INB Lunch-Seminar

Perceiving head pose and gaze direction for human-computer interaction:

Experiments and neural modeling

Prof. Dr. Heiko Neumann
Fakultät für Informatik, Abt. Neuroinformatik
Universität Ulm

We investigate basic mechanisms of social perception, in particular the detection of faces, their representation, and the estimation of the attentional state of a person from images. The latter is based on the estimation of head poses and the analysis of eye gaze patterns. In a first attempt we extend existing methods for head pose estimation and investigate the use of local image phase for gaze detection. We demonstrate that a simple implementation of methods without extensive training sessions or calibration is sufficient to accurately detect the view direction for, e.g., human-computer interaction. Using a database of real images with ground truth we analyze the proposed methods. Moreover, using video sequences, we demonstrate the robustness of tracking of facial regions based on a hierarchical scheme and how head pose is estimated from tracked facial regions.

In order to robustly detect facial image components we propose an iterative scheme of recurrent feature processing and grouping to generate sketch-like representations of oriented faces. This scheme is part of a more general processing architecture to extract in-variant properties of surface geometries. The computational modeling investigations were combined with psychophysical studies of gaze detection and eye tracking.

Zeit: Freitag, den 30. Juni 2006, 12 Uhr c.t.

Ort: Institut für Neuro- und Bioinformatik
        Seminarraum (1. OG, Raum 17),
        Ratzeburger Allee 160 (Geb. 64, 1. OG)