Human behaviour recognition in data-scarce domains

This paper presents the novel theory for performing multi-agent activity recognition without requiring large training corpora. The reduced need for data means that robust probabilistic recognition can be performed within domains where annotated datasets are traditionally unavailable. Complex human activities are composed from sequences of underlying primitive activities. We do not assume that the exact temporal ordering of primitives is necessary, so can represent complex activity using an unordered bag. Our three-tier architecture comprises low-level video tracking, event analysis and high-level inference. High-level inference is performed using a new, cascading extension of the Rao–Blackwellised Particle Filter. Simulated annealing is used to identify pairs of agents involved in multi-agent activity. We validate our framework using the benchmarked PETS 2006 video surveillance dataset and our own sequences, and achieve a mean recognition F-Score of 0.82. Our approach achieves a mean improvement of 17% over a Hidden Markov Model baseline.

The concept of transport capacity and its use in different geomorphic process domains

The notion of sediment-transport capacity has been engrained in geomorphological and related literature for over 50 years, although its earliest roots date back explicitly to Gilbert in fluvial geomorphology in the 1870s and implicitly to eighteenth to nineteenth century developments in engineering. Despite cross fertilization between different process domains, there seem to have been independent inventions of the idea in aeolian geomorphology by Bagnold in the 1930s and in hillslope studies by Ellison in the 1940s. Here we review the invention and development of the idea of transport capacity in the fluvial, aeolian, coastal, hillslope, débris flow, and glacial process domains. As these various developments have occurred, different definitions have been used, which makes it both a difficult concept to test, and one that may lead to poor communications between those working in different domains of geomorphology. We argue that the original relation between the power of a flow and its ability to transport sediment can be challenged for three reasons. First, as sediment becomes entrained in a flow, the nature of the flow changes and so it is unreasonable to link the capacity of the water or wind only to the ability of the fluid to move sediment. Secondly, environmental sediment transport is complicated, and the range of processes involved in most movements means that simple relationships are unlikely to hold, not least because the movement of sediment often changes the substrate, which in turn affects the flow conditions. Thirdly, the inherently stochastic nature of sediment transport means that any capacity relationships do not scale either in time or in space. Consequently, new theories of sediment transport are needed to improve understanding and prediction and to guide measurement and management of all geomorphic systems.