodes, and is defined by weights between the connected nodes and an activation function that activates the nodes.

The learning apparatus may train the first artificial neural network 600 by using a Gradient Decent (GD) technique or a Stochastic Gradient Descent (SGD) technique. The learning apparatus may use a loss function designed by the outputs and labels of the first artificial neural network 600 .

The learning apparatus may calculate a training error using a predefined loss function. The loss function may be predefined with a label, an output, and a parameter as input variables, where the parameter may be set by weights in the first artificial neural network 600 . For example, the loss function may be designed in a Mean Square Error (MSE) form, an entropy form, or the like, and various techniques or methods may be employed in an embodiment in which the loss function is designed.

The learning apparatus may find weights affecting the training error by using a backpropagation technique. Here, the weights are relationships between nodes in the first artificial neural network 600 . The learning apparatus may use the SGD technique using labels and outputs to optimize the weights found through the backpropagation technique. For example, the learning apparatus may update the weights of the loss function defined based on the labels, outputs, and weights using the SGD technique.

According to an embodiment, the learning apparatus may obtain the labeled training product images 701 from the database. The learning apparatus may obtain pre-labeled information on each of the product images 701 set for each type of product, and the product images 701 may be labeled according to the pre-classified product information.

According to an embodiment, the learning device may acquire 1000 labeled training product images 701 , and generate first training product vectors 702 based on the labeled training product images 701 . can Various schemes may be employed for extracting the first training product vectors 702 .

According to an embodiment, the learning apparatus may obtain first training outputs 703 by applying the first training product vectors 702 to the first artificial neural network 600 . The learning apparatus may train the first artificial neural network 600 based on the first training outputs 703 and the first labels 704 . The learning apparatus calculates training errors corresponding to the first training outputs 703 , and optimizes the connection relationship of nodes in the first artificial neural network 600 to minimize the training errors to create the first artificial neural network 600 . can learn The device 200 may identify the type of product in the image by using the first artificial neural network 600 on which the learning has been completed.

8 is a flowchart for explaining a process of managing a quantity taken out of a product according to an exemplary embodiment.

Referring to FIG. 8 , first, in step S801 , the device 200 may acquire a first image generated by photographing a product stored in a warehouse of a first seller at a first time point.

Specifically, a warehouse for storing products may be provided in a company operated by the first seller, and a storage box for storing products may be provided inside the warehouse. The first seller terminal 110 may generate a first image by performing a photographing of the product displayed in the storage box of the warehouse using the camera provided in the first seller terminal 110 at a first time point. When the first image is generated by photographing the product stored in the warehouse of the first seller at the first time point, the device 200 may acquire the first image from the first seller terminal 110 .

According to an embodiment, the camera included in the first seller terminal 110 is a full HD (High Definition) quality camera equipment for image capturing, and may be a filming equipment based on 30 to 60 frames. The camera can automatically adjust exposure, shutter speed, and sensitivity, and manual adjustment may be possible according to the seller's convenience.

In step S802 , the device 200 may generate a second input signal by encoding the first image.

Specifically, the device 200 may generate the first input signal by encoding the pixels of the first image with color information. The color information may include, but is not limited to, RGB color information, brightness information, and saturation information. The apparatus 200 may convert color information into a numerical value, and may encode an image in the form of a data sheet including this value.

In step S803 , the device 200 may input the second input signal to the second artificial neural network.

According to an embodiment, the second artificial neural network is implemented as a convolutional neural network, and the convolutional neural network is composed of a feature extraction neural network and a classification neural network, and the feature extraction neural network stacks an input signal with a convolutional layer and a pooling layer sequentially. do. The convolution layer includes a convolution operation, a convolution filter, and an activation function. The calculation of the convolution filter is adjusted according to the matrix size of the target input, but a 9X9 matrix is generally used. The activation function generally uses, but is not limited to, a ReLU function, a sigmoid function, and a tanh function. The pooling layer is a layer that reduces the size of the input matrix, and uses a method of extracting representative values by tying pixels in a specific area. In general, the average value or the maximum value is often used for the calculation of the pooling layer, but is not limited thereto. The operation is performed using a square matrix, usually a 9x9 matrix. The convolutional layer and the pooling layer are repeated alternately until the corresponding input becomes small enough while maintaining the difference.

According to one embodiment, the classification neural network has a hidden layer and an output layer. In a convolutional neural network for an AI-based product export management method, there are generally three or more hidden layers, and 100 nodes for each hidden layer are specified, but more can be specified in some cases. The activation function of the hidden layer uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. A total of 50 output layer nodes of a convolutional neural network can be made. A detailed description of the convolutional neural network will be described later with reference to FIG. 9 .

In operation S804 , the device 200 may obtain a second output signal that is an output value of the second artificial neural network based on the result of the input of the second artificial neural network that is the convolutional neural network.

According to an embodiment, 50 output layer nodes of the convolutional neural network may include upper 25 output layer nodes and lower 25 output layer nodes. Among the 50 output layer nodes of the convolutional neural network, the top 25 output layer nodes can indicate the type of product. The lower 25 nodes may indicate the inventory amount of products corresponding to the upper 25 nodes. A detailed description of the output of the convolutional neural network will be described later with reference to FIG. 9 .

In step S805 , the device 200 may analyze the type and inventory amount of the product stored at the first time point based on the second output signal. That is, the device 200 may analyze which types of products are stored in the warehouse at the first time point, and analyze how many stocks are there for each type of product.

In step S806 , the device 200 may acquire a second image generated by photographing the product stored in the warehouse of the first seller at the second time point. Here, the second time point means a time point after the reference period has elapsed from the first time point, and the reference period may be set differently according to embodiments. For example, when the first time point is 13:00 and the reference period is set to 5 hours, the second time point may be 18:00.

Specifically, the first seller terminal 110 may generate a second image by taking a picture of the product displayed in the storage box of the warehouse using the camera provided in the first seller terminal 110 at the second time point. have. When the second image is generated by photographing the product stored in the warehouse of the first seller at the second time point, the device 200 may acquire the second image from the first seller terminal 110 .

In step S807 , the device 200 may generate a third input signal by encoding the second image.

In step S808 , the device 200 may input the third input signal to the second artificial neural network.

In step S809 , the apparatus 200 may obtain a third output signal that is an output value of the second artificial neural network based on the result of the input of the second artificial neural network.

In step S810 , the device 200 may analyze the type and inventory amount of the product stored at the second time point based on the third output signal. That is, the device 200 may analyze which types of products are stored in the warehouse at the second time point, and analyze how many stocks are there for each type of product.

In step S811, the device 200 compares the type and inventory amount of the product stored at the first time point with the type and inventory amount of the product stored at the second time point, and there is a difference in the inventory amount between the first time point and the second time point. You can check whether the product is available or not.

For example, it is confirmed that 5 first products and 4 second products are stored at the first time point, and it is confirmed that 4 first products and 4 second products are stored at the second time point , the device 200 may determine that there is a difference in the stock amount of the first product between the first time point and the second time point.

If it is confirmed in step S811 that there is a difference in the stock amount of the first product, in step S812, the device 200 calculates the stock amount of the first product that is different between the first time point and the second time point as the amount of the first product out. have.

Specifically, when the stock amount of the first product stored in the warehouse at the first time point is the first quantity, and the inventory amount of the first product stored in the warehouse at the second time point is the second quantity, the device 200 may When it is confirmed that the first quantity and the second quantity are different, it can be confirmed that there is a difference in the inventory amount of the first product between the first time point and the second time point, and the quantity of the first product taken out by subtracting the second quantity from the first quantity can be calculated.

For example, if it is confirmed that five first products are stored in the warehouse at the first time point, and it is confirmed that four first products are stored in the warehouse at the second time point, the device 200 changes from five to five. By subtracting four, one can be calculated as the amount of transport of the first product.

9 is a diagram for explaining a second artificial neural network according to an embodiment.

According to an embodiment, the device 200 may generate an input signal by encoding an image generated by photographing through the camera of the first seller terminal 110 , and the second artificial neural network 901 inputs the input signal , the type 902 of the product and the stock amount 903 of the product may be output.

Encoding according to an embodiment may be performed by storing the color information for each pixel of the image in the form of a digitized data sheet, and the color information may include RGB color, brightness information, and saturation information of one pixel. , but not limited thereto.

According to an embodiment, the second artificial neural network 901 is implemented as a convolutional neural network, and the second artificial neural network 901 may be composed of a feature extraction neural network 910 and a classification neural network 920 , and a feature extraction neural network 910 may perform an operation of separating a product and a background from an image of the product, and the classification neural network 920 classifies the product by type therefrom, and performs an operation of identifying each stock amount by type of product can do. The method of the feature extraction neural network 910 to distinguish the product from the background is that the change in each value of color information from the data sheet of the input signal encoding the image is 30% or more in 6 or more of 8 pixels including one pixel. A group of pixels detected as having changed may be used as the boundary between the product and the background, but is not limited thereto.

According to an embodiment, the feature extraction neural network 910 stacks the input signal by sequentially stacking a convolutional layer and a pooling layer. The convolution layer includes a convolution operation, a convolution filter, and an activation function. The calculation of the convolution filter is adjusted according to the matrix size of the target input, but a 9X9 matrix is generally used. The activation function generally uses, but is not limited to, a ReLU function, a sigmoid function, and a tanh function. The pooling layer is a layer that reduces the size of the input matrix, and uses a method of extracting representative values by tying pixels in a specific area. In general, the average value or the maximum value is often used for the calculation of the pooling layer, but is not limited thereto. The operation is performed using a square matrix, usually a 9x9 matrix. The convolutional layer and the pooling layer are repeated alternately until the corresponding input becomes small enough while maintaining the difference.

The classification neural network 920 according to an embodiment classifies the product separated from the background through the feature extraction neural network 910, the product type 902 according to its shape and continuity, and the inventory amount 903 of the product. It can be identified by type. Information stored in the database can be used for product comparison. The classification neural network 920 prioritizes the task of identifying the type 902 of the product, and may facilitate the identification of the inventory amount 903 of the product according to the identified product shape and size.

According to one embodiment, the classification neural network 920 has a hidden layer and an output layer. In the second artificial neural network 901 in the device 200, there are generally 5 or more hidden layers, and 80 nodes of each hidden layer are designated, but more may be specified in some cases. The activation function of the hidden layer uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. The number of output layer nodes of the second artificial neural network 901 may be 50 in total.

As for the output of the second artificial neural network 901 according to an embodiment, the upper 25 nodes among 50 nodes of the output layer may indicate the type 902 of the product, and the lower 25 nodes may indicate products corresponding to the upper nodes, respectively. may indicate the inventory amount 903 of The method of matching the top 25 nodes with the bottom 25 nodes is a method of matching the top nth node and the bottom nth node, and the nth node in the whole corresponds to the 25+nth node in the whole can proceed with For example, the 1st node corresponds to the 26th node, the 2nd node corresponds to the 27th node, the 10th node corresponds to the 35th node, and the 25th node corresponds to the 50th node. can proceed in this way. The product type 902 may be output as code information corresponding to the product, but is not limited thereto. An output layer node having no output value among 50 output layer nodes of the second artificial neural network 901 may output the number '0' as its output value. Among the top 25 nodes, nodes containing this number ‘0’ can be considered as having no corresponding product. If there are more than 25 types of classified products, the remaining products may be automatically processed after all output values generated in advance are processed.

According to one embodiment, the second artificial neural network 901 learns by receiving the second learning signal generated by the corrected correct answer input by the user when the user discovers a problem in product identification according to the second artificial neural network 901 . can do. The problem of product identification according to the second artificial neural network 901 may mean a case in which there is a problem in the product type 902 and the inventory amount 903 of the product.

The second learning signal according to an embodiment is created based on the error between the correct answer and the output value, and in some cases, SGD using delta, a batch method, or a method following a backpropagation algorithm may be used. The second artificial neural network 901 performs learning by modifying the existing weights according to the second learning signal, and may use momentum in some cases. A cost function can be used to calculate the error, and a cross entropy function can be used as the cost function. The second artificial neural network 901 may be learned through the same method as that of the first artificial neural network 600 .

10 is a flowchart for explaining a process of confirming the amount of products taken out through product sales details according to an exemplary embodiment.

Referring to FIG. 10 , first, in step S1001 , the device 200 calculates the take-out amount of the first product, and when it is confirmed that the take-out amount of the first product is greater than 0, the product sales of the first seller sold during the reference period history can be obtained. Here, the reference period may mean a period from the first time point to the second time point, and the product sales details may be composed of a list indicating how many products were sold by the first seller at any time.

For example, when the first time point is 13:00 and the second time point is 18:00, the device 200 may obtain the product sales history of the first seller sold from 13:00 to 18:00.

In step S1002 , the device 200 may analyze the type and sales volume of products sold during the reference period, based on the sales details of the products of the first seller sold during the reference period.

For example, if the base period is from 13:00 to 18:00, in the sales history of products sold at 14:00, one first product and two second products are sold, and in the sales history of products sold at 15:00, When one second product and two third products are sold, the device 200 indicates that one first product, three second products, and two third products are sold during the reference period; You can analyze the types and sales volume of products sold.

In step S1003 , the device 200 may check the sales volume of the first product from the analysis result of the type and sales volume of the product sold during the reference period. That is, when it is confirmed that the first product is sold during the reference period, the device 200 may check the sales volume of the first product sold during the reference period.

In step S1004 , the device 200 may check whether the amount of the first product taken out and the sales volume of the first product match.

If it is confirmed in step S1004 that the amount of the first product taken out and the sales volume of the first product match, in step S1005 , the device 200 may determine that there is no problem in the amount of the first product taken out.

If it is determined in step S1004 that the amount of the first product taken out and the sales volume of the first product do not match, in step S1006 , the device 200 may determine that there is a problem in the amount of the first product taken out.

In step S1007 , the device 200 may transmit a notification message for confirming the amount of shipment for the first product to the first seller terminal 110 . Here, the carrying-out amount confirmation notification message for the first product is a notification message for informing that the carrying-out amount of the first product is different from the carrying-out amount and that confirmation of the carrying-out amount is required.

11 is an exemplary diagram of a configuration of an apparatus according to an embodiment.

The device 200 according to an embodiment includes a processor 210 and a memory 220 . The processor 210 may include at least one of the devices described above with reference to FIGS. 1 to 10 , or perform at least one method described above with reference to FIGS. 1 to 10 . A person or organization using the apparatus 200 may provide a service related to some or all of the methods described above with reference to FIGS. 1 to 10 .

The memory 220 may store information related to the above-described methods or a program in which methods to be described below are implemented. The memory 220 may be a volatile memory or a non-volatile memory.

The processor 210 may execute a program and control the device 200 . The code of the program executed by the processor 210 may be stored in the memory 220 . The device 200 may be connected to an external device (eg, a personal computer or a network) through an input/output device (not shown), and may exchange data through wired/wireless communication.

The device 200 may be used to train an artificial neural network or to use a learned artificial neural network. The memory 220 may include a learning or learned artificial neural network. The processor 210 may learn or execute an artificial neural network algorithm stored in the memory 220 . The apparatus 200 for learning an artificial neural network and the apparatus 200 for using the learned artificial neural network may be the same or may be separate.

The embodiments described above may be implemented by a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the apparatus, methods and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose or special purpose computers. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Software may comprise a computer program, code, instructions, or a combination of one or more of these, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

As described above, although the embodiments have been described with reference to the limited drawings, those skilled in the art may apply various technical modifications and variations based on the above. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (3)

A method for automatically uploading an overseas sales page of a product based on artificial intelligence, performed by a device, the method comprising:
receiving a request for generating an overseas sales page for the first product from the first seller terminal;
generating a first input signal by encoding the image of the first product;
inputting the first input signal to a first artificial neural network, and obtaining a first output signal based on a result of the input of the first artificial neural network;
identifying a first category that is a type of the first product based on the first output signal;
checking a first site operating the first category as a main category among overseas sales sites;
generating a first page that is a sales page of the first product based on a first template that is a template used in the first site;
uploading and registering the first page to the first site;
analyzing the color of the first product for each pixel based on the image of the first product, and classifying all pixels of the first product according to color series;
identifying a first color series having the largest number of pixels as a result of classifying all pixels of the first product according to color series;
calculating a first ratio through a value obtained by dividing the number of pixels of the first color series by the total number of pixels of the first product;
checking whether the first ratio is greater than a preset reference ratio;
When it is confirmed that the first ratio is greater than the reference ratio, the representative color of the first product is set as the first color series, and the first color series is displayed in the posting space for each color series provided on the first site. registering a thumbnail of the first page in a posting space;
when it is confirmed that the first ratio is equal to or less than the reference ratio, as a result of classifying all pixels of the first product according to color series, identifying a second color series in which the number of pixels is second to that of the first color series;
summing the number of pixels of the first color series and the number of pixels of the second color series, and calculating a second ratio through a value obtained by dividing the sum of the number of pixels by the total number of pixels of the first product;
determining whether the second ratio is greater than the reference ratio;
When it is confirmed that the second ratio is greater than the reference ratio, a third color is derived by mixing the first color and the second color, and a representative color of the first product is set as the third color series, registering the thumbnail of the first page in the posting space of the third color series among the posting spaces for each color series provided in the first site; and
When it is confirmed that the second ratio is less than or equal to the reference ratio, the representative color of the first product is set as unclassified, and the thumbnail of the first page is displayed in the unclassified posting space among the posting spaces for each color series provided on the first site. comprising the step of registering;
How to automatically upload overseas sales pages for AI-based products. The method of claim 1,
Creating the first page comprises:
checking whether the first product is being sold on the first site;
when it is confirmed that the first product is being sold at the first site, checking a first price that is the lowest price of the first product being sold at the first site;
checking whether a second price set by the first seller is higher than the first price;
when it is confirmed that the second price is higher than the first price, transmitting a notification message informing that the first product is being sold abroad at the first price to the first seller terminal;
generating the first page by setting the second price as the selling price of the first product when it is determined that the second price is not higher than the first price; and
generating the first page by setting the second price as the selling price of the first product when it is confirmed that the first product is not sold on the first site;
How to automatically upload overseas sales pages for AI-based products. The method of claim 1,
After the step of uploading and registering the first page to the first site,
The number of views by customers who viewed the first page, the number of interests of customers who added the first product to their shopping cart after viewing the first page, and the number of purchases by customers who purchased the first product through the first page checking each of them;
calculating a purchase ratio based on a value obtained by dividing the number of purchases by the number of inquiries, and calculating an interest ratio based on a value obtained by dividing the number of interests by the number of inquiries;
checking whether the purchase ratio is higher than a preset first reference value;
if it is confirmed that the purchase ratio is higher than the first reference value, providing a sales strategy for recommending maintaining the current sales policy of the first product to the first seller terminal;
when it is determined that the purchase ratio is not higher than the first reference value, checking whether the interest ratio is higher than a preset second reference value;
providing a sales strategy for recommending a discount event for the first product to the first seller terminal when it is confirmed that the interest rate is higher than the second reference value; and
If it is confirmed that the interest rate is not higher than the second reference value, providing a sales strategy for recommending a change to the detailed information of the first product registered on the first page to the first seller terminal containing,
How to automatically upload overseas sales pages for AI-based products.

Images