Jramos followface

From jderobot

Revision as of 01:40, 27 March 2009 by J.ramos (Talk | contribs)
(diff) ←Older revision | Current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search
  • Project Name: Face detection and tracking on jdec platform.
  • Academic Year: 2007-2008
  • Degree: Undergrad (Ingeniero técnico en informática de sistemas [1])
  • SVN Repository: [2]
  • Tags: face detection, face tracking, Adaboost cascade, Haar features, García Mateo's Integral Projections, Rowley's neural networks, Rob Hess' SIFT implementation.
  • Technology: OpenCV, jdec suite, C, SIFT.
  • State: Finished
  • Abstract:

In the recent years, great improvements in computing have made possible the development of new scientific areas such as computer vision. Nowadays, cameras are low-cost sensors whose images provide us with a huge amount of information from the environment where they may be used. Extracting that information and interpreting it let us create applications with intelligent behaviours, really useful to carry out hard or monotonous daily tasks.

Among the various areas aimed to study within computer vision, those related to facial processing become outstanding. Using these techniques with regard to security issues may be the most notable, but there are a wide variety of uses such as perceptual interfaces, video indexing, etc.

The goal of this project was the development of facial detection and tracking algorithms.

We studied several detection techniques in order to build some modules for jdec platform. Specifically, we worked on three: AdaBoost cascade based on Haar-like features, Rowley's neural network, and García Mateo's integral projections.

This project also dealt with the face-tracking problem, using low-cost methods as well as some higher-level ones. We studied their feasibility according to their flexibility, power and speed. We showed several experiments on an assorted and wide set of videos to evaluate the temporal continuity of the tracking method taken. Furthermore, we built a software prototype to command a PTZ camera, so that it could be self-positioned to track a face.

  • Documentation: 
Thesis: PDF, Spanish (latex code not available yet)
Presentation: not available yet (latex code not available either)


  • Videos:
Retrieved from "http://jde.gsyc.es/index.php/Jramos_followface"
Personal tools