Object Search

Share         

Face Recognition

Introduction

A facial recognition system is a computer application for automatically identifying or verifying a person from a digital image or a video frame from a video source. It can be used mainly in the areas of security, service, and entertainment.

 

  1. For security application: one direction is criminal searching, e.g., at customs, transportation, and shopping center. Another direction is alarm alert, e.g., for home, community, card cheating and casino.
  2. For service application: such as VIP service at hotel/bank/airport, etc, help to find the lost person, especially for young and elderly, and advertising push based on age/gender estimation.
  3. For entertainment: such as the application of which star looks like you, automatic photo management and personalized TV show.

 

Technical

For a facial recognition system, usually the following steps are needed. i.e., face detection, facial landmark localization, e.g., eye, face normalization, feature extraction and compared with registered database to give the results.The pictures below show two demos created by using our technology:

The first demo is to identify a query Image from a dataset of known persons. In this demo screen, the left side is the selected query image and the right side is the output result of identity with face detection and eye localization.

The second demo is for a given query image, images containing the query person’s face can be retrieved. These images could have single or multiple faces, etc.

 

 

Our system has the following technology key points:

  1. It is ROBUST to variations in lighting, face expressions and face sizes
  2. It is with LOW MEMORY RESOURCE, e.g. around 1MB model file
  3. It is HIGH SPEED for feature extraction, e.g., around 20ms on a PC with Core 3.2GHz CPU & 32GB RAM.

The Future

The target of the project is to make the application of face recognition on mobile platform and cloud service. The facial landmark localization for mobile application and surveillance will be also developed. Moreover, fast large scale database searching and high speed feature extraction will be considered.

People Involved

Professor: Xu Dong

Researcher: Zhang Wenchao

Video​​

​​
 
​​​​​​​