Passenger flow statistics and accurate advertising based on face recognition

Bohe's customer flow statistics solution based on face recognition, combining big data with offline stores accurately, quickly and efficiently, leading through big data of passenger flow, big data of member management, big data of product sales, big data of advertisements, leading The offline dealers have a new wave of change.

Demand analysis
       In recent years, the popularity of online e-commerce is gradually changing people's consumption habits. The competition between online and offline is becoming increasingly fierce. The operating costs under the line are higher and the coverage is smaller. In the face of the increasingly popular “Double Eleven” Season, how do offline store operators break through the current difficulties? According to Mattel's customer flow statistics solution based on face recognition, it combines big data with offline stores accurately, quickly and efficiently, and through customer flow big data, member management big data, product sales big data, and advertising big data. Leading the new wave of changes in the offline store.

Program overview
       Bohe's passenger flow statistics solution based on face recognition is based on Bohe MIPS multimedia information publishing software, integrated with Shangtang Technology's face recognition technology, and saves existing advertising business according to the actual operation of many business district stores. On the basis of the process, the advertising mode of big data statistics, analysis and entertainment interaction has been added to provide offline consumer information reference, perfect member user management and interesting interactive marketing promotion to further improve offline store operators. ROI.  

Introduction to face recognition technology

1, face detection tracking
       A millisecond-level face detection can be implemented on mobile devices and personal computers for complex low-quality images or surveillance videos for hundreds of people. This technology can adapt to various practical environments such as side face, occlusion, blur, and expression changes.

2, face key point positioning
       Microsecond level eye, mouth, nose contour and other 106 face positioning. This technology can adapt to a variety of real-world environments such as large-angle side faces, expression changes, occlusion, blur, and light and dark changes.

3, face identity authentication
       Given a face sample, the millisecond level retrieves a large-scale face database or surveillance video, giving identity authentication. When 96% of faces are certified, the false detection rate is less than one in 100,000.

4, face attributes
       Accurately identify more than 10 types of face attributes, such as gender, age, race, expression, jewelry, beard, facial movement status, etc. Can be used for ad targeting or customer information analysis, so that you can understand the customer in seconds.

5, face clustering
       Faces of hundreds of thousands of people are quickly clustered for smart face-based photo albums and group-based social network analysis. Make photo management more intuitive and make social connections clearer.

6, real person detection
       Detect whether the user is a real person before the camera, and cooperate with face authentication to provide real-life authentication for serious application scenarios such as financial security. Can effectively distinguish high-definition photos, PS, 3D models, face-changing and other counterfeiting fraud. We provide solutions for user collaboration and user mismatch scenarios.

7, portrait beauty / beauty


Based on the intelligent face detection and positioning technology, it creates a mobile beauty and beauty effect solution, which makes the mobile Internet entertainment era have “beauty”.

 

Solution construction

Bohe's IoT-3288A Ruixin micro 3288 motherboard; high-definition camera; as the United States MITA face recognition passenger flow statistics software.


Program characteristics
1, accurate passenger flow statistics


Install Bohe MIPS face recognition passenger flow statistics software, through face detection and face tracking recognition of incoming customers, you can achieve:
Real-time passenger flow statistics: Statistics of real-time passenger flow information per day, extracting facial feature values, and avoiding repeated counting.
Passenger flow trend statistics: Statistics of passenger flow trends can be queried in the most recent day, week, month, season, half year, year, and year, and presented in a column or linear graph.
Residual duration statistics: Statistics on the distribution of the duration of a single passenger flow in the total passenger flow before and after the camera.

2, passenger flow analysis
        Through the face recognition technology, it is possible to accurately calculate the male and female attributes and ages of a single passenger flow entering the store at any time. The accuracy of the male and female attributes is above 99.5%, the age accuracy is over 90%, and the error is no more than 3 years old. Through the ratio of male to female and age attribute of the shoppers, the store can quickly adjust the product structure and change the marketing strategy to achieve the maximum rate of return.

 3. Cultivate and tap loyal fans
        Through the face recognition technology, the VIP customers and repeat customers who have entered the purchase system are accurately found. The store staff receives the purchase history information of the push, and focuses on converting the returning customers into VIPs, turning the VIP into a loyal fan, and greatly increasing the store sales.
 

4, big data operation, optimization management
        Through the face recognition passenger flow big data, it is possible to analyze the length of time that different people reside in the storefront, which areas are more attractive, what products are most concerned, what products are of concern to different people, etc., and optimize the store management through the results of data analysis. Find out the most suitable business model for the store, the most concerned advertisements for consumers, the products with the highest conversion rate, analyze the hotspots according to the customer's shopping path, adjust the main product display to attract customer attention, and improve the deficiencies in the clerk service process. Optimize from macro to detail to maximize the benefits of the store.
 

5, accurate advertising, improve conversion efficiency
        Through face recognition, according to the male and female attributes of the shoppers and the age structure, switching more targeted advertising content within 1-2S can effectively improve the attractiveness of advertising, achieve accurate product promotion, and improve purchasing. Conversion rate.
 

Application scenario
        Shopping malls, commercial streets, gold and silver jewelry stores, automobile 4S stores, cosmetics chain stores, brand clothing chain stores, food specialty stores, eyeglasses specialty stores, drug counters, shoe bag stores and other public activities and crowd gathering places.