http://hdl.handle.net/11343/159 2019-11-14T00:11:12Z http://hdl.handle.net/11343/230885 Video surveillance based crowd activity analysis Yang, Meng Video-based crowd motion analysis is an important problem in surveillance applications. Tasks such as identifying anomalous crowd motion patterns, finding sudden changes in the size of a crowd, event detection, and monitoring in public places have gained considerable attention due to safety and security considerations. While crowd analysis covers a wide range of applications, we focus on anomalous event detection and crowd behaviour identification. Although several approaches have been proposed for crowd scenario analysis in recent years, there still remain limitations in the existing techniques as well as challenges such as handling complex scenarios in the context of real-world scenes. One of the challenges in anomalous event detection is how to define anomalies. To address this challenge, we propose a spatio-temporal feature representation scheme that can be combined with a hyperspherical clustering approach to perform online anomalous crowd analysis in video surveillance. This system can flag abnormal events such as people standing statically or loitering in crowded scenes. Our next challenge is to devise a technique that can perform short-term anomalous object detection, as well as performing long-term anomalous motion/behaviour detection, while guaranteeing high accuracy and efficiency. To achieve this goal, we propose a deep learning architecture that comprises of a stacked denoising auto-encoder (SDAE), deep belief network (DBN) and plane-based one class support vector machine (PSVM) for automatic feature extraction and learning. The DBN-PSVM combination helps achieve better detection accuracy by means of dimensionality reduction to obtain a few high level representative features from the DBN before applying the PSVM. Considering that deep learning based architectures still suffer from high computational complexity, we model the crowd activity as graphs and address the problems of abnormal event detention and crowd behaviour identification in video sequences by using graph mining techniques. We first introduce a scalable clustering-based method that can be used to detect changes in crowd movements, including sudden running, merging, and splitting events. Furthermore, we develop a novel architecture to translate the crowd events in videos into a graph stream analysis problem, where major flow regions are represented as nodes and the crowd density is modelled as the edge weights in the graph. Anomalous crowd behaviour events are identified as the change points in the time series of graphs. Finally, we propose a more general graph-based architecture to achieve both anomalous event detection and crowd behaviour identification, where we take each moving object as a node, and use similarity measures (such as direction, magnitude, velocity and geo-distance) as the edge weights. The video sequence can then be represented as a time-varying graph sequence. We use a max-flow/min-cut approach to model the changes in the network and identify anomalous incidents and objects in the scene. Throughout the thesis, we empirically evaluate our proposed methods on a variety of benchmark and real-life video datasets of crowd movements in public spaces. © 2019 Meng Yang 2019-01-01T00:00:00Z http://hdl.handle.net/11343/230875 Genomics and population biology of the anthracnose pathogen (Colletotrichum tanaceti) of pyrethrum in Australia Lelwala, Lelwala Liyana Arachchige Ruvini Vinodya Colletotrichum tanaceti, is an emerging fungal pathogen of pyrethrum (Tanacetum cinerariifolium) in Australia. This thesis investigated the evolutionary potential and the molecular bases of sexual reproduction and pathogenicity of C. tanaceti by combining genomics, comparative genomics, molecular genetics and population genetics, with molecular biology and mycology techniques. The whole genome of the holotype isolate of C. tanaceti, BRIP57314 was sequenced, assembled and annotated then supplemented with a RNA transcriptome thus providing the first public genome sequence of C. tanaceti. The repetitive elements in the genome were identified in silico. Comparative genomics identified a relatively high repeat content therefore, suggesting a high genomic plasticity of the C. tanaceti genome. The bipartite nature of the C. tanaceti genome with distinct gene-sparse repeat and A-T rich regions were identified using in silico methods. Signals of genome-wide repeat induced point mutation (RIP) were identified using RIP-indices. The phenomenon of RIP in C. tanaceti genome was therefore, hypothesized to cause the A-T rich regions that evolve at a higher rate than the rest of the genome, generating a ‘two-speed genome’. Genomics tools were used to compare the C. tanaceti genome to publicly available genomes of other Colletotrichum spp. Single-copy orthologs among the species were used to develop a phylogram. This was then converted to a chronogram to estimate the divergence of C. tanaceti from C. higginsianum at 9.97 million years ago. The genome sequence and the predicted proteome of C. tanaceti were utilized to identify the putative pathogenicity genes of C. tanaceti. A large array of putative genes involved in infection and colonization of the host tissue such as CAZymes, effectors, secondary metabolites, transporters, cytochrome P450 and proteases were identified in the C. tanaceti genome. This was achieved by comparing genome homology to database entries and bioinformatics tools for de novo identification of fungal pathogenicity genes. The comparison of the putative pathogenicity gene repertoire of C. tanaceti with other Colletotrichum spp. indicated C. tanaceti had a similar pathogenicity profile to Colletotrichum spp. with intermediate host ranges. Therefore, C. tanaceti was speculated to have potential hosts other than pyrethrum. Furthermore, the in silico analyses revealed that the pathogenicity genes of C. tanaceti are located significantly closer to the tandem repeats, transposable elements and the RIP-affected regions of the genome. This suggested the potential of these pathogenicity genes to diversify and evolve adaptively. The population biology of C. tanaceti was assessed using a neutral, polymorphic microsatellite marker library developed from the draft genome sequence of C. tanaceti. Symptomatic pyrethrum leaves were hierarchically sampled across pyrethrum growing regions of Australia over three years in order to establish a collection of C. tanaceti isolates. These isolates were genotyped using the microsatellite marker library. Analysis of the genotypic data indicated the high genotypic diversity and only little differentiation of the populations of C. tanaceti. The index of association and the genotypic data indicated the mixed mode of reproduction of C. tanaceti with both sexual and asexual reproduction. Coalescent–based analysis of isolation with migration of the 2016 populations from the two growing regions of pyrethrum in Australia, indicated that long-distance gene flow of C. tanaceti is possible between Victoria and Tasmania which are separated by a sea strait. Approximate Bayesian computation methods revealed that the Tasmanian population of C. tanaceti was the source of the disease introduced to pyrethrum growing in Victoria. Furthermore, the hypothesis that C. tanaceti is likely to have shifted hosts from another unidentified host which was already present in Australia was supported. Combined, these results indicated a high evolutionary potential for C. tanaceti which supports observations of the pathogen’s genome. The sexual recombination strategy of C. tanaceti was further evaluated in vitro. Classical mating experiments revealed the fully heterothallic nature and instability of mating in C. tanaceti. These mating experiments revealed the presence of more than two mating specificities in C. tanaceti, a departure from the typical unilocus bi-allelic mating system of ascomycetes. The genome sequence of an isolate that engages in successful reproduction with the holotype isolate: BRIP57315 was also sequenced, assembled and annotated. The two genome assemblies and their annotations were utilized for in silico homology-based identification of the putative mating type MAT locus, pheromone precursors, pheromone receptors, mating type switching genes and other HMG box genes that may be involved in the sexual reproduction of C. tanaceti. Both genomes only contained the MAT1-2 idiomorph with identical MAT1-2-1 coding regions in the two genomes. Comparative genomics approaches, such as sequence alignment and phylogeny, were utilized to estimate orthology to functionally validate mating genes of other fungal genera and thereby to predict the functions of putative mating genes of C. tanaceti. Based on these results, C. tanaceti was hypothesized to have a multilocus mating system. In summary, this thesis elucidated the high evolutionary potential of the anthracnose pathogen of pyrethrum. The work indicates the pathogen has a highly plastic genome, a large array of pathogenicity genes that evolve rapidly, a mixed reproduction strategy, the potential for long-distance gene flow and the potential for infecting alternative hosts. These observations should be taken into consideration to ensure durability of future disease management strategies of pyrethrum. Moreover, this study also improved our understanding of the molecular basis of pathogenicity and the sexual reproduction of the important plant pathogenic fungal genus, Colletotrichum. The two genome assemblies and the candidate genes available through this study provide a solid foundation into future research on ‘omics’ and molecular genetics of pathogenicity and sexual reproduction of Colletotrichum. © 2019 Lelwala Liyana Arachchige Ruvini Vinodya Lelwala 2019-01-01T00:00:00Z http://hdl.handle.net/11343/230874 Analysis of Volatile Organic Compound Emissions from Fragranced Consumer Products Nematollahi, Neda Fragranced consumer products—such as cleaning supplies, air fresheners, and personal care products— emit numerous volatile organic compounds (VOCs), some classified as potentially hazardous. In addition, emissions from fragranced products can be a primary contributor to indoor and outdoor air pollutants, and human exposure risks. Further, exposure to volatile emissions from fragranced products has been associated with adverse human health effects in approximately one-third of Australians. However, fragranced products are exempt from full ingredient disclosure, thus limiting information and awareness about emissions. Moreover, relatively little prior research has analysed emissions from these products. The overall aim of this thesis research is to investigate volatile emissions from a wide range of fragranced consumer products. As a three-part research study, it investigates VOCs emitted from 200 products in three groups: 42 baby products, 24 essential oils, and 134 consumer products. Products include both regular versions and those with claims of being green or related terms such as organic or natural. Volatile emissions from each individual product were analysed using headspace gas chromatography/mass spectrometry. Overall, the study found 2,817 VOCs, including 848 potentially hazardous VOCs, emitted from the 200 products. Specifically, for each group of products, the study found (1) 684 VOCs including 207 potentially hazardous VOCs in baby products; (2) 589 VOCs including 124 potentially hazardous VOCs in essential oils; and (3) 1,538 VOCs including 517 potentially hazardous VOCs in consumer products. Across the 200 products, fewer than 4% of VOCs and 9% of potentially hazardous VOCs were disclosed by being listed on product labels. For each of the three groups of products, emissions of potentially hazardous VOCs were not significantly different between regular and green fragranced products. This dissertation research provides contributions to knowledge through the analysis and comparison of relatively large number and diversity of products, a broad suite of VOCs, compounds classified as potentially hazardous, and both regular and green products. Results from this research provide new knowledge on volatile emissions from consumer products, which can improve awareness of potential exposures and effects on air quality, health, and societal well-being. © 2019 Neda Nematollahi 2019-01-01T00:00:00Z http://hdl.handle.net/11343/230873 Bike skills training for children with cerebral palsy Toovey, Rachel Ann Marjorie Training targeted towards goals that are meaningful to children with cerebral palsy (CP) and their families is needed to improve function and support participation in physical activities in this population. Riding a two-wheel bike is a common goal for ambulant children with CP, yet little specific evidence exists to guide clinicians and families. This thesis developed and tested a task-specific approach to training bike skills in this population through three studies: 1) a systematic review, 2) a practice survey, and 3) a randomised controlled trial (RCT). Each study used the International Classification of Functioning, Health and Disability (ICF) as a framework. While strong evidence exists for task-specific training (TST) for improved upper limb (UL) function in this population, prior to this thesis the literature regarding TST for gross motor skills, including bike riding, in ambulant children with CP had not been systematically appraised. Thus, Study 1 aimed to systematically evaluate the effectiveness of task-specific gross motor skills training for improving activity and participation outcomes in ambulant school-aged children with CP. This review involved 13 studies of low-to-moderate overall quality and found effects of TST were positive for participation-related outcomes, and mixed for specific skill performance and functional skills, while little or negative effects were found for general gross motor skills. This study identified the need for higher quality studies and reporting that enables evidence synthesis. Given the importance of understanding current practice when designing effectiveness studies, Study 2 involved a survey of 95 physiotherapists (PTs) and occupational therapist (OTs) in Australia about their practices when training two-wheel bike skills in children with CP. This study found that while functional approaches to training and goal-based assessment and evaluation were predominant, overall practices appear highly variable. Moreover, the need to develop and test bike-specific measures and interventions in this population was highlighted. The findings from Studies 1 and 2 informed the design of Study 3. This multi-site assessor-blind RCT aimed to determine if a task-specific approach was more effective than a parent-led home program for attaining individualised two-wheel bike riding goals in ambulant children with CP. Sixty-two children were randomly allocated to either the task-specific approach (n=31) or home program (n=31). The primary finding was that the task-specific program was more effective than the home program for goal attainment at one week post-intervention. Greater odds of goal attainment were retained at three months and evidence of better outcomes following the task-specific program were found for some outcomes related to participation in bike riding, physical activity and self-perception. In addition, there was evidence of mixed effects for functional skills, and little difference in bike skills and health-related quality of life. While each of these studies provides an original contribution to the literature, together they form a significant foundation for evidence on training bike skills in ambulant children with CP. Use of the ICF across the thesis meant findings could be synthesised and enhanced the clinical relevance of the research. Given that an effective approach for attaining two-wheel bike riding goals in this population now exists, training for clinicians to optimise knowledge translation should be developed. Future research should seek to understand relationships between bike skills training and a broader range of ICF domains and levels of function in CP, tailor interventions to individuals and determine longer-term outcomes. © 2019 Rachel Ann Marjorie Toovey 2019-01-01T00:00:00Z