Applied Deep Learning for Diverse Research Communities

Abstract

The Melbourne eResearch Group (www.eresearch.unimelb.edu.au) are involved in a multitude of projects, many of which are focused on big data and data analytics. Many researcher challenges have much to benefit from artificial intelligence and especially from the application of deep learning and convolutional neural networks (CNNs). This talk will provide an overview of a portfolio of projects that have benefited from recent advances in the deep learning domain. These include case studies related to: • pedestrian/crowd counting for the City of Melbourne; • (early) fruit counting on trees (for fruit growers to estimate yield); • tree volume canopy estimation (for fruit growers to estimate the amount of spraying needed); • truck and trailer classification for VicRoads; • feral cat classification for ecology researchers working in rural Victoria; • plant and flower classification for commercial agricultural companies, and • encroachment of vegetation on powerlines for a range of utility companies The talk will cover a brief background to deep learning and CNNs and focus on the results that are now possible, with specific focus on projects requiring image detection and classification. ABOUT THE AUTHOR Professor Richard O. Sinnott is the Director of eResearch at the University of Melbourne and Chair of Applied Computing Systems. In these roles he is responsible for all aspects of eResearch (research-oriented IT development) at the University. He has been lead software engineer/architect on an extensive portfolio of national and international projects, with specific focus on those research domains requiring finer-grained access control (security). He has over 400 peer reviewed publications across a range of applied computing research areas