Australian Centre for Visual Technologies Blog

Modelling Protector
November 20, 2013, 9:08 am
Filed under: ACVT, research | Tags: ,

HMCS Protector was the only boat in South Australia’s navy for much fo the time before federation.

We have a great project with the SA Maritime Museum to develop technologies for making 3D models of objects from historic photos, and much of the work centres on the Protector.  There are in fact a number of historical photos of the boat, but each one is in a different location, and with a different camera.  More interestingly, the shape of the boat changes over time too.

As part of the project I went to Heron Island, where Protector is now a breakwater, to capture some footage to make models from.

I’ve put some of the models we’ve made thus far on youtube.

ACVT Discovery Grant success
October 25, 2010, 4:55 pm
Filed under: news, research

Profs David Suter and Anton van den Hengel were both awarded ARC Discovery Grants to begin in 2011.  They are:

  • Computer vision from a multistructural analysis framework, Prof D. Suter.  Computer vision has applications in a wide variety of areas: security (video surveillance), entertainment (special effects), health care (medical imaging), and economy (improved automation and consumer products). This project will improve the accuracy and reliability of such applications. Advances will also lead to new products and industries.
  • Added depth: automated high level image interpretation, Prof Anton J van den Hengel, Prof Philip Torr, Dr Simon M Lucey, Humans are very good at understanding the world through imagery, but computers lack this fundamental capacity because they lack experience of what they might see. This project will provide this experience by combining the large volumes of imagery on the Internet with three dimensional information generated by humans for other purposes.

ISMAR Paper Prize
October 21, 2010, 9:40 am
Filed under: publications, research

The paper Real-time Modelling for AR Applications won second prize in the best paper awards at the International Symposium on Mixed and Augmented Reality 2010 in Seoul, Korea.

The paper describes a method for generating 3D models of objects which requires minimal user interaction.  The motivation was the desire to allow users to create special effects to video as it is captures.  The interface is thus suitable for implementation on a video camera or phone.

The video shows the technique being used to copy and paste real objects in live video.  The user can thus indicate the object they are interested in, and then indicate where they’d like it to appear.  The system models the object and renders it where it needs to be, all as the video is being captured.

The slides for the ISMAR talk are available here.

ACVT wins content-based image search CTD
August 18, 2010, 6:10 pm
Filed under: ACVT, news, research

The ACVT has been successful in attracting a Capability and Technology Demonstrator (CTD) Program award from the Federal Government and DSTO.

The award supports the development of a technology demonstrator based on ACVT’s existing capabilities in in content-based image retrieval.  Content-based image retrieval allows large image databases to be searched on the basis of the images themselves.   It means that you can use images to search for images.  If you have a picture of what you’re looking for, you can retrieve all of the similar images out of the database.

The effect of the capability is that it allows a small number of people to search a vast database of images for the content they are interested.  The alternative is really just that they look at random images hoping to find something.

There’s a story about it in the Australian.

CVPR Best Paper Award
June 17, 2010, 11:13 am
Filed under: news, publications, research

Our paper

Efficient Computation of Robust Low-Rank Matrix Approximations in the Presence of Missing Data using the L1 Norm, Anders Eriksson and Anton van den Hengel, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2010), June 2010, San Francisco, USA, IEEE 2009

won the CVPR’10 best paper award. There were almost 1750 submissions to the conference, of which 78, including ours were given oral presentations.

The paper presents a method for calculating a low rank factorisation of a matrix with missing elements on the basis of the L1 norm. Our approach is a generalisation of the Wiberg algorithm for achieving the same result using the L2 norm, which has recently been shown to be highly effective. A key step in the generalisation to the L1 case is to calculate the derivative of the solution to a linear programming problem, which allows the Jacobian of the over-arching minimisation problem to be calculated directly. Results for the new method show a significant performance advantage over competing approaches.

The pdf is available here.

The bibtex is

author = “Anders Eriksson and Anton van den Hengel”,
title = “Efficient Computation of Robust Low-Rank Matrix Approximations in the Presence of Missing Data using the L1 Norm”,
booktitle = “The Twenty-Third IEEE Conference on Computer Vision and Pattern Recognition (CVPR’10)”,
year = “2010”,
month = “June”,
address = “San Francisco, CA”,

Three ACVT papers at CVPR’10
June 17, 2010, 11:12 am
Filed under: news, publications, research

The ACVT had 3 papers accepted to CVPR this year.

They are

Two ACVT papers accepted to ECCV
June 17, 2010, 11:07 am
Filed under: news, publications, research

Tat-Jun Chin and Jin Yu had their papers accepted to ECCV 2010.

Details and pdfs soon.