[Παλιό] -- ComputerVision@AIT:AutomaticFaceRecognitionSystem

Εδώ μπορείτε να ενημερώνετε ή να ενημερώνεστε για τη διοργάνωση διαφόρων συνεδρίων και σεμιναρίων.
Post Reply
User avatar
Einherjar
Venus Project Founder
Venus Project Founder
Posts: 3751
Joined: Tue Jan 27, 2004 4:42 pm
Academic status: Alumnus/a
Gender:
Location: Washington DC, USA
Contact:

[Παλιό] -- ComputerVision@AIT:AutomaticFaceRecognitionSystem

Post by Einherjar » Sun Nov 07, 2004 12:36 am

We are pleased to invite you to our next Open Seminar organized by AIT on Monday, November the 8th, at 16:30. The Seminar that will take place in amphitheatre 2A (2nd Floor), is entitled “Computer Vision @ AIT: An Automatic Face Recognition System” and will be conducted by Dr. Aristodimos Pnevmatikakis, Assistant Professor at AIT.

You may register at cefs@ait.edu.gr until Monday noon, November the 8th.

Abstract

Face recognition has an ever growing number of applications; for example, pilot systems are evaluated at public transportation terminals. However, variations in illumination, background, visual angle and facial expressions degrade the performance of face recognition systems. If performance could be boosted to that achieved by other biometric systems, face recognition systems would rank among the most desirable, because of the non-intrusiveness and the simplicity of the infrastructure required.
The face recognition system being built at AIT is given image sequences and a database of normalized faces. The task of the system is to identify individuals in the images whose faces match those in the database. To do so the system detects faces in the input images, then detects the eyes in the facial segments. Faster versions of the detectors provide tracking of the faces and the eyes, based on a previous detection. Based on the eye positions, the faces are de-rotated and normalized. Then the normalized faces are compared against a face-space trained using PCA. Out of N frames, the K faces most closely matching the face-space are forwarded to the recognition algorithm. The face recognizer makes K decisions that are combined to yield the identity of the face in the sequence.

περισσότερες πληροφορίες: Athens Information Technology
[Better to understand a little than to misunderstand a lot]
Post Reply

Return to “Συνέδρια - Σεμινάρια”