Welcome to Andrew Mitchell who has recently joined the Centre for Invasive Species Solutions as our Automated Weed Identification Program Manager, to work on the weeds automated identification app project, announced by the Minister for Agriculture back in November 2019, as part of the Smart Farming Partnership stage 2 funding.
Andrew has more than 20 years’ experience in botany, vegetation mapping and surveying and strong skills in GIS. He has spent much of career being an environmental consultant, where has had previous roles developing software for environmental management purposes.
Through his work as an environmental consultant, Andrew has travelled throughout Australia and has documented both native vegetation and non-native plants including weeds. To cover such a geographic range and to remember so many species, Andrew long ago committed to building databases and virtual herbariums to record his observations. These records now account for more than 5000 species including most weed species. His work as a consultant also included consultation with landholders about the economic and environmental impact of weeds. Andrew has worked to raise awareness of many poorly known weedy species with weed control authorities.
Andrew will be based in Canberra at both the Centre for Invasive Species Solutions HQ and CSIRO Black Mountain Laboratories, however he will be spending a large amount of time in the field taking photos of invasive plants as part of the project in its developmental stages.
Citizen science projects, including weed monitoring programs often use photographs as data. Andrew’s experience in managing large collections of photos and his passion for databases will be used in an innovative program to automate the identification of weeds using a mobile phone app. Training the computer vision system contained within the app will require large sets of photos for each type of weed showing all the variations in shape, colour and life cycle that are encountered in the field. Andrew’s role will be photograph weed species in all their variety and to train the computer vision algorithm.
More information about this exciting project will be announced soon.