LAN Facial Recognition Pattern Sharing for Retail Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial recognition technologies in commercial or public areas face inefficiencies due to the need for mobile phone signals (Bluetooth or WiFi) and high processing resources, leading to reduced tracking efficiency and long processing times.
Innovation Solution
A method that generates unique facial patterns from images captured by cameras in a local area network, comparing and sharing them among computers to expedite facial recognition processing, reducing duplicate counting of individuals and enhancing tracking efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If mobile phone signals (Bluetooth or WiFi) are used for tracking, then tracking capability is provided, but tracking efficiency is significantly reduced
Solution Approach 1:
The patent replaces the mobile phone signal-based tracking mechanism (requiring Bluetooth/WiFi) with a facial recognition system using cameras. The system captures facial images, generates facial patterns, and compares them to identify individuals, eliminating the need for mobile devices to be enabled and significantly improving tracking efficiency while maintaining reliable identification.
2Measurement precision
If large processing resources are used for facial recognition, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The patent pre-generates and stores facial patterns for known individuals in a database before actual recognition is needed. When a new face is captured, the system only needs to compare the generated pattern against stored patterns rather than performing complex analysis from scratch, significantly reducing processing time while maintaining high accuracy.
Solution Approach 2:
The patent creates a simplified representation (facial pattern) of each person's face and stores it as a reference copy. During recognition, instead of analyzing the entire facial image in detail, the system compares the new facial pattern against the stored pattern copies, reducing computational resources and processing time while maintaining recognition accuracy.
3Area of stationary object
If multiple cameras detect the same person, then coverage is improved, but duplicate counting occurs
Solution Approach 1:
The patent implements a feedback mechanism where facial patterns and identification results from one camera are shared with other cameras in the network. When a camera detects a person, it generates a facial pattern and compares it against stored patterns; if a match is found, the identification result is fed back to other cameras, allowing them to recognize the same person without duplicating the count, thus maintaining comprehensive coverage while eliminating duplicates.
Data Source
AI summary
A method that expedites processing of facial recognition from facial images captured by different cameras in a local area network (LAN). The method generates a first facial pattern for a first facial image of a person and a first facial identification that is unique to the first facial pattern. A second facial pattern is generated from a second facial image and compared, by a computer in the LAN, with the first facial pattern to determine whether the first facial image and the second facial image are both from the same person. A second facial identification of the second facial pattern is generated when the second facial image is determined to be different from the first person. If the second facial pattern is determined to be the same as the first facial pattern generated from the first facial image of the first person, the second facial pattern is assigned the first facial identification. The first and second facial patterns and their facial identifications are shared among the computers in the LAN to expedite processing of facial recognition.