Master Thesis Seminar: Camera Auto Focus Controlled by Face Recognition on the GPU

Date: June 27, 2008 (Friday) at 13:00

Pär Gregersson och Karl Berggren presenterar sitt examensarbete "Camera Auto Focus Controlled by Face Recognition on the GPU".

Abstract

Embedded GPUs are coming into mobile phones. GPUs are powerful parallel processors designed for 3D graphics applications. With the advent of OpenGL2.0, GPUs also become programmable. This makes is possible to use the GPUs in many applications other than 3D graphics. The purpose of this masters thesis is to examine if face detection can be performed on a GPU instead of running it on dedicated hardware or on the CPU. We implemented a framework to train a face detector given an image set containing faces, and an image set not containing faces. This framework can be used to create simple GLSL ES 2.0 shaer calls that executes a face detection algorithm in fragment shader programs. We also created a reference implementation that runs on the CPU and detects faces in a video stream from the onboard camera of our target platform.

Room: Falsterbo, Ideon Beta

Last modified Dec 9, 2011 12:59 pm by Mikael.Antic@cs.lth.se

0310