Distributed Facial Recognition in Monitoring Cameras
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current monitoring systems face high processing loads and delays in facial recognition due to the transmission of still images from multiple cameras to a central host device for face checks, making rapid detection of individuals challenging.
Innovation Solution
The system distributes the processing load by having monitoring cameras perform initial feature checks and transmit results to a management device, which then directs time-based checks to other cameras, allowing the management device to rapidly detect specific persons by offloading image processing tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the host device performs face checks on still images from multiple monitoring cameras, then comprehensive facial recognition can be achieved, but the processing load on the host device becomes large causing process delays
Solution Approach 1:
The patent divides the facial recognition system into two segments: monitoring cameras that perform initial face detection and feature extraction on captured images, and a host device that receives only the extracted feature quantities for comprehensive matching. This segmentation reduces the processing burden on the host device while maintaining recognition accuracy through distributed computation.
Solution Approach 2:
The monitoring cameras perform preliminary face detection and feature extraction before transmitting data to the host device. By conducting initial processing at the camera level, the system prepares data in advance, reducing the computational load on the host device and accelerating the overall recognition process.
2Measurement precision
If still images are transmitted from multiple monitoring cameras to the host device for face checks, then facial recognition can be performed, but network traffic increases and processing time increases
Solution Approach 1:
The patent extracts only the essential feature quantities from captured images at the monitoring camera level, rather than transmitting complete image data to the host device. This extraction approach maintains face check capability while significantly reducing network traffic and processing time, as only compressed feature data needs to be transmitted and processed centrally.
3Ease of operation
If the host device performs all face checking processes, then centralized control is maintained, but the processing load on the host device becomes large
Solution Approach 1:
The system segments processing functions between monitoring cameras and the host device, with cameras handling initial detection and feature extraction, and the host device performing centralized matching and coordination. This segmentation maintains centralized management capabilities while distributing computational complexity, reducing the processing burden on the host device.
Solution Approach 2:
The monitoring cameras perform self-service by autonomously conducting face detection and feature extraction on captured images before transmitting results to the host device. This self-service capability reduces the processing burden on the host device while maintaining centralized coordination for overall system management.
Data Source
AI summary
There is provided a monitoring system which includes a plurality of monitoring cameras and a management device. The management device transmits feature information relevant to a person to one or more first monitoring cameras and receives check results, and transmits time information in which the person is captured to second monitoring cameras based on the check results. The management device specifies the person based on a check result acquired in such a way that the second monitoring cameras perform a check using the time information.