Spotlight
Ten Years of Pedestrian Detection, What Have we Learned?
Instead of using a large variety of classifier types, it is possible to obtain best results in shortest time by precisely applying the sliding windows technique.
Directors & Senior Researchers
Bjoern Andres
Klaus Berberich
Jasmin Christian Blanchette
Karl Bringmann
Andreas Bulling
Balakrishnan Chandrasekaran
Anja Feldmann, Director
Mario Fritz
Olga Kalinina
Andreas Karrenbauer
Thomas Lengauer, Director (em.)
Christoph Lenzen
Tobias Marschall
Kurt Mehlhorn, Director (em.)
Karol Myszkowski
Nico Pfeifer
Gerard Pons-Moll
Simon Razniewski
Rishiraj Saha Roy
Bernt Schiele, Director
Marcel Schulz
Hans-Peter Seidel, Director
Thomas Sturm
Christian Theobalt
Jilles Vreeken
Christoph Weidenbach
Gerhard Weikum, Director
Andrew Yates
Mission
While the acceleration of hardware has been a landmark of progress in computing technology in the past few decades, the computing enhancements that it provides is dwarfed by the increase in speed, performance, and robustness resulting from new algorithms.
As a point in case, the status of hardware and algorithms in 1970 allowed to compute an optimal tour of a traveling salesman (a classical optimization problem and accepted benchmark for computing power) through 120 cities. Increasing the number of cities from n to n+1 leads to a multiplicative increase of the number of possible tours by a factor of n. Thus, relying only on the increase of hardware speed, with today’s technology, and the algorithms of 1970 we could find optimal tours among only 135 cities. It is the progress in algorithms that, today, enables us to find optimal tours between many thousand of cities. Relying only on progress in hardware this performance would not be achievable in hundreds of years.
The Max-Planck Institute for Informatics is devoted to cutting-edge research in informatics with a focus on algorithms and their applications in a broad sense.