Whole image understanding by convolutions on graphs. This project seeks to develop technologies that will help computer vision interpret the whole visible scene, rather than just some of the objects therein. Existing automated methods for understanding images perform well at recognising specific objects in canonical poses, but the problem of whole image interpretation is far more challenging. Convolutional neural networks (CNN) have underpinned recent progress in object recognition, but whole-im ....Whole image understanding by convolutions on graphs. This project seeks to develop technologies that will help computer vision interpret the whole visible scene, rather than just some of the objects therein. Existing automated methods for understanding images perform well at recognising specific objects in canonical poses, but the problem of whole image interpretation is far more challenging. Convolutional neural networks (CNN) have underpinned recent progress in object recognition, but whole-image understanding cannot be tackled similarly because the number of possible combinations of objects is too large. The project thus proposes a graph-based generalisation of the CNN approach which allows scene structure to be learned explicitly. This would represent an important step towards providing computers with robust vision, allowing them to interact with their environment.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE170101259
Funder
Australian Research Council
Funding Amount
$360,000.00
Summary
Zero-shot and few-shot learning with deep knowledge transfer. This project aims to develop few-shot and zero-shot learning, visual recognition techniques that can learn a visual concept with few or no visual examples. Visual recognition is a major component in Artificial Intelligence and used in cybernetic security, robotic vision and medical image analysis. This project will use deep learning to enable the zero/few-shot learning to use and model previously unexplored information, making zero/fe ....Zero-shot and few-shot learning with deep knowledge transfer. This project aims to develop few-shot and zero-shot learning, visual recognition techniques that can learn a visual concept with few or no visual examples. Visual recognition is a major component in Artificial Intelligence and used in cybernetic security, robotic vision and medical image analysis. This project will use deep learning to enable the zero/few-shot learning to use and model previously unexplored information, making zero/few-shot learning more practical, scalable and flexible. The project is expected to advance the applicability of visual recognition in many challenging scenarios and provide effective tools to analyse the online visual data for supporting Australia’s cybernetic security.Read moreRead less
Interacting with visualisations of extremely large graph structures on large displays. The latest technological progressions have delivered very large data sets that can be modelled as graphs or networks. Examples include: social networks, biological data, and software structures. This project will develop techniques to allow users to visualise the graphs in the entirety and directly interact with data.
A hardware accellerated platform for high-fidelity, high speed simulation of infrared scenes. Missiles present a major military and terrorist threat to aircraft and ships. A possible method to detect them is an infrared imaging system which is sensitive to a missile's spectrally unique rocket propulsion exhaust. It is both dangerous and expensive to conduct field trials; so simulation is used extensively. This project aims to use high performance computing to accelerate the slowest parts of the ....A hardware accellerated platform for high-fidelity, high speed simulation of infrared scenes. Missiles present a major military and terrorist threat to aircraft and ships. A possible method to detect them is an infrared imaging system which is sensitive to a missile's spectrally unique rocket propulsion exhaust. It is both dangerous and expensive to conduct field trials; so simulation is used extensively. This project aims to use high performance computing to accelerate the slowest parts of the industrial partner's existing simulations: the generation of simulated infrared images.
This project will improve the competitiveness of the manufacturer of infrared threat and warning systems and provide research training in an area of high performance computing.
Read moreRead less
Statistical Methods of Model Fitting and Segmentation in Computer Vision. Electronic sensors such as cameras and lasers can provide a rich source of information about the position, shape, and motion of objects around us. However, to extract this information in a reliable, automatic, and accurate way requires a sophisticated statistical theory of the process. Example applications include: video surveillance (better automatic detection of moving people and vehicles and of characterising what those ....Statistical Methods of Model Fitting and Segmentation in Computer Vision. Electronic sensors such as cameras and lasers can provide a rich source of information about the position, shape, and motion of objects around us. However, to extract this information in a reliable, automatic, and accurate way requires a sophisticated statistical theory of the process. Example applications include: video surveillance (better automatic detection of moving people and vehicles and of characterising what those people and vehicles are doing), industrial prototyping and inspection (measuring the size and shape of objects), urban planning (laser scanning streetscapes to create computer models of cities), entertainment industry (movie special effects and games), etc. Read moreRead less
Efficient data mining methods for evidence-based decision making. This project aims to develop efficient data mining methods for causal predictions. Evidence-based decision making (EBD), such as evidence-based medicine and policy, is always preferable. To support EBD, causal predictions forecast how outcomes change when conditions are manipulated. Progress has been made in theoretical research on causal inference based on observational data, but few methods can automatically mine causal signals ....Efficient data mining methods for evidence-based decision making. This project aims to develop efficient data mining methods for causal predictions. Evidence-based decision making (EBD), such as evidence-based medicine and policy, is always preferable. To support EBD, causal predictions forecast how outcomes change when conditions are manipulated. Progress has been made in theoretical research on causal inference based on observational data, but few methods can automatically mine causal signals from the data and methods for efficient causal predictions based on data are even fewer. This project will apply its methods to biomedical problems. The outcomes could support smart and data-driven evidence based decision making in many areas, such as therapeutics and government policy making.Read moreRead less
Early detection of component incompatibility in time-dependent computer architectures. Complex real-time systems are increasingly being built by integrating off-the-shelf components. There are obvious benefits to this approach, but the hidden costs associated with integration are still a major problem. Our proposed approach will enable early detection of integration problems, and thus provide potential for large cost savings. This brings with it clear benefits to industry. One industry that woul ....Early detection of component incompatibility in time-dependent computer architectures. Complex real-time systems are increasingly being built by integrating off-the-shelf components. There are obvious benefits to this approach, but the hidden costs associated with integration are still a major problem. Our proposed approach will enable early detection of integration problems, and thus provide potential for large cost savings. This brings with it clear benefits to industry. One industry that would benefit by such technology is the Australian Navy, which is increasingly being confronted with the challenge of integrating off-the-shelf components in large Naval Combat Systems. Read moreRead less
Continuous process improvement through workstation feedback for General Practice medicine using experts-in-the-loop data mining. This project investigates the iterative use of data mining results to allow experts to construct feedback to influence subsequent production work. We explore the problem in the context of General Practice medicine by having General Practitioners (GPs) review emerging patterns from their own practice's electronic medical records and author feedback to discourage undesi ....Continuous process improvement through workstation feedback for General Practice medicine using experts-in-the-loop data mining. This project investigates the iterative use of data mining results to allow experts to construct feedback to influence subsequent production work. We explore the problem in the context of General Practice medicine by having General Practitioners (GPs) review emerging patterns from their own practice's electronic medical records and author feedback to discourage undesirable patterns. The work will have immediate applicability to medical practice and will drive innovation in data mining method, notably for efficient identification of temporal and complex niche patterns. More broadly, the work will extend the way data mining is used to create new expectations of workstation behaviour.Read moreRead less
Through Walls Collaboration to Support Command and Control Operations with Eyes and Ears in the Field. Australia is a geographically dispersed country with locations of high concentrations of technology resources. Australia requires the ability to gather real time intelligence information in the field to support planning and operational decisions by a command team for military and civil defence operations. Currently Australia supports such operations in remote areas of the country and numerous o ....Through Walls Collaboration to Support Command and Control Operations with Eyes and Ears in the Field. Australia is a geographically dispersed country with locations of high concentrations of technology resources. Australia requires the ability to gather real time intelligence information in the field to support planning and operational decisions by a command team for military and civil defence operations. Currently Australia supports such operations in remote areas of the country and numerous overseas operations. A major research outcome is the design and development of interaction techniques for the mobile users in through walls collaboration systems to control and manipulate augmented reality information in the field across a number of application domains, such as medical, maintenance, military, search and rescue, and GIS visualization.Read moreRead less
Linear programming approach to nonlinear deterministic and stochastic control problems: perturbations methods and numerical analysis. The proposed research will significantly advance knowledge by creating new analytical and numerical methods for tackling complex nonlinear control problems arising in many applications. The study's outputs will lead to a deeper understanding of fundamental issues in mathematical modelling. Collaboration with renowned researchers will further improve Australia's st ....Linear programming approach to nonlinear deterministic and stochastic control problems: perturbations methods and numerical analysis. The proposed research will significantly advance knowledge by creating new analytical and numerical methods for tackling complex nonlinear control problems arising in many applications. The study's outputs will lead to a deeper understanding of fundamental issues in mathematical modelling. Collaboration with renowned researchers will further improve Australia's standing in the international research community. Also their visits may further promote research both within and outside the host institution. In particular, lectures and seminars that they will deliver will be transmitted to Australian universities participating in the Access Grid Room Project.Read moreRead less