On Twitter

The Australian Bioinformatics Network was initiated by

ABACBS (pron.ˈabəkəs) runs the Australian Bioinformatics Network

The latest...
Handy links...
Open Jobs

International Workshop on Computer Vision for Accelerated Bioscience: CVAB 2013

At the International Conference on Computer Vision, ICCV 2013

Workshop Date: 2 December 2013

Workshop Venue: Room 201, Sydney Convention Centre

Topic/Call for Papers

This workshop aims to combine biological and computer vision research to enhance the scientific understanding of life.

Non-destructive visual measurement is a cornerstone of biology, underpinning some of our most basic understandings of form and function. Yet, there have been relatively few efforts to accelerate these processes through the use of Computer Vision techniques. Bottlenecks exists when getting information from the physical to the digital domain (acquisition), assembling the raw digital data into a more meaningful form (analysis), and when adding knowledge to the data to enhance its usefulness (annotation). In each case there are many opportunities to employ Computer Vision research.

The impacts of such efforts will be profound. Here are just two examples:

  • By relating genomic information to quantitative data extracted from images of growing plants, we stand to dramatically accelerate our understanding of gene function during plant development.
  • Characterising images from insect collections in morphologically meaningful ways would help create globally accessible taxonomic resources with applications in biosecurity, biodiversity and functional genomics.

This workshop will explore and showcase research efforts that apply novel Computer Vision techniques to better our understanding of natural organisms. Topics will include, but are not limited to:

  • 3D Reconstruction of Plants, Insects, or Animals
  • Microscopic Computer Vision
  • Multi-scale Computer Vision
  • Structural analysis of organisms
  • Tracking of animal motion or behaviour
  • Capturing the growth of individual organisms
  • Co-registration of optical images with volumetric scans
  • Capture of reflectance and transparency
  • New insight from existing scientific image collections and archives
  • Human-in-the-loop interfaces for classification and annotation
  • Computational photography of specimens
  • Computer Vision in harsh/outdoor environments
  • Case studies, including practical challenges faced by bioscientists




Workshop introduction

Invited Talk: Advancing Biodiversity Discovery with Computer Vision, John La Salle (Atlas of Living Australia and CSIRO)


Invited Talk: What there is to see: Imaging Spectroscopy for Scene Analysis, Antonio Robles-Kelly (National ICT Australia)


Virtual 3D Models of Insects for Accelerated Quarantine Control, Chuong Nguyen, David Lovell, Rolf Oberprieler, Debbie Jennings, Matt Adcock, Eleanor Gates-Stuart, John LaSalle (CSIRO)


Coffee Break


Invited Talk: Computer Vision: Can it help us digest Insect Soup?, Paul Flemons (Australian Museum)


Insect Soup Challenge - Segmentation, Counting, and Simple Classification, Katarina Mele (CSIRO)


3D Plant Modelling via Hyperspectral Imaging, Jie Liang, Ali Zia, Jun Zhou, Xavier Sirault (Australian National University, Griffith University and CSIRO)


Super-resolution 3D Reconstruction of thick biological samples: a computer vision perspective, Alessio Del Bue, Francesca Cella Zanacchi, Alberto Diaspro (Italian Institute of Technology)


Morning session discussion




Invited Talk: Imaging Less than Meets the Eye, Kyros Kutulakos (University of Toronto)


Extended Gaussian-filtered Local Binary Patterns for colonoscopy image classification, Siyamalan Manivannan, Ruixuan  Wang, Emanuele  Trucco, Adrian Hood (University of Dundee and University of Leeds)


Learning to Detect Basal Tubules of Nematocysts in SEM Images, Michael Lam, Janardhan Rao Doppa, Hu Xu, Abigail  Reft, Sinisa Todorovic, Thomas Dietterich, Marymegan Daly (Oregon State University)


Dirichlet Process Mixtures of Multinomials for Data Mining in Mice Behaviour  Analysis, Matteo Zanotto, Diego Sona, Francesco Papaleo, Vittorio Murino (Italian Institute of Technology)


Coffee Break


Invited Talk: A Framework for Ultra-High Resolution 3D Imaging, Michael S. Brown (National University of Singapore)


Zero-Shot Learning and Detection of Teeth in Images of Bat Skulls, Hu Xu, Michael Lam, Sinisa Todorovic, Thomas Dietterich, Andrea Cirranello, Paul Velazco, Nancy Simmons, Maureen O'Leary (Oregon State University and American Museum of Natural History)


High Precision Localization of bacterium and Scientific Visualization, Mohammadreza Hosseini, Arcot Sowmya, Pascal Vallotton, Tomasz Bednarz (University of New South Wales and CSIRO)


Invited Talk: Automatic flower categorization, Yuning Chai (University of Oxford)


Afternoon session discussion and review


Workshop concludes


Invited Speakers

Advancing Biodiversity Discovery with Computer Vision
(Dr John La Salle, Director, Atlas of Living Australia)

Recent advances in computer vision and accelerated phenomics techniques will help biologists face some of the big challenges of the next two to three decades: managing natural and agricultural landscapes under environmental change, emerging diseases, an increasing number of invasive species, and growing twice as much food on the same amount of arable land.  All of these challenges will rely on an extensive knowledge of biodiversity, and current methodology for documenting biodiversity is not keeping pace with the challenge.  This talk will explore whether computer vision can help supply the order of magnitude increase in the rate at which biodiversity information is captured necessary to inform biologists and environmental scientists as they face these challenges. 

What there is to see: Imaging Spectroscopy for Scene Analysis
(Dr Antonio Robles-Kelly, Principal Researcher and Research Leader at NICTA and Adjunct Associate Professor at ANU)

Imaging spectroscopy technology captures and processes image data in tens or hundreds of bands covering a broad spectral range. Compared to traditional monochrome and Red/Green/Blue (RGB) cameras, the multispectral and hyperspectral image sensors used for imaging spectroscopy can provide an information-rich representation of the spectral response of materials. Since imaging spectroscopy involves the ability to robustly encode material properties, object composition and concentrations of primordial components in the scene, it has a broad domain of application in the visual analysis of natural organisms. In this talk, I will ellaborate on the use of statistical pattern recognition techniques and physics-based computer vision for shape and compositional analysis and will give examples of how imaging spectroscopy can be used for biosecurity, quality control and image enhancement taking into account the materials in the scene and the object shape. This poses great opportunities for food security, health and precision agriculture where fruit can be graded with high accuracy in volume and pests can be detected before symptoms are apparent to the naked eye.

Computer Vision: Can it help us digest Insect Soup?
(Paul FlemonsManager of Collections Informatics, Australian Museum)

Museums hold many unsorted bulk entomology samples (termed soups). These samples are systematically collected and so contain scientifically valuable information on the distribution of biodiversity, and potentially new species.

The Australian Museum is exploring ways in which to harness the latent data from these soups and make available a range of data on insect distributions, diversity, abundance and for identifying new species and better understanding the distributions of others.

One of these ways is the use of computer vision to extract and classify insects from insect soup images. The resultant library of insect images could made more accessible and useful by then using crowdsourcing to tag the images so that scientists could more easily find related insects.

This will make the soups and the data derived from them accessible to the scientific community in a way never previously possible.

Imaging Less than Meets the Eye
(Professor Kyros Kutulakos, Department of Computer Science, University of Toronto)

When we snap a photo with a conventional camera, we record all incident light no matter how it got there. In this talk I will discuss a new family of cameras that gives us many more degrees of freedom: these cameras record just a fraction of the light coming from a controllable source, based on the actual 3D path followed. Photos and live video captured this way offer an unconventional view of everyday scenes in which the effects of scattering, refraction and other phenomena can be selectively blocked or enhanced, visual structures that are too subtle to notice with the naked eye can become apparent, and object appearance can depend on depth.

Although their basic operating principle is firmly rooted in 3D computer vision and computer graphics, the cameras themselves operate in the optical domain and produce images that require little or no computational post-processing. I will explain how these cameras work and show output from three prototypes we built, imaging a variety of common objects and people.

A Framework for Ultra-High Resolution 3D Imaging
(Associate Professor Michael S. Brown, Department of Computer Science, National University of Singapore)

This talk discusses an imaging framework to acquire 3D surface scans at ultra-high resolutions (exceeding 600 samples per sq mm).  The framework works by coupling a standard structure-light setup together with photometric stereo using a large-format ultra-high-resolution camera and can produce 3D scans with exceptional surface detail exceeding most existing technologies.

Automatic Flower Categorization
Yuning Chai, Department of Engineering Science, University of Oxford

Classifying flower species based on photographs is for most people a very challenging task. In this work, we developed a system which automatically discriminates between 272 different flower species. We started by collecting flower images from Flickr interest groups and had the images filtered by flower enthusiasts. For less represented species, we commissioned professional photographers. The algorithm, operates by first finding the foreground region in each photograph, followed by an embedding of the image into a high-dimensional descriptor, which is then fed into a standard linear classifier. We also created a website to demonstrate the categorization for uploaded images, and to extend the corpus to new flower species.


Workshop Registration

Workshop registration is via the ICCV 2013 registration websiteFor accepted submissions (only), presenters will be offered an extended 'early-bird' deadline of October 25.


  • Everyone needs a visa to Australia except Australians and New Zealanders
  • Contributors and attendees can get a letter of invitation once registration is completed
  • 6 weeks is the recommended time for obtaining a visa

Workshop Chairs

  • David Lovell, CSIRO, Australia
  • Matt Adcock, CSIRO, Australia

Program Chairs

  • Shahram Izadi, Microsoft Research, UK
  • Chuong Nguyen, CSIRO, Australia
  • Hongdong Li, Australian National University, Australia

Program Committee

  • Alexis Tindall, South Australian Museum
  • Andy Deans, Penn State University
  • Antonio Robles Kelly, NICTA
  • Changming Sun, CSIRO Computational Informatics
  • Chunhua Shen, University of Adelaide
  • Congtian Lin, Institute of Zoology, Chinese Academy of Sciences
  • Dadong Wang, CSIRO Computational Informatics
  • Donald Hobern, Global Biodiversity Information Facility (GBIF)
  • Graham Brown, Museum and Art Gallery of the Northern Territory
  • Jiangning Wang, Institute of Zoology, Chinese Academy of Sciences
  • John La Salle, Atlas of Living Australia
  • Jun Zhou, Griffith University
  • Ken Walker, Museum Victoria
  • Li Cheng, ASTAR Singapore
  • Nick Barnes, NICTA
  • Nicole Fisher, CSIRO Ecosystem Sciences
  • Pascal Vallotton, CSIRO Computational Informatics
  • Paul Flemons, Australian Museum
  • Ying Zheng, Duke University



Authors were invited to submit research, case study or position papers of 2 to 6 pages according to the (two-column) formatting guidelines available on the ICCV 2013 conference website at Submitted papers were reviewed in a double blind process. Only electronic submissions were accepted via the the online submission system:

Important Dates

Submissions due     7 October 2013

Notification of acceptance   18 October 2013

Submission of camera ready papers  29 October 2013

Workshop    2 December 2013

Insect Soup Challenge

This year's workshop also includes a Computer Vision Challenge, separate to the papers submission. This offers practitioners an additional way to participate. Further information can be found at:


To contact the workshop organisers, email CVAB[at]

Supported by...

  • The Australian Bioinformatics Network
  • Microsoft Research
  • The Australian National University