CSIRO's automated insect imaging system

CSIRO's automated insect imaging system extracts taxonomic information from hundreds of images.

Insect imaging system to accelerate species discovery

CSIRO scientists are developing an automated insect imaging system that will speed up the process of classifying and naming thousands of new species every year.

  • 9 April 2010 | Updated 26 March 2013

Meeting the need

Scientists around the world are looking for innovative ways to accelerate the identification of new species.

While there may only be three new bird species discovered each year, there are around seven thousand new insect species, leaving a significant backlog of work for insect taxonomists.

Image analysts from CSIRO Mathematics, Informatics and Statistics are working with taxonomists from the Australian National Insect Collection to develop an automated imaging tool to help them classify the large backlog of potentially new insect species.

The technology

CSIRO's imaging system will provide taxonomists with a digital tool that can capture and analyse images of insects and possibly other species into the future.

It has been designed specifically to image insects less than 1cm in size, which are the most difficult and time consuming for taxonomists to observe.

It comprises of a front-end imaging station and a software package that uses clever algorithms to automatically extract taxonomic information about the insect’s body features from the images.

CSIRO's imaging system will provide taxonomists with a digital tool that can capture and analyse images of insects and possibly other species into the future.

'There are certain features that taxonomists look for to distinguish species, for example in insects the pattern of veins on the wings is important,' said CSIRO Image analyst Dr Paul Jackway.

The system uses a high tech scientific camera to take hundreds of two-dimensional (2D) images of an insect specimen.

The insect is rotated 360 degrees on a stage under a series of red, green and blue LED lights.  Infrared and UV lights are also used to pick up tiny details or florescent features that may not be visible to the human eye.

'Currently we are building the software that automatically extracts information about the insect’s body structures from these images and measures them with the computer.' said Dr Jackway.

'The computer might identify a combination of features that have never been seen before, which could be of great interest to taxonomists because it’s potentially a new species.'


The system while still in its prototype phase is being developed to work completely automatically, so that the time taken to analyse each insect is kept to a minimum.

One of the challenges is to design an efficient way to load the insects for imaging, without requiring human handling which could damage delicate body structures.

'We are now working on software that will be able stitch 2D images from the device into a digital three dimensional (3D) model of the insect.' said Dr Jackway.

3D models are a very rich and compact way to store a record of what that species looks like, and could enhance existing efforts to compile large biodiversity databases, such as the Atlas of Living Australia, by providing high resolution images of individual species that can be stored digitally with taxonomic information.

Other applications of this work could include the fast identification of pest insects for quarantine.

Highly detailed 3D images of insects could be made available to quarantine inspection officers to help them accurately match insect pests.

The scientists

This work involves CSIRO imaging specialist Dr Paul Jackway and Dr John La Salle, Head of the Australian National Insect Collection.

The project is part of the Atlas of Living Australia, funded under the Australian Government's National Collaborative Research Infrastructure Strategy (NCRIS).

Find out more about Biotech imaging: imaging for research, diagnostics and drug development.


La Salle J, Wheeler Q, Jackway P, Winterton S, Hobern  D and Lovell D. 2009. Accelerating taxonomic discovery through automated character extraction [145KB PDF, external link]. Zootaxa. 2217: 43-55.