Tracking and surveillance in wide-area spatial environments using the abstract hidden markov model.

In this paper, we consider the problem of tracking an object and predicting the object's future trajectory in a wide-area environment, with complex spatial layout and the use of multiple sensors/cameras. To solve this problem, there is a need for representing the dynamic and noisy data in the tracking tasks, and dealing with them at different levels of detail. We employ the Abstract Hidden Markov Models (AHMM), an extension of the well-known Hidden Markov Model (HMM) and a special type of Dynamic Probabilistic Network (DPN), as our underlying representation framework. The AHMM allows us to explicitly encode the hierarchy of connected spatial locations, making it scalable to the size of the environment being modeled. We describe an application for tracking human movement in an office-like spatial layout where the AHMM is used to track and predict the evolution of object trajectories at different levels of detail.

Pedestrian detection for mobile bus surveillance

In this paper, we present a system for pedestrian detection involving scenes captured by mobile bus surveillance cameras in busy city streets. Our approach integrates scene localization, foreground and background separation, and pedestrian detection modules into a unified detection framework. The scene localization module performs a two stage clustering of the video data. In the first stage, SIFT Homography is applied to cluster frames in terms of their structural similarities and second stage further clusters these aligned frames in terms of lighting. This produces clusters of images which are differential in viewpoint and lighting. A kernel density estimation (KDE) method for colour and gradient foreground-background separation are then used to construct background model for each image cluster which is subsequently used to detect all foreground pixels. Finally, using a hierarchical template matching approach, pedestrians can be identified. We have tested our system on a set of real bus video datasets and the experimental results verify that our system works well in practice.

Context aware privacy in visual surveillance

In this paper we present preliminary work implementing dynamic privacy in public surveillance. The aim is to maximise the privacy of those under surveillance, while giving an observer access to sufficient information to perform their duties. As these aspects are in conflict, a dynamic approach to privacy is required to balance the system's purpose with the system's privacy. Dynamic privacy is achieved by accounting for the situation, or context, within the environment. The context is determined by a number of visual features that are combined and then used to determine an appropriate level of privacy.

Distributed query processing for mobile surveillance

Addressing core issues in mobile surveillance, we present an architecture for querying and retrieving distributed, semi-permanent multi-modal data through challenged networks with limited connectivity. The system provides a rich set of queries for spatio-temporal querying in a surveillance context, and uses the network availability to provide best quality of service. It incrementally and adaptively refines the query, using data already retrieved that exists on static platforms and on-demand data that it requests from mobile platforms. We demonstrate the system using a real surveillance system on a mobile 20 bus transport network coupled with static bus depot infrastructure. In addition, we show the robustness of the system in handling different conditions in the underlying infrastructure by running simulations on a real, but historic dataset collected in an offline manner.

Learning discriminative sequence models from partially labelled data for activity recognition

Recognising daily activity patterns of people from low-level sensory data is an important problem. Traditional approaches typically rely on generative models such as the hidden Markov models and training on fully labelled data. While activity data can be readily acquired from pervasive sensors, e.g. in smart environments, providing manual labels to support fully supervised learning is often expensive. In this paper, we propose a new approach based on partially-supervised training of discriminative sequence models such as the conditional random field (CRF) and the maximum entropy Markov model (MEMM). We show that the approach can reduce labelling effort, and at the same time, provides us with the flexibility and accuracy of the discriminative framework. Our experimental results in the video surveillance domain illustrate that these models can perform better than their generative counterpart (i.e. the partially hidden Markov model), even when a substantial amount of labels are unavailable.

Hierarchical semi-markov conditional random fields for recursive sequential data

Inspired by the hierarchical hidden Markov models (HHMM), we present the hierarchical semi-Markov conditional random field (HSCRF), a generalisation of embedded undirected Markov chains to model complex hierarchical, nested Markov processes. It is parameterised in a discriminative framework and has polynomial time algorithms for learning and inference. Importantly, we develop efficient algorithms for learning and constrained inference in a partially-supervised setting, which is important issue in practice where labels can only be obtained sparsely. We demonstrate the HSCRF in two applications: (i) recognising human activities of daily living (ADLs) from indoor surveillance cameras, and (ii) noun-phrase chunking. We show that the HSCRF is capable of learning rich hierarchical models with reasonable accuracy in both fully and partially observed data cases.

Recognising behaviours of multiple people with hierarchical probabilistic model and statistical data association

Recognising behaviours of multiple people, especially high-level behaviours, is an important task in surveillance systems. When the reliable assignment of people to the set of observations is unavailable, this task becomes complicated. To solve this task, we present an approach, in which the hierarchical hidden Markov model (HHMM) is used for modeling the behaviour of each person and the joint probabilistic data association filters (JPDAF) is applied for data association. The main contributions of this paper lie in the integration of multiple HHMMs for recognising high-level behaviours of multiple people and the construction of the Rao-Blackwellised particle filters (RBPF) for approximate inference. Preliminary experimental results in a real environment show the robustness of our integrated method in behaviour recognition and its advantage over the use of Kalman filter in tracking people.

Detecting deviations from known paths and speeds in a surveillance situation

This paper presents techniques for analysing human behaviour via video surveillance. In known scenes under surveillance, common paths of movement between entry and exit points are obtained and classified. These are used, together with a priori velocity data, to serve as a model of normal traffic flow in the scene. Surveillance sequences are then processed to extract and track the movement of people in the scene, which is compared with the models to enable detection of abnormal movement

Detection of cross-channel anomalies from multiple data channels

We identify and formulate a novel problem: crosschannel anomaly detection from multiple data channels. Cross channel anomalies are common amongst the individual channel anomalies, and are often portent of significant events. Using spectral approaches, we propose a two-stage detection method: anomaly detection at a single-channel level, followed by the detection of cross-channel anomalies from the amalgamation of single channel anomalies. Our mathematical analysis shows that our method is likely to reduce the false alarm rate. We demonstrate our method in two applications: document understanding with multiple text corpora, and detection of repeated anomalies in video surveillance. The experimental results consistently demonstrate the superior performance of our method compared with related state-of-art methods, including the one-class SVM and principal component pursuit. In addition, our framework can be deployed in a decentralized manner, lending itself for large scale data stream analysis.

Estimating the disease burden of alcohol : the problems arising when data and policy interact

While the risk of many disease or injury conditions is raised by the presence of health risk factors, it is usually impossible to identify individual risk-factor caused cases. However, the overall burden of disease and injury attributable to various health risks can be estimated if we know the prevalence of exposure to the risk factor in the community and the relative risk of each causally associated disease or injury for those exposed to the risk factor. This paper describes the estimation of the total number of deaths and hospital episodes attributable to alcohol consumption. It also describes how some methodologically trivial variations to the estimation methods raised some major issues in interpreting the results for the development of public policy.

The burden of disease and injury in Australia

This report provides an overview of results from the Australian Burden of Disease and Injury Study undertaken by the AIHW during 1998 and 1999. The Study uses the methods developed for the Global Burden of Disease Study, adapted to the Australian context and drawing extensively on Australian sources of population health data. It provides a comprehensive assessment of the amount of ill health and disability, the ‘burden of disease’ in Australia in 1996.

Mortality, disability, impairment, illness and injury arising from 176 diseases, injuries and risk factors are measured using a common metric, the Disability-Adjusted Life Year or DALY. One DALY is a lost year of ‘healthy’ life and is calculated as a combination of years of life lost due to premature mortality (YLL) and equivalent ‘healthy’ years of life lost due to disability (YLD). This report provides estimates of the contribution of fatal and non-fatal health outcomes to the total burden of disease and injury measured in DALYs in Australia in 1996.

Towards a standardised methodology for estimating alcohol-caused death, injury and illness in Australia

Two key methodological issues underlying different methods for calculating estimates of the number of alcohol-caused deaths are identified and recommendations suggested for future work.

1. How to adjust alcohol aetiologic fractions across time and place to reflect different levels of risky drinking. A common approach is outlined for both acute and chronic alcohol-related conditions. In the absence of consistent, reliable and regionally specific measures of the prevalence of risky alcohol consumption from national surveys, the use of per capita consumption data as a means of adjusting alcohol population aetiologic fractions over time and across regions is recommended.

2. Whether abstainers or low-risk drinkers should be used as the reference group when assessing the impact of alcohol consumption and how the resulting information is best presented. It is recommended that when abstainers are used as the reference group, the costs and benefits for both 'low-risk' and 'risky/high-risk' drinking should be identified. Using this approach, it was estimated that for Australia in 1998 there was a net benefit of 5,100 lives saved due to low-risk drinking, while there was a net loss of 2,737 lives due to risky/high-risk drinking. On its own, the figure of a net saving of 2,363 lives per year is a simplistic and potentially misleading picture of alcohol as a net benefit to public health and safety. For public health communications, there is still value in providing estimates using the low-risk drinking contrast, of the number of lives saved if risky/high-risk drinkers all became low-risk drinkers (n=3,292 in 1998). The use of the abstinence contrast, however, allows the more complex picture of alcohol's impact on public health to be apparent, e.g. including the estimated 1,505 deaths associated with low-risk drinking (mostly from cancer).

Multifaceted role of nitric oxide in an in vitro mouse neuronal injury model: transcriptomic profiling defines the temporal recruitment of death signalling cascades

Nitric oxide is implicated in the pathogenesis of various neuropathologies characterized by oxidative stress. Although nitric oxide has been reported to be involved in the exacerbation of oxidative stress observed in several neuropathologies, existent data fail to provide a holistic description of how nitrergic pathobiology elicits neuronal injury. Here we provide a comprehensive description of mechanisms contributing to nitric oxide induced neuronal injury by global transcriptomic profiling. Microarray analyses were undertaken on RNA from murine primary cortical neurons treated with the nitric oxide generator DETA-NONOate (NOC-18, 0.5 mM) for 8–24 hrs. Biological pathway analysis focused upon 3672 gene probes which demonstrated at least a ±1.5-fold expression in a minimum of one out of three time-points and passed statistical analysis (one-way anova, P < 0.05). Numerous enriched processes potentially determining nitric oxide mediated neuronal injury were identified from the transcriptomic profile: cell death, developmental growth and survival, cell cycle, calcium ion homeostasis, endoplasmic reticulum stress, oxidative stress, mitochondrial homeostasis, ubiquitin-mediated proteolysis, and GSH and nitric oxide metabolism. Our detailed time-course study of nitric oxide induced neuronal injury allowed us to provide the first time a holistic description of the temporal sequence of cellular events contributing to nitrergic injury. These data form a foundation for the development of screening platforms and define targets for intervention in nitric oxide neuropathologies where nitric oxide mediated injury is causative.

Multi-modal abnormality detection in video with unknown data segmentation

This paper examines a new problem in large scale stream data: abnormality detection which is localized to a data segmentation process. Unlike traditional abnormality detection methods which typically build one unified model across data stream, we propose that building multiple detection models focused on different coherent sections of the video stream would result in better detection performance. One key challenge is to segment the data into coherent sections as the number of segments is not known in advance and can vary greatly across cameras; and a principled way approach is required. To this end, we first employ the recently proposed infinite HMM and collapsed Gibbs inference to automatically infer data segmentation followed by constructing abnormality detection models which are localized to each segmentation. We demonstrate the superior performance of the proposed framework in a real-world surveillance camera data over 14 days.