Confidence-driven early object elimination in quality-aware sensor workflows

Distributed media rich systems, which can provide ubiquitous services to human users, require perceptive capabilities, transparently embedded in the surroundings, to continuously sense users' needs, status, and the context, filter and fuse a multitude of real-time media data, and react by adapting the environment to the user. Designing such real-time adaptivity into an open reactive system is challenging as run-time situations are partially known or unknown in the design phase and multiple, potentially conflicting, criteria have to be taken into account during the runtime. The ARIA media workflow architecture [4, 18, 19, 20], which is composed of adaptive media sensing, processing, and actuating units, processes, filters, and fuses sensory inputs and actuates responses in real-time. Unlike traditional workflows, a media processing workflow needs to capture inherent redundancy and imprecision in media, in terms of alternative ways of achieving a given goal. The object streams are only statistically accurate due to the inherent uncertainty of feature extractors. In this paper, we present a quality-aware early object elimination scheme to enable informed resource savings in continuous real-time media processing workflows.