Workshop on Computer Vision for Computer Games

Research into visual game-interaction technologies is becoming one of the core problem areas of the computer vision research community. The field enjoys well-funded national and international funding programmes, especially in collaboration with industrial partners boasting a plethora of companies. This work offers a novel human-computer interaction experience. Yet, despite the apparent interest, the core problems remain as complex and varied as any in computer vision. The deployment of truly intelligent and robust human-computer interaction systems face a number of specific challenges including the segmentation and tracking of individuals in badly illuminated scenes; extracting body pose; characterization of actions; reduction in false alarm rates; real time 3D scene reconstruction and motion analysis; and the semantic linkage between different sensors. To address these challenges, contributions are particularly welcome in the following areas:
* Real time (low complexity) object tracking and human action classification
* Event and activity modelling
* Behaviour analysis and recognition
* 3D scene reconstruction
* Real-time motion estimation
* Scalability – computational efficiency (cuda computing, shaders etc)
* Face and human body modelling
* Multi-camera and multi-sensor fusion