Species identification tools

Taxonomic research comprises both the naming and grouping of organisms (classification) and the provision of tools that allow non-specialist end-users to find out the correct name for an unknown specimen (identification). Reliable and timely identification of specimens is critical for all biological research, conservation management, and biosecurity.

Traditional identification keys are paper-based, single entry, and require the end-user to understand technical terminology. With the advent of computers, multi-entry, electronic tools such as Lucid keys became available. The newest generation of identification tools uses artificial intelligence, specifically computer vision.

In 2018, we developed a prototype smartphone app for the identification of weed seeds, which was co-funded by CSIRO and Microsoft and coded by Altis Consulting. In 2019/2020 and 2021/22 we received funding from the Biosecurity Innovation Program to expand this approach to stink bugs, a group that poses a major biosecurity threat, in collaboration with the Australian National Insect Collection, and in 2022/23 we expanded the training image library for weed seeds. The Australian Government Department of Agriculture, Fisheries, and Forestry is currently building its own system on the lines of our app prototype to deploy models trained on our images to their staff.

Our team is one of the key partners behind Weedscan, building a national identification and reporting app for a priority list of weeds. The project was funded by the National Landcare Program, coordinated by the Centre for Invasive Species Solutions, and supported by stakeholders from state governments.

Horama ID is a system that allows taxonomists to deploy image classification models for species identification to end users. It has recently been published as an app for Android.