About
We are the Imaging and Computer Vision (ICV) Research Group, part of the Cyber-Physical Systems Research Program at CSIRO’s Data61 Business Unit.
Our vision
We aim to harness the benefits of imaging and computer vision science for Australia.
Imaging sources are becoming increasingly available. They include sensing the electromagnetic spectrum (e.g. ubiquitous RGB cameras, infrared, hyperspectral, or X-ray), ultrasonic imaging, and computed images (e.g. MRI, PET and CT) most commonly used for medical imaging. This deluge of images requires sophisticated analysis to extract useful information and augment images for human assistance and decision-making.
Image analysis has been at the core of the Artificial Intelligence (AI) revolution enabled by the rise in computer power affordability (driven to a large extent by GPU technology).
AI methods are thus a key capability of our group, in particular, machine learning techniques such as deep learning.
Our people
With over 70 staff, PhD students and visiting scientists based at Brisbane, Canberra, and Sydney in Australia, we have strong collaborations with several Australian Universities where our students are enrolled. We also enjoy working with interns through their Honours and Masters degrees.
Our capabilities
We are currently working on four research themes:
Fixed vision in known area
Based on the newest camera technologies, using infra-red, or multiple standard RGB cameras we are developing new AI techniques for improved security, health, and productivity. Applications include mental health detection, customer analytics, and crowd counting. Areas of interest include privacy-preserving technologies and efficient edge computing.
Moving vision in known and unknown area
Many moving platforms include video and lidar sensing. In collaboration with the CSIRO Data61 Robotics and Autonomous Systems Group, we are pushing the envelope in terms of efficient embedded AI for object/human detection, and 3D scenes understanding. Applications include autonomous driving, photo content tagging, navigation for robots, and precision farming.
Specialised imaging of known objects
Using video for measuring and detecting objects is a core sensing technique. We partner with external companies and CSIRO’s experts across multiple domains to develop AI-based techniques for quantifying and measuring information from videos and images. This includes several medical applications (e.g. MRI or Xray) in collaboration with our colleagues at the Australian e-Health Research Centre.
Augmented reality and virtual reality
Augmented reality is fast becoming an important technique to help computers and humans work seamlessly. We are investigating several applications with high potential to create economic value and augment human capability, including manufacturing, safety, and healthcare.
Our partners
Image analysis has a very broad range of applications across many domains, so we work with a wide diversity of partners, particularly in health and agriculture.
Our collaboration range from working with multinationals (e.g. Perkin Elmer and Microsoft) through to start-up companies such as Maxwell+ (medical imaging), Laava (supply chains), and BVT; apart from collaboration with numerous industry advisory and standards bodies and Universities.
Within CSIRO, the ICV group maintains long-term collaborations, particularly with the Health & Biosecurity and Agriculture & Food Business Units.
Highlights and key scientific achievements
Illustrating the diversity of the group’s high-impact work in health are its:
- Work with BVT to restore partial sight (some of the blind patients are now in clinical trials).
- Algorithms and software for analysis of cell damage for drug discovery (1200 licences sold).
- Ongoing research to better measure neurodegeneration from MRI images.
Other work examples are:
- Estimation of agricultural yield and quality by measuring volume, morphology and hyperspectral signatures for farmer decision support.
- Work with Boeing to automatically detect, locate and document aircraft damage.
- Work with Saltor to measure the engagement of school children to improve educational outcomes. This work uses a platform technology developed for human behaviour analysis. Such work also has application in public safety, workplace productivity and early detection of mental health issues.
Contact us to learn more.