Facial Recognition-Based Customer Counting and Targeted Advertising
Bohe’s facial recognition-based customer counting solution accurately, quickly, and efficiently integrates big data with offline stores. Leveraging big data on customer flow, membership management, product sales, and advertising, it is leading a new wave of transformation for offline stores.
Demand Analysis
In recent years, the popularity of online e-commerce has been gradually changing consumer habits. Competition between online and offline is intensifying, while offline operations are facing higher costs and smaller reach. Faced with the increasingly popular “Double 11” shopping festival, how can offline stores overcome this predicament? Visiontech’s facial recognition-based customer counting solution accurately, quickly, and efficiently integrates big data with offline stores. By leveraging big data on customer flow, membership management, product sales, and advertising, it is leading a new wave of transformation for offline stores.
Solution Overview
Bohe’s facial recognition-based customer counting solution, based on Bohe’s MIPS multimedia information publishing software and incorporating SenseTime’s facial recognition technology, leverages the actual operational performance of numerous stores in commercial districts. While preserving existing advertising business processes, it adds big data statistics, analysis, and interactive advertising models. This solution provides offline stores with accurate consumer data references, comprehensive membership management, and engaging interactive marketing promotions, further improving their return on investment.
Facial Recognition Technology Overview
1. Face Detection and Tracking
This technology achieves millisecond-level face detection on mobile devices and personal computers, even for low-quality images with complex backgrounds or surveillance videos of crowds of hundreds of people. This technology adapts to various real-world conditions, including profile faces, occlusion, blur, and changes in facial expression.
2. Facial Keypoint Localization
This technology locates 106 key points on the face, including the eye, mouth, and nose contour, within microseconds. This technology adapts to various real-world conditions, including profile faces at wide angles, changes in facial expression, occlusion, blur, and changes in brightness.
3. Facial Identity Verification
Given a face sample, this technology searches a large-scale face database or surveillance video in milliseconds to verify the identity. With a false positive rate of less than 1 in 100,000, the technology achieves 96% face verification.
4. Facial Attributes
This technology accurately identifies over 10 major facial attribute categories, such as gender, age, ethnicity, facial expression, accessories, facial hair, and facial gestures. This technology can be used for targeted advertising or customer analysis, allowing you to instantly understand your customers.
5. Face Clustering
Rapidly clustering hundreds of thousands of faces can be used for face-based smart albums and social network analysis based on group photos. This makes photo management more intuitive and social relationships clearer.
6. Real Person Detection
Detecting whether the user in front of the camera is a real person. Combined with facial authentication, this provides real-person authentication for critical applications such as finance with high security requirements. It effectively identifies counterfeit and fraudulent images, including high-definition photos, photoshopped images, 3D models, and face swaps. We provide solutions for both cooperative and uncooperative user scenarios.
7. Portrait Beauty/Makeup
Based on intelligent face detection and positioning technology, we create mobile beauty and makeup solutions, bringing beauty to the mobile internet entertainment era.
Solution Construction
Bohe’s IoT-3288A, Rockchip 3288 motherboard; HD camera; and Visiontech MIPS facial recognition and customer flow counting software.
Solution Features
1. Accurate Customer Flow Statistics
By installing Bohe MIPS facial recognition customer flow statistics software, you can achieve the following through facial detection and tracking of incoming customers:
Real-time customer flow statistics: Counts daily and hourly real-time customer flow information, extracts facial features, and avoids double counting.
Customer flow trend statistics: Query customer flow trend statistics by the most recent day, week, month, quarter, half year, year, or three years, and presents them in bar or line graphs.
Dwell time statistics: Analyzes the duration of each customer’s appearance in front of the camera until their departure.

2. Customer Crowd Analysis
Using facial recognition technology, we can accurately determine the gender and age of each customer entering the store during any given time period. The accuracy for gender is over 99.5%, and the accuracy for age is over 90%, with an error of no more than three years. By analyzing the gender ratio and age of incoming customers, retailers can quickly adjust their product mix and marketing strategies to maximize returns.

3. Cultivate and tap into loyal fans
Using facial recognition technology, we can accurately identify VIP customers and repeat customers registered in our purchasing system. Store staff will receive push notifications about their purchase history, focusing on converting repeat customers into VIPs and VIPs into loyal fans, significantly increasing store sales.

4. Big Data Operations for Optimized Management
Facial recognition traffic data can be used to analyze how long different demographics spend in-store, which areas attract the most attention, which products attract the most attention, and what products attract different demographics. This data analysis can be used to optimize store management, identifying the most suitable business model, the ads that consumers most engage with, and the products with the highest conversion rates. By analyzing hot spots based on customer shopping paths, we can adjust the display of featured products to attract customer attention and address any deficiencies in store staff service. Optimization from the macro to the micro level maximizes store profits.

5. Targeted Advertising to Improve Conversion Efficiency
Using facial recognition, we can tailor advertising content to the gender and age of customers entering the store within 1-2 seconds. This effectively increases the appeal of advertising, precisely promotes products, and improves purchase conversion rates.

Application Scenarios
Shopping malls, commercial streets, jewelry stores, auto dealerships, cosmetics chains, branded clothing chains, food stores, eyewear stores, pharmacies, shoe and bag stores, and other public venues where people gather.

 
			 
				 
				 
		 
		 
		