Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Face Detection and Tracking in Video Lecture


F. Gentile

Semester Thesis, FS16 (10491)

The importance of the computer vision technology is rapidly growing and computer vision systems are already deployed in many application domains, such as surveillance, texture analysis and image recognition.

In this project we focus on a new possible application of computer vision, namely detection and tracking of lectures. With a system composed by two steerable cameras (a wide-angle camera and a zoom-in camera), it is possible to detect and track the lecturer: first the wide-angle camera locates the position of the lecturer in the stage, then the zoom-in camera is directed towards the region of interest by the motors. The zoom-in view allows us to have a clear view of the lecturer. From this view we can perform face detection, using the reliable Viola-Jones face detector. Moreover, we investigated how to improve the detection performance discarding false detections. We tried out two methods: the first one, based on matching of key-points descriptors, did not return satisfying improvements; the second approach, that relies on typical intensity colors of the human skin, met instead our expectations. The output of the face detector is fed into a tracking system, responsible to track the desired target throughout different frames. We used the Lukas-Canade tracker and extended its functionality enabling simultaneous tracking of multiple objects; finally we proposed a solution to drifting.

Supervisors: Nikolaos Kariotoglu, John Lygeros


Type of Publication:

(13)Semester/Bachelor Thesis

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2016:IFA_5428
Permanent link