Microsoft’s green push rewards six local AI-backed projects

By on
Microsoft’s green push rewards six local AI-backed projects

Bringing Azure's grunt to conservation projects.

The latest round of Microsoft’s US$50 million AI for Earth program has resulted in six Australian artificial intelligence-based projects securing grants to further their work.

The recipients this year include Monash University, Griffith University, Queensland University of Technology, the Australian Wildlife Conservancy, Bush Heritage Australia, and startup InFarm.

The dollar value of the individual grants was not disclosed and iTnews is seeking further detail on the actual amounts distributed in Australia.

Microsoft bills AI for Earth as “more than just grants”, offering successful applicants “transformative solutions” and open source APIs to bring their projects up to commercial scale.

Participants also receive a mix of Azure cloud computing resources, including AI tools, data labelling services and access to the training on the company’s data science, machine learning and visualisation tools.

The grants program, which kicked off in 2017, is targetted at organisations looking to tackle climate change and its effects on agriculture, biodiversity and water resources.

Australian projects to receive grants and support this year include:

Mapping species distributions: This Faculty of Information Technology project at Monash University uses social network geotagged photos and harnesses Azure cognitive services to provide insight on key ecological phenomena, including insect pollinator distributions and flowering plant activity, to understand how they are impacted by climate change.

PhD candidate Moataz Medhat ElQadi and associate professor Alan Dorin said, “Access to Microsoft’s Azure Cognitive Services is enabling us to continue to build an understanding of ecosystems key to food security and the sustainability of earth’s natural environments.”

Below ground carbon level prediction: Using Azure machine learning, Griffith University is developing a prediction model providing greater insights regarding carbon sequestration of green stormwater infrastructure.

Integrating different levels of relevant ecological data with state-of-the-art AI methods to facilitate the better prediction of the urban carbon budget. “Traditionally, we monitor a few influencing factors of soil carbon accumulation while we know that there are more factors involved,” PhD student Emad Kavehei said.

Drone-based reef monitoring: A Queensland University of Technology project exploring how data collected by drones fitted with advanced sensors can be interpreted by AI to aid with the classification and restoration of coral.

One of the challenges is the scale of the reef – which is roughly the size of Japan, requiring an innovative approach to data collection. Associate professor Felipe Gonzalez explained that the team has built an AI system that interprets images captured by hyperspectral cameras mounted on drones, to get a better understanding of the health of the reef.

Weed identification and classification: Startup business InFarm is using AI to help identify weed species in fallow fields – specifically those that are chemical resistant – and also to provide application maps for use in autonomous, variable spray tractors.

InFarm managing director Jerome Leray said, “We use drones to survey fields, AI to interpret the images, and the resulting insight to allow targeted weed-spraying that we estimate can reduce farmers’ herbicide costs by 95 per cent and also rein in the amount of chemicals used overall.”

Invasive predator detection in the outback: Using 60,000 camera trap images, conservation organisation Bush Heritage Australia and data scientists Jonathan Bourne and Anindya Basu are using Azure virtual machines to develop a species recognition algorithm to estimate the predator population on their properties, which can help optimise predator detection and protection of threatened species.

Feral and native fauna identification: Using AI and image recognition, the Australian Wildlife Conservancy is exploring how to automate animal identification from 90,000 camera trap images.

Microsoft’s chief environmental officer, Lucas Joppa, said that although advancements in AI and cloud technologies are increasingly being leveraged to tackle the world’s biggest problems, the uptake of these solutions to protect the planet is proceeding slowly.

“And as such, we are essentially flying blind when it comes to understanding how our planet is changing and how to best solve environmental challenges. AI can change that.

“By putting AI in the hands of researchers and organisations we can use important data insights to help solve issues related to water, agriculture, biodiversity and climate change.”

Got a news tip for our journalists? Share it with us anonymously here.
Copyright © . All rights reserved.

Most Read Articles

Log In

  |  Forgot your password?