KR102108549B1 - Method and apparatus for distribution management of tile and sanitary ware - Google Patents

Abstract

Disclosed are a method for managing distribution of tiles and sanitary ware, which applies deep learning technology to increase correctness and efficiency, and an apparatus thereof. According to one embodiment of the present invention, a method for managing distribution of tiles and sanitary ware based on artificial intelligence comprises: a step of scanning a unit QR code of a target tile and sanitary ware pallet and loading the pallet to each position of a carrier, wherein the pallet includes the tiles and sanitary ware of the same maker, size, and kind; when a motion sensor installed in the carrier detects abnormal movement, capturing the target tile and sanitary ware with five real-time monitoring cameras installed on four sides and an upper surface, wherein the abnormal movement includes movement of the tile and sanitary ware and contact and impact with an abnormal object; generating a first input signal based on a capturing result and inputting the first input signal to a convolutional neural network (CNN) in a blockchain network; acquiring a first output signal based on an input result; informing a user of the damaged pallet based on the first output signal; generating a learning signal in accordance with a user′s input based on a pallet checking result of the user; training the CNN based on the learning signal; and transmitting an order signal to a maker address stored in a first network storage device based on a training result.

Description

METHOD AND APPARATUS FOR DISTRIBUTION MANAGEMENT OF TILE AND SANITARY WARE}

The embodiments below relate to a method of managing distribution of tiles and sanitary ware.

The distribution of tiles and sanitary ware involves the importation of products from overseas into the country through ships. In this process, some breakage of the product may occur due to the shaking of the ship, and due to the nature of the product packaged in bulk on a pallet basis, it may be difficult to determine whether the product is damaged. Therefore, research on a system that can track damaged products and learn by itself is required.

KR101595538 KR1020190098865 KR102039351 KR1020190110370

The embodiments are intended to increase accuracy and efficiency by applying deep learning technology to the distribution management method of tiles and sanitary ware.

The embodiments are intended to increase the speed of the distribution process by automatically generating a reorder signal for the damaged product.

The embodiments are intended to provide continuous feedback by using errors in damage discrimination errors in the distribution management method of tiles and sanitary ware.

The embodiments intend to make information big data through a blockchain network and use enhanced security.

A method of managing distribution of tiles and sanitary ware according to an embodiment is a method of managing distribution of tiles and sanitary ware based on artificial intelligence, wherein the pallets of the target tiles and sanitary ware are the same manufacturer, the same Including size and type of tile and sanitary ware-scanning unit QR code and shipping by location; Photographing the target tile and sanitary ware with a total of 5 real-time monitoring cameras including the north-west, north-south, and top surfaces in the shipped distribution line; Based on the photographing result, the motion sensor detects abnormal movement-the abnormal movement includes movement of the target tile and sanitary ware or contact and impact of an abnormal object-upon detection, encodes the captured image 10 seconds before and after the detection point Generating a first input signal; Inputting the first input signal into a convolutional neural network in a blockchain network; Obtaining a first output signal based on the input result; Informing the user of the pallet information of the target tile and the sanitary ware, which has been determined to be damaged, based on the first output signal; Generating a learning signal according to a user input based on a result of the user's palette confirmation; Training the convolutional neural network based on the learning signal; And transmitting an order signal to a producer in the first network storage device based on the user's input.

According to an embodiment, the QR code includes route information capable of distinguishing the type of the target tile and the sanitary ware regardless of discoloration in the first quadrant based on the quadrant divided by the horizontal axis and the vertical axis; The second quadrant includes the amount of the target tile and sanitary ware included for each pallet; The third quadrant includes the size of the target tile and sanitary ware; And a manufacturer of the target tile and the sanitary ware in the fourth quadrant, and by registering the location information of the pallet holding each QR code when scanning the QR code, it is possible to estimate the location of the damaged pallet.

According to an embodiment, the blockchain network includes blocks including a database including captured image information, coordinate information, and damage determination information in a distribution line and the pre-trained convolutional neural network; Chains connecting each block in chronological order; And the network storage devices storing the respective blocks and chains, and the network storage devices include the producers producing the target tiles and sanitary ware in the distribution management method of the tiles and sanitary ware. The first network storage device; A second network storage device including a distributor for distributing the target tile and sanitary ware in the distribution management method of the tile and sanitary ware; A third network storage device including a seller who receives the target tile and the sanitary ware of the distribution management method of the tile and the sanitary ware and performs sales to an actual consumer; And a high-speed Internet connection network connecting each network storage device.

According to an embodiment, the database may include the photographed image information of preliminarily photographing tile and sanitary ware for a pallet unit in variously layered states in the distribution line; Color information capable of distinguishing the tile and the sanitary ware targeted from the pallet unit from the captured image information; The coordinate information representing the angle of view of the real-time monitoring cameras of the north-west, north-south, and top surfaces in X and Y axes; The damage discrimination information including comparison image information when the actual damage occurs when the abnormal movement is detected and the actual damage does not occur; Ordering information containing the ordering of the target tile and sanitary ware; And reorder information on the target tile and sanitary ware for which the damage is determined.

According to one embodiment, the convolutional neural network divides the shooting information of the real-time monitoring camera into 60 frames per second, and color information for each frame per pixel-the color information includes RGB information, brightness, and saturation per pixel- A first input signal recorded in the form of a digitized data sheet is used as an input, and based on the color information, whether damage and its coordinates are classified through the feature extraction neural network and the classification neural network of the convolutional neural network, and based on the classification , A first learning generated from an input of a result of re-checking the user regarding whether the damaged pallet is damaged or not as an output, and outputting a first output signal including location information of the damaged pallet determined from the QR code. You can learn through signals.

The apparatus according to an embodiment may be controlled by a computer program stored in a medium in order to execute the method of any one of the above-described methods in combination with hardware.

Embodiments can increase accuracy and efficiency by applying deep learning technology to the distribution management method of tiles and sanitary ware.

Embodiments can increase the speed of the distribution process by automatically generating a re-order signal for the product that is damaged.

Embodiments can provide continuous feedback by using errors in damage discrimination errors in the distribution management method of tiles and sanitary ware.

Embodiments can make information big data through a blockchain network and use enhanced security.

1 is a flow chart for explaining a method of managing distribution of tiles and sanitary ware according to an embodiment.
2 is a view for explaining a QR code according to an embodiment.
3 is a view for explaining a blockchain network according to an embodiment.
4 is a view for explaining a database according to an embodiment.
5 is a diagram for describing a convolutional neural network according to an embodiment.
6 is an exemplary view of a configuration of an apparatus according to an embodiment according to an embodiment.

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

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

The terms first or second may be used to describe various components, but these terms should be interpreted only for the purpose of distinguishing one component from other components. For example, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component.

When an element is said to be "connected" to another element, it should be understood that other elements may be present, either directly connected to or connected to the other element.

The terms used in the examples are for illustrative purposes only and should not be construed as limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, the terms "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts or combinations thereof described herein, one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance.

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

In addition, in the description with reference to the accompanying drawings, the same reference numerals are assigned to the same components regardless of reference numerals, and redundant descriptions thereof will be omitted. In describing the embodiments, when it is determined that detailed descriptions of related known technologies may unnecessarily obscure the subject matter of the embodiments, detailed descriptions thereof will be omitted.

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, and wearable devices.

1 is a flow chart for explaining a method of managing distribution of tiles and sanitary ware according to an embodiment.

In one embodiment, in a method for managing distribution of tiles and sanitary ware based on artificial intelligence, a control device (hereinafter referred to as a control device) for distribution management of tiles and sanitary ware is a unit of pallets of target tiles and sanitary ware QR The code can be scanned and shipped by location (101). The control device is a control device for manufacturing an insole for golf shoes, and may be implemented by, for example, a software module, a hardware module, or a combination thereof. The pallet may include tiles and sanitary ware of the same manufacturer, of the same size and type.

According to an embodiment, the control device may be an electronic device including a communication function. For example, the electronic device includes a smart phone, a tablet personal computer (PC), a mobile phone, a video phone, an e-book reader, a desktop personal computer (PC), and a laptop. A laptop personal computer (PC), a netbook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), an MP3 player, a mobile medical device, a camera, or a wearable device (eg It may include at least one of a head-mounted-device (HMD) such as electronic glasses, an electronic garment, an electronic bracelet, an electronic necklace, an electronic appcessory, an electronic tattoo, a smart car, or a smart watch. You can.

According to one embodiment, the electronic device may be a smart home appliance with a communication function. For smart appliances, for example, electronic devices are televisions, digital video disk (DVD) players, audio, refrigerators, air conditioners, vacuum cleaners, ovens, microwave ovens, washing machines, air cleaners, set-top boxes, TVs. It may include at least one of a box (eg, Samsung HomeSyncTM, Apple TVTM, or Google TVTM), game consoles, electronic dictionary, electronic keys, camcorder, or electronic picture frame.

According to an embodiment, the electronic device includes various medical devices (eg, magnetic resonance angiography (MRA), magnetic resonance imaging (MRI), computed tomography (CT), photographing device, ultrasound device, etc.), navigation device, GPS receiver ( Global positioning system receiver (EDR), event data recorder (EDR), flight data recorder (FDR), automotive infotainment devices, marine electronic equipment (e.g. navigational devices and gyro compasses, etc.), avionics, security It may include at least one of a device, a head unit for a vehicle, an industrial or household robot, an automated teller's machine (ATM) of a financial institution, or a point of sales (POS) of a store.

According to one embodiment, the electronic device is a furniture or part of a building / structure including a communication function, an electronic board, an electronic signature receiving device, a projector, or various measurement devices. It may include at least one of a device (for example, water, electricity, gas, or radio wave measurement devices). The electronic device according to an embodiment may be a combination of one or more of the various devices described above. Also, the electronic device according to an embodiment may be a flexible device. Also, it is apparent to those skilled in the art that the electronic device according to an embodiment is not limited to the above-described devices. The term user used in various embodiments may refer to a person using an electronic device or a device using an electronic device (eg, an artificial intelligence electronic device).

An electronic device according to an embodiment includes a processor, a memory, a user interface, and a communication interface, and may be connected to another electronic device through a network. The communication interface may transmit and receive data to and from other electronic devices within a predetermined distance through a wired, wireless network, or wired serial communication. The network enables wired and wireless communication between electronic devices and various entities according to an embodiment. The electronic device can communicate with various entities over the network, and the network can use standard communication technologies and / or protocols. At this time, the network (network) includes, but is not limited to, the Internet (Internet), a local area network (LAN), a wireless local area network (LAN), a wide area network (WAN), a personal area network (PAN), and the like. Anyone who has ordinary knowledge in the field of communication technology can know that it may be another kind of network capable of transmitting and receiving information.

According to one embodiment, the QR code may contain product information of the target tile and sanitary ware. The detailed description of the QR code will be described later with reference to FIG. 2. Products can be shipped in a stacked form at each location in a pallet. QR code scanning in the process of shipping allows you to store where each product is placed. The location information according to the result of the QR code scanning can be used to determine the location of the damaged pallet in the future.

According to one embodiment, the control device may photograph the target tile and the sanitary ware with a total of 5 real-time monitoring cameras, including north and south and north and upper sides of the distribution line that is shipped.

Real-time monitoring camera according to an embodiment is a video recording equipment capable of high-definition (HD) -class shooting at 60 frames per second that can automatically adjust ISO (International Standardization Organization), exposure, shutter speed, etc. A total of 5 units can be located on the north-west, north-south, and top, excluding the bottom. The real-time monitoring camera continuously captures an image. At this time, an image 4 hours old may be automatically deleted except for the image included in the first input signal according to the size of the memory capacity.

According to an embodiment, the control device may generate a first input signal by encoding the captured image of 10 seconds before and after the detection time when the motion detection sensor detects an abnormal motion based on the shooting result (103).

According to an embodiment, the motion detection sensor is a sensor that can detect an object when motion of the object is detected, and may use an infrared sensor or an ultrasonic sensor, but is not limited thereto. The abnormal movement detected by the motion detection sensor may include movement of the target tile and the sanitary ware or contact and impact of the abnormal object.

According to an embodiment, the control device may store a captured image 10 seconds before and after the abnormal motion detection time point based on a signal transmitted from the motion detection sensor. The control device may generate a first input signal based on pictures for each frame that converts the stored images into 30 to 60 frames per second. The first input signal may be re-digitized with the color information to distinguish the pictures as described above and encoded into a standardized size suitable for input to the convolutional neural network.

According to one embodiment, the control device may input the first input signal to the convolutional neural network in the blockchain network (104).

The convolutional neural network according to an embodiment may be designed to determine whether the target tile and the sanitary ware are damaged based on the results of the real-time monitoring camera. Such a convolutional neural network is composed of a feature extraction neural network and a classification neural network, and the feature extraction neural network proceeds by stacking the input signal sequentially in a convolutional layer and a pooling layer. The convolution layer includes a convolution operation, a convolution filter, and an active 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 active function generally uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. The pooling layer is a layer that serves to reduce the matrix size of the input, and uses a method of extracting representative values by grouping pixels in a specific area. In the calculation of the pooling layer, an average value or a maximum value is generally used, but is not limited thereto. The operation is performed using a square matrix, and generally uses a 9X9 matrix. The convolutional layer and the pooling layer alternately repeat until the input is small enough while maintaining the difference.

According to one embodiment, the classification neural network has a hidden layer and an output layer. In the convolutional neural network for the distribution management method of tiles and sanitary ware, there are generally three or more hidden layers, and the number of nodes in each hidden layer is designated as 100, but may be determined as more or less depending on the case. The active function of the hidden layer uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. The output layer node of the convolutional neural network can be 10 in total. The output layer activation function of the classification neural network uses a softmax function. The softmax function is a representative function of one-hot encoding, and the sum of all output nodes is 1 in total, and the output of the output node having the largest value is 1, and the output of the remaining output nodes is 0. . It is possible to select only one of the 10 outputs through the Softmax function.

The blockchain network according to an embodiment includes blocks storing a database and a chain connecting them in chronological order, and the database can store data such as photographed image information, coordinate information, and damage determination information. have. The stored database is managed as big data, and the blockchain network can help securely store information in this database and make it available to limited users. The description of the blockchain network and the convolutional neural network will be described later with reference to FIGS. 3, 4, and 5.

According to one embodiment, the control device may obtain a first output signal based on the input result (105). The control device may notify the user of the target tile and the pallet information of the sanitary ware, which are determined to be damaged, based on the first output signal (106).

According to an embodiment, the convolutional neural network may have 10 outputs, each of which may include information on whether the pallet is damaged and location based on a captured image of a real-time monitoring camera. When it is determined that there is damage in the damage determination of the pallet, the control device may generate a voice signal or a lighting signal to inform the user.

According to an embodiment, the control device may generate a learning signal according to the user's input based on the user's palette confirmation result (107). The control device can train the convolutional neural network based on the learning signal (108).

According to one embodiment, the control device may generate a learning signal based on the input value when the user checks the pallet and then inputs whether the pallet is actually damaged. If the manager or the manufacturer does not write such feedback, the convolutional neural network can progress in the direction of increasing the confidence level of the previously generated first output signal. The user input is based on 24 hours after the first output signal of the convolutional neural network is generated, but may be more or less depending on the case. The control device can apply accuracy and efficiency by applying deep learning technology to the distribution management method of tiles and sanitary ware.

According to one embodiment, the control device may transmit an order signal to the producer in the first network storage device, based on the user's input (109)

According to an embodiment, the control device may automatically generate and transmit an order signal corresponding to the amount of damage if there is actual damage, according to a value entered after the user checks for damage, or if there is no damage, the order signal May not generate. If there is no user input, the control device determines whether or not damage occurs according to the result of the first output signal 24 hours after generating the first output signal, and the degree of damage can be determined through the ratio of the changed volume from the existing volume of the corresponding pallet. have. The control device can automatically generate a re-order signal for the damaged product to increase the speed of the distribution process.

2 is a view for explaining a QR code according to an embodiment.

Referring to FIG. 2, the QR code includes route information 201 that can distinguish a target tile and a type of sanitary ware regardless of discoloration in a first quadrant based on a quadrant divided by a horizontal axis and a vertical axis; The second quadrant includes the target tile and sanitary ware quantity 202 included for each pallet; The third quadrant includes the size of the target tile and sanitary ware 203; And the fourth quadrant may include a manufacturer 204 of the target tile and sanitary ware.

The route information 201 according to an embodiment may cause color changes in a fine manner according to a temperature change in a manufacturing and transportation process, and may be an index for evaluating that the color is the same tile. The route information 201 is generally composed of numbers, and it is possible to determine the type of a unique tile through this value.

The quantity 202 according to an embodiment may refer to the quantity of each tile and the sanitary ware included in each pallet, and accordingly, the quantity may be included in the quantity for each route information 201. Since one pallet contains the same product, it may be sufficient to store only the quantity of one type of product. The quantity of products is based on the quantity of boxes, but the total quantity value including the quantity of tiles for each box can also be displayed.

단위로 표현할 수 있다.The size 203 according to an embodiment may include size information of individual target products, and thus, the size of each route information 201 may be included as content. Additionally, it may include size information according to a state packaged in a pallet, and the size information may be stored as a result of scanning through a 3D scanner in a distribution process. Size 203 means volume, and accordingly

It can be expressed in units.

The manufacturer 204 according to an embodiment may contain manufacturer 204 information of individual target products. Since the same products belong to the pallet, it may be sufficient to include information of one manufacturer 204 for each pallet. The manufacturer 204 information may include the country, city, address, etc. of the manufacturer 204.

According to an embodiment, the control device may estimate the location of the pallet determined to be damaged by registering the location information of the pallet holding each QR code when scanning the QR code. When shipping the product, the user can scan the QR code through the QR code scanner of the control unit.In the process of real-time monitoring camera shooting the pallet that entered the distribution line with the scan of the QR code, the location information of the pallet is stored. You can.

3 is a view for explaining a blockchain network according to an embodiment.

Referring to FIG. 3, a block 301 including a database including pre-trained convolutional neural networks and a database including captured image information, coordinate information, and damage determination information in a distribution line; Chains 302 connecting each block 301 in chronological order; And it may be a private blockchain network including the first (310), second (320), and third network storage device 330 for storing each blockchain.

According to one embodiment, the block 301 of the blockchain network may include a database including captured image information, coordinate information, and damage determination information in a distribution line and a pre-trained convolutional neural network that deals with the database. Each block 301 is generally connected in chronological order, so that new blocks 301 can be produced at 10 minute intervals. The contents of the already generated block 301 are stored in all network storage devices, and cannot be changed unless the majority of the contents are changed in time. For example, in a blockchain network having a total of three network storage devices, when it is assumed that one block 301 is provided for each network storage device, if the contents of two or more blocks 301 cannot be changed within a limited time, Each block 301 may change the value of the block 301 having a different content from the majority to be equal to the majority by verification. Accordingly, it is possible to maintain high security. In fact, the number of network storage devices participating in the blockchain network may reach tens to hundreds of thousands, and thus, it may exhibit higher security.

Chains 302 according to one embodiment may consist of hash values. The chains 302 allow the blocks 301 to be continuous in chronological order, and each block 301 can be connected using a hash value. The block 301 may store a database and a hash value of the database, a previous header, and a header of the current block. Here, the header of the current block functions as a previous header in the next block, so that each block 301 can be organically connected. In addition, the hash value is transformed into a completely different form when the contents of the block are changed even slightly, which effectively prevents an attempt to change the contents of the database. Since the header of the current block becomes the total hash value including the database, the hash value of the database, and the previous header, it is difficult to successfully attempt to compromise security by effectively modifying the database and the hash value. Because when the database and hash values are modified, the contents of the header are changed, and accordingly, the previous header that enters the next block is also changed, and accordingly, the header of the block is also changed, so that the previous header of the next block can be changed again. That is, all subsequent blocks 301 must be hacked. Therefore, it is possible to increase the security of the blockchain with the chain 302 through the hash value. The control device can make the information big data and use enhanced security through the blockchain network.

According to an embodiment, the first network storage device 310 including a producer of a target tile and a sanitary ware in a distribution management method of the tile and a sanitary ware; A second network storage device 320 including a distributor distributing the tile and the sanitary ware targeted for the distribution management method of the tile and the sanitary ware; A third network storage device 330 including a seller who receives a target tile and sanitary ware of the distribution management method of the tile and sanitary ware and performs sales to an actual consumer; And it may include a high-speed Internet connection network 303 for connecting each network storage device. The network storage devices classified as the first, second, and third may be determined according to the number of practitioners included, the number of users, and the number of storage devices.

According to one embodiment, the first network storage device 310 may include a producer that produces the target tile and sanitary ware, and the producers may be one company or more. The number of first network storage devices 310 may be determined according to the number of storage devices used by each company. Producers of the target tile and sanitary ware of the first network storage device 310 can make efforts to reduce damage by grasping the reason for the damage occurring in the distribution process based on the database and convolutional neural network of the blockchain network. New products may be provided again based on an order signal obtained according to the products.

According to an embodiment, the second network storage device 320 may include a distributor that distributes target tiles and sanitary ware. The number of the second network storage device 320 may be determined according to the number of companies and users that distribute the target tiles and sanitary ware. The distributor of the second network storage device 320 may also be referred to as a real user (hereinafter referred to as a user) using the method and apparatus for distribution and management of tiles and sanitary ware. Users of the distribution management device of the second network storage device 320 may facilitate learning of the convolutional neural network by manually inputting information about errors occurring during actual use.

According to an embodiment, the third network storage device 330 may include a seller who receives target tiles and sanitary ware and performs sales to actual consumers. The seller of the third network storage device 330 may be authorized as a user of the third network storage device 330 when the order of the tile and the sanitary ware ordered by the user is ordered, and the current situation of the target tile and the sanitary ware ordered by the user. You can check the update information on the blockchain network.

The high-speed Internet connection network 303 according to an embodiment generally refers to an Internet connection network having a speed of 10 Mb / s or more, and includes a local area network (LAN), a wireless LAN (Wireless) as a connection network including wired, wireless, and optical cable technologies. Local Area Network (WAN), Wide Area Network (WAN), Personal Area Network (PAN), and the like.

4 is a view for explaining a database according to an embodiment.

Referring to FIG. 4, the database includes photographed image information 401 pre-photographed the tile and the sanitary ware of the pallet unit in variously layered states in the distribution line; Color information 402 capable of distinguishing the tile and the sanitary ware targeted from the pallet unit from the captured image information and its background; Coordinate information 403 representing the angle of view of the real-time monitoring cameras of the north-south north and south and the upper surface in X and Y axes; Damage discrimination information 404 including comparison image information when a real damage occurs when an abnormal movement is detected and when no real damage has occurred; Group order information 405 containing the order contents of the target tile and sanitary ware; And reorder information 406 for the damaged target tile and sanitary ware.

The photographed image information 401 according to an embodiment is photographed images of various cases according to various stacking methods of products shipped in a pallet unit, and is used as data for comparing the photographed image when detecting abnormal motion. You can. Therefore, the photographed image information 401 is based on a result photographed by the same cameras in the distribution line, and the same can be captured at 60 frames per second to distinguish frames.

The color information 402 according to an embodiment is information for distinguishing background and palette unit target products from a picture for each frame of the captured image information 401, and collects and compares the color information 402 in units of pixels Can be distinguished through. The color information basically expresses the values of RGB color, brightness, and saturation as numbers, and when two or more of each value changes significantly, it can be viewed as a border line. Through the collected color information 402, the background and the target product can be distinguished from the frames of the first input signal, and based on this, it is possible to check what changes the target product undergoes in the image in the first input signal.

The coordinate information 403 according to an embodiment means determining coordinates by placing the X-axis and the Y-axis based on the lower left corner of the camera's angle of view for each camera. Coordinate values may be determined through pixels, and up to (1920,1080) coordinates may be generated based on the HD-class camera resolution. Coordinate information 403 may be used to determine the location of the target product shipped to the distribution line after sensing the QR code, and may also be used to trace back the location of the damaged pallet.

The damage discrimination information 404 according to an embodiment may be generated based on pallets that have already suffered from actual damage. The convolutional neural network can determine the possibility that the change of target products occurring in the first input signal is the actual damage based on the damage determination information 404 in the database. The damage may be determined by movement of the target product, hitting the target product of another object, or volume change of the target product itself, but is not limited thereto.

The pre-order information 405 according to an embodiment indicates order information that a seller who uses a third network storage device puts in a producer who uses the first network storage device, wherein the pre-order information 405 is the route information of the item, It can contain the contents of the QR code, such as quantity, size, and manufacturer. The pre-order information 405 may serve as a criterion for determining whether the number of products moved through actual transportation in the shipping process and the unloading process is consistent.

The reorder information 406 according to an embodiment may be determined by a value manually input after the user checks whether the damage is made. When the user enters the damage, the reorder signal automatically generated according to the degree of damage is blocked in the blockchain. It can mean recording and storing in a network. In addition, even if the user does not directly input, the reorder signal automatically formed accordingly can be recorded and stored in the blockchain network.

5 is a diagram for describing a convolutional neural network according to an embodiment.

Referring to FIG. 5, blocks in a blockchain network may include a pre-trained convolutional neural network 510 and a database 520, using the first input signal 501 as an input, and the first output signal 502 ) As an output, and can be learned through the learning signal 503.

According to one embodiment, the first input signal 501, which is the input of the convolutional neural network 510, classifies the shooting information of the real-time monitoring camera into 60 frames per second, and forms a data sheet in which each frame is numerically color-coded for each pixel. Can be recorded as The color information may include RGB information for each pixel, brightness, and saturation. The first input signal 501 distinguishes the target product from the background based on each value of color information and provides the information so as to check the boundary of the product, from which it is possible to detect a change in the target product. The color information included in the first input signal 501 may be damaged or not, and the coordinate classification may be performed through the convolutional neural network 510.

According to an embodiment, the convolutional neural network 510 may classify whether the damage and its coordinates are based on the color information in the first input signal through the feature extraction neural network and the classification neural network of the convolutional neural network. The feature extraction neural network proceeds by stacking the convolutional layer and the pooling layer in order of the input signal. The convolution layer includes a convolution operation, a convolution filter, and an active 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 active function generally uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. The pooling layer is a layer that serves to reduce the matrix size of the input, and uses a method of extracting representative values by grouping pixels in a specific area. In the calculation of the pooling layer, an average value or a maximum value is generally used, but is not limited thereto. The operation is performed using a square matrix, and generally uses a 9X9 matrix. The convolutional layer and the pooling layer may be alternately repeated until the input is small enough while maintaining a difference. The classification neural network has a hidden layer and an output layer. In the convolutional neural network 510 for the distribution management method of tiles and sanitary ware, there are generally three or more hidden layers, and the number of nodes of each hidden layer is designated as 100, but may be set as more or less depending on the case. The active function of the hidden layer uses a ReLU function, a sigmoid function, and a tanh function, but is not limited thereto. The output layer node of the convolutional neural network 510 may be 10 in total. The output layer activation function of the classification neural network uses a softmax function. The softmax function is a representative function of one-hot encoding, and the sum of all output nodes is 1 in total, and the output of the output node having the largest value is 1, and the output of the remaining output nodes is 0. . It may be possible to select only one of the 10 outputs through the Softmax function. Ten outputs show whether or not the damage and its coordinates, various information on this may be stored on the database 520.

According to an embodiment, the first output signal 502 that is the output of the convolutional neural network 510 may include location information of the damaged discriminated palette obtained from the QR code based on the classification of the convolutional neural network. The first output signal 502 can show a combination of damage information and coordinate information, where the classification of the combination can generally be based on information stored on the database 520. The data stored on the database 520 may increase or decrease the number of data according to the learning process of the convolutional neural network 510, and accordingly, the number of output nodes may be changed to 10 or more.

According to an embodiment, the learning signal 503 for learning of the convolutional neural network 510 may be generated from a user's re-checking result input on whether or not the damage determined palette is damaged. The user may manually input information to be corrected based on the determination of the need to generate a learning signal, and through this, reinforcement or weakening of database information may be achieved according to the type of the learning signal 503.

The learning signal 503 according to an embodiment is made based on an error between a correct answer and an output value. In some cases, an SGD using a delta, a batch method, or a method using a back propagation algorithm may be used. According to the learning signal, the convolutional neural network 510 performs learning by modifying an existing weight, and may use momentum in some cases. The cost function can be used to calculate the error, and the cross entropy function can be used as the cost function.

According to an embodiment, the convolutional neural network 510 may learn to correct coordinate information and damage determination information in the database 520 based on the learning signal 503. The pre-trained convolutional neural network 510 has three or more hidden layers, and each hidden layer may have 50 or more hidden nodes. ReLU function, sigmoid function and tanh function can be used as the active function of each hidden node, but is not limited thereto. As a function of the output node, a softmax function using one-hot encoding can be used. The output selects only one classification according to the one-hot encoding technique, and can execute commands from the selected classification.

According to one embodiment, the database 520 may store information as big data, and may be continuously updated through a high-speed Internet connection network connected to a blockchain network.

The control device according to an embodiment may continuously provide feedback by using an error of a damage discrimination error in the distribution management method of tiles and sanitary ware.

6 is an exemplary view of a configuration of an apparatus according to an embodiment.

The device 1201 according to an embodiment includes a processor 1202 and a memory 1203. The device 1201 according to an embodiment may be the above-described server or terminal. The processor may include at least one of the devices described above with reference to FIGS. 1 to 5, or may perform the at least one method described with reference to FIGS. 1 to 5. The memory 1203 may store information related to the above-described method or a program implemented with the above-described method. The memory 1203 may be volatile memory or nonvolatile memory.

According to an embodiment, the processor 1202 may execute a program and control the device 1201. The code of the program executed by the processor 1202 may be stored in the memory 1203. The device 1201 is connected to an external device (for example, a personal computer or a network) through an input / output device (not shown), and can exchange data.

The embodiments described above may be implemented with hardware components, software components, and / or combinations of hardware components and software components. For example, the devices, methods, and components described in the embodiments include, for example, a processor, controller, arithmetic logic unit (ALU), digital signal processor (micro signal processor), microcomputer, field programmable gate (FPGA). It can be implemented using one or more general purpose computers or special purpose computers, such as arrays, programmable logic units (PLUs), microprocessors, or any other device capable of executing and responding to instructions. The processing device may run an operating system (OS) and one or more software applications running on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For convenience of understanding, a processing device may be described as one being used, but a person having ordinary skill in the art, the processing device may include 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 a processor and a controller. In addition, other processing configurations, such as parallel processors, are possible.

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 on a computer readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiments or may be known and usable by those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic media such as floptical disks. -Hardware devices specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language code that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

The software may include a computer program, code, instruction, or a combination of one or more of these, and configure the processing device to operate as desired, or process independently or collectively You can command the device. Software and / or data may be interpreted by a processing device, or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodied in the transmitted signal wave. The software may be distributed on 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 limited drawings, those skilled in the art can 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 components of the described system, structure, device, circuit, etc. are combined or combined in a different form from the described method, or other components Alternatively, even if replaced or substituted by equivalents, appropriate results can be achieved.

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

Claims (5)

In the method of managing the distribution of tiles and sanitary ware based on artificial intelligence,
Pallet of target tile and sanitary ware-The pallet includes tiles and sanitary ware of the same manufacturer, the same size and type-scanning unit QR codes and shipping by location;
Photographing the target tile and sanitary ware with a total of 5 real-time monitoring cameras including the north-west, north-south, and top surfaces in the shipped distribution line;
Based on the photographing result, the motion sensor detects abnormal movement-the abnormal movement includes movement of the target tile and sanitary ware or contact and impact of an abnormal object-upon detection, encodes the captured image 10 seconds before and after the detection point Generating a first input signal;
Inputting the first input signal into a convolutional neural network in a blockchain network;
Obtaining a first output signal based on the input result;
Informing the user of the pallet information of the target tile and the sanitary ware, which is determined to be damaged, based on the first output signal;
Generating a learning signal according to a user input based on a result of the user's palette confirmation;
Training the convolutional neural network based on the learning signal; And
Transmitting an order signal to a producer in the first network storage device based on the user's input
Including,
The blockchain network is
Blocks including a database containing pre-trained convolutional neural networks and a database including captured image information, coordinate information, and damage determination information in the distribution line;
Chains connecting each block in chronological order; And
The network storage devices storing each block and chain
It is a private blockchain network that includes,
The network storage devices
The first network storage device including the producer for producing the target tile and the sanitary ware in the distribution management method of the tile and the sanitary ware;
A second network storage device including a distributor for distributing the target tile and sanitary ware in the distribution management method of the tile and sanitary ware;
A third network storage device including a seller who receives the target tile and the sanitary ware of the distribution management method of the tile and the sanitary ware and performs sales to an actual consumer; And
High-speed Internet connection network connecting each network storage device
Containing
How to manage the distribution of tiles and sanitary ware.
delete delete According to claim 1,
The database
The photographed image information of pre-imaged tiles and sanitary ware targeted to pallet units in variously layered states within the distribution line;
Color information capable of distinguishing the tile and the sanitary ware targeted from the pallet unit from the captured image information;
The coordinate information representing the angle of view of the real-time monitoring cameras of the north-west, north-south, and top surfaces in X and Y axes;
The damage discrimination information including comparison image information when the actual damage occurs when the abnormal movement is detected and the actual damage does not occur;
Ordering information containing the ordering of the target tile and sanitary ware; And
Reorder information for the damaged target tile and sanitary ware
Containing
How to manage the distribution of tiles and sanitary ware.
According to claim 1,
The convolutional neural network
The first to divide the shooting information of the real-time monitoring camera into 60 frames per second, and record each frame as color information for each pixel-the color information includes RGB information, brightness, and saturation per pixel-in the form of a numerical data sheet. With the input signal as input,
Based on the color information, whether the damage and its coordinates are classified through the feature extraction neural network and the classification neural network of the convolutional neural network,
Based on the classification, a first output signal including location information of the damaged discriminated pallet obtained from the QR code is used as an output,
Learning through the first learning signal generated from the input as a result of the user's re-check whether the damage determined pallet is damaged
How to manage the distribution of tiles and sanitary ware.

Images