TWM443899U - An intelligent point-of-sales system with active identification of customer group - Google Patents

Abstract

The disclosed is an intelligent point-of-sales system with active identification of customer group. This system is composed with integrated software and hardware modules. The hardware modules consist of IP-cameras, POS machines and electric billboards. The software modules consist of image identification, business intelligence and data mining. The image identification recognizes the customer's gender and age automatically. The identification result is combined with his purchase to analyse the favorite merchandise for the specific customer group through the cross-over analysis in database. The commercials of the favorite merchandise are shown on the electric billboards to stimulate the purchase motivation. The improvement in marketing mode will surely promote the sale efficiency.

Description

M443899 October 5, 101, new description: [New technology field] [0001] The technical field of this new creation belongs to the POS server to close the periphery, such as the lock sale point (simplified) advertisement Machine H advertising machine, electric board and camera materials ", set, and borrowed the image age technology, fuck the instant image of the customer before the electronic advertising board" and then analyze the customer attribute group by face recognition and gender classification technology, to f Library cross-analysis to analyze product consumption behavior and marketing advertising appropriateness, in order to strengthen the sales of goods, and send the data to the headquarters to store intelligent attribute identification data and consumer information and other multi-dimensional benefit analysis information software and hardware integration Lu Technology [Previous Technology] [0002] The chain-operated end-point marketing checkout system is a passive type. When the consumer checks out the counter, the clerk must judge the customer's customer level and number of people according to the rules of thumb. Enter the customer's gender and age and then check the bill, and the small information screen set on the counter is only looped. Put the pre-set ads. Allowing customers to watch, it is difficult to automatically and accurately grasp the preference of Lu customers for product categories, and automatically adjust the advertising content, advertising effectiveness - difficult to assess, can not know the attention of the product after the shelves, can not pin. Give appropriate recommendations to specific customer groups The service of goods does not help much in promoting the sales of goods. Therefore, it is necessary to analyze the customer's group to adjust the advertising content, so as to promote the marketing of goods and create the sales performance of the store', even if there is retail using the p〇S system. The utilization rate of access and analysis technology is still low, so the active identification of the customer layer system can effectively improve the marketing efficiency of the sales endpoint. β 10120349 Production order number Α 0101 Page 4 / Total 19 pages 1013390967-0 October 12日俊正换页 [New content] 圃相县_ 客客层智慧(四)h (pGSu age is to be sold by the point of sale (P0S), camera, P〇s advertising machine' shelf advertising machine, electronic signage and headquarters ship Part of the hardware system (see the second (5) and the software system part of the processing of the wisdom and technology, and the material information of the material (four) materials, Part of the hardware function is as follows: cameras located on the P0S advertising machine, shelf advertising machine and electronic billboard automatically take customer image data, and the camera lens that detects the flow of people also transmits the number of people to the point of sale (POSH) The server, in the point of sale (10)), applies the smart female technology automatically _ the customer's face, gender and age' and sends the appropriate advertisement to the advertising machine for the customer to watch to increase the sales performance. Send the headquarters server to store the intelligent attribute identification data and enter the m-dimensional fee. The dimension of the software is pure; the software function is described as follows: the response speed from the special acquisition to the identification is very fast. Hybrid π depends on the skin color area, so it does not affect the interference of the Vatican and the background of the skin color area, and considers the global and regional characteristics, and is not affected by factors such as glasses and hairstyle. After automatic detection, identify the gender number and age of the consumers and combine the automatic record analysis with the purchased products. At the same time, through the I's "knife analysis technology", recommend the applicable products related to the customer group, and improve the traditional advertising. The method, that is, the data, the heavy impact factor, and the 7th extraction, the more suitable products are recommended to the customer, and the information for the individual consumers enters (4) the product is reduced, so that the advertisement becomes __for a smart marketing method' The practicality of advertising machines and digital signage in commercial applications 'by collecting and communicating and analyzing the propensity of potential customers to increase the consumer's motivation. 1013390967-0 1〇12.349Bill order number deletion 1 Page 5 / Total 19 pages M443899 _- October 12, 101 Press the replacement page [Implementation] [0004] This creation "actively identify the customer-level intelligent point of sale The embodiment of the (POS) system is mainly a hardware function system consisting of a point-of-sale (POS) advertising machine, a shelf advertising machine, an electronic signboard, and an automatically captured image data camera disposed on each device (see the first figure). And a software functional system component for processing the intelligent attribute identification service, the customer purchase behavior analysis service, the product suitability recommendation service, and the product advertisement exposure service, and the embodiments of the present invention will be described in detail below with reference to the accompanying drawings, and these schematic diagrams are only used. To illustrate the basic architecture and implementation methods of this creation, the systemic connotation of its actual implementation will be tailored to different application objects. < In this creation, the front hardware function system part (see the second figure), the store or the circulation operator can use the installed store server (10 3) as the central control host of each store, located at the point of sale (POS) The camera (101) of each screen of the advertising machine (105), the shelf advertising machine (106) and the electronic signage (107) can transmit attribute data such as customer images to the store server (103), and is disposed on the shelf advertising machine (106). The hemispherical lens (1 02 ) that is vertically shot above transmits the flow count information to the server (103), and the server (103) integrates information such as the number of people detected, facial texture features, etc., to identify the age of the character 4 No, customer traffic and product attention, and then the processed data are classified into the face identification database (201), gender identification database (202) 'age identification database (203) and flow count database (204) ), so when the customer is close to the store base, the advertisement content of the advertisement machine is viewed through the electronic billboard (107) outside the store, and the broadcasted content is extracted through the analysis technology of the face detection and the database result. The customer's customer group is the most suitable recommended product for purchase; when the customer enters the store, the internal shelf advertising machine (106) will recommend it to the customer according to the shelf product in the area, 1〇12_^职1 1 page/total 19 pages 1013390967-0 M443899 On October 12, 101, when the check-out page was checked out, the advertisements were provided through the two-way screen point-of-sale (P0S) advertising machine (including the camera) (105), mainly to automatically establish the store. A two-way communication channel with customers to increase the purchase rate of goods in order to increase the sales performance of the store.

In the latter part of the hardware function system (see the third figure), the point-of-sale (P0S) server (103) determines the appropriate advertisement playing strategy based on the customer attribute information, and transmits the adaptive advertising information to the point-of-sale (P0S) advertising machine (105). The shelf advertising machine (106) and the electronic billboard (107) are screened, and the customer consumption data collection process is transmitted back to the headquarters server (104) for intelligent attribute identification data storage, data exploration and product consumption behavior advertising benefits and the like. analysis. The software function system part includes: 1 · Intelligent attribute identification service:

Use visual capture technology to capture real-time images before kanban, and use face recognition and gender classification techniques to analyze viewers' genders, and use instant and multi-person face recognition technology to simultaneously detect viewers facing kanban And identify and track the movement in the identification screen, and the gender identification technology adopts the facial features, textures and geometric changes of the face image. After the feature reduction, the extracted features are input into the classifier. The gender identification rate coefficient of the face image is obtained, and is used as a basis for identifying gender. After analyzing the gender of the viewer in the surveillance image, the customer's product data is cross-analyzed by the external influence factor and the identification completion factor, and combined with the back-end database extraction, the customer's most suitable product information is recommended. The factors affecting include: gender, age, special festivals, seasonal weather and past consumption records. When the amount of recorded data is accumulated, 'more accurate recommendation of products that consumers may like, increase purchases 10120349#麟 A〇101 Page 7 of 19 1013390967-0 M443899 _ On October 12, 101, Shuttle is replacing the page machine to promote the procurement rate of shelf goods in a certain area of the store or store, and the field detection is adopted by the field. The hemispherical lens above the electronic signage detection area is photographed vertically downwards to analyze the flow of people in order to improve the sales performance of the goods. 2. Customer purchase behavior analysis service: According to the information of intelligent attribute identification, multi-dimensional analysis is carried out, and the reference information of single store, regional store and head office is effectively provided to cope with the development of marketing promotion plan. 3. Product suitability recommendation service:

According to the customer's browsing behavior, sales records and attribute detection data, the I data exploration process is recommended, and the best product information is recommended to be played on the point of sale (P0S) advertising machine (105), the shelf advertising machine (106) or the electronic signboard (107). Provide customers with the best results for browsing. 4. Product advertisement exposure service: Through the service platform of the point-of-sale (P0S) system, the store provides the store's promotional information and advertising activities, and the exposure of the terminal device increases the exposure of the product, and the product advertisement is exposed through the data. After the library extraction analysis, it is provided to consumers for viewing and viewing. < Functional specifications are mainly divided into two parts: "Smart Marketing Point of Sale (P0S) Management System" and "Product Consumer Behavior Analysis System", which are described as follows: 1. Smart Marketing Point of Sale (POS) Management System 1. Advertising and promotional message management: The headquarters management center of the retail endpoint can be used to create an advertising media library based on the advertisements provided by the manufacturer, and set according to the product suitability recommended by the manufacturer (target group age, gender), in advance Set the appropriate customer level of the product, if there is a promotion period activity of the product HU20349# single number deletion 1 page 8 / total 19 pages 1013390967-0 M443899 October 12, 101 revised replacement page and promotional advertising media files, The promotion period of the product promotion, the applicable area (the store), and the corresponding promotional advertisement file can be established in advance. Through the consolidation, a corresponding regular advertisement playlist and a promotional advertisement playlist are generated and deployed to the respective outlets through transmission management. When the advertisement is broadcast, whether it is a regular carousel (no customer stops/checkout), or a suitable push/promotional advertisement, the corresponding module of the store sales point (P0S) is recorded in detail, and The relevant information will be recorded and transmitted back to the headquarters management center through transmission management for subsequent check and report analysis.

2. Active push-on shelf electronic signage: The product suitability setting and advertising multimedia file information can be defined in advance at the headquarters management center, plus the definition of the product display position (shelf number) in the store, through the playlist collection, transmission to each store initiative Push the shelf on the electronic board. The shelf electronic signboard integrates the webcam, the advertisement player, the control firmware and the face recognition application interface, and when the customer stops to watch, the corresponding suitable advertisement in the shelf item can be searched according to the identified age layer and gender. Instantly play and record timestamps in the background, identify the customer layer and advertising attention, etc.

Then, the relevant data is transmitted back through transmission management for subsequent analysis. 3. Appropriate delivery and recycling of merchandise voucher: If the manufacturer provides merchandise repurchase promotion discount for a specific product, the identification code of the merchandise voucher, corresponding merchandise, applicable area/market, discount amount, etc. can be set at the headquarters management center. Information is transmitted to the point of sale (P0S) front desk through the transmission management mechanism. When the face recognition interface identifies the customer's age layer and gender information, it can preferentially search whether there is a suitable product promotion coupon to be sent, and if so, control the thermal printer to output the merchandise sales coupon, and invoice and credit card We will hand it over to the customer. If the customer carries the goods at the next purchase, 1〇12〇349, the production number, Xie 01, 9 pages, 19 pages, 1013390967-0, M443899, October 12, 1101, can be sold at the point of sale (p〇s) The front desk enters the identification code of the voucher, and the data is collected and transmitted back to the headquarters management center through the transmission management module for subsequent analysis and reconciliation. 4. Product suitability: The suitability conditions of the goods can be set at the headquarters management center. Wait for the basic settings, and through the playlist summary and transmission management, deployed to the sales point (P0S) front desk of each store. Through the point-of-sale (P0S) front-end device, the second screen of the camera and the face recognition interface module operate to identify the age layer and gender of the checkout customer, play suitable advertisements, and centrally manage the identification and playback records. The transmission management is returned to the headquarters management center for subsequent analysis. 5. Customer consumption data collection and processing (see the fourth picture): The system will use the visual operation technology to capture the instant image (402) before the electronic signage/smart advertising machine and use face recognition technology and gender classification technology. (403), analyze the gender of the audience. Instant and multi-person face recognition technology that detects viewers facing kanban, detects faces in the scene and tracks the movement of faces. Using the characteristics of the image intensity, texture, and geometric changes of the face, the viewer's gender is analyzed instantly. In the surveillance image, the viewer's number and gender are indicated. For image recognition, the back-end personal data security protection mechanism (405) is as follows (see Figure 5): (1) When face detection, only the area of the face contour (5〇2) is captured. (2) When the background function of the consumer is taken in the background (5〇3), the encryption action will be performed during transmission to prevent the interception during transmission. (3) When the backend server receives the encrypted data, it will delete the image directly and will not save it. (4) In addition, after decryption, the characteristics of the unrecognized individual will be extracted, and then the event will be saved into a log file via the “attribute identifier” (5〇4) (eg, the interval of the age is 1013390967-0 10120349. The first page / a total of 19 pages M443899 | ~ι〇ι year iq month 12 grain _ page material, surname characteristics), the above will not involve the consumer's personal information outside the situation, to ensure the implementation of the application. 6 point-of-sale (POS) customer identification system: mainly the shutter control of the network camera, the face recognition core technology and the identification result storage method modularization and componentization, making it easy to refer to in the software development process. Through the definition and method of component design and the test procedure after the component is produced, the controllability of the identification can be effectively improved, and the stability of the identification process and the result can be ensured. 7. Least Point of Sale (POS) interface integration: firmware point of sale (p〇s) interface

The process is mainly to control the hardware "sales point (p〇S) second screen and "electronic signage" webcam shutter start, multimedia scheduling, marquee message playback, and for playback time, content, etc. Into a record. The firmware that controls playback can receive information from the application, strings & and variables and other information and play it according to its control purpose to achieve image manipulation and adaptive playback. Second, the product consumption behavior analysis system 1. Potential customer data speculation: Conducting the lead customer data collection - the amount of data is based on the product consumption behavior analysis system planning and measurement, through the multi-consumer gaze training related speculation β sneak in the customer data The speculative training procedure is as follows: (1) Set the relevant format of the database rotten and attribute. (2) According to the face detection program, the consumer facial contour sampling tears, ▲ (3) According to the gender identification program for consumer texture measurement 1 , testing. (4) Consumer facial features are based on the age-level identification program; ^ List number Α ' : 0349 ϊ (5) The consumer viewing database of the above test is tested and tested. : .0101 Page 11 of 19 Data Library 1013390967-0 1012 M443899 _ October 12, 2011 Correction Replacement Page Line cross-analysis, recording for viewing goods, and commodity records purchased at checkout. Through service verification, the consumer face detection training generates identification data for gender and age detection, and the consumption records and browsing records are cross-recorded into the database to generate potential data that can be analyzed later. 2. Analysis of customer consumption behavior: According to the sales information recorded by the point of sale (P0S) system, such as: product category, store location difference, consumption time, consumption amount, etc., plus consumer analysis information (age, gender), Use business intelligence

(Bus iness Intel 1 i gence, BI) analysis technology, obtain customer's consumption behavior analysis data for the store's channel supplier to stock the basis, or the channel provider can feedback to the product manufacturer/distributor to reach the point of sale (P0S) And the final purpose of a sales data collection. Business Intelligence (BI) technology application level and method, generally from the analysis of enterprise needs, to the data loading of business intelligence preparation procedures, and even the back-end security and performance management. The focus of this creation is to provide a total solution for the active identification of the customer layer using the point-of-sale (P0S) system channel provider. This work item focuses on the historical sales data of the beta based on the demonstration point, and estimates the age level of the customer. Correlation of customer gender, sales volume, geographic distinction, product category, etc. to establish the applicability of the business intelligence model. The sales figures before and after the creation and verification of this creative demonstration point are continuously accumulated, and can also be reflected in the established business intelligence model and become the customer consumption behavior analysis library. Since the source of the analysis data for this creation is the point-of-sale (P0S) system, the main point of construction is to build a suitable block with the sales point (P0S) system data for subsequent analysis. 1〇12〇349f 单编号Α0101 Page 12/Total 19 Page 1013390967-0 M443899 October 12, 2011 Shuttle Replacement Page 3. Product Pusher Layer Recommendation: Based on the multi-dimensional analysis of business intelligence (BI), The actual sales status at the time of implementation of this creative demonstration can be used as a sampling model to analyze the age layer and gender of the actual purchase customer. This information can be provided to the product manufacturer/distributor for reference, and the product can be re-set according to the result. Appropriately push the target audience, it is expected to achieve more effective, closer to the demand for active marketing. The product launcher recommendations are generated as follows: (1) Import a list of suitable push products. (2) According to the list of products, the sales information is obtained, and the information of the region and time period is obtained.

(3) According to the product list and the sales transaction time, the customer identification data is obtained, and the customer's age layer and gender information are obtained. (4) According to the merchandise list, the merchandise sales coupons are collected and the relevant sales materials are obtained. (5) Obtain statistics on the age, gender, region and time period of the actual purchase of the commodity" (6) Generate a customer-specific analysis of the actual consumption of the commodity. (7) It is up to the manufacturer/router to decide whether to change the push settings.

In summary, the preferred embodiment of the present invention is used to illustrate the basic architecture and implementation method of the present invention, and is not intended to limit the hardware and software components or structures of the present invention, although the present invention has been described in the preferred embodiment. The above disclosure, however, is not intended to limit the creation of this work. Any simple modifications, equivalence changes, and modifications made by those who are familiar with this software and hardware integration technology are still within the scope of this creation. Within the technical scope. This creation "actively identifys the customer-level smart point-of-sale (P0S) system" has excellent industrial applicability, and is not found in publications and public use, as well as having better convenience, practicability and originality, in line with new patents. 12〇349f Single No. A0101 Page 13/Total 19 Page 1013390967-0 M443899 On October 12, 101, according to the regular page, the new patent application was filed according to law. [Simple description of the diagram] [0005] The first picture: This creation actively recognizes the system composition structure of the customer-level intelligent point-of-sale (POS) system. The second picture: a schematic diagram of the hardware function system before the creation. The third picture: a schematic diagram of the hardware function system in the latter part of the creation. Figure 4: Schematic diagram of the identification process and protection mechanism of the image capture of this creation. Figure 5: Schematic diagram of the image extraction and machine learning identification program of this creation [0006] 10120349^^^^ [Description of main component symbols] (101) A camera provided on a display device. (102) To calculate the human flow hemisphere lens. (103) is a sales point server. (104) is a headquarters management server. (105) is a point-of-sale advertising machine. (106) is a shelf advertising machine. (107) is an electronic signage. (201) is a face recognition database. (202) is a gender identification database. (203) An identification database for an age. (204) Count the database for one person flow. (301) Identify a database for an intelligent attribute. (302) is a data exploration module. (303) is a multi-dimensional analysis module. A0101 Page 14 of 19

1013390967-0 October 12, 101 Shuttle Replacement Page M443899 (401) is an image capture and pre-processing module. (402) Capture the area for a person's face. (403) is an identification module. (404) is a processing server. (405) A protection mechanism for the identification of a back end image. (501) is an image capture and pre-processing program. (502) is a face detection program. (503) is a feature extraction procedure. (504) is a machine learning and classification program. • HU20349 production order number group 01 Page 15 of 19 1013390967-0

Claims (1)

M443899 _ October 12, 101 revised replacement page VI, the scope of application for patents: 1. An active identification of the customer-level intelligent point-of-sale system, which includes: a sales point server for processing customer attribute data; a point of sale advertising Machines, a number of shelf advertising machines and electronic billboards for playing adaptive advertising information; a number of cameras located at point-of-sale advertising machines and electronic billboards to capture customer images for the outlet point-of-sale server to analyze customer attributes and develop suitability Advertising strategy; a number of hemispherical lenses placed vertically above the shelf advertising machine to capture the flow of information for the store's point-of-sale server to calculate and analyze; a headquarters management server for storing customer attribute identification data and performing multiple data I Dimensional Analysis; the aforementioned camera set at the point of sale advertising machine, the shelf advertising machine and the electronic signboard, and the hemispherical lens set vertically above the shelf advertising machine transmit the customer image and the flow information to the sales point server to analyze the customer attribute. The sales point server sends the appropriate advertising information to the customer attribute. The corresponding point of sale advertising, shelf advertising or signage players, servers store point of sale and the customer attribute data is sent to the headquarters of the servo control related information management multi-dimensional analysis and storage. 2. The active identification of the customer-level intelligent point of sale system as described in claim 1 of the patent application, wherein the function of the sales point server of the store can actively intelligently identify the face, gender and age of the customer, and summarizes The customer group, and the management of advertising, promotional messages, merchandise sales coupons and recycling, through the advertising machine and electronic billboards set in the store to push the appropriate product information, collect customer consumption data, and conduct product consumption behavior analysis, The consumption information collected by the point-of-sale system is constructed to construct a business intelligence model, analyzing customer consumption behavior characteristics, potential customer data speculation, and product push-layer recommendations. 3. The active identification of the customer-level intelligent point-of-sale system as described in claim 1 of the patent application, wherein the headquarters management server is used to store the customer attribute identification Zi Li Yi ^ single number A0101 1013390967-0 page 16 / total On the 19th page, on October 12, 101, the misplaced materials and the #t❹ material analysis, the fresh consumption f line ^ - the sales information recorded by the point-of-sale system, plus the information of the customer layer analysis, using business intelligence analysis Technology, get the customer's consumption behavior analysis data for the store's channel supplier to stock the basis, can also be channelized by the channel to the upstream dealers, to achieve the purpose of sales data collection' This creation focuses on the historical sales data based on the point of sale, estimated and According to the multi-dimensional analysis results of the business intelligence analysis technology, the actual sales status can be used as a sample model to analyze the age level of the actual purchase customer and the correlation between the customer's age layer, gender, sales volume, geographical difference and product category. Gender, this information can be provided to the upstream Weis as a reference, and based on this result, the product suitability target audience can be re-set. 4. The active identification of the customer-smart intelligent point-of-sale system as described in claim 1, wherein the shelf advertising machine and the electronic signboard can integrate a webcam, an advertisement player, a control body, and a face recognition application interface. When the customer stops to watch, the user can search for the corresponding suitable advertisement in the shelf item according to the identified age layer and gender, and actively push and record the background time stamp, identify the customer layer and the advertisement attention degree #: New, The relevant data is then transmitted back to the sales point server through the transmission management. 5. The active identification of the customer-smart intelligent point-of-sale system as described in claim 1, wherein the cameras located on the point-of-sale advertising machine and the electronic billboard are used to capture the customers before the advertising machine and the electronic signage. The image is transmitted to the store sales point word server. The store point-of-sale server uses instant and multi-person face recognition technology and gender classification technology to detect and track the viewers facing the kanban, using the image intensity, texture and geometry of the face. The characteristics of the change, the gender and age of the viewer are analyzed in real time, and the number and gender of the viewer are marked in the surveillance image. 1〇兰49Product No. A〇101 Page/19 pages 1013390967-0