• An app using image analysis algorithms co-developed by Charles Sturt University researchers and partners allows users to quickly diagnose grape vine nutrient deficiencies
• The app has been commercialised with global agtech start-up Deep Planet to enable winegrape growers to diagnose plant deficiencies and access remedial actions
• Vine nutrition can impact yield and quality and is a significant cost to the management of a vineyard if not handled correctly

Charles Sturt University researchers have contributed to an innovative Australian-developed smartphone app that uses artificial intelligence to help winegrape growers easily diagnose nutrient disorders in grapevines.

A research partnership between Charles Sturt University and NSW Department of Primary Industries (NSW DPI), with funding from Wine Australia has seen a team of viticulturists, plant physiologists and machine learning specialists develop a prototype app which assesses images of vine leaf symptoms captured using a standard smartphone camera.

Dr Suzy Rogiers (NSW DPI) led the project, and Professor in Computer Science Manoranjan Paul and Associate Professor in Computer Science Lihong Zheng in the Charles Sturt School of Computing, Mathematics and Engineering led the image processing / machine learning research.

Professor of Oenology and former Director of the National Wine and Grape Industry Centre Leigh Schmidtke in the Charles Sturt Gulbali Institute of Agriculture, Water and Environment was another contributing investigator.

Use of the app is set to grow following a new commercialisation agreement with global agtech start-up Deep Planet which has obtained exclusive licensing of the technology to add to its capabilities to further benefit grape and wine producers.

Vine nutrition is a significant cost to the management of a vineyard, and if not handled correctly, yield and quality can suffer. The prototype app was developed as a technical solution to help winegrape growers deal with symptom confusion of vine nutritional disorders.

Charles Sturt Pro Vice-Chancellor (Research and Innovation) Professor Michael Friend, said the University’s image analysis experts worked with NSW DPI researchers to develop disease image libraries for artificial intelligence assessment of vines in the field.

“The detection and diagnostic capability of the app works through the image analysis algorithms developed by the team and allows users to quickly identify vine nutrient deficiencies and access remedial actions based upon the diagnosis,” Professor Friend said.

The research was partly funded by Associate Professor Zheng’s Charles Sturt research fellowship fund.

“I invented several ways to reduce the model size and automate the manual annotation procedure resulting in a super-fast diagnosis tool, and I worked closely with the programmer to finalise the VitiDoc model that is the core part of the app,” Professor Zheng said.

“It was a great experience of knowledge discovery and sharing among a multi-disciplinary team.

“I’m currently working with Deep Planet colleagues and facilitating the transfer of the apps, and look forward to continuing to work collaboratively and productively.”

Deep Planet CEO Mr David Carter is excited at the prospect of taking on the technology and adding it to the suite of viticulture remote sensing, monitoring and prediction features offered by their industry-leading VineSignal platform.

“We’ve been working closely with a number of Australian producers to help them manage their vine health, irrigation, yield and maturity using satellite imagery combined with our machine learning and AI capabilities,” Mr Carter said.

“Adding and improving this technology as a practical tool for on the ground nutrition monitoring is an obvious next step to enhance the impact we can offer our clients.”

NSW DPI Deputy Director General Dr Adrian Zammit said the commercialisation of research through government, industry and corporate agreements can help farmers address practical challenges.

“It is encouraging to see these research bodies working together with a global agtech company, and together recognising the value of this key co-operative research project,” Dr Zammit said.

“NSW DPI is committed to understanding and researching the challenges of our agriculture sector and ultimately delivering practical outcomes that add capability and competitiveness to the farm gate.”

Charles Sturt personnel Dr Motiur Rahman (Research Fellow), Dr Tintu Baby (Research Fellow) and Mr Alex Oczkowski (programmer) contributed to the research project.