New Paradigms for Robust Fitting: Kernelisation and Polyhedral Search. Outliers inevitably exist in visual data due to imperfect data acquisition or preprocessing. To enable computer vision applications that can perform reliably, robust fitting algorithms are necessary to counter the biasing influence of outliers. However, current robust algorithms are unsatisfactory: they are unreliable (due to using randomisation) or too computationally costly (due to using exhaustive search). This project wil ....New Paradigms for Robust Fitting: Kernelisation and Polyhedral Search. Outliers inevitably exist in visual data due to imperfect data acquisition or preprocessing. To enable computer vision applications that can perform reliably, robust fitting algorithms are necessary to counter the biasing influence of outliers. However, current robust algorithms are unsatisfactory: they are unreliable (due to using randomisation) or too computationally costly (due to using exhaustive search). This project will develop new robust algorithms to mitigate these shortcomings. It will do so by investigating two new paradigms of kernelisation and polyhedral search, which offer unprecedented theoretical insights into the problem. The outcomes will contribute towards computer vision applications that are more practical and reliable.Read moreRead less
Continuously learning to see. The ultimate goal of computer vision is to make a machine able to understand the world through analysis of images or videos. The new machine learning techniques developed in this project will enable previously impossible methods of computer vision and help strengthen Australia's competitiveness in this important area.
Defense against adversarial attacks on deep learning in computer vision. Computer vision applications rely heavily on deep learning, which is highly vulnerable to being fooled by adding subtle perturbations to object/image textures that are imperceptible to humans. This project aims to develop defense mechanisms to detect and remove adversarial patterns from the input images. The project expects to advance knowledge in understanding the vulnerabilities of deep learning, and to design deep learni ....Defense against adversarial attacks on deep learning in computer vision. Computer vision applications rely heavily on deep learning, which is highly vulnerable to being fooled by adding subtle perturbations to object/image textures that are imperceptible to humans. This project aims to develop defense mechanisms to detect and remove adversarial patterns from the input images. The project expects to advance knowledge in understanding the vulnerabilities of deep learning, and to design deep learning architectures that are inherently robust. The outcomes of this project will increase the security and reliability of computer vision by detecting, reporting and nullifying such attacks and will benefit the general public and industry on many fronts.Read moreRead less
How the brain generates robust behaviour in noisy sensory environments. This project aims to investigate the origins of variability in the control of movements. This project expects to generate new knowledge in the area of sensory and motor neuroscience by determining how variability in the activity of sensory and motor neurons accounts for variability in the initiation and control of eye movements. Expected outcomes of this project include international collaboration, development of new methods ....How the brain generates robust behaviour in noisy sensory environments. This project aims to investigate the origins of variability in the control of movements. This project expects to generate new knowledge in the area of sensory and motor neuroscience by determining how variability in the activity of sensory and motor neurons accounts for variability in the initiation and control of eye movements. Expected outcomes of this project include international collaboration, development of new methods for imaging neural activity in vivo, and refinement of theories concerning the cause and implications of noise in the brain. This should provide significant benefits such as a better understanding of why our movements are variable, and whether it is desirable or possible to minimise this variability. Read moreRead less
Deep Learning that Scales. Deep learning has dramatically improved the accuracy of a breathtaking variety of tasks in AI such as image understanding and natural language processing. This project addresses fundamental bottlenecks when attempting to develop deep learning applications at scale. First, this project proposes efficient neural architecture search that is orders of magnitude faster than previously reported, abstracting away the most complex part of deep learning. Second, we will desig ....Deep Learning that Scales. Deep learning has dramatically improved the accuracy of a breathtaking variety of tasks in AI such as image understanding and natural language processing. This project addresses fundamental bottlenecks when attempting to develop deep learning applications at scale. First, this project proposes efficient neural architecture search that is orders of magnitude faster than previously reported, abstracting away the most complex part of deep learning. Second, we will design very efficient binary networks, enabling large-scale deployment of deep learning to mobile devices. Thus this project will overcome two primary limitations of deep learning generally, however, and will greatly increase its already impressive domain of practical application.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE120100995
Funder
Australian Research Council
Funding Amount
$375,000.00
Summary
Visual navigation for sunny summer days and stormy winter nights. This project will develop innovative techniques for camera-based navigation that recognise locations under a wide range of environmental conditions caused by day-night cycles, weather and seasonal change. These techniques will enable the widespread use of cheap and lightweight cameras in robot and personal navigation systems.
Learning Robotic Navigation and Interaction from Object-based Semantic Maps. Our project aims to develop new learning algorithms that enable robots to perform high-complexity tasks that are currently impossible. Compared to existing methods that rely on low-level sensor data, we aim to achieve this by learning from a high-level graph representation of the environment that captures semantics, affordances, and geometry. The outcome would be robots capable of using human instructions to efficiently ....Learning Robotic Navigation and Interaction from Object-based Semantic Maps. Our project aims to develop new learning algorithms that enable robots to perform high-complexity tasks that are currently impossible. Compared to existing methods that rely on low-level sensor data, we aim to achieve this by learning from a high-level graph representation of the environment that captures semantics, affordances, and geometry. The outcome would be robots capable of using human instructions to efficiently learn complex interaction and navigation behaviours that transfer to unseen environments. Our research should benefit new applications in domains of economic and societal importance that are currently too complex, unsafe, and uncertain for robot assistants, such as aged care, advanced manufacturing and domestic robotics.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE220101390
Funder
Australian Research Council
Funding Amount
$402,900.00
Summary
Towards Human-like Machine Perception for Embodied AI. This project aims to investigate human-like visual perception, whereby AI machines can see and interpret the world like a human. The expected outputs will empower AI machines with the abilities of human-centered visual recognition and annotation-efficient learning through a set of deep learning techniques, and the ability to actively gather visual information through a reinforcement learning methodology (for decision support). This research ....Towards Human-like Machine Perception for Embodied AI. This project aims to investigate human-like visual perception, whereby AI machines can see and interpret the world like a human. The expected outputs will empower AI machines with the abilities of human-centered visual recognition and annotation-efficient learning through a set of deep learning techniques, and the ability to actively gather visual information through a reinforcement learning methodology (for decision support). This research is fundamental to the creation of embodied AI machines, which are expected to provide assistance to humans in industry, education and health. It thus will indicate immediate applications embracing autonomous vehicles and domestic robotics, providing scientific, social and economic benefits for Australia.Read moreRead less
Linkage Infrastructure, Equipment And Facilities - Grant ID: LE130100156
Funder
Australian Research Council
Funding Amount
$210,000.00
Summary
Computational infrastructure for machine learning in computer vision. The many trillions of images stored on computers around the world, including more than 100 billion on Facebook alone, represent exactly the information needed to develop artificial vision. All we need do is extract it. This project will develop the computational infrastructure required to allow Australian researchers to achieve this goal.
Discovery Early Career Researcher Award - Grant ID: DE120102948
Funder
Australian Research Council
Funding Amount
$375,000.00
Summary
Interactive computer vision for image interpretation. This project aims at pushing forward the fundamental research in interactive computer vision. The outcome of this project will enable reliable and efficient solutions to real world image interpretation tasks, such as medical image analysis, financial document processing, and impact evaluation from natural disasters.