Face Detection and Tracking with Web Camera

Ruslana Makovetsky and Gayane Petrosyan

Comp-558 course project(Winter 2012)





Introduction

Human-robot interaction receives an increasing attention among the researches in different areas of interest during last years. In the current project we address one of the computer vision tasks, involved in developing of such an interactive system. Namely, we are interested in detecting human user presence and tracking of his attention. Possible applications for this would include receptionist robot, which needs to perform particular actions when human approaches or leaves the system, interactive games, commercials etc.
The goal of the current project was to implement a commercial player, having the following specifications:
  • automatically start playing a movie if human approaches the display,
  • pause a movie if human loses his attention (for example, turns away),
  • stop a movie if human leaves the location.

As a a result, we were looking for a face detection and tracking algorithm, which would enable us to detect faces in frames, forwarded from a web camera, and meeting the following requirements:
  • real-time performance,
  • variability in scale (person may move back and forward),
  • in-plane rotations (person may shake his head),
  • possible occlusions (person may perform actions, such as bringing a cup to his face, partially cover his face by hands, etc.)
  • ability to distinguish between profile and non-profile faces - this requirement is particularly important for our application, since it enables to detect situations when user loses his attention.

The algorithm presented by Viola and Jones for real-time face detection and its extension for multiple views were adopted for the system implementation. The performance of the implemented system is reasonably good for the real time system - the systems processes about 8 frames per second.

The full paper can be found here



Demo

Program demonstration for single face

The Camtasia Studio video content presented here requires JavaScript to be enabled and the latest version of the Adobe Flash Player. If you are using a browser with JavaScript disabled please enable it now. Otherwise, please update your version of the free Adobe Flash Player by downloading here.


Program demonstration for multi face

The Camtasia Studio video content presented here requires JavaScript to be enabled and the latest version of the Adobe Flash Player. If you are using a browser with JavaScript disabled please enable it now. Otherwise, please update your version of the free Adobe Flash Player by downloading here.




Downloads

Demo videos of the program Single face demo, Multi face demo and test video
Source code can be found here


Gayane Petrosyan, Ruslana Makovetsky (May 2012)