The paper describes a method for an off-line analysis of human motion using views from multiple cameras, which enables reliable localization and identification of persons without the possibility of cascading errors and total failure of the tracking. The method is based on fusion of multiple sources of visual information, where relatively basic and noisy low level features are extracted at the bottom of the hierarchy. They are processed in a manner which models increasingly complex interactions, resulting in significantly more reliable results. Acceptance rate for Asian Conference on Computer Vision in 2012: 26.8%
Paper describes a parallel asynchronous master-slave implementation of DEMO, an evolutionary algorithm for multiobjective optimization. The implementation extends the use of DEMO from single processor use, to multiple interconnected multi-processor computers. It achieves high efficiency even on heterogeneous computer architectures. The paper describes a parallel algorithm, its differences from the serial algorithm and introduces a new measure of parallelism efficiency for the evolutionary algorithms.