Tuesday, 19 May 2009

Mobile Automatic Pedestrian Detection - ETH Zuerich

At ETH Zuerich a mobile version of pedestrian detection was developed. It is demonstrated in a clip that featured in a blog post over at technology review. It is stunning how accurate the software is able to distinguish between individuals in these rather crowded scenes. Plus, all is mobile, the camera is mounted in a car and presumably the software runs the analysis live, including these nice little trails the pedestrians leave behind.
Technology was developed by researches from the Computer Vision Laboratory at ETH Zuerich in a collaboration with Toyota.

stills taken from http://link.brightcove.com/services/player/bcpid1460879066?bctid=22789618001

Two papers related to the project above are available online:
Coupled Object Detection and Tracking from Static Cameras and Moving Vehicles, by Bastian Leibe, Konrad Schindler, Nico Cornelis, and Luc Van Gool
A Mobile Vision System for Robust Multi-Person Tracking, by Andreas Ess1 Bastian Leibe Konrad Schindler Luc Van Gool1, ETH Zurich, Switzerland 2KU Leuven, Belgium

No comments: