University of Twente Student Theses

Login

Facial landmark tracking on a mobile device

Wettum, Y.C. van (2017) Facial landmark tracking on a mobile device.

[img] PDF
1MB
Abstract:Face recognition is an important research topic in computer vision and biometrics and has become an important component in society. Face recognition is used in crime fighting and person identification at airports. With the increasing interest in applications of face recognition, the need for real-time face recognition has arisen. In order to perform face recognition, the facial features (e.g. eyes, nose, mouth) have to be localised. Usually, this is done by ’detection’ in each individual frame. This is an inefficient method for dynamic footage, since the location information of facial features from previous frames is not used. A more efficient method is ’tracking’. This paper is about the feasibility of facial feature localisation in dynamic footage using video tracking algorithms. This research is conducted with the aim of improving the speed of a 2D face recognition system on a smartphone. A C++ framework is realised in order to compare 4 video tracking algorithms and 2 widely used facial landmark localisation software libraries. The region-of-interest of the different video tracking algorithms is set automatically. A database is composed to evaluate the different algorithms on accuracy, precision, robustness and speed. With an ideal landmark detector, 2 of the 4 video tracking algorithms are able to outperform a state-of-the-art facial landmark localisation method on precision and accuracy.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:53 electrotechnology, 54 computer science
Programme:Electrical Engineering BSc (56953)
Link to this item:https://purl.utwente.nl/essays/71696
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page